You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@iceberg.apache.org by gi...@apache.org on 2025/11/13 09:38:48 UTC

(iceberg-rust) branch gh-pages updated: deploy: 8d4851f0db824050c0419bd20d939e5dd3a5a80b

This is an automated email from the ASF dual-hosted git repository.

github-bot pushed a commit to branch gh-pages
in repository https://gitbox.apache.org/repos/asf/iceberg-rust.git


The following commit(s) were added to refs/heads/gh-pages by this push:
     new df5aa928 deploy: 8d4851f0db824050c0419bd20d939e5dd3a5a80b
df5aa928 is described below

commit df5aa9283a7181549552c63965e33aa651277778
Author: liurenjie1024 <li...@users.noreply.github.com>
AuthorDate: Thu Nov 13 09:38:43 2025 +0000

    deploy: 8d4851f0db824050c0419bd20d939e5dd3a5a80b
---
 api/iceberg/arrow/struct.ArrowFileReader.html      |   14 +-
 api/iceberg/arrow/struct.ArrowReader.html          |    2 +-
 api/iceberg/scan/index.html                        |    2 +-
 api/iceberg/scan/struct.FileScanTask.html          |   32 +-
 .../scan/struct.FileScanTaskDeleteFile.html        |   10 +-
 api/iceberg/scan/type.FileScanTaskStream.html      |    2 +-
 api/search-index.js                                |    4 +-
 api/search.desc/iceberg/iceberg-desc-0-.js         |    2 +-
 .../arrow/caching_delete_file_loader.rs.html       |   45 +-
 api/src/iceberg/arrow/delete_file_loader.rs.html   |    4 +-
 api/src/iceberg/arrow/delete_filter.rs.html        |   74 +-
 api/src/iceberg/arrow/reader.rs.html               | 7239 ++++++++++----------
 .../iceberg/arrow/record_batch_transformer.rs.html | 2066 ++++--
 api/src/iceberg/scan/context.rs.html               |  275 +-
 api/src/iceberg/scan/mod.rs.html                   |   42 +-
 api/src/iceberg/scan/task.rs.html                  |  244 +-
 16 files changed, 5600 insertions(+), 4457 deletions(-)

diff --git a/api/iceberg/arrow/struct.ArrowFileReader.html b/api/iceberg/arrow/struct.ArrowFileReader.html
index a363ab7f..4f874000 100644
--- a/api/iceberg/arrow/struct.ArrowFileReader.html
+++ b/api/iceberg/arrow/struct.ArrowFileReader.html
@@ -1,12 +1,12 @@
-<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="ArrowFileReader is a wrapper around a FileRead that impls parquets AsyncFileReader."><title>ArrowFileReader in iceberg::arrow - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic- [...]
-</div></details><h2 id="implementations" class="section-header">Implementations<a href="#implementations" class="anchor">§</a></h2><div id="implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-ArrowFileReader%3CR%3E" class="impl"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1527-1566">Source</a><a href="#impl-ArrowFileReader%3CR%3E" class="anchor">§</a><h3 class="code-header">impl&lt;R: <a class="trait" href="../io/tra [...]
-</div></details><details class="toggle method-toggle" open><summary><section id="method.with_preload_column_index" class="method"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1541-1544">Source</a><h4 class="code-header">pub fn <a href="#method.with_preload_column_index" class="fn">with_preload_column_index</a>(self, preload: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.bool.html">bool</a>) -&gt; Self</h4></section></summary><div clas [...]
-</div></details><details class="toggle method-toggle" open><summary><section id="method.with_preload_offset_index" class="method"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1547-1550">Source</a><h4 class="code-header">pub fn <a href="#method.with_preload_offset_index" class="fn">with_preload_offset_index</a>(self, preload: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.bool.html">bool</a>) -&gt; Self</h4></section></summary><div clas [...]
-</div></details><details class="toggle method-toggle" open><summary><section id="method.with_preload_page_index" class="method"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1553-1556">Source</a><h4 class="code-header">pub fn <a href="#method.with_preload_page_index" class="fn">with_preload_page_index</a>(self, preload: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.bool.html">bool</a>) -&gt; Self</h4></section></summary><div class="doc [...]
-</div></details><details class="toggle method-toggle" open><summary><section id="method.with_metadata_size_hint" class="method"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1562-1565">Source</a><h4 class="code-header">pub fn <a href="#method.with_metadata_size_hint" class="fn">with_metadata_size_hint</a>(self, hint: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.usize.html">usize</a>) -&gt; Self</h4></section></summary><div class="docb [...]
+<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="ArrowFileReader is a wrapper around a FileRead that impls parquets AsyncFileReader."><title>ArrowFileReader in iceberg::arrow - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic- [...]
+</div></details><h2 id="implementations" class="section-header">Implementations<a href="#implementations" class="anchor">§</a></h2><div id="implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-ArrowFileReader%3CR%3E" class="impl"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1640-1679">Source</a><a href="#impl-ArrowFileReader%3CR%3E" class="anchor">§</a><h3 class="code-header">impl&lt;R: <a class="trait" href="../io/tra [...]
+</div></details><details class="toggle method-toggle" open><summary><section id="method.with_preload_column_index" class="method"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1654-1657">Source</a><h4 class="code-header">pub fn <a href="#method.with_preload_column_index" class="fn">with_preload_column_index</a>(self, preload: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.bool.html">bool</a>) -&gt; Self</h4></section></summary><div clas [...]
+</div></details><details class="toggle method-toggle" open><summary><section id="method.with_preload_offset_index" class="method"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1660-1663">Source</a><h4 class="code-header">pub fn <a href="#method.with_preload_offset_index" class="fn">with_preload_offset_index</a>(self, preload: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.bool.html">bool</a>) -&gt; Self</h4></section></summary><div clas [...]
+</div></details><details class="toggle method-toggle" open><summary><section id="method.with_preload_page_index" class="method"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1666-1669">Source</a><h4 class="code-header">pub fn <a href="#method.with_preload_page_index" class="fn">with_preload_page_index</a>(self, preload: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.bool.html">bool</a>) -&gt; Self</h4></section></summary><div class="doc [...]
+</div></details><details class="toggle method-toggle" open><summary><section id="method.with_metadata_size_hint" class="method"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1675-1678">Source</a><h4 class="code-header">pub fn <a href="#method.with_metadata_size_hint" class="fn">with_metadata_size_hint</a>(self, hint: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.usize.html">usize</a>) -&gt; Self</h4></section></summary><div class="docb [...]
 <p>This hint can help reduce the number of fetch requests. For more details see the
 <a href="https://docs.rs/parquet/latest/parquet/file/metadata/struct.ParquetMetaDataReader.html#method.with_prefetch_hint">ParquetMetaDataReader documentation</a>.</p>
-</div></details></div></details></div><h2 id="trait-implementations" class="section-header">Trait Implementations<a href="#trait-implementations" class="anchor">§</a></h2><div id="trait-implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-AsyncFileReader-for-ArrowFileReader%3CR%3E" class="impl"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1568-1597">Source</a><a href="#impl-AsyncFileReader-for-ArrowFileReader%3CR%3E" c [...]
+</div></details></div></details></div><h2 id="trait-implementations" class="section-header">Trait Implementations<a href="#trait-implementations" class="anchor">§</a></h2><div id="trait-implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-AsyncFileReader-for-ArrowFileReader%3CR%3E" class="impl"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#1681-1710">Source</a><a href="#impl-AsyncFileReader-for-ArrowFileReader%3CR%3E" c [...]
     &amp;mut self,
     _options: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;&amp;ArrowReaderOptions&gt;,
 ) -&gt; BoxFuture&lt;'_, Result&lt;<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/sync/struct.Arc.html" title="struct alloc::sync::Arc">Arc</a>&lt;ParquetMetaData&gt;&gt;&gt;</h4></section></summary><div class='docblock'>Return a future which results in the [<code>ParquetMetaData</code>] for this Parquet file. <a>Read more</a></div></details><details class="toggle method-toggle" open><summary><section id="method.get_byte_ranges" class="method trait-impl"><a href="#method [...]
diff --git a/api/iceberg/arrow/struct.ArrowReader.html b/api/iceberg/arrow/struct.ArrowReader.html
index 26815b2f..ee79f333 100644
--- a/api/iceberg/arrow/struct.ArrowReader.html
+++ b/api/iceberg/arrow/struct.ArrowReader.html
@@ -1,5 +1,5 @@
 <!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Reads data from Parquet files"><title>ArrowReader in iceberg::arrow - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSans-Me [...]
-</div></details><h2 id="implementations" class="section-header">Implementations<a href="#implementations" class="anchor">§</a></h2><div id="implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-ArrowReader" class="impl"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#140-878">Source</a><a href="#impl-ArrowReader" class="anchor">§</a><h3 class="code-header">impl <a class="struct" href="struct.ArrowReader.html" title="struct [...]
+</div></details><h2 id="implementations" class="section-header">Implementations<a href="#implementations" class="anchor">§</a></h2><div id="implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-ArrowReader" class="impl"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#140-920">Source</a><a href="#impl-ArrowReader" class="anchor">§</a><h3 class="code-header">impl <a class="struct" href="struct.ArrowReader.html" title="struct [...]
 Returns a stream of Arrow RecordBatches containing the data from the files</p>
 </div></details></div></details></div><h2 id="trait-implementations" class="section-header">Trait Implementations<a href="#trait-implementations" class="anchor">§</a></h2><div id="trait-implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-Clone-for-ArrowReader" class="impl"><a class="src rightside" href="../../src/iceberg/arrow/reader.rs.html#127">Source</a><a href="#impl-Clone-for-ArrowReader" class="anchor">§</a><h3 class="code-header">impl  [...]
     T: 'static + ?<a class="trait" href="https://doc.rust-lang.org/nightly/core/marker/trait.Sized.html" title="trait core::marker::Sized">Sized</a>,</div></h3></section></summary><div class="impl-items"><details class="toggle method-toggle" open><summary><section id="method.type_id" class="method trait-impl"><a class="src rightside" href="https://doc.rust-lang.org/nightly/src/core/any.rs.html#139">Source</a><a href="#method.type_id" class="anchor">§</a><h4 class="code-header">fn <a href [...]
diff --git a/api/iceberg/scan/index.html b/api/iceberg/scan/index.html
index 6742119d..4145bd6d 100644
--- a/api/iceberg/scan/index.html
+++ b/api/iceberg/scan/index.html
@@ -1,2 +1,2 @@
-<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Table scan api."><title>iceberg::scan - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSans-MediumItalic-ccf7e434.woff2,Fira [...]
+<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Table scan api."><title>iceberg::scan - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSans-MediumItalic-ccf7e434.woff2,Fira [...]
 </div></details><h2 id="structs" class="section-header">Structs<a href="#structs" class="anchor">§</a></h2><dl class="item-table"><dt><a class="struct" href="struct.FileScanTask.html" title="struct iceberg::scan::FileScanTask">File<wbr>Scan<wbr>Task</a></dt><dd>A task to scan part of file.</dd><dt><a class="struct" href="struct.FileScanTaskDeleteFile.html" title="struct iceberg::scan::FileScanTaskDeleteFile">File<wbr>Scan<wbr>Task<wbr>Delete<wbr>File</a></dt><dd>A task to scan part of fi [...]
\ No newline at end of file
diff --git a/api/iceberg/scan/struct.FileScanTask.html b/api/iceberg/scan/struct.FileScanTask.html
index c5c81740..85133c77 100644
--- a/api/iceberg/scan/struct.FileScanTask.html
+++ b/api/iceberg/scan/struct.FileScanTask.html
@@ -1,4 +1,4 @@
-<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="A task to scan part of file."><title>FileScanTask in iceberg::scan - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSans-Med [...]
+<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="A task to scan part of file."><title>FileScanTask in iceberg::scan - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSans-Med [...]
     pub start: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.u64.html">u64</a>,
     pub length: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.u64.html">u64</a>,
     pub record_count: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.u64.html">u64</a>&gt;,
@@ -8,6 +8,9 @@
     pub project_field_ids: <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/vec/struct.Vec.html" title="struct alloc::vec::Vec">Vec</a>&lt;<a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.i32.html">i32</a>&gt;,
     pub predicate: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="enum" href="../expr/enum.BoundPredicate.html" title="enum iceberg::expr::BoundPredicate">BoundPredicate</a>&gt;,
     pub deletes: <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/vec/struct.Vec.html" title="struct alloc::vec::Vec">Vec</a>&lt;<a class="struct" href="struct.FileScanTaskDeleteFile.html" title="struct iceberg::scan::FileScanTaskDeleteFile">FileScanTaskDeleteFile</a>&gt;,
+    pub partition: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="struct" href="../spec/struct.Struct.html" title="struct iceberg::spec::Struct">Struct</a>&gt;,
+    pub partition_spec: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/sync/struct.Arc.html" title="struct alloc::sync::Arc">Arc</a>&lt;<a class="struct" href="../spec/struct.PartitionSpec.html" title="struct iceberg::spec::PartitionSpec">PartitionSpec</a>&gt;&gt;,
+    pub name_mapping: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/sync/struct.Arc.html" title="struct alloc::sync::Arc">Arc</a>&lt;<a class="struct" href="../spec/struct.NameMapping.html" title="struct iceberg::spec::NameMapping">NameMapping</a>&gt;&gt;,
 }</code></pre><details class="toggle top-doc" open><summary class="hideme"><span>Expand description</span></summary><div class="docblock"><p>A task to scan part of file.</p>
 </div></details><h2 id="fields" class="fields section-header">Fields<a href="#fields" class="anchor">§</a></h2><span id="structfield.start" class="structfield section-header"><a href="#structfield.start" class="anchor field">§</a><code>start: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.u64.html">u64</a></code></span><div class="docblock"><p>The start offset of the file to scan.</p>
 </div><span id="structfield.length" class="structfield section-header"><a href="#structfield.length" class="anchor field">§</a><code>length: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.u64.html">u64</a></code></span><div class="docblock"><p>The length of the file to scan.</p>
@@ -20,15 +23,24 @@ reading the entire data file.</p>
 </div><span id="structfield.project_field_ids" class="structfield section-header"><a href="#structfield.project_field_ids" class="anchor field">§</a><code>project_field_ids: <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/vec/struct.Vec.html" title="struct alloc::vec::Vec">Vec</a>&lt;<a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.i32.html">i32</a>&gt;</code></span><div class="docblock"><p>The field ids to project.</p>
 </div><span id="structfield.predicate" class="structfield section-header"><a href="#structfield.predicate" class="anchor field">§</a><code>predicate: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="enum" href="../expr/enum.BoundPredicate.html" title="enum iceberg::expr::BoundPredicate">BoundPredicate</a>&gt;</code></span><div class="docblock"><p>The predicate to filter.</p>
 </div><span id="structfield.deletes" class="structfield section-header"><a href="#structfield.deletes" class="anchor field">§</a><code>deletes: <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/vec/struct.Vec.html" title="struct alloc::vec::Vec">Vec</a>&lt;<a class="struct" href="struct.FileScanTaskDeleteFile.html" title="struct iceberg::scan::FileScanTaskDeleteFile">FileScanTaskDeleteFile</a>&gt;</code></span><div class="docblock"><p>The list of delete files that may need  [...]
-</div><h2 id="implementations" class="section-header">Implementations<a href="#implementations" class="anchor">§</a></h2><div id="implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-FileScanTask" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#59-84">Source</a><a href="#impl-FileScanTask" class="anchor">§</a><h3 class="code-header">impl <a class="struct" href="struct.FileScanTask.html" title="struct iceberg::sc [...]
-</div></details><details class="toggle method-toggle" open><summary><section id="method.project_field_ids" class="method"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#66-68">Source</a><h4 class="code-header">pub fn <a href="#method.project_field_ids" class="fn">project_field_ids</a>(&amp;self) -&gt; &amp;[<a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.i32.html">i32</a>]</h4></section></summary><div class="docblock"><p>Returns the project  [...]
-</div></details><details class="toggle method-toggle" open><summary><section id="method.predicate" class="method"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#71-73">Source</a><h4 class="code-header">pub fn <a href="#method.predicate" class="fn">predicate</a>(&amp;self) -&gt; <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;&amp;<a class="enum" href="../expr/enum.BoundPredicate.html [...]
-</div></details><details class="toggle method-toggle" open><summary><section id="method.schema" class="method"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#76-78">Source</a><h4 class="code-header">pub fn <a href="#method.schema" class="fn">schema</a>(&amp;self) -&gt; &amp;<a class="struct" href="../spec/struct.Schema.html" title="struct iceberg::spec::Schema">Schema</a></h4></section></summary><div class="docblock"><p>Returns the schema of this file scan task as a r [...]
-</div></details><details class="toggle method-toggle" open><summary><section id="method.schema_ref" class="method"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#81-83">Source</a><h4 class="code-header">pub fn <a href="#method.schema_ref" class="fn">schema_ref</a>(&amp;self) -&gt; <a class="type" href="../spec/type.SchemaRef.html" title="type iceberg::spec::SchemaRef">SchemaRef</a></h4></section></summary><div class="docblock"><p>Returns the schema of this file scan t [...]
-</div></details></div></details></div><h2 id="trait-implementations" class="section-header">Trait Implementations<a href="#trait-implementations" class="anchor">§</a></h2><div id="trait-implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-Clone-for-FileScanTask" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#29">Source</a><a href="#impl-Clone-for-FileScanTask" class="anchor">§</a><h3 class="code-header">impl <a [...]
-    __D: <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/de/trait.Deserializer.html" title="trait serde_core::de::Deserializer">Deserializer</a>&lt;'de&gt;,</div></h4></section></summary><div class='docblock'>Deserialize this value from the given Serde deserializer. <a href="https://docs.rs/serde_core/1.0.228/serde_core/de/trait.Deserialize.html#tymethod.deserialize">Read more</a></div></details></div></details><details class="toggle implementors-toggle" open><summar [...]
-and should not be overridden without very good reason.</div></details></div></details><details class="toggle implementors-toggle" open><summary><section id="impl-Serialize-for-FileScanTask" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#29">Source</a><a href="#impl-Serialize-for-FileScanTask" class="anchor">§</a><h3 class="code-header">impl <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serialize.html" title="trait serde_co [...]
-    __S: <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serializer.html" title="trait serde_core::ser::Serializer">Serializer</a>,</div></h4></section></summary><div class='docblock'>Serialize this value into the given Serde serializer. <a href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serialize.html#tymethod.serialize">Read more</a></div></details></div></details><section id="impl-StructuralPartialEq-for-FileScanTask" class="impl"><a class= [...]
+</div><span id="structfield.partition" class="structfield section-header"><a href="#structfield.partition" class="anchor field">§</a><code>partition: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="struct" href="../spec/struct.Struct.html" title="struct iceberg::spec::Struct">Struct</a>&gt;</code></span><div class="docblock"><p>Partition data from the manifest entry, used to identify which col [...]
+constant values from partition metadata vs. reading from the data file.
+Per the Iceberg spec, only identity-transformed partition fields should use constants.</p>
+</div><span id="structfield.partition_spec" class="structfield section-header"><a href="#structfield.partition_spec" class="anchor field">§</a><code>partition_spec: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/sync/struct.Arc.html" title="struct alloc::sync::Arc">Arc</a>&lt;<a class="struct" href="../spec/struct.PartitionSpec.html" title [...]
+(which use partition metadata constants) from non-identity transforms like
+bucket/truncate (which must read source columns from the data file).</p>
+</div><span id="structfield.name_mapping" class="structfield section-header"><a href="#structfield.name_mapping" class="anchor field">§</a><code>name_mapping: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/sync/struct.Arc.html" title="struct alloc::sync::Arc">Arc</a>&lt;<a class="struct" href="../spec/struct.NameMapping.html" title="struct [...]
+used to resolve field IDs from column names when Parquet files lack field IDs
+or have field ID conflicts.</p>
+</div><h2 id="implementations" class="section-header">Implementations<a href="#implementations" class="anchor">§</a></h2><div id="implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-FileScanTask" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#109-134">Source</a><a href="#impl-FileScanTask" class="anchor">§</a><h3 class="code-header">impl <a class="struct" href="struct.FileScanTask.html" title="struct iceberg:: [...]
+</div></details><details class="toggle method-toggle" open><summary><section id="method.project_field_ids" class="method"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#116-118">Source</a><h4 class="code-header">pub fn <a href="#method.project_field_ids" class="fn">project_field_ids</a>(&amp;self) -&gt; &amp;[<a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.i32.html">i32</a>]</h4></section></summary><div class="docblock"><p>Returns the projec [...]
+</div></details><details class="toggle method-toggle" open><summary><section id="method.predicate" class="method"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#121-123">Source</a><h4 class="code-header">pub fn <a href="#method.predicate" class="fn">predicate</a>(&amp;self) -&gt; <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;&amp;<a class="enum" href="../expr/enum.BoundPredicate.ht [...]
+</div></details><details class="toggle method-toggle" open><summary><section id="method.schema" class="method"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#126-128">Source</a><h4 class="code-header">pub fn <a href="#method.schema" class="fn">schema</a>(&amp;self) -&gt; &amp;<a class="struct" href="../spec/struct.Schema.html" title="struct iceberg::spec::Schema">Schema</a></h4></section></summary><div class="docblock"><p>Returns the schema of this file scan task as a [...]
+</div></details><details class="toggle method-toggle" open><summary><section id="method.schema_ref" class="method"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#131-133">Source</a><h4 class="code-header">pub fn <a href="#method.schema_ref" class="fn">schema_ref</a>(&amp;self) -&gt; <a class="type" href="../spec/type.SchemaRef.html" title="type iceberg::spec::SchemaRef">SchemaRef</a></h4></section></summary><div class="docblock"><p>Returns the schema of this file scan [...]
+</div></details></div></details></div><h2 id="trait-implementations" class="section-header">Trait Implementations<a href="#trait-implementations" class="anchor">§</a></h2><div id="trait-implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-Clone-for-FileScanTask" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#52">Source</a><a href="#impl-Clone-for-FileScanTask" class="anchor">§</a><h3 class="code-header">impl <a [...]
+    __D: <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/de/trait.Deserializer.html" title="trait serde_core::de::Deserializer">Deserializer</a>&lt;'de&gt;,</div></h4></section></summary><div class='docblock'>Deserialize this value from the given Serde deserializer. <a href="https://docs.rs/serde_core/1.0.228/serde_core/de/trait.Deserialize.html#tymethod.deserialize">Read more</a></div></details></div></details><details class="toggle implementors-toggle" open><summar [...]
+and should not be overridden without very good reason.</div></details></div></details><details class="toggle implementors-toggle" open><summary><section id="impl-Serialize-for-FileScanTask" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#52">Source</a><a href="#impl-Serialize-for-FileScanTask" class="anchor">§</a><h3 class="code-header">impl <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serialize.html" title="trait serde_co [...]
+    __S: <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serializer.html" title="trait serde_core::ser::Serializer">Serializer</a>,</div></h4></section></summary><div class='docblock'>Serialize this value into the given Serde serializer. <a href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serialize.html#tymethod.serialize">Read more</a></div></details></div></details><section id="impl-StructuralPartialEq-for-FileScanTask" class="impl"><a class= [...]
     T: 'static + ?<a class="trait" href="https://doc.rust-lang.org/nightly/core/marker/trait.Sized.html" title="trait core::marker::Sized">Sized</a>,</div></h3></section></summary><div class="impl-items"><details class="toggle method-toggle" open><summary><section id="method.type_id" class="method trait-impl"><a class="src rightside" href="https://doc.rust-lang.org/nightly/src/core/any.rs.html#139">Source</a><a href="#method.type_id" class="anchor">§</a><h4 class="code-header">fn <a href [...]
     T: <a class="trait" href="https://doc.rust-lang.org/nightly/core/any/trait.Any.html" title="trait core::any::Any">Any</a>,</div></h3></section></summary><div class="impl-items"><section id="method.as_any" class="method trait-impl"><a href="#method.as_any" class="anchor">§</a><h4 class="code-header">fn <a class="fn">as_any</a>(&amp;self) -&gt; &amp;(dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/any/trait.Any.html" title="trait core::any::Any">Any</a> + 'static)</h4 [...]
     T: ?<a class="trait" href="https://doc.rust-lang.org/nightly/core/marker/trait.Sized.html" title="trait core::marker::Sized">Sized</a>,</div></h3></section></summary><div class="impl-items"><details class="toggle method-toggle" open><summary><section id="method.borrow" class="method trait-impl"><a class="src rightside" href="https://doc.rust-lang.org/nightly/src/core/borrow.rs.html#211">Source</a><a href="#method.borrow" class="anchor">§</a><h4 class="code-header">fn <a href="https:/ [...]
diff --git a/api/iceberg/scan/struct.FileScanTaskDeleteFile.html b/api/iceberg/scan/struct.FileScanTaskDeleteFile.html
index 746571bb..2a7dec2f 100644
--- a/api/iceberg/scan/struct.FileScanTaskDeleteFile.html
+++ b/api/iceberg/scan/struct.FileScanTaskDeleteFile.html
@@ -1,4 +1,4 @@
-<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="A task to scan part of file."><title>FileScanTaskDeleteFile in iceberg::scan - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,Fi [...]
+<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="A task to scan part of file."><title>FileScanTaskDeleteFile in iceberg::scan - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,Fi [...]
     pub file_path: <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>,
     pub file_type: <a class="enum" href="../spec/enum.DataContentType.html" title="enum iceberg::spec::DataContentType">DataContentType</a>,
     pub partition_spec_id: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.i32.html">i32</a>,
@@ -8,10 +8,10 @@
 </div><span id="structfield.file_type" class="structfield section-header"><a href="#structfield.file_type" class="anchor field">§</a><code>file_type: <a class="enum" href="../spec/enum.DataContentType.html" title="enum iceberg::spec::DataContentType">DataContentType</a></code></span><div class="docblock"><p>delete file type</p>
 </div><span id="structfield.partition_spec_id" class="structfield section-header"><a href="#structfield.partition_spec_id" class="anchor field">§</a><code>partition_spec_id: <a class="primitive" href="https://doc.rust-lang.org/nightly/std/primitive.i32.html">i32</a></code></span><div class="docblock"><p>partition id</p>
 </div><span id="structfield.equality_ids" class="structfield section-header"><a href="#structfield.equality_ids" class="anchor field">§</a><code>equality_ids: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a>&lt;<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/vec/struct.Vec.html" title="struct alloc::vec::Vec">Vec</a>&lt;<a class="primitive" href="https://doc.rust-lang.org/nightly/std/primiti [...]
-</div><h2 id="trait-implementations" class="section-header">Trait Implementations<a href="#trait-implementations" class="anchor">§</a></h2><div id="trait-implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-Clone-for-FileScanTaskDeleteFile" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#104">Source</a><a href="#impl-Clone-for-FileScanTaskDeleteFile" class="anchor">§</a><h3 class="code-header">impl <a class="tra [...]
-    __D: <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/de/trait.Deserializer.html" title="trait serde_core::de::Deserializer">Deserializer</a>&lt;'de&gt;,</div></h4></section></summary><div class='docblock'>Deserialize this value from the given Serde deserializer. <a href="https://docs.rs/serde_core/1.0.228/serde_core/de/trait.Deserialize.html#tymethod.deserialize">Read more</a></div></details></div></details><details class="toggle implementors-toggle" open><summar [...]
-and should not be overridden without very good reason.</div></details></div></details><details class="toggle implementors-toggle" open><summary><section id="impl-Serialize-for-FileScanTaskDeleteFile" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#104">Source</a><a href="#impl-Serialize-for-FileScanTaskDeleteFile" class="anchor">§</a><h3 class="code-header">impl <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serialize.html"  [...]
-    __S: <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serializer.html" title="trait serde_core::ser::Serializer">Serializer</a>,</div></h4></section></summary><div class='docblock'>Serialize this value into the given Serde serializer. <a href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serialize.html#tymethod.serialize">Read more</a></div></details></div></details><section id="impl-StructuralPartialEq-for-FileScanTaskDeleteFile" class="impl" [...]
+</div><h2 id="trait-implementations" class="section-header">Trait Implementations<a href="#trait-implementations" class="anchor">§</a></h2><div id="trait-implementations-list"><details class="toggle implementors-toggle" open><summary><section id="impl-Clone-for-FileScanTaskDeleteFile" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#154">Source</a><a href="#impl-Clone-for-FileScanTaskDeleteFile" class="anchor">§</a><h3 class="code-header">impl <a class="tra [...]
+    __D: <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/de/trait.Deserializer.html" title="trait serde_core::de::Deserializer">Deserializer</a>&lt;'de&gt;,</div></h4></section></summary><div class='docblock'>Deserialize this value from the given Serde deserializer. <a href="https://docs.rs/serde_core/1.0.228/serde_core/de/trait.Deserialize.html#tymethod.deserialize">Read more</a></div></details></div></details><details class="toggle implementors-toggle" open><summar [...]
+and should not be overridden without very good reason.</div></details></div></details><details class="toggle implementors-toggle" open><summary><section id="impl-Serialize-for-FileScanTaskDeleteFile" class="impl"><a class="src rightside" href="../../src/iceberg/scan/task.rs.html#154">Source</a><a href="#impl-Serialize-for-FileScanTaskDeleteFile" class="anchor">§</a><h3 class="code-header">impl <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serialize.html"  [...]
+    __S: <a class="trait" href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serializer.html" title="trait serde_core::ser::Serializer">Serializer</a>,</div></h4></section></summary><div class='docblock'>Serialize this value into the given Serde serializer. <a href="https://docs.rs/serde_core/1.0.228/serde_core/ser/trait.Serialize.html#tymethod.serialize">Read more</a></div></details></div></details><section id="impl-StructuralPartialEq-for-FileScanTaskDeleteFile" class="impl" [...]
     T: 'static + ?<a class="trait" href="https://doc.rust-lang.org/nightly/core/marker/trait.Sized.html" title="trait core::marker::Sized">Sized</a>,</div></h3></section></summary><div class="impl-items"><details class="toggle method-toggle" open><summary><section id="method.type_id" class="method trait-impl"><a class="src rightside" href="https://doc.rust-lang.org/nightly/src/core/any.rs.html#139">Source</a><a href="#method.type_id" class="anchor">§</a><h4 class="code-header">fn <a href [...]
     T: <a class="trait" href="https://doc.rust-lang.org/nightly/core/any/trait.Any.html" title="trait core::any::Any">Any</a>,</div></h3></section></summary><div class="impl-items"><section id="method.as_any" class="method trait-impl"><a href="#method.as_any" class="anchor">§</a><h4 class="code-header">fn <a class="fn">as_any</a>(&amp;self) -&gt; &amp;(dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/any/trait.Any.html" title="trait core::any::Any">Any</a> + 'static)</h4 [...]
     T: ?<a class="trait" href="https://doc.rust-lang.org/nightly/core/marker/trait.Sized.html" title="trait core::marker::Sized">Sized</a>,</div></h3></section></summary><div class="impl-items"><details class="toggle method-toggle" open><summary><section id="method.borrow" class="method trait-impl"><a class="src rightside" href="https://doc.rust-lang.org/nightly/src/core/borrow.rs.html#211">Source</a><a href="#method.borrow" class="anchor">§</a><h4 class="code-header">fn <a href="https:/ [...]
diff --git a/api/iceberg/scan/type.FileScanTaskStream.html b/api/iceberg/scan/type.FileScanTaskStream.html
index 870e755b..7c218900 100644
--- a/api/iceberg/scan/type.FileScanTaskStream.html
+++ b/api/iceberg/scan/type.FileScanTaskStream.html
@@ -1,2 +1,2 @@
-<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="A stream of `FileScanTask`."><title>FileScanTaskStream in iceberg::scan - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSan [...]
+<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="A stream of `FileScanTask`."><title>FileScanTaskStream in iceberg::scan - Rust</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de.woff2,FiraSans-Regular-0fe48ade.woff2,FiraSan [...]
 </div></details><h2 id="aliased-type" class="section-header">Aliased Type<a href="#aliased-type" class="anchor">§</a></h2><pre class="rust item-decl"><code>struct FileScanTaskStream {}</code></pre></section></div></main></body></html>
\ No newline at end of file
diff --git a/api/search-index.js b/api/search-index.js
index f1a29718..0be7613b 100644
--- a/api/search-index.js
+++ b/api/search-index.js
@@ -1,4 +1,4 @@
-var searchIndex = new Map(JSON.parse('[["iceberg",{"t":"PPPPPPPPPRKKPPPPPPFGPPPFFFPFPPPPPPPPPPPPPPIPPPPPPPPPPPPFFFPGGPPPPFGNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNNNNNNNNNNNNNNNNMNMNOONNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNMNMNQNNNNNNNNNNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNNNNNONNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNMNNNNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNCNMNMNMMNOOCNNNOOONOMNNNNNNNNNNNNNNNNNNNNNNONOOCMNMNONCOONNNNNONCOCMNNNCNNNNNNNNNNNNNNCCNNNNNNNNNNN [...]
+var searchIndex = new Map(JSON.parse('[["iceberg",{"t":"PPPPPPPPPRKKPPPPPPFGPPPFFFPFPPPPPPPPPPPPPPIPPPPPPPPPPPPFFFPGGPPPPFGNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNNNNNNNNNNNNNNNNMNMNOONNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNMNMNQNNNNNNNNNNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNNNNNONNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNMNNNNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNCNMNMNMMNOOCNNNOOONOMNNNNNNNNNNNNNNNNNNNNNNONOOCMNMNONCOONNNNNONCOCMNNNCNNNNNNNNNNNNNNCCNNNNNNNNNNN [...]
 if (typeof exports !== 'undefined') exports.searchIndex = searchIndex;
 else if (window.initSearch) window.initSearch(searchIndex);
-//{"start":39,"fragment_lengths":[169191,1613,3562,3755,1767,3412,3464,4249,8092,1325,2013,3213,1116]}
\ No newline at end of file
+//{"start":39,"fragment_lengths":[169303,1613,3562,3755,1767,3412,3464,4249,8092,1325,2013,3213,1116]}
\ No newline at end of file
diff --git a/api/search.desc/iceberg/iceberg-desc-0-.js b/api/search.desc/iceberg/iceberg-desc-0-.js
index f59323eb..3e6726af 100644
--- a/api/search.desc/iceberg/iceberg-desc-0-.js
+++ b/api/search.desc/iceberg/iceberg-desc-0-.js
@@ -1 +1 @@
-searchState.loadedDescShard("iceberg", 0, "Apache Iceberg Official Native Rust Implementation\nAdd an encryption key\nAdd a new schema to the table\nAdd a new schema to the view\nAdd snapshot to table.\nAdd sort order to table.\nAdd a new partition spec to the table\nAdd a new version to the view\nAssign a new UUID to the table\nAssign a new UUID to the view\nThe catalog type that this builder creates.\nThe catalog API for Iceberg Rust.\nCommon interface for all catalog builders.\nCatalo [...]
\ No newline at end of file
+searchState.loadedDescShard("iceberg", 0, "Apache Iceberg Official Native Rust Implementation\nAdd an encryption key\nAdd a new schema to the table\nAdd a new schema to the view\nAdd snapshot to table.\nAdd sort order to table.\nAdd a new partition spec to the table\nAdd a new version to the view\nAssign a new UUID to the table\nAssign a new UUID to the view\nThe catalog type that this builder creates.\nThe catalog API for Iceberg Rust.\nCommon interface for all catalog builders.\nCatalo [...]
\ No newline at end of file
diff --git a/api/src/iceberg/arrow/caching_delete_file_loader.rs.html b/api/src/iceberg/arrow/caching_delete_file_loader.rs.html
index 17b82561..b08fadea 100644
--- a/api/src/iceberg/arrow/caching_delete_file_loader.rs.html
+++ b/api/src/iceberg/arrow/caching_delete_file_loader.rs.html
@@ -881,24 +881,27 @@
 <a href=#881 id=881 data-nosnippet>881</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">3</span>],
 <a href=#882 id=882 data-nosnippet>882</a>            predicate: <span class="prelude-val">None</span>,
 <a href=#883 id=883 data-nosnippet>883</a>            deletes: <span class="macro">vec!</span>[pos_del, eq_del],
-<a href=#884 id=884 data-nosnippet>884</a>        };
-<a href=#885 id=885 data-nosnippet>885</a>
-<a href=#886 id=886 data-nosnippet>886</a>        <span class="comment">// Load the deletes - should handle both types without error
-<a href=#887 id=887 data-nosnippet>887</a>        </span><span class="kw">let </span>delete_file_loader = CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>);
-<a href=#888 id=888 data-nosnippet>888</a>        <span class="kw">let </span>delete_filter = delete_file_loader
-<a href=#889 id=889 data-nosnippet>889</a>            .load_deletes(<span class="kw-2">&amp;</span>file_scan_task.deletes, file_scan_task.schema_ref())
-<a href=#890 id=890 data-nosnippet>890</a>            .<span class="kw">await
-<a href=#891 id=891 data-nosnippet>891</a>            </span>.unwrap()
-<a href=#892 id=892 data-nosnippet>892</a>            .unwrap();
-<a href=#893 id=893 data-nosnippet>893</a>
-<a href=#894 id=894 data-nosnippet>894</a>        <span class="comment">// Verify both delete types can be processed together
-<a href=#895 id=895 data-nosnippet>895</a>        </span><span class="kw">let </span>result = delete_filter
-<a href=#896 id=896 data-nosnippet>896</a>            .build_equality_delete_predicate(<span class="kw-2">&amp;</span>file_scan_task)
-<a href=#897 id=897 data-nosnippet>897</a>            .<span class="kw">await</span>;
-<a href=#898 id=898 data-nosnippet>898</a>        <span class="macro">assert!</span>(
-<a href=#899 id=899 data-nosnippet>899</a>            result.is_ok(),
-<a href=#900 id=900 data-nosnippet>900</a>            <span class="string">"Failed to build equality delete predicate: {:?}"</span>,
-<a href=#901 id=901 data-nosnippet>901</a>            result.err()
-<a href=#902 id=902 data-nosnippet>902</a>        );
-<a href=#903 id=903 data-nosnippet>903</a>    }
-<a href=#904 id=904 data-nosnippet>904</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file
+<a href=#884 id=884 data-nosnippet>884</a>            partition: <span class="prelude-val">None</span>,
+<a href=#885 id=885 data-nosnippet>885</a>            partition_spec: <span class="prelude-val">None</span>,
+<a href=#886 id=886 data-nosnippet>886</a>            name_mapping: <span class="prelude-val">None</span>,
+<a href=#887 id=887 data-nosnippet>887</a>        };
+<a href=#888 id=888 data-nosnippet>888</a>
+<a href=#889 id=889 data-nosnippet>889</a>        <span class="comment">// Load the deletes - should handle both types without error
+<a href=#890 id=890 data-nosnippet>890</a>        </span><span class="kw">let </span>delete_file_loader = CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>);
+<a href=#891 id=891 data-nosnippet>891</a>        <span class="kw">let </span>delete_filter = delete_file_loader
+<a href=#892 id=892 data-nosnippet>892</a>            .load_deletes(<span class="kw-2">&amp;</span>file_scan_task.deletes, file_scan_task.schema_ref())
+<a href=#893 id=893 data-nosnippet>893</a>            .<span class="kw">await
+<a href=#894 id=894 data-nosnippet>894</a>            </span>.unwrap()
+<a href=#895 id=895 data-nosnippet>895</a>            .unwrap();
+<a href=#896 id=896 data-nosnippet>896</a>
+<a href=#897 id=897 data-nosnippet>897</a>        <span class="comment">// Verify both delete types can be processed together
+<a href=#898 id=898 data-nosnippet>898</a>        </span><span class="kw">let </span>result = delete_filter
+<a href=#899 id=899 data-nosnippet>899</a>            .build_equality_delete_predicate(<span class="kw-2">&amp;</span>file_scan_task)
+<a href=#900 id=900 data-nosnippet>900</a>            .<span class="kw">await</span>;
+<a href=#901 id=901 data-nosnippet>901</a>        <span class="macro">assert!</span>(
+<a href=#902 id=902 data-nosnippet>902</a>            result.is_ok(),
+<a href=#903 id=903 data-nosnippet>903</a>            <span class="string">"Failed to build equality delete predicate: {:?}"</span>,
+<a href=#904 id=904 data-nosnippet>904</a>            result.err()
+<a href=#905 id=905 data-nosnippet>905</a>        );
+<a href=#906 id=906 data-nosnippet>906</a>    }
+<a href=#907 id=907 data-nosnippet>907</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file
diff --git a/api/src/iceberg/arrow/delete_file_loader.rs.html b/api/src/iceberg/arrow/delete_file_loader.rs.html
index 074bc2d1..b21fb6b4 100644
--- a/api/src/iceberg/arrow/delete_file_loader.rs.html
+++ b/api/src/iceberg/arrow/delete_file_loader.rs.html
@@ -20,7 +20,7 @@
 <a href=#20 id=20 data-nosnippet>20</a><span class="kw">use </span>futures::{StreamExt, TryStreamExt};
 <a href=#21 id=21 data-nosnippet>21</a>
 <a href=#22 id=22 data-nosnippet>22</a><span class="kw">use </span><span class="kw">crate</span>::arrow::ArrowReader;
-<a href=#23 id=23 data-nosnippet>23</a><span class="kw">use </span><span class="kw">crate</span>::arrow::record_batch_transformer::RecordBatchTransformer;
+<a href=#23 id=23 data-nosnippet>23</a><span class="kw">use </span><span class="kw">crate</span>::arrow::record_batch_transformer::RecordBatchTransformerBuilder;
 <a href=#24 id=24 data-nosnippet>24</a><span class="kw">use </span><span class="kw">crate</span>::io::FileIO;
 <a href=#25 id=25 data-nosnippet>25</a><span class="kw">use </span><span class="kw">crate</span>::scan::{ArrowRecordBatchStream, FileScanTaskDeleteFile};
 <a href=#26 id=26 data-nosnippet>26</a><span class="kw">use </span><span class="kw">crate</span>::spec::{Schema, SchemaRef};
@@ -82,7 +82,7 @@
 <a href=#82 id=82 data-nosnippet>82</a>        equality_ids: <span class="kw-2">&amp;</span>[i32],
 <a href=#83 id=83 data-nosnippet>83</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ArrowRecordBatchStream&gt; {
 <a href=#84 id=84 data-nosnippet>84</a>        <span class="kw">let </span><span class="kw-2">mut </span>record_batch_transformer =
-<a href=#85 id=85 data-nosnippet>85</a>            RecordBatchTransformer::build(target_schema.clone(), equality_ids);
+<a href=#85 id=85 data-nosnippet>85</a>            RecordBatchTransformerBuilder::new(target_schema.clone(), equality_ids).build();
 <a href=#86 id=86 data-nosnippet>86</a>
 <a href=#87 id=87 data-nosnippet>87</a>        <span class="kw">let </span>record_batch_stream = record_batch_stream.map(<span class="kw">move </span>|record_batch| {
 <a href=#88 id=88 data-nosnippet>88</a>            record_batch.and_then(|record_batch| {
diff --git a/api/src/iceberg/arrow/delete_filter.rs.html b/api/src/iceberg/arrow/delete_filter.rs.html
index 52414c23..4ca86be1 100644
--- a/api/src/iceberg/arrow/delete_filter.rs.html
+++ b/api/src/iceberg/arrow/delete_filter.rs.html
@@ -341,37 +341,43 @@
 <a href=#341 id=341 data-nosnippet>341</a>                project_field_ids: <span class="macro">vec!</span>[],
 <a href=#342 id=342 data-nosnippet>342</a>                predicate: <span class="prelude-val">None</span>,
 <a href=#343 id=343 data-nosnippet>343</a>                deletes: <span class="macro">vec!</span>[pos_del_1, pos_del_2.clone()],
-<a href=#344 id=344 data-nosnippet>344</a>            },
-<a href=#345 id=345 data-nosnippet>345</a>            FileScanTask {
-<a href=#346 id=346 data-nosnippet>346</a>                start: <span class="number">0</span>,
-<a href=#347 id=347 data-nosnippet>347</a>                length: <span class="number">0</span>,
-<a href=#348 id=348 data-nosnippet>348</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#349 id=349 data-nosnippet>349</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/2.parquet"</span>, table_location.to_str().unwrap()),
-<a href=#350 id=350 data-nosnippet>350</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#351 id=351 data-nosnippet>351</a>                schema: data_file_schema.clone(),
-<a href=#352 id=352 data-nosnippet>352</a>                project_field_ids: <span class="macro">vec!</span>[],
-<a href=#353 id=353 data-nosnippet>353</a>                predicate: <span class="prelude-val">None</span>,
-<a href=#354 id=354 data-nosnippet>354</a>                deletes: <span class="macro">vec!</span>[pos_del_3],
-<a href=#355 id=355 data-nosnippet>355</a>            },
-<a href=#356 id=356 data-nosnippet>356</a>        ];
-<a href=#357 id=357 data-nosnippet>357</a>
-<a href=#358 id=358 data-nosnippet>358</a>        file_scan_tasks
-<a href=#359 id=359 data-nosnippet>359</a>    }
-<a href=#360 id=360 data-nosnippet>360</a>
-<a href=#361 id=361 data-nosnippet>361</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>create_pos_del_schema() -&gt; ArrowSchemaRef {
-<a href=#362 id=362 data-nosnippet>362</a>        <span class="kw">let </span>fields = <span class="macro">vec!</span>[
-<a href=#363 id=363 data-nosnippet>363</a>            arrow_schema::Field::new(<span class="string">"file_path"</span>, arrow_schema::DataType::Utf8, <span class="bool-val">false</span>)
-<a href=#364 id=364 data-nosnippet>364</a>                .with_metadata(HashMap::from([(
-<a href=#365 id=365 data-nosnippet>365</a>                    PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#366 id=366 data-nosnippet>366</a>                    FIELD_ID_POSITIONAL_DELETE_FILE_PATH.to_string(),
-<a href=#367 id=367 data-nosnippet>367</a>                )])),
-<a href=#368 id=368 data-nosnippet>368</a>            arrow_schema::Field::new(<span class="string">"pos"</span>, arrow_schema::DataType::Int64, <span class="bool-val">false</span>).with_metadata(
-<a href=#369 id=369 data-nosnippet>369</a>                HashMap::from([(
-<a href=#370 id=370 data-nosnippet>370</a>                    PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#371 id=371 data-nosnippet>371</a>                    FIELD_ID_POSITIONAL_DELETE_POS.to_string(),
-<a href=#372 id=372 data-nosnippet>372</a>                )]),
-<a href=#373 id=373 data-nosnippet>373</a>            ),
-<a href=#374 id=374 data-nosnippet>374</a>        ];
-<a href=#375 id=375 data-nosnippet>375</a>        Arc::new(arrow_schema::Schema::new(fields))
-<a href=#376 id=376 data-nosnippet>376</a>    }
-<a href=#377 id=377 data-nosnippet>377</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file
+<a href=#344 id=344 data-nosnippet>344</a>                partition: <span class="prelude-val">None</span>,
+<a href=#345 id=345 data-nosnippet>345</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#346 id=346 data-nosnippet>346</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#347 id=347 data-nosnippet>347</a>            },
+<a href=#348 id=348 data-nosnippet>348</a>            FileScanTask {
+<a href=#349 id=349 data-nosnippet>349</a>                start: <span class="number">0</span>,
+<a href=#350 id=350 data-nosnippet>350</a>                length: <span class="number">0</span>,
+<a href=#351 id=351 data-nosnippet>351</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#352 id=352 data-nosnippet>352</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/2.parquet"</span>, table_location.to_str().unwrap()),
+<a href=#353 id=353 data-nosnippet>353</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#354 id=354 data-nosnippet>354</a>                schema: data_file_schema.clone(),
+<a href=#355 id=355 data-nosnippet>355</a>                project_field_ids: <span class="macro">vec!</span>[],
+<a href=#356 id=356 data-nosnippet>356</a>                predicate: <span class="prelude-val">None</span>,
+<a href=#357 id=357 data-nosnippet>357</a>                deletes: <span class="macro">vec!</span>[pos_del_3],
+<a href=#358 id=358 data-nosnippet>358</a>                partition: <span class="prelude-val">None</span>,
+<a href=#359 id=359 data-nosnippet>359</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#360 id=360 data-nosnippet>360</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#361 id=361 data-nosnippet>361</a>            },
+<a href=#362 id=362 data-nosnippet>362</a>        ];
+<a href=#363 id=363 data-nosnippet>363</a>
+<a href=#364 id=364 data-nosnippet>364</a>        file_scan_tasks
+<a href=#365 id=365 data-nosnippet>365</a>    }
+<a href=#366 id=366 data-nosnippet>366</a>
+<a href=#367 id=367 data-nosnippet>367</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>create_pos_del_schema() -&gt; ArrowSchemaRef {
+<a href=#368 id=368 data-nosnippet>368</a>        <span class="kw">let </span>fields = <span class="macro">vec!</span>[
+<a href=#369 id=369 data-nosnippet>369</a>            arrow_schema::Field::new(<span class="string">"file_path"</span>, arrow_schema::DataType::Utf8, <span class="bool-val">false</span>)
+<a href=#370 id=370 data-nosnippet>370</a>                .with_metadata(HashMap::from([(
+<a href=#371 id=371 data-nosnippet>371</a>                    PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#372 id=372 data-nosnippet>372</a>                    FIELD_ID_POSITIONAL_DELETE_FILE_PATH.to_string(),
+<a href=#373 id=373 data-nosnippet>373</a>                )])),
+<a href=#374 id=374 data-nosnippet>374</a>            arrow_schema::Field::new(<span class="string">"pos"</span>, arrow_schema::DataType::Int64, <span class="bool-val">false</span>).with_metadata(
+<a href=#375 id=375 data-nosnippet>375</a>                HashMap::from([(
+<a href=#376 id=376 data-nosnippet>376</a>                    PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#377 id=377 data-nosnippet>377</a>                    FIELD_ID_POSITIONAL_DELETE_POS.to_string(),
+<a href=#378 id=378 data-nosnippet>378</a>                )]),
+<a href=#379 id=379 data-nosnippet>379</a>            ),
+<a href=#380 id=380 data-nosnippet>380</a>        ];
+<a href=#381 id=381 data-nosnippet>381</a>        Arc::new(arrow_schema::Schema::new(fields))
+<a href=#382 id=382 data-nosnippet>382</a>    }
+<a href=#383 id=383 data-nosnippet>383</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file
diff --git a/api/src/iceberg/arrow/reader.rs.html b/api/src/iceberg/arrow/reader.rs.html
index ac1178bd..440f5a61 100644
--- a/api/src/iceberg/arrow/reader.rs.html
+++ b/api/src/iceberg/arrow/reader.rs.html
@@ -45,7 +45,7 @@
 <a href=#45 id=45 data-nosnippet>45</a><span class="kw">use </span>parquet::schema::types::{SchemaDescriptor, Type <span class="kw">as </span>ParquetType};
 <a href=#46 id=46 data-nosnippet>46</a>
 <a href=#47 id=47 data-nosnippet>47</a><span class="kw">use </span><span class="kw">crate</span>::arrow::caching_delete_file_loader::CachingDeleteFileLoader;
-<a href=#48 id=48 data-nosnippet>48</a><span class="kw">use </span><span class="kw">crate</span>::arrow::record_batch_transformer::RecordBatchTransformer;
+<a href=#48 id=48 data-nosnippet>48</a><span class="kw">use </span><span class="kw">crate</span>::arrow::record_batch_transformer::RecordBatchTransformerBuilder;
 <a href=#49 id=49 data-nosnippet>49</a><span class="kw">use </span><span class="kw">crate</span>::arrow::{arrow_schema_to_schema, get_arrow_datum};
 <a href=#50 id=50 data-nosnippet>50</a><span class="kw">use </span><span class="kw">crate</span>::delete_vector::DeleteVector;
 <a href=#51 id=51 data-nosnippet>51</a><span class="kw">use </span><span class="kw">crate</span>::error::Result;
@@ -55,7 +55,7 @@
 <a href=#55 id=55 data-nosnippet>55</a><span class="kw">use </span><span class="kw">crate</span>::expr::{BoundPredicate, BoundReference};
 <a href=#56 id=56 data-nosnippet>56</a><span class="kw">use </span><span class="kw">crate</span>::io::{FileIO, FileMetadata, FileRead};
 <a href=#57 id=57 data-nosnippet>57</a><span class="kw">use </span><span class="kw">crate</span>::scan::{ArrowRecordBatchStream, FileScanTask, FileScanTaskStream};
-<a href=#58 id=58 data-nosnippet>58</a><span class="kw">use </span><span class="kw">crate</span>::spec::{Datum, NestedField, PrimitiveType, Schema, Type};
+<a href=#58 id=58 data-nosnippet>58</a><span class="kw">use </span><span class="kw">crate</span>::spec::{Datum, NameMapping, NestedField, PrimitiveType, Schema, Type};
 <a href=#59 id=59 data-nosnippet>59</a><span class="kw">use </span><span class="kw">crate</span>::utils::available_parallelism;
 <a href=#60 id=60 data-nosnippet>60</a><span class="kw">use crate</span>::{Error, ErrorKind};
 <a href=#61 id=61 data-nosnippet>61</a>
@@ -181,3531 +181,3844 @@
 <a href=#181 id=181 data-nosnippet>181</a>        <span class="kw">let </span>should_load_page_index =
 <a href=#182 id=182 data-nosnippet>182</a>            (row_selection_enabled &amp;&amp; task.predicate.is_some()) || !task.deletes.is_empty();
 <a href=#183 id=183 data-nosnippet>183</a>
-<a href=#184 id=184 data-nosnippet>184</a>        <span class="kw">let </span>delete_filter_rx = delete_file_loader.load_deletes(<span class="kw-2">&amp;</span>task.deletes, task.schema.clone());
-<a href=#185 id=185 data-nosnippet>185</a>
-<a href=#186 id=186 data-nosnippet>186</a>        <span class="comment">// Migrated tables lack field IDs, requiring us to inspect the schema to choose
-<a href=#187 id=187 data-nosnippet>187</a>        // between field-ID-based or position-based projection
-<a href=#188 id=188 data-nosnippet>188</a>        </span><span class="kw">let </span>initial_stream_builder = <span class="self">Self</span>::create_parquet_record_batch_stream_builder(
-<a href=#189 id=189 data-nosnippet>189</a>            <span class="kw-2">&amp;</span>task.data_file_path,
-<a href=#190 id=190 data-nosnippet>190</a>            file_io.clone(),
-<a href=#191 id=191 data-nosnippet>191</a>            should_load_page_index,
-<a href=#192 id=192 data-nosnippet>192</a>            <span class="prelude-val">None</span>,
-<a href=#193 id=193 data-nosnippet>193</a>        )
-<a href=#194 id=194 data-nosnippet>194</a>        .<span class="kw">await</span><span class="question-mark">?</span>;
-<a href=#195 id=195 data-nosnippet>195</a>
-<a href=#196 id=196 data-nosnippet>196</a>        <span class="comment">// Parquet files from Hive/Spark migrations lack field IDs in their metadata
-<a href=#197 id=197 data-nosnippet>197</a>        </span><span class="kw">let </span>missing_field_ids = initial_stream_builder
-<a href=#198 id=198 data-nosnippet>198</a>            .schema()
-<a href=#199 id=199 data-nosnippet>199</a>            .fields()
-<a href=#200 id=200 data-nosnippet>200</a>            .iter()
-<a href=#201 id=201 data-nosnippet>201</a>            .next()
-<a href=#202 id=202 data-nosnippet>202</a>            .is_some_and(|f| f.metadata().get(PARQUET_FIELD_ID_META_KEY).is_none());
-<a href=#203 id=203 data-nosnippet>203</a>
-<a href=#204 id=204 data-nosnippet>204</a>        <span class="comment">// Adding position-based fallback IDs at schema level (not per-batch) enables projection
-<a href=#205 id=205 data-nosnippet>205</a>        // on files that lack embedded field IDs. We recreate the builder to apply the modified schema.
-<a href=#206 id=206 data-nosnippet>206</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>record_batch_stream_builder = <span class="kw">if </span>missing_field_ids {
-<a href=#207 id=207 data-nosnippet>207</a>            <span class="kw">let </span>arrow_schema =
-<a href=#208 id=208 data-nosnippet>208</a>                add_fallback_field_ids_to_arrow_schema(initial_stream_builder.schema());
-<a href=#209 id=209 data-nosnippet>209</a>            <span class="kw">let </span>options = ArrowReaderOptions::new().with_schema(arrow_schema);
-<a href=#210 id=210 data-nosnippet>210</a>
-<a href=#211 id=211 data-nosnippet>211</a>            <span class="self">Self</span>::create_parquet_record_batch_stream_builder(
-<a href=#212 id=212 data-nosnippet>212</a>                <span class="kw-2">&amp;</span>task.data_file_path,
-<a href=#213 id=213 data-nosnippet>213</a>                file_io.clone(),
-<a href=#214 id=214 data-nosnippet>214</a>                should_load_page_index,
-<a href=#215 id=215 data-nosnippet>215</a>                <span class="prelude-val">Some</span>(options),
-<a href=#216 id=216 data-nosnippet>216</a>            )
-<a href=#217 id=217 data-nosnippet>217</a>            .<span class="kw">await</span><span class="question-mark">?
-<a href=#218 id=218 data-nosnippet>218</a>        </span>} <span class="kw">else </span>{
-<a href=#219 id=219 data-nosnippet>219</a>            initial_stream_builder
-<a href=#220 id=220 data-nosnippet>220</a>        };
-<a href=#221 id=221 data-nosnippet>221</a>
-<a href=#222 id=222 data-nosnippet>222</a>        <span class="comment">// Fallback IDs don't match Parquet's embedded field IDs (since they don't exist),
-<a href=#223 id=223 data-nosnippet>223</a>        // so we must use position-based projection instead of field-ID matching
-<a href=#224 id=224 data-nosnippet>224</a>        </span><span class="kw">let </span>projection_mask = <span class="self">Self</span>::get_arrow_projection_mask(
-<a href=#225 id=225 data-nosnippet>225</a>            <span class="kw-2">&amp;</span>task.project_field_ids,
-<a href=#226 id=226 data-nosnippet>226</a>            <span class="kw-2">&amp;</span>task.schema,
-<a href=#227 id=227 data-nosnippet>227</a>            record_batch_stream_builder.parquet_schema(),
-<a href=#228 id=228 data-nosnippet>228</a>            record_batch_stream_builder.schema(),
-<a href=#229 id=229 data-nosnippet>229</a>            missing_field_ids, <span class="comment">// Whether to use position-based (true) or field-ID-based (false) projection
-<a href=#230 id=230 data-nosnippet>230</a>        </span>)<span class="question-mark">?</span>;
-<a href=#231 id=231 data-nosnippet>231</a>
-<a href=#232 id=232 data-nosnippet>232</a>        record_batch_stream_builder =
-<a href=#233 id=233 data-nosnippet>233</a>            record_batch_stream_builder.with_projection(projection_mask.clone());
-<a href=#234 id=234 data-nosnippet>234</a>
-<a href=#235 id=235 data-nosnippet>235</a>        <span class="comment">// RecordBatchTransformer performs any transformations required on the RecordBatches
-<a href=#236 id=236 data-nosnippet>236</a>        // that come back from the file, such as type promotion, default column insertion
-<a href=#237 id=237 data-nosnippet>237</a>        // and column re-ordering
-<a href=#238 id=238 data-nosnippet>238</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>record_batch_transformer =
-<a href=#239 id=239 data-nosnippet>239</a>            RecordBatchTransformer::build(task.schema_ref(), task.project_field_ids());
+<a href=#184 id=184 data-nosnippet>184</a>        <span class="kw">let </span>delete_filter_rx =
+<a href=#185 id=185 data-nosnippet>185</a>            delete_file_loader.load_deletes(<span class="kw-2">&amp;</span>task.deletes, Arc::clone(<span class="kw-2">&amp;</span>task.schema));
+<a href=#186 id=186 data-nosnippet>186</a>
+<a href=#187 id=187 data-nosnippet>187</a>        <span class="comment">// Migrated tables lack field IDs, requiring us to inspect the schema to choose
+<a href=#188 id=188 data-nosnippet>188</a>        // between field-ID-based or position-based projection
+<a href=#189 id=189 data-nosnippet>189</a>        </span><span class="kw">let </span>initial_stream_builder = <span class="self">Self</span>::create_parquet_record_batch_stream_builder(
+<a href=#190 id=190 data-nosnippet>190</a>            <span class="kw-2">&amp;</span>task.data_file_path,
+<a href=#191 id=191 data-nosnippet>191</a>            file_io.clone(),
+<a href=#192 id=192 data-nosnippet>192</a>            should_load_page_index,
+<a href=#193 id=193 data-nosnippet>193</a>            <span class="prelude-val">None</span>,
+<a href=#194 id=194 data-nosnippet>194</a>        )
+<a href=#195 id=195 data-nosnippet>195</a>        .<span class="kw">await</span><span class="question-mark">?</span>;
+<a href=#196 id=196 data-nosnippet>196</a>
+<a href=#197 id=197 data-nosnippet>197</a>        <span class="comment">// Check if Parquet file has embedded field IDs
+<a href=#198 id=198 data-nosnippet>198</a>        // Corresponds to Java's ParquetSchemaUtil.hasIds()
+<a href=#199 id=199 data-nosnippet>199</a>        // Reference: parquet/src/main/java/org/apache/iceberg/parquet/ParquetSchemaUtil.java:118
+<a href=#200 id=200 data-nosnippet>200</a>        </span><span class="kw">let </span>missing_field_ids = initial_stream_builder
+<a href=#201 id=201 data-nosnippet>201</a>            .schema()
+<a href=#202 id=202 data-nosnippet>202</a>            .fields()
+<a href=#203 id=203 data-nosnippet>203</a>            .iter()
+<a href=#204 id=204 data-nosnippet>204</a>            .next()
+<a href=#205 id=205 data-nosnippet>205</a>            .is_some_and(|f| f.metadata().get(PARQUET_FIELD_ID_META_KEY).is_none());
+<a href=#206 id=206 data-nosnippet>206</a>
+<a href=#207 id=207 data-nosnippet>207</a>        <span class="comment">// Three-branch schema resolution strategy matching Java's ReadConf constructor
+<a href=#208 id=208 data-nosnippet>208</a>        //
+<a href=#209 id=209 data-nosnippet>209</a>        // Per Iceberg spec Column Projection rules:
+<a href=#210 id=210 data-nosnippet>210</a>        // "Columns in Iceberg data files are selected by field id. The table schema's column
+<a href=#211 id=211 data-nosnippet>211</a>        //  names and order may change after a data file is written, and projection must be done
+<a href=#212 id=212 data-nosnippet>212</a>        //  using field ids."
+<a href=#213 id=213 data-nosnippet>213</a>        // https://iceberg.apache.org/spec/#column-projection
+<a href=#214 id=214 data-nosnippet>214</a>        //
+<a href=#215 id=215 data-nosnippet>215</a>        // When Parquet files lack field IDs (e.g., Hive/Spark migrations via add_files),
+<a href=#216 id=216 data-nosnippet>216</a>        // we must assign field IDs BEFORE reading data to enable correct projection.
+<a href=#217 id=217 data-nosnippet>217</a>        //
+<a href=#218 id=218 data-nosnippet>218</a>        // Java's ReadConf determines field ID strategy:
+<a href=#219 id=219 data-nosnippet>219</a>        // - Branch 1: hasIds(fileSchema) → trust embedded field IDs, use pruneColumns()
+<a href=#220 id=220 data-nosnippet>220</a>        // - Branch 2: nameMapping present → applyNameMapping(), then pruneColumns()
+<a href=#221 id=221 data-nosnippet>221</a>        // - Branch 3: fallback → addFallbackIds(), then pruneColumnsFallback()
+<a href=#222 id=222 data-nosnippet>222</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>record_batch_stream_builder = <span class="kw">if </span>missing_field_ids {
+<a href=#223 id=223 data-nosnippet>223</a>            <span class="comment">// Parquet file lacks field IDs - must assign them before reading
+<a href=#224 id=224 data-nosnippet>224</a>            </span><span class="kw">let </span>arrow_schema = <span class="kw">if let </span><span class="prelude-val">Some</span>(name_mapping) = <span class="kw-2">&amp;</span>task.name_mapping {
+<a href=#225 id=225 data-nosnippet>225</a>                <span class="comment">// Branch 2: Apply name mapping to assign correct Iceberg field IDs
+<a href=#226 id=226 data-nosnippet>226</a>                // Per spec rule #2: "Use schema.name-mapping.default metadata to map field id
+<a href=#227 id=227 data-nosnippet>227</a>                // to columns without field id"
+<a href=#228 id=228 data-nosnippet>228</a>                // Corresponds to Java's ParquetSchemaUtil.applyNameMapping()
+<a href=#229 id=229 data-nosnippet>229</a>                </span>apply_name_mapping_to_arrow_schema(
+<a href=#230 id=230 data-nosnippet>230</a>                    Arc::clone(initial_stream_builder.schema()),
+<a href=#231 id=231 data-nosnippet>231</a>                    name_mapping,
+<a href=#232 id=232 data-nosnippet>232</a>                )<span class="question-mark">?
+<a href=#233 id=233 data-nosnippet>233</a>            </span>} <span class="kw">else </span>{
+<a href=#234 id=234 data-nosnippet>234</a>                <span class="comment">// Branch 3: No name mapping - use position-based fallback IDs
+<a href=#235 id=235 data-nosnippet>235</a>                // Corresponds to Java's ParquetSchemaUtil.addFallbackIds()
+<a href=#236 id=236 data-nosnippet>236</a>                </span>add_fallback_field_ids_to_arrow_schema(initial_stream_builder.schema())
+<a href=#237 id=237 data-nosnippet>237</a>            };
+<a href=#238 id=238 data-nosnippet>238</a>
+<a href=#239 id=239 data-nosnippet>239</a>            <span class="kw">let </span>options = ArrowReaderOptions::new().with_schema(arrow_schema);
 <a href=#240 id=240 data-nosnippet>240</a>
-<a href=#241 id=241 data-nosnippet>241</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(batch_size) = batch_size {
-<a href=#242 id=242 data-nosnippet>242</a>            record_batch_stream_builder = record_batch_stream_builder.with_batch_size(batch_size);
-<a href=#243 id=243 data-nosnippet>243</a>        }
-<a href=#244 id=244 data-nosnippet>244</a>
-<a href=#245 id=245 data-nosnippet>245</a>        <span class="kw">let </span>delete_filter = delete_filter_rx.<span class="kw">await</span>.unwrap()<span class="question-mark">?</span>;
-<a href=#246 id=246 data-nosnippet>246</a>        <span class="kw">let </span>delete_predicate = delete_filter.build_equality_delete_predicate(<span class="kw-2">&amp;</span>task).<span class="kw">await</span><span class="question-mark">?</span>;
-<a href=#247 id=247 data-nosnippet>247</a>
-<a href=#248 id=248 data-nosnippet>248</a>        <span class="comment">// In addition to the optional predicate supplied in the `FileScanTask`,
-<a href=#249 id=249 data-nosnippet>249</a>        // we also have an optional predicate resulting from equality delete files.
-<a href=#250 id=250 data-nosnippet>250</a>        // If both are present, we logical-AND them together to form a single filter
-<a href=#251 id=251 data-nosnippet>251</a>        // predicate that we can pass to the `RecordBatchStreamBuilder`.
-<a href=#252 id=252 data-nosnippet>252</a>        </span><span class="kw">let </span>final_predicate = <span class="kw">match </span>(<span class="kw-2">&amp;</span>task.predicate, delete_predicate) {
-<a href=#253 id=253 data-nosnippet>253</a>            (<span class="prelude-val">None</span>, <span class="prelude-val">None</span>) =&gt; <span class="prelude-val">None</span>,
-<a href=#254 id=254 data-nosnippet>254</a>            (<span class="prelude-val">Some</span>(predicate), <span class="prelude-val">None</span>) =&gt; <span class="prelude-val">Some</span>(predicate.clone()),
-<a href=#255 id=255 data-nosnippet>255</a>            (<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="kw-2">ref </span>predicate)) =&gt; <span class="prelude-val">Some</span>(predicate.clone()),
-<a href=#256 id=256 data-nosnippet>256</a>            (<span class="prelude-val">Some</span>(filter_predicate), <span class="prelude-val">Some</span>(delete_predicate)) =&gt; {
-<a href=#257 id=257 data-nosnippet>257</a>                <span class="prelude-val">Some</span>(filter_predicate.clone().and(delete_predicate))
-<a href=#258 id=258 data-nosnippet>258</a>            }
-<a href=#259 id=259 data-nosnippet>259</a>        };
-<a href=#260 id=260 data-nosnippet>260</a>
-<a href=#261 id=261 data-nosnippet>261</a>        <span class="comment">// There are three possible sources for potential lists of selected RowGroup indices,
-<a href=#262 id=262 data-nosnippet>262</a>        // and two for `RowSelection`s.
-<a href=#263 id=263 data-nosnippet>263</a>        // Selected RowGroup index lists can come from three sources:
-<a href=#264 id=264 data-nosnippet>264</a>        //   * When task.start and task.length specify a byte range (file splitting);
-<a href=#265 id=265 data-nosnippet>265</a>        //   * When there are equality delete files that are applicable;
-<a href=#266 id=266 data-nosnippet>266</a>        //   * When there is a scan predicate and row_group_filtering_enabled = true.
-<a href=#267 id=267 data-nosnippet>267</a>        // `RowSelection`s can be created in either or both of the following cases:
-<a href=#268 id=268 data-nosnippet>268</a>        //   * When there are positional delete files that are applicable;
-<a href=#269 id=269 data-nosnippet>269</a>        //   * When there is a scan predicate and row_selection_enabled = true
-<a href=#270 id=270 data-nosnippet>270</a>        // Note that row group filtering from predicates only happens when
-<a href=#271 id=271 data-nosnippet>271</a>        // there is a scan predicate AND row_group_filtering_enabled = true,
-<a href=#272 id=272 data-nosnippet>272</a>        // but we perform row selection filtering if there are applicable
-<a href=#273 id=273 data-nosnippet>273</a>        // equality delete files OR (there is a scan predicate AND row_selection_enabled),
-<a href=#274 id=274 data-nosnippet>274</a>        // since the only implemented method of applying positional deletes is
-<a href=#275 id=275 data-nosnippet>275</a>        // by using a `RowSelection`.
-<a href=#276 id=276 data-nosnippet>276</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>selected_row_group_indices = <span class="prelude-val">None</span>;
-<a href=#277 id=277 data-nosnippet>277</a>        <span class="kw">let </span><span class="kw-2">mut </span>row_selection = <span class="prelude-val">None</span>;
-<a href=#278 id=278 data-nosnippet>278</a>
-<a href=#279 id=279 data-nosnippet>279</a>        <span class="comment">// Filter row groups based on byte range from task.start and task.length.
-<a href=#280 id=280 data-nosnippet>280</a>        // If both start and length are 0, read the entire file (backwards compatibility).
-<a href=#281 id=281 data-nosnippet>281</a>        </span><span class="kw">if </span>task.start != <span class="number">0 </span>|| task.length != <span class="number">0 </span>{
-<a href=#282 id=282 data-nosnippet>282</a>            <span class="kw">let </span>byte_range_filtered_row_groups = <span class="self">Self</span>::filter_row_groups_by_byte_range(
-<a href=#283 id=283 data-nosnippet>283</a>                record_batch_stream_builder.metadata(),
-<a href=#284 id=284 data-nosnippet>284</a>                task.start,
-<a href=#285 id=285 data-nosnippet>285</a>                task.length,
-<a href=#286 id=286 data-nosnippet>286</a>            )<span class="question-mark">?</span>;
-<a href=#287 id=287 data-nosnippet>287</a>            selected_row_group_indices = <span class="prelude-val">Some</span>(byte_range_filtered_row_groups);
-<a href=#288 id=288 data-nosnippet>288</a>        }
+<a href=#241 id=241 data-nosnippet>241</a>            <span class="self">Self</span>::create_parquet_record_batch_stream_builder(
+<a href=#242 id=242 data-nosnippet>242</a>                <span class="kw-2">&amp;</span>task.data_file_path,
+<a href=#243 id=243 data-nosnippet>243</a>                file_io.clone(),
+<a href=#244 id=244 data-nosnippet>244</a>                should_load_page_index,
+<a href=#245 id=245 data-nosnippet>245</a>                <span class="prelude-val">Some</span>(options),
+<a href=#246 id=246 data-nosnippet>246</a>            )
+<a href=#247 id=247 data-nosnippet>247</a>            .<span class="kw">await</span><span class="question-mark">?
+<a href=#248 id=248 data-nosnippet>248</a>        </span>} <span class="kw">else </span>{
+<a href=#249 id=249 data-nosnippet>249</a>            <span class="comment">// Branch 1: File has embedded field IDs - trust them
+<a href=#250 id=250 data-nosnippet>250</a>            </span>initial_stream_builder
+<a href=#251 id=251 data-nosnippet>251</a>        };
+<a href=#252 id=252 data-nosnippet>252</a>
+<a href=#253 id=253 data-nosnippet>253</a>        <span class="comment">// Create projection mask based on field IDs
+<a href=#254 id=254 data-nosnippet>254</a>        // - If file has embedded IDs: field-ID-based projection (missing_field_ids=false)
+<a href=#255 id=255 data-nosnippet>255</a>        // - If name mapping applied: field-ID-based projection (missing_field_ids=true but IDs now match)
+<a href=#256 id=256 data-nosnippet>256</a>        // - If fallback IDs: position-based projection (missing_field_ids=true)
+<a href=#257 id=257 data-nosnippet>257</a>        </span><span class="kw">let </span>projection_mask = <span class="self">Self</span>::get_arrow_projection_mask(
+<a href=#258 id=258 data-nosnippet>258</a>            <span class="kw-2">&amp;</span>task.project_field_ids,
+<a href=#259 id=259 data-nosnippet>259</a>            <span class="kw-2">&amp;</span>task.schema,
+<a href=#260 id=260 data-nosnippet>260</a>            record_batch_stream_builder.parquet_schema(),
+<a href=#261 id=261 data-nosnippet>261</a>            record_batch_stream_builder.schema(),
+<a href=#262 id=262 data-nosnippet>262</a>            missing_field_ids, <span class="comment">// Whether to use position-based (true) or field-ID-based (false) projection
+<a href=#263 id=263 data-nosnippet>263</a>        </span>)<span class="question-mark">?</span>;
+<a href=#264 id=264 data-nosnippet>264</a>
+<a href=#265 id=265 data-nosnippet>265</a>        record_batch_stream_builder =
+<a href=#266 id=266 data-nosnippet>266</a>            record_batch_stream_builder.with_projection(projection_mask.clone());
+<a href=#267 id=267 data-nosnippet>267</a>
+<a href=#268 id=268 data-nosnippet>268</a>        <span class="comment">// RecordBatchTransformer performs any transformations required on the RecordBatches
+<a href=#269 id=269 data-nosnippet>269</a>        // that come back from the file, such as type promotion, default column insertion
+<a href=#270 id=270 data-nosnippet>270</a>        // and column re-ordering.
+<a href=#271 id=271 data-nosnippet>271</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>record_batch_transformer_builder =
+<a href=#272 id=272 data-nosnippet>272</a>            RecordBatchTransformerBuilder::new(task.schema_ref(), task.project_field_ids());
+<a href=#273 id=273 data-nosnippet>273</a>
+<a href=#274 id=274 data-nosnippet>274</a>        <span class="kw">if let </span>(<span class="prelude-val">Some</span>(partition_spec), <span class="prelude-val">Some</span>(partition_data)) =
+<a href=#275 id=275 data-nosnippet>275</a>            (task.partition_spec.clone(), task.partition.clone())
+<a href=#276 id=276 data-nosnippet>276</a>        {
+<a href=#277 id=277 data-nosnippet>277</a>            record_batch_transformer_builder =
+<a href=#278 id=278 data-nosnippet>278</a>                record_batch_transformer_builder.with_partition(partition_spec, partition_data);
+<a href=#279 id=279 data-nosnippet>279</a>        }
+<a href=#280 id=280 data-nosnippet>280</a>
+<a href=#281 id=281 data-nosnippet>281</a>        <span class="kw">let </span><span class="kw-2">mut </span>record_batch_transformer = record_batch_transformer_builder.build();
+<a href=#282 id=282 data-nosnippet>282</a>
+<a href=#283 id=283 data-nosnippet>283</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(batch_size) = batch_size {
+<a href=#284 id=284 data-nosnippet>284</a>            record_batch_stream_builder = record_batch_stream_builder.with_batch_size(batch_size);
+<a href=#285 id=285 data-nosnippet>285</a>        }
+<a href=#286 id=286 data-nosnippet>286</a>
+<a href=#287 id=287 data-nosnippet>287</a>        <span class="kw">let </span>delete_filter = delete_filter_rx.<span class="kw">await</span>.unwrap()<span class="question-mark">?</span>;
+<a href=#288 id=288 data-nosnippet>288</a>        <span class="kw">let </span>delete_predicate = delete_filter.build_equality_delete_predicate(<span class="kw-2">&amp;</span>task).<span class="kw">await</span><span class="question-mark">?</span>;
 <a href=#289 id=289 data-nosnippet>289</a>
-<a href=#290 id=290 data-nosnippet>290</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(predicate) = final_predicate {
-<a href=#291 id=291 data-nosnippet>291</a>            <span class="kw">let </span>(iceberg_field_ids, field_id_map) = <span class="self">Self</span>::build_field_id_set_and_map(
-<a href=#292 id=292 data-nosnippet>292</a>                record_batch_stream_builder.parquet_schema(),
-<a href=#293 id=293 data-nosnippet>293</a>                <span class="kw-2">&amp;</span>predicate,
-<a href=#294 id=294 data-nosnippet>294</a>            )<span class="question-mark">?</span>;
-<a href=#295 id=295 data-nosnippet>295</a>
-<a href=#296 id=296 data-nosnippet>296</a>            <span class="kw">let </span>row_filter = <span class="self">Self</span>::get_row_filter(
-<a href=#297 id=297 data-nosnippet>297</a>                <span class="kw-2">&amp;</span>predicate,
-<a href=#298 id=298 data-nosnippet>298</a>                record_batch_stream_builder.parquet_schema(),
-<a href=#299 id=299 data-nosnippet>299</a>                <span class="kw-2">&amp;</span>iceberg_field_ids,
-<a href=#300 id=300 data-nosnippet>300</a>                <span class="kw-2">&amp;</span>field_id_map,
-<a href=#301 id=301 data-nosnippet>301</a>            )<span class="question-mark">?</span>;
-<a href=#302 id=302 data-nosnippet>302</a>            record_batch_stream_builder = record_batch_stream_builder.with_row_filter(row_filter);
-<a href=#303 id=303 data-nosnippet>303</a>
-<a href=#304 id=304 data-nosnippet>304</a>            <span class="kw">if </span>row_group_filtering_enabled {
-<a href=#305 id=305 data-nosnippet>305</a>                <span class="kw">let </span>predicate_filtered_row_groups = <span class="self">Self</span>::get_selected_row_group_indices(
-<a href=#306 id=306 data-nosnippet>306</a>                    <span class="kw-2">&amp;</span>predicate,
-<a href=#307 id=307 data-nosnippet>307</a>                    record_batch_stream_builder.metadata(),
-<a href=#308 id=308 data-nosnippet>308</a>                    <span class="kw-2">&amp;</span>field_id_map,
-<a href=#309 id=309 data-nosnippet>309</a>                    <span class="kw-2">&amp;</span>task.schema,
-<a href=#310 id=310 data-nosnippet>310</a>                )<span class="question-mark">?</span>;
-<a href=#311 id=311 data-nosnippet>311</a>
-<a href=#312 id=312 data-nosnippet>312</a>                <span class="comment">// Merge predicate-based filtering with byte range filtering (if present)
-<a href=#313 id=313 data-nosnippet>313</a>                // by taking the intersection of both filters
-<a href=#314 id=314 data-nosnippet>314</a>                </span>selected_row_group_indices = <span class="kw">match </span>selected_row_group_indices {
-<a href=#315 id=315 data-nosnippet>315</a>                    <span class="prelude-val">Some</span>(byte_range_filtered) =&gt; {
-<a href=#316 id=316 data-nosnippet>316</a>                        <span class="comment">// Keep only row groups that are in both filters
-<a href=#317 id=317 data-nosnippet>317</a>                        </span><span class="kw">let </span>intersection: Vec&lt;usize&gt; = byte_range_filtered
-<a href=#318 id=318 data-nosnippet>318</a>                            .into_iter()
-<a href=#319 id=319 data-nosnippet>319</a>                            .filter(|idx| predicate_filtered_row_groups.contains(idx))
-<a href=#320 id=320 data-nosnippet>320</a>                            .collect();
-<a href=#321 id=321 data-nosnippet>321</a>                        <span class="prelude-val">Some</span>(intersection)
-<a href=#322 id=322 data-nosnippet>322</a>                    }
-<a href=#323 id=323 data-nosnippet>323</a>                    <span class="prelude-val">None </span>=&gt; <span class="prelude-val">Some</span>(predicate_filtered_row_groups),
-<a href=#324 id=324 data-nosnippet>324</a>                };
-<a href=#325 id=325 data-nosnippet>325</a>            }
-<a href=#326 id=326 data-nosnippet>326</a>
-<a href=#327 id=327 data-nosnippet>327</a>            <span class="kw">if </span>row_selection_enabled {
-<a href=#328 id=328 data-nosnippet>328</a>                row_selection = <span class="prelude-val">Some</span>(<span class="self">Self</span>::get_row_selection_for_filter_predicate(
-<a href=#329 id=329 data-nosnippet>329</a>                    <span class="kw-2">&amp;</span>predicate,
-<a href=#330 id=330 data-nosnippet>330</a>                    record_batch_stream_builder.metadata(),
-<a href=#331 id=331 data-nosnippet>331</a>                    <span class="kw-2">&amp;</span>selected_row_group_indices,
-<a href=#332 id=332 data-nosnippet>332</a>                    <span class="kw-2">&amp;</span>field_id_map,
-<a href=#333 id=333 data-nosnippet>333</a>                    <span class="kw-2">&amp;</span>task.schema,
-<a href=#334 id=334 data-nosnippet>334</a>                )<span class="question-mark">?</span>);
-<a href=#335 id=335 data-nosnippet>335</a>            }
-<a href=#336 id=336 data-nosnippet>336</a>        }
+<a href=#290 id=290 data-nosnippet>290</a>        <span class="comment">// In addition to the optional predicate supplied in the `FileScanTask`,
+<a href=#291 id=291 data-nosnippet>291</a>        // we also have an optional predicate resulting from equality delete files.
+<a href=#292 id=292 data-nosnippet>292</a>        // If both are present, we logical-AND them together to form a single filter
+<a href=#293 id=293 data-nosnippet>293</a>        // predicate that we can pass to the `RecordBatchStreamBuilder`.
+<a href=#294 id=294 data-nosnippet>294</a>        </span><span class="kw">let </span>final_predicate = <span class="kw">match </span>(<span class="kw-2">&amp;</span>task.predicate, delete_predicate) {
+<a href=#295 id=295 data-nosnippet>295</a>            (<span class="prelude-val">None</span>, <span class="prelude-val">None</span>) =&gt; <span class="prelude-val">None</span>,
+<a href=#296 id=296 data-nosnippet>296</a>            (<span class="prelude-val">Some</span>(predicate), <span class="prelude-val">None</span>) =&gt; <span class="prelude-val">Some</span>(predicate.clone()),
+<a href=#297 id=297 data-nosnippet>297</a>            (<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="kw-2">ref </span>predicate)) =&gt; <span class="prelude-val">Some</span>(predicate.clone()),
+<a href=#298 id=298 data-nosnippet>298</a>            (<span class="prelude-val">Some</span>(filter_predicate), <span class="prelude-val">Some</span>(delete_predicate)) =&gt; {
+<a href=#299 id=299 data-nosnippet>299</a>                <span class="prelude-val">Some</span>(filter_predicate.clone().and(delete_predicate))
+<a href=#300 id=300 data-nosnippet>300</a>            }
+<a href=#301 id=301 data-nosnippet>301</a>        };
+<a href=#302 id=302 data-nosnippet>302</a>
+<a href=#303 id=303 data-nosnippet>303</a>        <span class="comment">// There are three possible sources for potential lists of selected RowGroup indices,
+<a href=#304 id=304 data-nosnippet>304</a>        // and two for `RowSelection`s.
+<a href=#305 id=305 data-nosnippet>305</a>        // Selected RowGroup index lists can come from three sources:
+<a href=#306 id=306 data-nosnippet>306</a>        //   * When task.start and task.length specify a byte range (file splitting);
+<a href=#307 id=307 data-nosnippet>307</a>        //   * When there are equality delete files that are applicable;
+<a href=#308 id=308 data-nosnippet>308</a>        //   * When there is a scan predicate and row_group_filtering_enabled = true.
+<a href=#309 id=309 data-nosnippet>309</a>        // `RowSelection`s can be created in either or both of the following cases:
+<a href=#310 id=310 data-nosnippet>310</a>        //   * When there are positional delete files that are applicable;
+<a href=#311 id=311 data-nosnippet>311</a>        //   * When there is a scan predicate and row_selection_enabled = true
+<a href=#312 id=312 data-nosnippet>312</a>        // Note that row group filtering from predicates only happens when
+<a href=#313 id=313 data-nosnippet>313</a>        // there is a scan predicate AND row_group_filtering_enabled = true,
+<a href=#314 id=314 data-nosnippet>314</a>        // but we perform row selection filtering if there are applicable
+<a href=#315 id=315 data-nosnippet>315</a>        // equality delete files OR (there is a scan predicate AND row_selection_enabled),
+<a href=#316 id=316 data-nosnippet>316</a>        // since the only implemented method of applying positional deletes is
+<a href=#317 id=317 data-nosnippet>317</a>        // by using a `RowSelection`.
+<a href=#318 id=318 data-nosnippet>318</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>selected_row_group_indices = <span class="prelude-val">None</span>;
+<a href=#319 id=319 data-nosnippet>319</a>        <span class="kw">let </span><span class="kw-2">mut </span>row_selection = <span class="prelude-val">None</span>;
+<a href=#320 id=320 data-nosnippet>320</a>
+<a href=#321 id=321 data-nosnippet>321</a>        <span class="comment">// Filter row groups based on byte range from task.start and task.length.
+<a href=#322 id=322 data-nosnippet>322</a>        // If both start and length are 0, read the entire file (backwards compatibility).
+<a href=#323 id=323 data-nosnippet>323</a>        </span><span class="kw">if </span>task.start != <span class="number">0 </span>|| task.length != <span class="number">0 </span>{
+<a href=#324 id=324 data-nosnippet>324</a>            <span class="kw">let </span>byte_range_filtered_row_groups = <span class="self">Self</span>::filter_row_groups_by_byte_range(
+<a href=#325 id=325 data-nosnippet>325</a>                record_batch_stream_builder.metadata(),
+<a href=#326 id=326 data-nosnippet>326</a>                task.start,
+<a href=#327 id=327 data-nosnippet>327</a>                task.length,
+<a href=#328 id=328 data-nosnippet>328</a>            )<span class="question-mark">?</span>;
+<a href=#329 id=329 data-nosnippet>329</a>            selected_row_group_indices = <span class="prelude-val">Some</span>(byte_range_filtered_row_groups);
+<a href=#330 id=330 data-nosnippet>330</a>        }
+<a href=#331 id=331 data-nosnippet>331</a>
+<a href=#332 id=332 data-nosnippet>332</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(predicate) = final_predicate {
+<a href=#333 id=333 data-nosnippet>333</a>            <span class="kw">let </span>(iceberg_field_ids, field_id_map) = <span class="self">Self</span>::build_field_id_set_and_map(
+<a href=#334 id=334 data-nosnippet>334</a>                record_batch_stream_builder.parquet_schema(),
+<a href=#335 id=335 data-nosnippet>335</a>                <span class="kw-2">&amp;</span>predicate,
+<a href=#336 id=336 data-nosnippet>336</a>            )<span class="question-mark">?</span>;
 <a href=#337 id=337 data-nosnippet>337</a>
-<a href=#338 id=338 data-nosnippet>338</a>        <span class="kw">let </span>positional_delete_indexes = delete_filter.get_delete_vector(<span class="kw-2">&amp;</span>task);
-<a href=#339 id=339 data-nosnippet>339</a>
-<a href=#340 id=340 data-nosnippet>340</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(positional_delete_indexes) = positional_delete_indexes {
-<a href=#341 id=341 data-nosnippet>341</a>            <span class="kw">let </span>delete_row_selection = {
-<a href=#342 id=342 data-nosnippet>342</a>                <span class="kw">let </span>positional_delete_indexes = positional_delete_indexes.lock().unwrap();
-<a href=#343 id=343 data-nosnippet>343</a>
-<a href=#344 id=344 data-nosnippet>344</a>                <span class="self">Self</span>::build_deletes_row_selection(
-<a href=#345 id=345 data-nosnippet>345</a>                    record_batch_stream_builder.metadata().row_groups(),
-<a href=#346 id=346 data-nosnippet>346</a>                    <span class="kw-2">&amp;</span>selected_row_group_indices,
-<a href=#347 id=347 data-nosnippet>347</a>                    <span class="kw-2">&amp;</span>positional_delete_indexes,
-<a href=#348 id=348 data-nosnippet>348</a>                )
-<a href=#349 id=349 data-nosnippet>349</a>            }<span class="question-mark">?</span>;
-<a href=#350 id=350 data-nosnippet>350</a>
-<a href=#351 id=351 data-nosnippet>351</a>            <span class="comment">// merge the row selection from the delete files with the row selection
-<a href=#352 id=352 data-nosnippet>352</a>            // from the filter predicate, if there is one from the filter predicate
-<a href=#353 id=353 data-nosnippet>353</a>            </span>row_selection = <span class="kw">match </span>row_selection {
-<a href=#354 id=354 data-nosnippet>354</a>                <span class="prelude-val">None </span>=&gt; <span class="prelude-val">Some</span>(delete_row_selection),
-<a href=#355 id=355 data-nosnippet>355</a>                <span class="prelude-val">Some</span>(filter_row_selection) =&gt; {
-<a href=#356 id=356 data-nosnippet>356</a>                    <span class="prelude-val">Some</span>(filter_row_selection.intersection(<span class="kw-2">&amp;</span>delete_row_selection))
-<a href=#357 id=357 data-nosnippet>357</a>                }
-<a href=#358 id=358 data-nosnippet>358</a>            };
-<a href=#359 id=359 data-nosnippet>359</a>        }
-<a href=#360 id=360 data-nosnippet>360</a>
-<a href=#361 id=361 data-nosnippet>361</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(row_selection) = row_selection {
-<a href=#362 id=362 data-nosnippet>362</a>            record_batch_stream_builder =
-<a href=#363 id=363 data-nosnippet>363</a>                record_batch_stream_builder.with_row_selection(row_selection);
-<a href=#364 id=364 data-nosnippet>364</a>        }
-<a href=#365 id=365 data-nosnippet>365</a>
-<a href=#366 id=366 data-nosnippet>366</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_group_indices) = selected_row_group_indices {
-<a href=#367 id=367 data-nosnippet>367</a>            record_batch_stream_builder =
-<a href=#368 id=368 data-nosnippet>368</a>                record_batch_stream_builder.with_row_groups(selected_row_group_indices);
-<a href=#369 id=369 data-nosnippet>369</a>        }
-<a href=#370 id=370 data-nosnippet>370</a>
-<a href=#371 id=371 data-nosnippet>371</a>        <span class="comment">// Build the batch stream and send all the RecordBatches that it generates
-<a href=#372 id=372 data-nosnippet>372</a>        // to the requester.
-<a href=#373 id=373 data-nosnippet>373</a>        </span><span class="kw">let </span>record_batch_stream =
-<a href=#374 id=374 data-nosnippet>374</a>            record_batch_stream_builder
-<a href=#375 id=375 data-nosnippet>375</a>                .build()<span class="question-mark">?
-<a href=#376 id=376 data-nosnippet>376</a>                </span>.map(<span class="kw">move </span>|batch| <span class="kw">match </span>batch {
-<a href=#377 id=377 data-nosnippet>377</a>                    <span class="prelude-val">Ok</span>(batch) =&gt; record_batch_transformer.process_record_batch(batch),
-<a href=#378 id=378 data-nosnippet>378</a>                    <span class="prelude-val">Err</span>(err) =&gt; <span class="prelude-val">Err</span>(err.into()),
-<a href=#379 id=379 data-nosnippet>379</a>                });
-<a href=#380 id=380 data-nosnippet>380</a>
-<a href=#381 id=381 data-nosnippet>381</a>        <span class="prelude-val">Ok</span>(Box::pin(record_batch_stream) <span class="kw">as </span>ArrowRecordBatchStream)
-<a href=#382 id=382 data-nosnippet>382</a>    }
-<a href=#383 id=383 data-nosnippet>383</a>
-<a href=#384 id=384 data-nosnippet>384</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">async fn </span>create_parquet_record_batch_stream_builder(
-<a href=#385 id=385 data-nosnippet>385</a>        data_file_path: <span class="kw-2">&amp;</span>str,
-<a href=#386 id=386 data-nosnippet>386</a>        file_io: FileIO,
-<a href=#387 id=387 data-nosnippet>387</a>        should_load_page_index: bool,
-<a href=#388 id=388 data-nosnippet>388</a>        arrow_reader_options: <span class="prelude-ty">Option</span>&lt;ArrowReaderOptions&gt;,
-<a href=#389 id=389 data-nosnippet>389</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ParquetRecordBatchStreamBuilder&lt;ArrowFileReader&lt;<span class="kw">impl </span>FileRead + Sized&gt;&gt;&gt; {
-<a href=#390 id=390 data-nosnippet>390</a>        <span class="comment">// Get the metadata for the Parquet file we need to read and build
-<a href=#391 id=391 data-nosnippet>391</a>        // a reader for the data within
-<a href=#392 id=392 data-nosnippet>392</a>        </span><span class="kw">let </span>parquet_file = file_io.new_input(data_file_path)<span class="question-mark">?</span>;
-<a href=#393 id=393 data-nosnippet>393</a>        <span class="kw">let </span>(parquet_metadata, parquet_reader) =
-<a href=#394 id=394 data-nosnippet>394</a>            <span class="macro">try_join!</span>(parquet_file.metadata(), parquet_file.reader())<span class="question-mark">?</span>;
-<a href=#395 id=395 data-nosnippet>395</a>        <span class="kw">let </span>parquet_file_reader = ArrowFileReader::new(parquet_metadata, parquet_reader)
-<a href=#396 id=396 data-nosnippet>396</a>            .with_preload_column_index(<span class="bool-val">true</span>)
-<a href=#397 id=397 data-nosnippet>397</a>            .with_preload_offset_index(<span class="bool-val">true</span>)
-<a href=#398 id=398 data-nosnippet>398</a>            .with_preload_page_index(should_load_page_index);
-<a href=#399 id=399 data-nosnippet>399</a>
-<a href=#400 id=400 data-nosnippet>400</a>        <span class="comment">// Create the record batch stream builder, which wraps the parquet file reader
-<a href=#401 id=401 data-nosnippet>401</a>        </span><span class="kw">let </span>options = arrow_reader_options.unwrap_or_default();
-<a href=#402 id=402 data-nosnippet>402</a>        <span class="kw">let </span>record_batch_stream_builder =
-<a href=#403 id=403 data-nosnippet>403</a>            ParquetRecordBatchStreamBuilder::new_with_options(parquet_file_reader, options).<span class="kw">await</span><span class="question-mark">?</span>;
-<a href=#404 id=404 data-nosnippet>404</a>        <span class="prelude-val">Ok</span>(record_batch_stream_builder)
-<a href=#405 id=405 data-nosnippet>405</a>    }
-<a href=#406 id=406 data-nosnippet>406</a>
-<a href=#407 id=407 data-nosnippet>407</a>    <span class="doccomment">/// computes a `RowSelection` from positional delete indices.
-<a href=#408 id=408 data-nosnippet>408</a>    ///
-<a href=#409 id=409 data-nosnippet>409</a>    /// Using the Parquet page index, we build a `RowSelection` that rejects rows that are indicated
-<a href=#410 id=410 data-nosnippet>410</a>    /// as having been deleted by a positional delete, taking into account any row groups that have
-<a href=#411 id=411 data-nosnippet>411</a>    /// been skipped entirely by the filter predicate
-<a href=#412 id=412 data-nosnippet>412</a>    </span><span class="kw">fn </span>build_deletes_row_selection(
-<a href=#413 id=413 data-nosnippet>413</a>        row_group_metadata_list: <span class="kw-2">&amp;</span>[RowGroupMetaData],
-<a href=#414 id=414 data-nosnippet>414</a>        selected_row_groups: <span class="kw-2">&amp;</span><span class="prelude-ty">Option</span>&lt;Vec&lt;usize&gt;&gt;,
-<a href=#415 id=415 data-nosnippet>415</a>        positional_deletes: <span class="kw-2">&amp;</span>DeleteVector,
-<a href=#416 id=416 data-nosnippet>416</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;RowSelection&gt; {
-<a href=#417 id=417 data-nosnippet>417</a>        <span class="kw">let </span><span class="kw-2">mut </span>results: Vec&lt;RowSelector&gt; = Vec::new();
-<a href=#418 id=418 data-nosnippet>418</a>        <span class="kw">let </span><span class="kw-2">mut </span>selected_row_groups_idx = <span class="number">0</span>;
-<a href=#419 id=419 data-nosnippet>419</a>        <span class="kw">let </span><span class="kw-2">mut </span>current_row_group_base_idx: u64 = <span class="number">0</span>;
-<a href=#420 id=420 data-nosnippet>420</a>        <span class="kw">let </span><span class="kw-2">mut </span>delete_vector_iter = positional_deletes.iter();
-<a href=#421 id=421 data-nosnippet>421</a>        <span class="kw">let </span><span class="kw-2">mut </span>next_deleted_row_idx_opt = delete_vector_iter.next();
+<a href=#338 id=338 data-nosnippet>338</a>            <span class="kw">let </span>row_filter = <span class="self">Self</span>::get_row_filter(
+<a href=#339 id=339 data-nosnippet>339</a>                <span class="kw-2">&amp;</span>predicate,
+<a href=#340 id=340 data-nosnippet>340</a>                record_batch_stream_builder.parquet_schema(),
+<a href=#341 id=341 data-nosnippet>341</a>                <span class="kw-2">&amp;</span>iceberg_field_ids,
+<a href=#342 id=342 data-nosnippet>342</a>                <span class="kw-2">&amp;</span>field_id_map,
+<a href=#343 id=343 data-nosnippet>343</a>            )<span class="question-mark">?</span>;
+<a href=#344 id=344 data-nosnippet>344</a>            record_batch_stream_builder = record_batch_stream_builder.with_row_filter(row_filter);
+<a href=#345 id=345 data-nosnippet>345</a>
+<a href=#346 id=346 data-nosnippet>346</a>            <span class="kw">if </span>row_group_filtering_enabled {
+<a href=#347 id=347 data-nosnippet>347</a>                <span class="kw">let </span>predicate_filtered_row_groups = <span class="self">Self</span>::get_selected_row_group_indices(
+<a href=#348 id=348 data-nosnippet>348</a>                    <span class="kw-2">&amp;</span>predicate,
+<a href=#349 id=349 data-nosnippet>349</a>                    record_batch_stream_builder.metadata(),
+<a href=#350 id=350 data-nosnippet>350</a>                    <span class="kw-2">&amp;</span>field_id_map,
+<a href=#351 id=351 data-nosnippet>351</a>                    <span class="kw-2">&amp;</span>task.schema,
+<a href=#352 id=352 data-nosnippet>352</a>                )<span class="question-mark">?</span>;
+<a href=#353 id=353 data-nosnippet>353</a>
+<a href=#354 id=354 data-nosnippet>354</a>                <span class="comment">// Merge predicate-based filtering with byte range filtering (if present)
+<a href=#355 id=355 data-nosnippet>355</a>                // by taking the intersection of both filters
+<a href=#356 id=356 data-nosnippet>356</a>                </span>selected_row_group_indices = <span class="kw">match </span>selected_row_group_indices {
+<a href=#357 id=357 data-nosnippet>357</a>                    <span class="prelude-val">Some</span>(byte_range_filtered) =&gt; {
+<a href=#358 id=358 data-nosnippet>358</a>                        <span class="comment">// Keep only row groups that are in both filters
+<a href=#359 id=359 data-nosnippet>359</a>                        </span><span class="kw">let </span>intersection: Vec&lt;usize&gt; = byte_range_filtered
+<a href=#360 id=360 data-nosnippet>360</a>                            .into_iter()
+<a href=#361 id=361 data-nosnippet>361</a>                            .filter(|idx| predicate_filtered_row_groups.contains(idx))
+<a href=#362 id=362 data-nosnippet>362</a>                            .collect();
+<a href=#363 id=363 data-nosnippet>363</a>                        <span class="prelude-val">Some</span>(intersection)
+<a href=#364 id=364 data-nosnippet>364</a>                    }
+<a href=#365 id=365 data-nosnippet>365</a>                    <span class="prelude-val">None </span>=&gt; <span class="prelude-val">Some</span>(predicate_filtered_row_groups),
+<a href=#366 id=366 data-nosnippet>366</a>                };
+<a href=#367 id=367 data-nosnippet>367</a>            }
+<a href=#368 id=368 data-nosnippet>368</a>
+<a href=#369 id=369 data-nosnippet>369</a>            <span class="kw">if </span>row_selection_enabled {
+<a href=#370 id=370 data-nosnippet>370</a>                row_selection = <span class="prelude-val">Some</span>(<span class="self">Self</span>::get_row_selection_for_filter_predicate(
+<a href=#371 id=371 data-nosnippet>371</a>                    <span class="kw-2">&amp;</span>predicate,
+<a href=#372 id=372 data-nosnippet>372</a>                    record_batch_stream_builder.metadata(),
+<a href=#373 id=373 data-nosnippet>373</a>                    <span class="kw-2">&amp;</span>selected_row_group_indices,
+<a href=#374 id=374 data-nosnippet>374</a>                    <span class="kw-2">&amp;</span>field_id_map,
+<a href=#375 id=375 data-nosnippet>375</a>                    <span class="kw-2">&amp;</span>task.schema,
+<a href=#376 id=376 data-nosnippet>376</a>                )<span class="question-mark">?</span>);
+<a href=#377 id=377 data-nosnippet>377</a>            }
+<a href=#378 id=378 data-nosnippet>378</a>        }
+<a href=#379 id=379 data-nosnippet>379</a>
+<a href=#380 id=380 data-nosnippet>380</a>        <span class="kw">let </span>positional_delete_indexes = delete_filter.get_delete_vector(<span class="kw-2">&amp;</span>task);
+<a href=#381 id=381 data-nosnippet>381</a>
+<a href=#382 id=382 data-nosnippet>382</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(positional_delete_indexes) = positional_delete_indexes {
+<a href=#383 id=383 data-nosnippet>383</a>            <span class="kw">let </span>delete_row_selection = {
+<a href=#384 id=384 data-nosnippet>384</a>                <span class="kw">let </span>positional_delete_indexes = positional_delete_indexes.lock().unwrap();
+<a href=#385 id=385 data-nosnippet>385</a>
+<a href=#386 id=386 data-nosnippet>386</a>                <span class="self">Self</span>::build_deletes_row_selection(
+<a href=#387 id=387 data-nosnippet>387</a>                    record_batch_stream_builder.metadata().row_groups(),
+<a href=#388 id=388 data-nosnippet>388</a>                    <span class="kw-2">&amp;</span>selected_row_group_indices,
+<a href=#389 id=389 data-nosnippet>389</a>                    <span class="kw-2">&amp;</span>positional_delete_indexes,
+<a href=#390 id=390 data-nosnippet>390</a>                )
+<a href=#391 id=391 data-nosnippet>391</a>            }<span class="question-mark">?</span>;
+<a href=#392 id=392 data-nosnippet>392</a>
+<a href=#393 id=393 data-nosnippet>393</a>            <span class="comment">// merge the row selection from the delete files with the row selection
+<a href=#394 id=394 data-nosnippet>394</a>            // from the filter predicate, if there is one from the filter predicate
+<a href=#395 id=395 data-nosnippet>395</a>            </span>row_selection = <span class="kw">match </span>row_selection {
+<a href=#396 id=396 data-nosnippet>396</a>                <span class="prelude-val">None </span>=&gt; <span class="prelude-val">Some</span>(delete_row_selection),
+<a href=#397 id=397 data-nosnippet>397</a>                <span class="prelude-val">Some</span>(filter_row_selection) =&gt; {
+<a href=#398 id=398 data-nosnippet>398</a>                    <span class="prelude-val">Some</span>(filter_row_selection.intersection(<span class="kw-2">&amp;</span>delete_row_selection))
+<a href=#399 id=399 data-nosnippet>399</a>                }
+<a href=#400 id=400 data-nosnippet>400</a>            };
+<a href=#401 id=401 data-nosnippet>401</a>        }
+<a href=#402 id=402 data-nosnippet>402</a>
+<a href=#403 id=403 data-nosnippet>403</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(row_selection) = row_selection {
+<a href=#404 id=404 data-nosnippet>404</a>            record_batch_stream_builder =
+<a href=#405 id=405 data-nosnippet>405</a>                record_batch_stream_builder.with_row_selection(row_selection);
+<a href=#406 id=406 data-nosnippet>406</a>        }
+<a href=#407 id=407 data-nosnippet>407</a>
+<a href=#408 id=408 data-nosnippet>408</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_group_indices) = selected_row_group_indices {
+<a href=#409 id=409 data-nosnippet>409</a>            record_batch_stream_builder =
+<a href=#410 id=410 data-nosnippet>410</a>                record_batch_stream_builder.with_row_groups(selected_row_group_indices);
+<a href=#411 id=411 data-nosnippet>411</a>        }
+<a href=#412 id=412 data-nosnippet>412</a>
+<a href=#413 id=413 data-nosnippet>413</a>        <span class="comment">// Build the batch stream and send all the RecordBatches that it generates
+<a href=#414 id=414 data-nosnippet>414</a>        // to the requester.
+<a href=#415 id=415 data-nosnippet>415</a>        </span><span class="kw">let </span>record_batch_stream =
+<a href=#416 id=416 data-nosnippet>416</a>            record_batch_stream_builder
+<a href=#417 id=417 data-nosnippet>417</a>                .build()<span class="question-mark">?
+<a href=#418 id=418 data-nosnippet>418</a>                </span>.map(<span class="kw">move </span>|batch| <span class="kw">match </span>batch {
+<a href=#419 id=419 data-nosnippet>419</a>                    <span class="prelude-val">Ok</span>(batch) =&gt; record_batch_transformer.process_record_batch(batch),
+<a href=#420 id=420 data-nosnippet>420</a>                    <span class="prelude-val">Err</span>(err) =&gt; <span class="prelude-val">Err</span>(err.into()),
+<a href=#421 id=421 data-nosnippet>421</a>                });
 <a href=#422 id=422 data-nosnippet>422</a>
-<a href=#423 id=423 data-nosnippet>423</a>        <span class="kw">for </span>(idx, row_group_metadata) <span class="kw">in </span>row_group_metadata_list.iter().enumerate() {
-<a href=#424 id=424 data-nosnippet>424</a>            <span class="kw">let </span>row_group_num_rows = row_group_metadata.num_rows() <span class="kw">as </span>u64;
-<a href=#425 id=425 data-nosnippet>425</a>            <span class="kw">let </span>next_row_group_base_idx = current_row_group_base_idx + row_group_num_rows;
-<a href=#426 id=426 data-nosnippet>426</a>
-<a href=#427 id=427 data-nosnippet>427</a>            <span class="comment">// if row group selection is enabled,
-<a href=#428 id=428 data-nosnippet>428</a>            </span><span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_groups) = selected_row_groups {
-<a href=#429 id=429 data-nosnippet>429</a>                <span class="comment">// if we've consumed all the selected row groups, we're done
-<a href=#430 id=430 data-nosnippet>430</a>                </span><span class="kw">if </span>selected_row_groups_idx == selected_row_groups.len() {
-<a href=#431 id=431 data-nosnippet>431</a>                    <span class="kw">break</span>;
-<a href=#432 id=432 data-nosnippet>432</a>                }
-<a href=#433 id=433 data-nosnippet>433</a>
-<a href=#434 id=434 data-nosnippet>434</a>                <span class="kw">if </span>idx == selected_row_groups[selected_row_groups_idx] {
-<a href=#435 id=435 data-nosnippet>435</a>                    <span class="comment">// we're in a selected row group. Increment selected_row_groups_idx
-<a href=#436 id=436 data-nosnippet>436</a>                    // so that next time around the for loop we're looking for the next
-<a href=#437 id=437 data-nosnippet>437</a>                    // selected row group
-<a href=#438 id=438 data-nosnippet>438</a>                    </span>selected_row_groups_idx += <span class="number">1</span>;
-<a href=#439 id=439 data-nosnippet>439</a>                } <span class="kw">else </span>{
-<a href=#440 id=440 data-nosnippet>440</a>                    <span class="comment">// Advance iterator past all deletes in the skipped row group.
-<a href=#441 id=441 data-nosnippet>441</a>                    // advance_to() positions the iterator to the first delete &gt;= next_row_group_base_idx.
-<a href=#442 id=442 data-nosnippet>442</a>                    // However, if our cached next_deleted_row_idx_opt is in the skipped range,
-<a href=#443 id=443 data-nosnippet>443</a>                    // we need to call next() to update the cache with the newly positioned value.
-<a href=#444 id=444 data-nosnippet>444</a>                    </span>delete_vector_iter.advance_to(next_row_group_base_idx);
-<a href=#445 id=445 data-nosnippet>445</a>                    <span class="comment">// Only update the cache if the cached value is stale (in the skipped range)
-<a href=#446 id=446 data-nosnippet>446</a>                    </span><span class="kw">if let </span><span class="prelude-val">Some</span>(cached_idx) = next_deleted_row_idx_opt {
-<a href=#447 id=447 data-nosnippet>447</a>                        <span class="kw">if </span>cached_idx &lt; next_row_group_base_idx {
-<a href=#448 id=448 data-nosnippet>448</a>                            next_deleted_row_idx_opt = delete_vector_iter.next();
-<a href=#449 id=449 data-nosnippet>449</a>                        }
-<a href=#450 id=450 data-nosnippet>450</a>                    }
-<a href=#451 id=451 data-nosnippet>451</a>
-<a href=#452 id=452 data-nosnippet>452</a>                    <span class="comment">// still increment the current page base index but then skip to the next row group
-<a href=#453 id=453 data-nosnippet>453</a>                    // in the file
-<a href=#454 id=454 data-nosnippet>454</a>                    </span>current_row_group_base_idx += row_group_num_rows;
-<a href=#455 id=455 data-nosnippet>455</a>                    <span class="kw">continue</span>;
-<a href=#456 id=456 data-nosnippet>456</a>                }
-<a href=#457 id=457 data-nosnippet>457</a>            }
-<a href=#458 id=458 data-nosnippet>458</a>
-<a href=#459 id=459 data-nosnippet>459</a>            <span class="kw">let </span><span class="kw-2">mut </span>next_deleted_row_idx = <span class="kw">match </span>next_deleted_row_idx_opt {
-<a href=#460 id=460 data-nosnippet>460</a>                <span class="prelude-val">Some</span>(next_deleted_row_idx) =&gt; {
-<a href=#461 id=461 data-nosnippet>461</a>                    <span class="comment">// if the index of the next deleted row is beyond this row group, add a selection for
-<a href=#462 id=462 data-nosnippet>462</a>                    // the remainder of this row group and skip to the next row group
-<a href=#463 id=463 data-nosnippet>463</a>                    </span><span class="kw">if </span>next_deleted_row_idx &gt;= next_row_group_base_idx {
-<a href=#464 id=464 data-nosnippet>464</a>                        results.push(RowSelector::select(row_group_num_rows <span class="kw">as </span>usize));
-<a href=#465 id=465 data-nosnippet>465</a>                        current_row_group_base_idx += row_group_num_rows;
-<a href=#466 id=466 data-nosnippet>466</a>                        <span class="kw">continue</span>;
-<a href=#467 id=467 data-nosnippet>467</a>                    }
+<a href=#423 id=423 data-nosnippet>423</a>        <span class="prelude-val">Ok</span>(Box::pin(record_batch_stream) <span class="kw">as </span>ArrowRecordBatchStream)
+<a href=#424 id=424 data-nosnippet>424</a>    }
+<a href=#425 id=425 data-nosnippet>425</a>
+<a href=#426 id=426 data-nosnippet>426</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">async fn </span>create_parquet_record_batch_stream_builder(
+<a href=#427 id=427 data-nosnippet>427</a>        data_file_path: <span class="kw-2">&amp;</span>str,
+<a href=#428 id=428 data-nosnippet>428</a>        file_io: FileIO,
+<a href=#429 id=429 data-nosnippet>429</a>        should_load_page_index: bool,
+<a href=#430 id=430 data-nosnippet>430</a>        arrow_reader_options: <span class="prelude-ty">Option</span>&lt;ArrowReaderOptions&gt;,
+<a href=#431 id=431 data-nosnippet>431</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ParquetRecordBatchStreamBuilder&lt;ArrowFileReader&lt;<span class="kw">impl </span>FileRead + Sized&gt;&gt;&gt; {
+<a href=#432 id=432 data-nosnippet>432</a>        <span class="comment">// Get the metadata for the Parquet file we need to read and build
+<a href=#433 id=433 data-nosnippet>433</a>        // a reader for the data within
+<a href=#434 id=434 data-nosnippet>434</a>        </span><span class="kw">let </span>parquet_file = file_io.new_input(data_file_path)<span class="question-mark">?</span>;
+<a href=#435 id=435 data-nosnippet>435</a>        <span class="kw">let </span>(parquet_metadata, parquet_reader) =
+<a href=#436 id=436 data-nosnippet>436</a>            <span class="macro">try_join!</span>(parquet_file.metadata(), parquet_file.reader())<span class="question-mark">?</span>;
+<a href=#437 id=437 data-nosnippet>437</a>        <span class="kw">let </span>parquet_file_reader = ArrowFileReader::new(parquet_metadata, parquet_reader)
+<a href=#438 id=438 data-nosnippet>438</a>            .with_preload_column_index(<span class="bool-val">true</span>)
+<a href=#439 id=439 data-nosnippet>439</a>            .with_preload_offset_index(<span class="bool-val">true</span>)
+<a href=#440 id=440 data-nosnippet>440</a>            .with_preload_page_index(should_load_page_index);
+<a href=#441 id=441 data-nosnippet>441</a>
+<a href=#442 id=442 data-nosnippet>442</a>        <span class="comment">// Create the record batch stream builder, which wraps the parquet file reader
+<a href=#443 id=443 data-nosnippet>443</a>        </span><span class="kw">let </span>options = arrow_reader_options.unwrap_or_default();
+<a href=#444 id=444 data-nosnippet>444</a>        <span class="kw">let </span>record_batch_stream_builder =
+<a href=#445 id=445 data-nosnippet>445</a>            ParquetRecordBatchStreamBuilder::new_with_options(parquet_file_reader, options).<span class="kw">await</span><span class="question-mark">?</span>;
+<a href=#446 id=446 data-nosnippet>446</a>        <span class="prelude-val">Ok</span>(record_batch_stream_builder)
+<a href=#447 id=447 data-nosnippet>447</a>    }
+<a href=#448 id=448 data-nosnippet>448</a>
+<a href=#449 id=449 data-nosnippet>449</a>    <span class="doccomment">/// computes a `RowSelection` from positional delete indices.
+<a href=#450 id=450 data-nosnippet>450</a>    ///
+<a href=#451 id=451 data-nosnippet>451</a>    /// Using the Parquet page index, we build a `RowSelection` that rejects rows that are indicated
+<a href=#452 id=452 data-nosnippet>452</a>    /// as having been deleted by a positional delete, taking into account any row groups that have
+<a href=#453 id=453 data-nosnippet>453</a>    /// been skipped entirely by the filter predicate
+<a href=#454 id=454 data-nosnippet>454</a>    </span><span class="kw">fn </span>build_deletes_row_selection(
+<a href=#455 id=455 data-nosnippet>455</a>        row_group_metadata_list: <span class="kw-2">&amp;</span>[RowGroupMetaData],
+<a href=#456 id=456 data-nosnippet>456</a>        selected_row_groups: <span class="kw-2">&amp;</span><span class="prelude-ty">Option</span>&lt;Vec&lt;usize&gt;&gt;,
+<a href=#457 id=457 data-nosnippet>457</a>        positional_deletes: <span class="kw-2">&amp;</span>DeleteVector,
+<a href=#458 id=458 data-nosnippet>458</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;RowSelection&gt; {
+<a href=#459 id=459 data-nosnippet>459</a>        <span class="kw">let </span><span class="kw-2">mut </span>results: Vec&lt;RowSelector&gt; = Vec::new();
+<a href=#460 id=460 data-nosnippet>460</a>        <span class="kw">let </span><span class="kw-2">mut </span>selected_row_groups_idx = <span class="number">0</span>;
+<a href=#461 id=461 data-nosnippet>461</a>        <span class="kw">let </span><span class="kw-2">mut </span>current_row_group_base_idx: u64 = <span class="number">0</span>;
+<a href=#462 id=462 data-nosnippet>462</a>        <span class="kw">let </span><span class="kw-2">mut </span>delete_vector_iter = positional_deletes.iter();
+<a href=#463 id=463 data-nosnippet>463</a>        <span class="kw">let </span><span class="kw-2">mut </span>next_deleted_row_idx_opt = delete_vector_iter.next();
+<a href=#464 id=464 data-nosnippet>464</a>
+<a href=#465 id=465 data-nosnippet>465</a>        <span class="kw">for </span>(idx, row_group_metadata) <span class="kw">in </span>row_group_metadata_list.iter().enumerate() {
+<a href=#466 id=466 data-nosnippet>466</a>            <span class="kw">let </span>row_group_num_rows = row_group_metadata.num_rows() <span class="kw">as </span>u64;
+<a href=#467 id=467 data-nosnippet>467</a>            <span class="kw">let </span>next_row_group_base_idx = current_row_group_base_idx + row_group_num_rows;
 <a href=#468 id=468 data-nosnippet>468</a>
-<a href=#469 id=469 data-nosnippet>469</a>                    next_deleted_row_idx
-<a href=#470 id=470 data-nosnippet>470</a>                }
-<a href=#471 id=471 data-nosnippet>471</a>
-<a href=#472 id=472 data-nosnippet>472</a>                <span class="comment">// If there are no more pos deletes, add a selector for the entirety of this row group.
-<a href=#473 id=473 data-nosnippet>473</a>                </span><span class="kw">_ </span>=&gt; {
-<a href=#474 id=474 data-nosnippet>474</a>                    results.push(RowSelector::select(row_group_num_rows <span class="kw">as </span>usize));
-<a href=#475 id=475 data-nosnippet>475</a>                    current_row_group_base_idx += row_group_num_rows;
-<a href=#476 id=476 data-nosnippet>476</a>                    <span class="kw">continue</span>;
-<a href=#477 id=477 data-nosnippet>477</a>                }
-<a href=#478 id=478 data-nosnippet>478</a>            };
-<a href=#479 id=479 data-nosnippet>479</a>
-<a href=#480 id=480 data-nosnippet>480</a>            <span class="kw">let </span><span class="kw-2">mut </span>current_idx = current_row_group_base_idx;
-<a href=#481 id=481 data-nosnippet>481</a>            <span class="lifetime">'chunks</span>: <span class="kw">while </span>next_deleted_row_idx &lt; next_row_group_base_idx {
-<a href=#482 id=482 data-nosnippet>482</a>                <span class="comment">// `select` all rows that precede the next delete index
-<a href=#483 id=483 data-nosnippet>483</a>                </span><span class="kw">if </span>current_idx &lt; next_deleted_row_idx {
-<a href=#484 id=484 data-nosnippet>484</a>                    <span class="kw">let </span>run_length = next_deleted_row_idx - current_idx;
-<a href=#485 id=485 data-nosnippet>485</a>                    results.push(RowSelector::select(run_length <span class="kw">as </span>usize));
-<a href=#486 id=486 data-nosnippet>486</a>                    current_idx += run_length;
-<a href=#487 id=487 data-nosnippet>487</a>                }
-<a href=#488 id=488 data-nosnippet>488</a>
-<a href=#489 id=489 data-nosnippet>489</a>                <span class="comment">// `skip` all consecutive deleted rows in the current row group
-<a href=#490 id=490 data-nosnippet>490</a>                </span><span class="kw">let </span><span class="kw-2">mut </span>run_length = <span class="number">0</span>;
-<a href=#491 id=491 data-nosnippet>491</a>                <span class="kw">while </span>next_deleted_row_idx == current_idx
-<a href=#492 id=492 data-nosnippet>492</a>                    &amp;&amp; next_deleted_row_idx &lt; next_row_group_base_idx
-<a href=#493 id=493 data-nosnippet>493</a>                {
-<a href=#494 id=494 data-nosnippet>494</a>                    run_length += <span class="number">1</span>;
-<a href=#495 id=495 data-nosnippet>495</a>                    current_idx += <span class="number">1</span>;
-<a href=#496 id=496 data-nosnippet>496</a>
-<a href=#497 id=497 data-nosnippet>497</a>                    next_deleted_row_idx_opt = delete_vector_iter.next();
-<a href=#498 id=498 data-nosnippet>498</a>                    next_deleted_row_idx = <span class="kw">match </span>next_deleted_row_idx_opt {
-<a href=#499 id=499 data-nosnippet>499</a>                        <span class="prelude-val">Some</span>(next_deleted_row_idx) =&gt; next_deleted_row_idx,
-<a href=#500 id=500 data-nosnippet>500</a>                        <span class="kw">_ </span>=&gt; {
-<a href=#501 id=501 data-nosnippet>501</a>                            <span class="comment">// We've processed the final positional delete.
-<a href=#502 id=502 data-nosnippet>502</a>                            // Conclude the skip and then break so that we select the remaining
-<a href=#503 id=503 data-nosnippet>503</a>                            // rows in the row group and move on to the next row group
-<a href=#504 id=504 data-nosnippet>504</a>                            </span>results.push(RowSelector::skip(run_length));
-<a href=#505 id=505 data-nosnippet>505</a>                            <span class="kw">break </span><span class="lifetime">'chunks</span>;
-<a href=#506 id=506 data-nosnippet>506</a>                        }
-<a href=#507 id=507 data-nosnippet>507</a>                    };
-<a href=#508 id=508 data-nosnippet>508</a>                }
-<a href=#509 id=509 data-nosnippet>509</a>                <span class="kw">if </span>run_length &gt; <span class="number">0 </span>{
-<a href=#510 id=510 data-nosnippet>510</a>                    results.push(RowSelector::skip(run_length));
-<a href=#511 id=511 data-nosnippet>511</a>                }
-<a href=#512 id=512 data-nosnippet>512</a>            }
+<a href=#469 id=469 data-nosnippet>469</a>            <span class="comment">// if row group selection is enabled,
+<a href=#470 id=470 data-nosnippet>470</a>            </span><span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_groups) = selected_row_groups {
+<a href=#471 id=471 data-nosnippet>471</a>                <span class="comment">// if we've consumed all the selected row groups, we're done
+<a href=#472 id=472 data-nosnippet>472</a>                </span><span class="kw">if </span>selected_row_groups_idx == selected_row_groups.len() {
+<a href=#473 id=473 data-nosnippet>473</a>                    <span class="kw">break</span>;
+<a href=#474 id=474 data-nosnippet>474</a>                }
+<a href=#475 id=475 data-nosnippet>475</a>
+<a href=#476 id=476 data-nosnippet>476</a>                <span class="kw">if </span>idx == selected_row_groups[selected_row_groups_idx] {
+<a href=#477 id=477 data-nosnippet>477</a>                    <span class="comment">// we're in a selected row group. Increment selected_row_groups_idx
+<a href=#478 id=478 data-nosnippet>478</a>                    // so that next time around the for loop we're looking for the next
+<a href=#479 id=479 data-nosnippet>479</a>                    // selected row group
+<a href=#480 id=480 data-nosnippet>480</a>                    </span>selected_row_groups_idx += <span class="number">1</span>;
+<a href=#481 id=481 data-nosnippet>481</a>                } <span class="kw">else </span>{
+<a href=#482 id=482 data-nosnippet>482</a>                    <span class="comment">// Advance iterator past all deletes in the skipped row group.
+<a href=#483 id=483 data-nosnippet>483</a>                    // advance_to() positions the iterator to the first delete &gt;= next_row_group_base_idx.
+<a href=#484 id=484 data-nosnippet>484</a>                    // However, if our cached next_deleted_row_idx_opt is in the skipped range,
+<a href=#485 id=485 data-nosnippet>485</a>                    // we need to call next() to update the cache with the newly positioned value.
+<a href=#486 id=486 data-nosnippet>486</a>                    </span>delete_vector_iter.advance_to(next_row_group_base_idx);
+<a href=#487 id=487 data-nosnippet>487</a>                    <span class="comment">// Only update the cache if the cached value is stale (in the skipped range)
+<a href=#488 id=488 data-nosnippet>488</a>                    </span><span class="kw">if let </span><span class="prelude-val">Some</span>(cached_idx) = next_deleted_row_idx_opt {
+<a href=#489 id=489 data-nosnippet>489</a>                        <span class="kw">if </span>cached_idx &lt; next_row_group_base_idx {
+<a href=#490 id=490 data-nosnippet>490</a>                            next_deleted_row_idx_opt = delete_vector_iter.next();
+<a href=#491 id=491 data-nosnippet>491</a>                        }
+<a href=#492 id=492 data-nosnippet>492</a>                    }
+<a href=#493 id=493 data-nosnippet>493</a>
+<a href=#494 id=494 data-nosnippet>494</a>                    <span class="comment">// still increment the current page base index but then skip to the next row group
+<a href=#495 id=495 data-nosnippet>495</a>                    // in the file
+<a href=#496 id=496 data-nosnippet>496</a>                    </span>current_row_group_base_idx += row_group_num_rows;
+<a href=#497 id=497 data-nosnippet>497</a>                    <span class="kw">continue</span>;
+<a href=#498 id=498 data-nosnippet>498</a>                }
+<a href=#499 id=499 data-nosnippet>499</a>            }
+<a href=#500 id=500 data-nosnippet>500</a>
+<a href=#501 id=501 data-nosnippet>501</a>            <span class="kw">let </span><span class="kw-2">mut </span>next_deleted_row_idx = <span class="kw">match </span>next_deleted_row_idx_opt {
+<a href=#502 id=502 data-nosnippet>502</a>                <span class="prelude-val">Some</span>(next_deleted_row_idx) =&gt; {
+<a href=#503 id=503 data-nosnippet>503</a>                    <span class="comment">// if the index of the next deleted row is beyond this row group, add a selection for
+<a href=#504 id=504 data-nosnippet>504</a>                    // the remainder of this row group and skip to the next row group
+<a href=#505 id=505 data-nosnippet>505</a>                    </span><span class="kw">if </span>next_deleted_row_idx &gt;= next_row_group_base_idx {
+<a href=#506 id=506 data-nosnippet>506</a>                        results.push(RowSelector::select(row_group_num_rows <span class="kw">as </span>usize));
+<a href=#507 id=507 data-nosnippet>507</a>                        current_row_group_base_idx += row_group_num_rows;
+<a href=#508 id=508 data-nosnippet>508</a>                        <span class="kw">continue</span>;
+<a href=#509 id=509 data-nosnippet>509</a>                    }
+<a href=#510 id=510 data-nosnippet>510</a>
+<a href=#511 id=511 data-nosnippet>511</a>                    next_deleted_row_idx
+<a href=#512 id=512 data-nosnippet>512</a>                }
 <a href=#513 id=513 data-nosnippet>513</a>
-<a href=#514 id=514 data-nosnippet>514</a>            <span class="kw">if </span>current_idx &lt; next_row_group_base_idx {
-<a href=#515 id=515 data-nosnippet>515</a>                results.push(RowSelector::select(
-<a href=#516 id=516 data-nosnippet>516</a>                    (next_row_group_base_idx - current_idx) <span class="kw">as </span>usize,
-<a href=#517 id=517 data-nosnippet>517</a>                ));
-<a href=#518 id=518 data-nosnippet>518</a>            }
-<a href=#519 id=519 data-nosnippet>519</a>
-<a href=#520 id=520 data-nosnippet>520</a>            current_row_group_base_idx += row_group_num_rows;
-<a href=#521 id=521 data-nosnippet>521</a>        }
-<a href=#522 id=522 data-nosnippet>522</a>
-<a href=#523 id=523 data-nosnippet>523</a>        <span class="prelude-val">Ok</span>(results.into())
-<a href=#524 id=524 data-nosnippet>524</a>    }
-<a href=#525 id=525 data-nosnippet>525</a>
-<a href=#526 id=526 data-nosnippet>526</a>    <span class="kw">fn </span>build_field_id_set_and_map(
-<a href=#527 id=527 data-nosnippet>527</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
-<a href=#528 id=528 data-nosnippet>528</a>        predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#529 id=529 data-nosnippet>529</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;(HashSet&lt;i32&gt;, HashMap&lt;i32, usize&gt;)&gt; {
-<a href=#530 id=530 data-nosnippet>530</a>        <span class="comment">// Collects all Iceberg field IDs referenced in the filter predicate
-<a href=#531 id=531 data-nosnippet>531</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>collector = CollectFieldIdVisitor {
-<a href=#532 id=532 data-nosnippet>532</a>            field_ids: HashSet::default(),
-<a href=#533 id=533 data-nosnippet>533</a>        };
-<a href=#534 id=534 data-nosnippet>534</a>        visit(<span class="kw-2">&amp;mut </span>collector, predicate)<span class="question-mark">?</span>;
-<a href=#535 id=535 data-nosnippet>535</a>
-<a href=#536 id=536 data-nosnippet>536</a>        <span class="kw">let </span>iceberg_field_ids = collector.field_ids();
-<a href=#537 id=537 data-nosnippet>537</a>
-<a href=#538 id=538 data-nosnippet>538</a>        <span class="comment">// Without embedded field IDs, we fall back to position-based mapping for compatibility
-<a href=#539 id=539 data-nosnippet>539</a>        </span><span class="kw">let </span>field_id_map = <span class="kw">match </span>build_field_id_map(parquet_schema)<span class="question-mark">? </span>{
-<a href=#540 id=540 data-nosnippet>540</a>            <span class="prelude-val">Some</span>(map) =&gt; map,
-<a href=#541 id=541 data-nosnippet>541</a>            <span class="prelude-val">None </span>=&gt; build_fallback_field_id_map(parquet_schema),
-<a href=#542 id=542 data-nosnippet>542</a>        };
-<a href=#543 id=543 data-nosnippet>543</a>
-<a href=#544 id=544 data-nosnippet>544</a>        <span class="prelude-val">Ok</span>((iceberg_field_ids, field_id_map))
-<a href=#545 id=545 data-nosnippet>545</a>    }
-<a href=#546 id=546 data-nosnippet>546</a>
-<a href=#547 id=547 data-nosnippet>547</a>    <span class="doccomment">/// Recursively extract leaf field IDs because Parquet projection works at the leaf column level.
-<a href=#548 id=548 data-nosnippet>548</a>    /// Nested types (struct/list/map) are flattened in Parquet's columnar format.
-<a href=#549 id=549 data-nosnippet>549</a>    </span><span class="kw">fn </span>include_leaf_field_id(field: <span class="kw-2">&amp;</span>NestedField, field_ids: <span class="kw-2">&amp;mut </span>Vec&lt;i32&gt;) {
-<a href=#550 id=550 data-nosnippet>550</a>        <span class="kw">match </span>field.field_type.as_ref() {
-<a href=#551 id=551 data-nosnippet>551</a>            Type::Primitive(<span class="kw">_</span>) =&gt; {
-<a href=#552 id=552 data-nosnippet>552</a>                field_ids.push(field.id);
-<a href=#553 id=553 data-nosnippet>553</a>            }
-<a href=#554 id=554 data-nosnippet>554</a>            Type::Struct(struct_type) =&gt; {
-<a href=#555 id=555 data-nosnippet>555</a>                <span class="kw">for </span>nested_field <span class="kw">in </span>struct_type.fields() {
-<a href=#556 id=556 data-nosnippet>556</a>                    <span class="self">Self</span>::include_leaf_field_id(nested_field, field_ids);
-<a href=#557 id=557 data-nosnippet>557</a>                }
-<a href=#558 id=558 data-nosnippet>558</a>            }
-<a href=#559 id=559 data-nosnippet>559</a>            Type::List(list_type) =&gt; {
-<a href=#560 id=560 data-nosnippet>560</a>                <span class="self">Self</span>::include_leaf_field_id(<span class="kw-2">&amp;</span>list_type.element_field, field_ids);
-<a href=#561 id=561 data-nosnippet>561</a>            }
-<a href=#562 id=562 data-nosnippet>562</a>            Type::Map(map_type) =&gt; {
-<a href=#563 id=563 data-nosnippet>563</a>                <span class="self">Self</span>::include_leaf_field_id(<span class="kw-2">&amp;</span>map_type.key_field, field_ids);
-<a href=#564 id=564 data-nosnippet>564</a>                <span class="self">Self</span>::include_leaf_field_id(<span class="kw-2">&amp;</span>map_type.value_field, field_ids);
-<a href=#565 id=565 data-nosnippet>565</a>            }
-<a href=#566 id=566 data-nosnippet>566</a>        }
-<a href=#567 id=567 data-nosnippet>567</a>    }
-<a href=#568 id=568 data-nosnippet>568</a>
-<a href=#569 id=569 data-nosnippet>569</a>    <span class="kw">fn </span>get_arrow_projection_mask(
-<a href=#570 id=570 data-nosnippet>570</a>        field_ids: <span class="kw-2">&amp;</span>[i32],
-<a href=#571 id=571 data-nosnippet>571</a>        iceberg_schema_of_task: <span class="kw-2">&amp;</span>Schema,
-<a href=#572 id=572 data-nosnippet>572</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
-<a href=#573 id=573 data-nosnippet>573</a>        arrow_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
-<a href=#574 id=574 data-nosnippet>574</a>        use_fallback: bool, <span class="comment">// Whether file lacks embedded field IDs (e.g., migrated from Hive/Spark)
-<a href=#575 id=575 data-nosnippet>575</a>    </span>) -&gt; <span class="prelude-ty">Result</span>&lt;ProjectionMask&gt; {
-<a href=#576 id=576 data-nosnippet>576</a>        <span class="kw">fn </span>type_promotion_is_valid(
-<a href=#577 id=577 data-nosnippet>577</a>            file_type: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PrimitiveType&gt;,
-<a href=#578 id=578 data-nosnippet>578</a>            projected_type: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PrimitiveType&gt;,
-<a href=#579 id=579 data-nosnippet>579</a>        ) -&gt; bool {
-<a href=#580 id=580 data-nosnippet>580</a>            <span class="kw">match </span>(file_type, projected_type) {
-<a href=#581 id=581 data-nosnippet>581</a>                (<span class="prelude-val">Some</span>(lhs), <span class="prelude-val">Some</span>(rhs)) <span class="kw">if </span>lhs == rhs =&gt; <span class="bool-val">true</span>,
-<a href=#582 id=582 data-nosnippet>582</a>                (<span class="prelude-val">Some</span>(PrimitiveType::Int), <span class="prelude-val">Some</span>(PrimitiveType::Long)) =&gt; <span class="bool-val">true</span>,
-<a href=#583 id=583 data-nosnippet>583</a>                (<span class="prelude-val">Some</span>(PrimitiveType::Float), <span class="prelude-val">Some</span>(PrimitiveType::Double)) =&gt; <span class="bool-val">true</span>,
-<a href=#584 id=584 data-nosnippet>584</a>                (
-<a href=#585 id=585 data-nosnippet>585</a>                    <span class="prelude-val">Some</span>(PrimitiveType::Decimal {
-<a href=#586 id=586 data-nosnippet>586</a>                        precision: file_precision,
-<a href=#587 id=587 data-nosnippet>587</a>                        scale: file_scale,
-<a href=#588 id=588 data-nosnippet>588</a>                    }),
-<a href=#589 id=589 data-nosnippet>589</a>                    <span class="prelude-val">Some</span>(PrimitiveType::Decimal {
-<a href=#590 id=590 data-nosnippet>590</a>                        precision: requested_precision,
-<a href=#591 id=591 data-nosnippet>591</a>                        scale: requested_scale,
-<a href=#592 id=592 data-nosnippet>592</a>                    }),
-<a href=#593 id=593 data-nosnippet>593</a>                ) <span class="kw">if </span>requested_precision &gt;= file_precision &amp;&amp; file_scale == requested_scale =&gt; <span class="bool-val">true</span>,
-<a href=#594 id=594 data-nosnippet>594</a>                <span class="comment">// Uuid will be store as Fixed(16) in parquet file, so the read back type will be Fixed(16).
-<a href=#595 id=595 data-nosnippet>595</a>                </span>(<span class="prelude-val">Some</span>(PrimitiveType::Fixed(<span class="number">16</span>)), <span class="prelude-val">Some</span>(PrimitiveType::Uuid)) =&gt; <span class="bool-val">true</span>,
-<a href=#596 id=596 data-nosnippet>596</a>                <span class="kw">_ </span>=&gt; <span class="bool-val">false</span>,
-<a href=#597 id=597 data-nosnippet>597</a>            }
-<a href=#598 id=598 data-nosnippet>598</a>        }
-<a href=#599 id=599 data-nosnippet>599</a>
-<a href=#600 id=600 data-nosnippet>600</a>        <span class="kw">if </span>field_ids.is_empty() {
-<a href=#601 id=601 data-nosnippet>601</a>            <span class="kw">return </span><span class="prelude-val">Ok</span>(ProjectionMask::all());
-<a href=#602 id=602 data-nosnippet>602</a>        }
-<a href=#603 id=603 data-nosnippet>603</a>
-<a href=#604 id=604 data-nosnippet>604</a>        <span class="kw">if </span>use_fallback {
-<a href=#605 id=605 data-nosnippet>605</a>            <span class="comment">// Position-based projection necessary because file lacks embedded field IDs
-<a href=#606 id=606 data-nosnippet>606</a>            </span><span class="self">Self</span>::get_arrow_projection_mask_fallback(field_ids, parquet_schema)
-<a href=#607 id=607 data-nosnippet>607</a>        } <span class="kw">else </span>{
-<a href=#608 id=608 data-nosnippet>608</a>            <span class="comment">// Field-ID-based projection using embedded field IDs from Parquet metadata
-<a href=#609 id=609 data-nosnippet>609</a>
-<a href=#610 id=610 data-nosnippet>610</a>            // Parquet's columnar format requires leaf-level (not top-level struct/list/map) projection
-<a href=#611 id=611 data-nosnippet>611</a>            </span><span class="kw">let </span><span class="kw-2">mut </span>leaf_field_ids = <span class="macro">vec!</span>[];
-<a href=#612 id=612 data-nosnippet>612</a>            <span class="kw">for </span>field_id <span class="kw">in </span>field_ids {
-<a href=#613 id=613 data-nosnippet>613</a>                <span class="kw">let </span>field = iceberg_schema_of_task.field_by_id(<span class="kw-2">*</span>field_id);
-<a href=#614 id=614 data-nosnippet>614</a>                <span class="kw">if let </span><span class="prelude-val">Some</span>(field) = field {
-<a href=#615 id=615 data-nosnippet>615</a>                    <span class="self">Self</span>::include_leaf_field_id(field, <span class="kw-2">&amp;mut </span>leaf_field_ids);
-<a href=#616 id=616 data-nosnippet>616</a>                }
-<a href=#617 id=617 data-nosnippet>617</a>            }
-<a href=#618 id=618 data-nosnippet>618</a>
-<a href=#619 id=619 data-nosnippet>619</a>            <span class="self">Self</span>::get_arrow_projection_mask_with_field_ids(
-<a href=#620 id=620 data-nosnippet>620</a>                <span class="kw-2">&amp;</span>leaf_field_ids,
-<a href=#621 id=621 data-nosnippet>621</a>                iceberg_schema_of_task,
-<a href=#622 id=622 data-nosnippet>622</a>                parquet_schema,
-<a href=#623 id=623 data-nosnippet>623</a>                arrow_schema,
-<a href=#624 id=624 data-nosnippet>624</a>                type_promotion_is_valid,
-<a href=#625 id=625 data-nosnippet>625</a>            )
-<a href=#626 id=626 data-nosnippet>626</a>        }
-<a href=#627 id=627 data-nosnippet>627</a>    }
-<a href=#628 id=628 data-nosnippet>628</a>
-<a href=#629 id=629 data-nosnippet>629</a>    <span class="doccomment">/// Standard projection using embedded field IDs from Parquet metadata.
-<a href=#630 id=630 data-nosnippet>630</a>    /// For iceberg-java compatibility with ParquetSchemaUtil.pruneColumns().
-<a href=#631 id=631 data-nosnippet>631</a>    </span><span class="kw">fn </span>get_arrow_projection_mask_with_field_ids(
-<a href=#632 id=632 data-nosnippet>632</a>        leaf_field_ids: <span class="kw-2">&amp;</span>[i32],
-<a href=#633 id=633 data-nosnippet>633</a>        iceberg_schema_of_task: <span class="kw-2">&amp;</span>Schema,
-<a href=#634 id=634 data-nosnippet>634</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
-<a href=#635 id=635 data-nosnippet>635</a>        arrow_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
-<a href=#636 id=636 data-nosnippet>636</a>        type_promotion_is_valid: <span class="kw">fn</span>(<span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PrimitiveType&gt;, <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PrimitiveType&gt;) -&gt; bool,
-<a href=#637 id=637 data-nosnippet>637</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ProjectionMask&gt; {
-<a href=#638 id=638 data-nosnippet>638</a>        <span class="kw">let </span><span class="kw-2">mut </span>column_map = HashMap::new();
-<a href=#639 id=639 data-nosnippet>639</a>        <span class="kw">let </span>fields = arrow_schema.fields();
-<a href=#640 id=640 data-nosnippet>640</a>
-<a href=#641 id=641 data-nosnippet>641</a>        <span class="comment">// Pre-project only the fields that have been selected, possibly avoiding converting
-<a href=#642 id=642 data-nosnippet>642</a>        // some Arrow types that are not yet supported.
-<a href=#643 id=643 data-nosnippet>643</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>projected_fields: HashMap&lt;FieldRef, i32&gt; = HashMap::new();
-<a href=#644 id=644 data-nosnippet>644</a>        <span class="kw">let </span>projected_arrow_schema = ArrowSchema::new_with_metadata(
-<a href=#645 id=645 data-nosnippet>645</a>            fields.filter_leaves(|<span class="kw">_</span>, f| {
-<a href=#646 id=646 data-nosnippet>646</a>                f.metadata()
-<a href=#647 id=647 data-nosnippet>647</a>                    .get(PARQUET_FIELD_ID_META_KEY)
-<a href=#648 id=648 data-nosnippet>648</a>                    .and_then(|field_id| i32::from_str(field_id).ok())
-<a href=#649 id=649 data-nosnippet>649</a>                    .is_some_and(|field_id| {
-<a href=#650 id=650 data-nosnippet>650</a>                        projected_fields.insert((<span class="kw-2">*</span>f).clone(), field_id);
-<a href=#651 id=651 data-nosnippet>651</a>                        leaf_field_ids.contains(<span class="kw-2">&amp;</span>field_id)
-<a href=#652 id=652 data-nosnippet>652</a>                    })
-<a href=#653 id=653 data-nosnippet>653</a>            }),
-<a href=#654 id=654 data-nosnippet>654</a>            arrow_schema.metadata().clone(),
-<a href=#655 id=655 data-nosnippet>655</a>        );
-<a href=#656 id=656 data-nosnippet>656</a>        <span class="kw">let </span>iceberg_schema = arrow_schema_to_schema(<span class="kw-2">&amp;</span>projected_arrow_schema)<span class="question-mark">?</span>;
-<a href=#657 id=657 data-nosnippet>657</a>
-<a href=#658 id=658 data-nosnippet>658</a>        fields.filter_leaves(|idx, field| {
-<a href=#659 id=659 data-nosnippet>659</a>            <span class="kw">let </span><span class="prelude-val">Some</span>(field_id) = projected_fields.get(field).cloned() <span class="kw">else </span>{
-<a href=#660 id=660 data-nosnippet>660</a>                <span class="kw">return </span><span class="bool-val">false</span>;
-<a href=#661 id=661 data-nosnippet>661</a>            };
-<a href=#662 id=662 data-nosnippet>662</a>
-<a href=#663 id=663 data-nosnippet>663</a>            <span class="kw">let </span>iceberg_field = iceberg_schema_of_task.field_by_id(field_id);
-<a href=#664 id=664 data-nosnippet>664</a>            <span class="kw">let </span>parquet_iceberg_field = iceberg_schema.field_by_id(field_id);
-<a href=#665 id=665 data-nosnippet>665</a>
-<a href=#666 id=666 data-nosnippet>666</a>            <span class="kw">if </span>iceberg_field.is_none() || parquet_iceberg_field.is_none() {
-<a href=#667 id=667 data-nosnippet>667</a>                <span class="kw">return </span><span class="bool-val">false</span>;
-<a href=#668 id=668 data-nosnippet>668</a>            }
-<a href=#669 id=669 data-nosnippet>669</a>
-<a href=#670 id=670 data-nosnippet>670</a>            <span class="kw">if </span>!type_promotion_is_valid(
-<a href=#671 id=671 data-nosnippet>671</a>                parquet_iceberg_field
-<a href=#672 id=672 data-nosnippet>672</a>                    .unwrap()
-<a href=#673 id=673 data-nosnippet>673</a>                    .field_type
-<a href=#674 id=674 data-nosnippet>674</a>                    .as_primitive_type(),
-<a href=#675 id=675 data-nosnippet>675</a>                iceberg_field.unwrap().field_type.as_primitive_type(),
-<a href=#676 id=676 data-nosnippet>676</a>            ) {
-<a href=#677 id=677 data-nosnippet>677</a>                <span class="kw">return </span><span class="bool-val">false</span>;
-<a href=#678 id=678 data-nosnippet>678</a>            }
-<a href=#679 id=679 data-nosnippet>679</a>
-<a href=#680 id=680 data-nosnippet>680</a>            column_map.insert(field_id, idx);
-<a href=#681 id=681 data-nosnippet>681</a>            <span class="bool-val">true
-<a href=#682 id=682 data-nosnippet>682</a>        </span>});
-<a href=#683 id=683 data-nosnippet>683</a>
-<a href=#684 id=684 data-nosnippet>684</a>        <span class="comment">// Schema evolution: New columns may not exist in old Parquet files.
-<a href=#685 id=685 data-nosnippet>685</a>        // We only project existing columns; RecordBatchTransformer adds default/NULL values.
-<a href=#686 id=686 data-nosnippet>686</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>indices = <span class="macro">vec!</span>[];
-<a href=#687 id=687 data-nosnippet>687</a>        <span class="kw">for </span>field_id <span class="kw">in </span>leaf_field_ids {
-<a href=#688 id=688 data-nosnippet>688</a>            <span class="kw">if let </span><span class="prelude-val">Some</span>(col_idx) = column_map.get(field_id) {
-<a href=#689 id=689 data-nosnippet>689</a>                indices.push(<span class="kw-2">*</span>col_idx);
-<a href=#690 id=690 data-nosnippet>690</a>            }
-<a href=#691 id=691 data-nosnippet>691</a>        }
-<a href=#692 id=692 data-nosnippet>692</a>
-<a href=#693 id=693 data-nosnippet>693</a>        <span class="kw">if </span>indices.is_empty() {
-<a href=#694 id=694 data-nosnippet>694</a>            <span class="comment">// Edge case: All requested columns are new (don't exist in file).
-<a href=#695 id=695 data-nosnippet>695</a>            // Project all columns so RecordBatchTransformer has a batch to transform.
-<a href=#696 id=696 data-nosnippet>696</a>            </span><span class="prelude-val">Ok</span>(ProjectionMask::all())
-<a href=#697 id=697 data-nosnippet>697</a>        } <span class="kw">else </span>{
-<a href=#698 id=698 data-nosnippet>698</a>            <span class="prelude-val">Ok</span>(ProjectionMask::leaves(parquet_schema, indices))
-<a href=#699 id=699 data-nosnippet>699</a>        }
-<a href=#700 id=700 data-nosnippet>700</a>    }
-<a href=#701 id=701 data-nosnippet>701</a>
-<a href=#702 id=702 data-nosnippet>702</a>    <span class="doccomment">/// Fallback projection for Parquet files without field IDs.
-<a href=#703 id=703 data-nosnippet>703</a>    /// Uses position-based matching: field ID N → column position N-1.
-<a href=#704 id=704 data-nosnippet>704</a>    /// Projects entire top-level columns (including nested content) for iceberg-java compatibility.
-<a href=#705 id=705 data-nosnippet>705</a>    </span><span class="kw">fn </span>get_arrow_projection_mask_fallback(
-<a href=#706 id=706 data-nosnippet>706</a>        field_ids: <span class="kw-2">&amp;</span>[i32],
-<a href=#707 id=707 data-nosnippet>707</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
-<a href=#708 id=708 data-nosnippet>708</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ProjectionMask&gt; {
-<a href=#709 id=709 data-nosnippet>709</a>        <span class="comment">// Position-based: field_id N → column N-1 (field IDs are 1-indexed)
-<a href=#710 id=710 data-nosnippet>710</a>        </span><span class="kw">let </span>parquet_root_fields = parquet_schema.root_schema().get_fields();
-<a href=#711 id=711 data-nosnippet>711</a>        <span class="kw">let </span><span class="kw-2">mut </span>root_indices = <span class="macro">vec!</span>[];
-<a href=#712 id=712 data-nosnippet>712</a>
-<a href=#713 id=713 data-nosnippet>713</a>        <span class="kw">for </span>field_id <span class="kw">in </span>field_ids.iter() {
-<a href=#714 id=714 data-nosnippet>714</a>            <span class="kw">let </span>parquet_pos = (<span class="kw-2">*</span>field_id - <span class="number">1</span>) <span class="kw">as </span>usize;
-<a href=#715 id=715 data-nosnippet>715</a>
-<a href=#716 id=716 data-nosnippet>716</a>            <span class="kw">if </span>parquet_pos &lt; parquet_root_fields.len() {
-<a href=#717 id=717 data-nosnippet>717</a>                root_indices.push(parquet_pos);
-<a href=#718 id=718 data-nosnippet>718</a>            }
-<a href=#719 id=719 data-nosnippet>719</a>            <span class="comment">// RecordBatchTransformer adds missing columns with NULL values
-<a href=#720 id=720 data-nosnippet>720</a>        </span>}
+<a href=#514 id=514 data-nosnippet>514</a>                <span class="comment">// If there are no more pos deletes, add a selector for the entirety of this row group.
+<a href=#515 id=515 data-nosnippet>515</a>                </span><span class="kw">_ </span>=&gt; {
+<a href=#516 id=516 data-nosnippet>516</a>                    results.push(RowSelector::select(row_group_num_rows <span class="kw">as </span>usize));
+<a href=#517 id=517 data-nosnippet>517</a>                    current_row_group_base_idx += row_group_num_rows;
+<a href=#518 id=518 data-nosnippet>518</a>                    <span class="kw">continue</span>;
+<a href=#519 id=519 data-nosnippet>519</a>                }
+<a href=#520 id=520 data-nosnippet>520</a>            };
+<a href=#521 id=521 data-nosnippet>521</a>
+<a href=#522 id=522 data-nosnippet>522</a>            <span class="kw">let </span><span class="kw-2">mut </span>current_idx = current_row_group_base_idx;
+<a href=#523 id=523 data-nosnippet>523</a>            <span class="lifetime">'chunks</span>: <span class="kw">while </span>next_deleted_row_idx &lt; next_row_group_base_idx {
+<a href=#524 id=524 data-nosnippet>524</a>                <span class="comment">// `select` all rows that precede the next delete index
+<a href=#525 id=525 data-nosnippet>525</a>                </span><span class="kw">if </span>current_idx &lt; next_deleted_row_idx {
+<a href=#526 id=526 data-nosnippet>526</a>                    <span class="kw">let </span>run_length = next_deleted_row_idx - current_idx;
+<a href=#527 id=527 data-nosnippet>527</a>                    results.push(RowSelector::select(run_length <span class="kw">as </span>usize));
+<a href=#528 id=528 data-nosnippet>528</a>                    current_idx += run_length;
+<a href=#529 id=529 data-nosnippet>529</a>                }
+<a href=#530 id=530 data-nosnippet>530</a>
+<a href=#531 id=531 data-nosnippet>531</a>                <span class="comment">// `skip` all consecutive deleted rows in the current row group
+<a href=#532 id=532 data-nosnippet>532</a>                </span><span class="kw">let </span><span class="kw-2">mut </span>run_length = <span class="number">0</span>;
+<a href=#533 id=533 data-nosnippet>533</a>                <span class="kw">while </span>next_deleted_row_idx == current_idx
+<a href=#534 id=534 data-nosnippet>534</a>                    &amp;&amp; next_deleted_row_idx &lt; next_row_group_base_idx
+<a href=#535 id=535 data-nosnippet>535</a>                {
+<a href=#536 id=536 data-nosnippet>536</a>                    run_length += <span class="number">1</span>;
+<a href=#537 id=537 data-nosnippet>537</a>                    current_idx += <span class="number">1</span>;
+<a href=#538 id=538 data-nosnippet>538</a>
+<a href=#539 id=539 data-nosnippet>539</a>                    next_deleted_row_idx_opt = delete_vector_iter.next();
+<a href=#540 id=540 data-nosnippet>540</a>                    next_deleted_row_idx = <span class="kw">match </span>next_deleted_row_idx_opt {
+<a href=#541 id=541 data-nosnippet>541</a>                        <span class="prelude-val">Some</span>(next_deleted_row_idx) =&gt; next_deleted_row_idx,
+<a href=#542 id=542 data-nosnippet>542</a>                        <span class="kw">_ </span>=&gt; {
+<a href=#543 id=543 data-nosnippet>543</a>                            <span class="comment">// We've processed the final positional delete.
+<a href=#544 id=544 data-nosnippet>544</a>                            // Conclude the skip and then break so that we select the remaining
+<a href=#545 id=545 data-nosnippet>545</a>                            // rows in the row group and move on to the next row group
+<a href=#546 id=546 data-nosnippet>546</a>                            </span>results.push(RowSelector::skip(run_length));
+<a href=#547 id=547 data-nosnippet>547</a>                            <span class="kw">break </span><span class="lifetime">'chunks</span>;
+<a href=#548 id=548 data-nosnippet>548</a>                        }
+<a href=#549 id=549 data-nosnippet>549</a>                    };
+<a href=#550 id=550 data-nosnippet>550</a>                }
+<a href=#551 id=551 data-nosnippet>551</a>                <span class="kw">if </span>run_length &gt; <span class="number">0 </span>{
+<a href=#552 id=552 data-nosnippet>552</a>                    results.push(RowSelector::skip(run_length));
+<a href=#553 id=553 data-nosnippet>553</a>                }
+<a href=#554 id=554 data-nosnippet>554</a>            }
+<a href=#555 id=555 data-nosnippet>555</a>
+<a href=#556 id=556 data-nosnippet>556</a>            <span class="kw">if </span>current_idx &lt; next_row_group_base_idx {
+<a href=#557 id=557 data-nosnippet>557</a>                results.push(RowSelector::select(
+<a href=#558 id=558 data-nosnippet>558</a>                    (next_row_group_base_idx - current_idx) <span class="kw">as </span>usize,
+<a href=#559 id=559 data-nosnippet>559</a>                ));
+<a href=#560 id=560 data-nosnippet>560</a>            }
+<a href=#561 id=561 data-nosnippet>561</a>
+<a href=#562 id=562 data-nosnippet>562</a>            current_row_group_base_idx += row_group_num_rows;
+<a href=#563 id=563 data-nosnippet>563</a>        }
+<a href=#564 id=564 data-nosnippet>564</a>
+<a href=#565 id=565 data-nosnippet>565</a>        <span class="prelude-val">Ok</span>(results.into())
+<a href=#566 id=566 data-nosnippet>566</a>    }
+<a href=#567 id=567 data-nosnippet>567</a>
+<a href=#568 id=568 data-nosnippet>568</a>    <span class="kw">fn </span>build_field_id_set_and_map(
+<a href=#569 id=569 data-nosnippet>569</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
+<a href=#570 id=570 data-nosnippet>570</a>        predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#571 id=571 data-nosnippet>571</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;(HashSet&lt;i32&gt;, HashMap&lt;i32, usize&gt;)&gt; {
+<a href=#572 id=572 data-nosnippet>572</a>        <span class="comment">// Collects all Iceberg field IDs referenced in the filter predicate
+<a href=#573 id=573 data-nosnippet>573</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>collector = CollectFieldIdVisitor {
+<a href=#574 id=574 data-nosnippet>574</a>            field_ids: HashSet::default(),
+<a href=#575 id=575 data-nosnippet>575</a>        };
+<a href=#576 id=576 data-nosnippet>576</a>        visit(<span class="kw-2">&amp;mut </span>collector, predicate)<span class="question-mark">?</span>;
+<a href=#577 id=577 data-nosnippet>577</a>
+<a href=#578 id=578 data-nosnippet>578</a>        <span class="kw">let </span>iceberg_field_ids = collector.field_ids();
+<a href=#579 id=579 data-nosnippet>579</a>
+<a href=#580 id=580 data-nosnippet>580</a>        <span class="comment">// Without embedded field IDs, we fall back to position-based mapping for compatibility
+<a href=#581 id=581 data-nosnippet>581</a>        </span><span class="kw">let </span>field_id_map = <span class="kw">match </span>build_field_id_map(parquet_schema)<span class="question-mark">? </span>{
+<a href=#582 id=582 data-nosnippet>582</a>            <span class="prelude-val">Some</span>(map) =&gt; map,
+<a href=#583 id=583 data-nosnippet>583</a>            <span class="prelude-val">None </span>=&gt; build_fallback_field_id_map(parquet_schema),
+<a href=#584 id=584 data-nosnippet>584</a>        };
+<a href=#585 id=585 data-nosnippet>585</a>
+<a href=#586 id=586 data-nosnippet>586</a>        <span class="prelude-val">Ok</span>((iceberg_field_ids, field_id_map))
+<a href=#587 id=587 data-nosnippet>587</a>    }
+<a href=#588 id=588 data-nosnippet>588</a>
+<a href=#589 id=589 data-nosnippet>589</a>    <span class="doccomment">/// Recursively extract leaf field IDs because Parquet projection works at the leaf column level.
+<a href=#590 id=590 data-nosnippet>590</a>    /// Nested types (struct/list/map) are flattened in Parquet's columnar format.
+<a href=#591 id=591 data-nosnippet>591</a>    </span><span class="kw">fn </span>include_leaf_field_id(field: <span class="kw-2">&amp;</span>NestedField, field_ids: <span class="kw-2">&amp;mut </span>Vec&lt;i32&gt;) {
+<a href=#592 id=592 data-nosnippet>592</a>        <span class="kw">match </span>field.field_type.as_ref() {
+<a href=#593 id=593 data-nosnippet>593</a>            Type::Primitive(<span class="kw">_</span>) =&gt; {
+<a href=#594 id=594 data-nosnippet>594</a>                field_ids.push(field.id);
+<a href=#595 id=595 data-nosnippet>595</a>            }
+<a href=#596 id=596 data-nosnippet>596</a>            Type::Struct(struct_type) =&gt; {
+<a href=#597 id=597 data-nosnippet>597</a>                <span class="kw">for </span>nested_field <span class="kw">in </span>struct_type.fields() {
+<a href=#598 id=598 data-nosnippet>598</a>                    <span class="self">Self</span>::include_leaf_field_id(nested_field, field_ids);
+<a href=#599 id=599 data-nosnippet>599</a>                }
+<a href=#600 id=600 data-nosnippet>600</a>            }
+<a href=#601 id=601 data-nosnippet>601</a>            Type::List(list_type) =&gt; {
+<a href=#602 id=602 data-nosnippet>602</a>                <span class="self">Self</span>::include_leaf_field_id(<span class="kw-2">&amp;</span>list_type.element_field, field_ids);
+<a href=#603 id=603 data-nosnippet>603</a>            }
+<a href=#604 id=604 data-nosnippet>604</a>            Type::Map(map_type) =&gt; {
+<a href=#605 id=605 data-nosnippet>605</a>                <span class="self">Self</span>::include_leaf_field_id(<span class="kw-2">&amp;</span>map_type.key_field, field_ids);
+<a href=#606 id=606 data-nosnippet>606</a>                <span class="self">Self</span>::include_leaf_field_id(<span class="kw-2">&amp;</span>map_type.value_field, field_ids);
+<a href=#607 id=607 data-nosnippet>607</a>            }
+<a href=#608 id=608 data-nosnippet>608</a>        }
+<a href=#609 id=609 data-nosnippet>609</a>    }
+<a href=#610 id=610 data-nosnippet>610</a>
+<a href=#611 id=611 data-nosnippet>611</a>    <span class="kw">fn </span>get_arrow_projection_mask(
+<a href=#612 id=612 data-nosnippet>612</a>        field_ids: <span class="kw-2">&amp;</span>[i32],
+<a href=#613 id=613 data-nosnippet>613</a>        iceberg_schema_of_task: <span class="kw-2">&amp;</span>Schema,
+<a href=#614 id=614 data-nosnippet>614</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
+<a href=#615 id=615 data-nosnippet>615</a>        arrow_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
+<a href=#616 id=616 data-nosnippet>616</a>        use_fallback: bool, <span class="comment">// Whether file lacks embedded field IDs (e.g., migrated from Hive/Spark)
+<a href=#617 id=617 data-nosnippet>617</a>    </span>) -&gt; <span class="prelude-ty">Result</span>&lt;ProjectionMask&gt; {
+<a href=#618 id=618 data-nosnippet>618</a>        <span class="kw">fn </span>type_promotion_is_valid(
+<a href=#619 id=619 data-nosnippet>619</a>            file_type: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PrimitiveType&gt;,
+<a href=#620 id=620 data-nosnippet>620</a>            projected_type: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PrimitiveType&gt;,
+<a href=#621 id=621 data-nosnippet>621</a>        ) -&gt; bool {
+<a href=#622 id=622 data-nosnippet>622</a>            <span class="kw">match </span>(file_type, projected_type) {
+<a href=#623 id=623 data-nosnippet>623</a>                (<span class="prelude-val">Some</span>(lhs), <span class="prelude-val">Some</span>(rhs)) <span class="kw">if </span>lhs == rhs =&gt; <span class="bool-val">true</span>,
+<a href=#624 id=624 data-nosnippet>624</a>                (<span class="prelude-val">Some</span>(PrimitiveType::Int), <span class="prelude-val">Some</span>(PrimitiveType::Long)) =&gt; <span class="bool-val">true</span>,
+<a href=#625 id=625 data-nosnippet>625</a>                (<span class="prelude-val">Some</span>(PrimitiveType::Float), <span class="prelude-val">Some</span>(PrimitiveType::Double)) =&gt; <span class="bool-val">true</span>,
+<a href=#626 id=626 data-nosnippet>626</a>                (
+<a href=#627 id=627 data-nosnippet>627</a>                    <span class="prelude-val">Some</span>(PrimitiveType::Decimal {
+<a href=#628 id=628 data-nosnippet>628</a>                        precision: file_precision,
+<a href=#629 id=629 data-nosnippet>629</a>                        scale: file_scale,
+<a href=#630 id=630 data-nosnippet>630</a>                    }),
+<a href=#631 id=631 data-nosnippet>631</a>                    <span class="prelude-val">Some</span>(PrimitiveType::Decimal {
+<a href=#632 id=632 data-nosnippet>632</a>                        precision: requested_precision,
+<a href=#633 id=633 data-nosnippet>633</a>                        scale: requested_scale,
+<a href=#634 id=634 data-nosnippet>634</a>                    }),
+<a href=#635 id=635 data-nosnippet>635</a>                ) <span class="kw">if </span>requested_precision &gt;= file_precision &amp;&amp; file_scale == requested_scale =&gt; <span class="bool-val">true</span>,
+<a href=#636 id=636 data-nosnippet>636</a>                <span class="comment">// Uuid will be store as Fixed(16) in parquet file, so the read back type will be Fixed(16).
+<a href=#637 id=637 data-nosnippet>637</a>                </span>(<span class="prelude-val">Some</span>(PrimitiveType::Fixed(<span class="number">16</span>)), <span class="prelude-val">Some</span>(PrimitiveType::Uuid)) =&gt; <span class="bool-val">true</span>,
+<a href=#638 id=638 data-nosnippet>638</a>                <span class="kw">_ </span>=&gt; <span class="bool-val">false</span>,
+<a href=#639 id=639 data-nosnippet>639</a>            }
+<a href=#640 id=640 data-nosnippet>640</a>        }
+<a href=#641 id=641 data-nosnippet>641</a>
+<a href=#642 id=642 data-nosnippet>642</a>        <span class="kw">if </span>field_ids.is_empty() {
+<a href=#643 id=643 data-nosnippet>643</a>            <span class="kw">return </span><span class="prelude-val">Ok</span>(ProjectionMask::all());
+<a href=#644 id=644 data-nosnippet>644</a>        }
+<a href=#645 id=645 data-nosnippet>645</a>
+<a href=#646 id=646 data-nosnippet>646</a>        <span class="kw">if </span>use_fallback {
+<a href=#647 id=647 data-nosnippet>647</a>            <span class="comment">// Position-based projection necessary because file lacks embedded field IDs
+<a href=#648 id=648 data-nosnippet>648</a>            </span><span class="self">Self</span>::get_arrow_projection_mask_fallback(field_ids, parquet_schema)
+<a href=#649 id=649 data-nosnippet>649</a>        } <span class="kw">else </span>{
+<a href=#650 id=650 data-nosnippet>650</a>            <span class="comment">// Field-ID-based projection using embedded field IDs from Parquet metadata
+<a href=#651 id=651 data-nosnippet>651</a>
+<a href=#652 id=652 data-nosnippet>652</a>            // Parquet's columnar format requires leaf-level (not top-level struct/list/map) projection
+<a href=#653 id=653 data-nosnippet>653</a>            </span><span class="kw">let </span><span class="kw-2">mut </span>leaf_field_ids = <span class="macro">vec!</span>[];
+<a href=#654 id=654 data-nosnippet>654</a>            <span class="kw">for </span>field_id <span class="kw">in </span>field_ids {
+<a href=#655 id=655 data-nosnippet>655</a>                <span class="kw">let </span>field = iceberg_schema_of_task.field_by_id(<span class="kw-2">*</span>field_id);
+<a href=#656 id=656 data-nosnippet>656</a>                <span class="kw">if let </span><span class="prelude-val">Some</span>(field) = field {
+<a href=#657 id=657 data-nosnippet>657</a>                    <span class="self">Self</span>::include_leaf_field_id(field, <span class="kw-2">&amp;mut </span>leaf_field_ids);
+<a href=#658 id=658 data-nosnippet>658</a>                }
+<a href=#659 id=659 data-nosnippet>659</a>            }
+<a href=#660 id=660 data-nosnippet>660</a>
+<a href=#661 id=661 data-nosnippet>661</a>            <span class="self">Self</span>::get_arrow_projection_mask_with_field_ids(
+<a href=#662 id=662 data-nosnippet>662</a>                <span class="kw-2">&amp;</span>leaf_field_ids,
+<a href=#663 id=663 data-nosnippet>663</a>                iceberg_schema_of_task,
+<a href=#664 id=664 data-nosnippet>664</a>                parquet_schema,
+<a href=#665 id=665 data-nosnippet>665</a>                arrow_schema,
+<a href=#666 id=666 data-nosnippet>666</a>                type_promotion_is_valid,
+<a href=#667 id=667 data-nosnippet>667</a>            )
+<a href=#668 id=668 data-nosnippet>668</a>        }
+<a href=#669 id=669 data-nosnippet>669</a>    }
+<a href=#670 id=670 data-nosnippet>670</a>
+<a href=#671 id=671 data-nosnippet>671</a>    <span class="doccomment">/// Standard projection using embedded field IDs from Parquet metadata.
+<a href=#672 id=672 data-nosnippet>672</a>    /// For iceberg-java compatibility with ParquetSchemaUtil.pruneColumns().
+<a href=#673 id=673 data-nosnippet>673</a>    </span><span class="kw">fn </span>get_arrow_projection_mask_with_field_ids(
+<a href=#674 id=674 data-nosnippet>674</a>        leaf_field_ids: <span class="kw-2">&amp;</span>[i32],
+<a href=#675 id=675 data-nosnippet>675</a>        iceberg_schema_of_task: <span class="kw-2">&amp;</span>Schema,
+<a href=#676 id=676 data-nosnippet>676</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
+<a href=#677 id=677 data-nosnippet>677</a>        arrow_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
+<a href=#678 id=678 data-nosnippet>678</a>        type_promotion_is_valid: <span class="kw">fn</span>(<span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PrimitiveType&gt;, <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PrimitiveType&gt;) -&gt; bool,
+<a href=#679 id=679 data-nosnippet>679</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ProjectionMask&gt; {
+<a href=#680 id=680 data-nosnippet>680</a>        <span class="kw">let </span><span class="kw-2">mut </span>column_map = HashMap::new();
+<a href=#681 id=681 data-nosnippet>681</a>        <span class="kw">let </span>fields = arrow_schema.fields();
+<a href=#682 id=682 data-nosnippet>682</a>
+<a href=#683 id=683 data-nosnippet>683</a>        <span class="comment">// Pre-project only the fields that have been selected, possibly avoiding converting
+<a href=#684 id=684 data-nosnippet>684</a>        // some Arrow types that are not yet supported.
+<a href=#685 id=685 data-nosnippet>685</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>projected_fields: HashMap&lt;FieldRef, i32&gt; = HashMap::new();
+<a href=#686 id=686 data-nosnippet>686</a>        <span class="kw">let </span>projected_arrow_schema = ArrowSchema::new_with_metadata(
+<a href=#687 id=687 data-nosnippet>687</a>            fields.filter_leaves(|<span class="kw">_</span>, f| {
+<a href=#688 id=688 data-nosnippet>688</a>                f.metadata()
+<a href=#689 id=689 data-nosnippet>689</a>                    .get(PARQUET_FIELD_ID_META_KEY)
+<a href=#690 id=690 data-nosnippet>690</a>                    .and_then(|field_id| i32::from_str(field_id).ok())
+<a href=#691 id=691 data-nosnippet>691</a>                    .is_some_and(|field_id| {
+<a href=#692 id=692 data-nosnippet>692</a>                        projected_fields.insert((<span class="kw-2">*</span>f).clone(), field_id);
+<a href=#693 id=693 data-nosnippet>693</a>                        leaf_field_ids.contains(<span class="kw-2">&amp;</span>field_id)
+<a href=#694 id=694 data-nosnippet>694</a>                    })
+<a href=#695 id=695 data-nosnippet>695</a>            }),
+<a href=#696 id=696 data-nosnippet>696</a>            arrow_schema.metadata().clone(),
+<a href=#697 id=697 data-nosnippet>697</a>        );
+<a href=#698 id=698 data-nosnippet>698</a>        <span class="kw">let </span>iceberg_schema = arrow_schema_to_schema(<span class="kw-2">&amp;</span>projected_arrow_schema)<span class="question-mark">?</span>;
+<a href=#699 id=699 data-nosnippet>699</a>
+<a href=#700 id=700 data-nosnippet>700</a>        fields.filter_leaves(|idx, field| {
+<a href=#701 id=701 data-nosnippet>701</a>            <span class="kw">let </span><span class="prelude-val">Some</span>(field_id) = projected_fields.get(field).cloned() <span class="kw">else </span>{
+<a href=#702 id=702 data-nosnippet>702</a>                <span class="kw">return </span><span class="bool-val">false</span>;
+<a href=#703 id=703 data-nosnippet>703</a>            };
+<a href=#704 id=704 data-nosnippet>704</a>
+<a href=#705 id=705 data-nosnippet>705</a>            <span class="kw">let </span>iceberg_field = iceberg_schema_of_task.field_by_id(field_id);
+<a href=#706 id=706 data-nosnippet>706</a>            <span class="kw">let </span>parquet_iceberg_field = iceberg_schema.field_by_id(field_id);
+<a href=#707 id=707 data-nosnippet>707</a>
+<a href=#708 id=708 data-nosnippet>708</a>            <span class="kw">if </span>iceberg_field.is_none() || parquet_iceberg_field.is_none() {
+<a href=#709 id=709 data-nosnippet>709</a>                <span class="kw">return </span><span class="bool-val">false</span>;
+<a href=#710 id=710 data-nosnippet>710</a>            }
+<a href=#711 id=711 data-nosnippet>711</a>
+<a href=#712 id=712 data-nosnippet>712</a>            <span class="kw">if </span>!type_promotion_is_valid(
+<a href=#713 id=713 data-nosnippet>713</a>                parquet_iceberg_field
+<a href=#714 id=714 data-nosnippet>714</a>                    .unwrap()
+<a href=#715 id=715 data-nosnippet>715</a>                    .field_type
+<a href=#716 id=716 data-nosnippet>716</a>                    .as_primitive_type(),
+<a href=#717 id=717 data-nosnippet>717</a>                iceberg_field.unwrap().field_type.as_primitive_type(),
+<a href=#718 id=718 data-nosnippet>718</a>            ) {
+<a href=#719 id=719 data-nosnippet>719</a>                <span class="kw">return </span><span class="bool-val">false</span>;
+<a href=#720 id=720 data-nosnippet>720</a>            }
 <a href=#721 id=721 data-nosnippet>721</a>
-<a href=#722 id=722 data-nosnippet>722</a>        <span class="kw">if </span>root_indices.is_empty() {
-<a href=#723 id=723 data-nosnippet>723</a>            <span class="prelude-val">Ok</span>(ProjectionMask::all())
-<a href=#724 id=724 data-nosnippet>724</a>        } <span class="kw">else </span>{
-<a href=#725 id=725 data-nosnippet>725</a>            <span class="prelude-val">Ok</span>(ProjectionMask::roots(parquet_schema, root_indices))
-<a href=#726 id=726 data-nosnippet>726</a>        }
-<a href=#727 id=727 data-nosnippet>727</a>    }
-<a href=#728 id=728 data-nosnippet>728</a>
-<a href=#729 id=729 data-nosnippet>729</a>    <span class="kw">fn </span>get_row_filter(
-<a href=#730 id=730 data-nosnippet>730</a>        predicates: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#731 id=731 data-nosnippet>731</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
-<a href=#732 id=732 data-nosnippet>732</a>        iceberg_field_ids: <span class="kw-2">&amp;</span>HashSet&lt;i32&gt;,
-<a href=#733 id=733 data-nosnippet>733</a>        field_id_map: <span class="kw-2">&amp;</span>HashMap&lt;i32, usize&gt;,
-<a href=#734 id=734 data-nosnippet>734</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;RowFilter&gt; {
-<a href=#735 id=735 data-nosnippet>735</a>        <span class="comment">// Collect Parquet column indices from field ids.
-<a href=#736 id=736 data-nosnippet>736</a>        // If the field id is not found in Parquet schema, it will be ignored due to schema evolution.
-<a href=#737 id=737 data-nosnippet>737</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>column_indices = iceberg_field_ids
-<a href=#738 id=738 data-nosnippet>738</a>            .iter()
-<a href=#739 id=739 data-nosnippet>739</a>            .filter_map(|field_id| field_id_map.get(field_id).cloned())
-<a href=#740 id=740 data-nosnippet>740</a>            .collect::&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;();
-<a href=#741 id=741 data-nosnippet>741</a>        column_indices.sort();
-<a href=#742 id=742 data-nosnippet>742</a>
-<a href=#743 id=743 data-nosnippet>743</a>        <span class="comment">// The converter that converts `BoundPredicates` to `ArrowPredicates`
-<a href=#744 id=744 data-nosnippet>744</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>converter = PredicateConverter {
-<a href=#745 id=745 data-nosnippet>745</a>            parquet_schema,
-<a href=#746 id=746 data-nosnippet>746</a>            column_map: field_id_map,
-<a href=#747 id=747 data-nosnippet>747</a>            column_indices: <span class="kw-2">&amp;</span>column_indices,
-<a href=#748 id=748 data-nosnippet>748</a>        };
-<a href=#749 id=749 data-nosnippet>749</a>
-<a href=#750 id=750 data-nosnippet>750</a>        <span class="comment">// After collecting required leaf column indices used in the predicate,
-<a href=#751 id=751 data-nosnippet>751</a>        // creates the projection mask for the Arrow predicates.
-<a href=#752 id=752 data-nosnippet>752</a>        </span><span class="kw">let </span>projection_mask = ProjectionMask::leaves(parquet_schema, column_indices.clone());
-<a href=#753 id=753 data-nosnippet>753</a>        <span class="kw">let </span>predicate_func = visit(<span class="kw-2">&amp;mut </span>converter, predicates)<span class="question-mark">?</span>;
-<a href=#754 id=754 data-nosnippet>754</a>        <span class="kw">let </span>arrow_predicate = ArrowPredicateFn::new(projection_mask, predicate_func);
-<a href=#755 id=755 data-nosnippet>755</a>        <span class="prelude-val">Ok</span>(RowFilter::new(<span class="macro">vec!</span>[Box::new(arrow_predicate)]))
-<a href=#756 id=756 data-nosnippet>756</a>    }
+<a href=#722 id=722 data-nosnippet>722</a>            column_map.insert(field_id, idx);
+<a href=#723 id=723 data-nosnippet>723</a>            <span class="bool-val">true
+<a href=#724 id=724 data-nosnippet>724</a>        </span>});
+<a href=#725 id=725 data-nosnippet>725</a>
+<a href=#726 id=726 data-nosnippet>726</a>        <span class="comment">// Schema evolution: New columns may not exist in old Parquet files.
+<a href=#727 id=727 data-nosnippet>727</a>        // We only project existing columns; RecordBatchTransformer adds default/NULL values.
+<a href=#728 id=728 data-nosnippet>728</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>indices = <span class="macro">vec!</span>[];
+<a href=#729 id=729 data-nosnippet>729</a>        <span class="kw">for </span>field_id <span class="kw">in </span>leaf_field_ids {
+<a href=#730 id=730 data-nosnippet>730</a>            <span class="kw">if let </span><span class="prelude-val">Some</span>(col_idx) = column_map.get(field_id) {
+<a href=#731 id=731 data-nosnippet>731</a>                indices.push(<span class="kw-2">*</span>col_idx);
+<a href=#732 id=732 data-nosnippet>732</a>            }
+<a href=#733 id=733 data-nosnippet>733</a>        }
+<a href=#734 id=734 data-nosnippet>734</a>
+<a href=#735 id=735 data-nosnippet>735</a>        <span class="kw">if </span>indices.is_empty() {
+<a href=#736 id=736 data-nosnippet>736</a>            <span class="comment">// Edge case: All requested columns are new (don't exist in file).
+<a href=#737 id=737 data-nosnippet>737</a>            // Project all columns so RecordBatchTransformer has a batch to transform.
+<a href=#738 id=738 data-nosnippet>738</a>            </span><span class="prelude-val">Ok</span>(ProjectionMask::all())
+<a href=#739 id=739 data-nosnippet>739</a>        } <span class="kw">else </span>{
+<a href=#740 id=740 data-nosnippet>740</a>            <span class="prelude-val">Ok</span>(ProjectionMask::leaves(parquet_schema, indices))
+<a href=#741 id=741 data-nosnippet>741</a>        }
+<a href=#742 id=742 data-nosnippet>742</a>    }
+<a href=#743 id=743 data-nosnippet>743</a>
+<a href=#744 id=744 data-nosnippet>744</a>    <span class="doccomment">/// Fallback projection for Parquet files without field IDs.
+<a href=#745 id=745 data-nosnippet>745</a>    /// Uses position-based matching: field ID N → column position N-1.
+<a href=#746 id=746 data-nosnippet>746</a>    /// Projects entire top-level columns (including nested content) for iceberg-java compatibility.
+<a href=#747 id=747 data-nosnippet>747</a>    </span><span class="kw">fn </span>get_arrow_projection_mask_fallback(
+<a href=#748 id=748 data-nosnippet>748</a>        field_ids: <span class="kw-2">&amp;</span>[i32],
+<a href=#749 id=749 data-nosnippet>749</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
+<a href=#750 id=750 data-nosnippet>750</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ProjectionMask&gt; {
+<a href=#751 id=751 data-nosnippet>751</a>        <span class="comment">// Position-based: field_id N → column N-1 (field IDs are 1-indexed)
+<a href=#752 id=752 data-nosnippet>752</a>        </span><span class="kw">let </span>parquet_root_fields = parquet_schema.root_schema().get_fields();
+<a href=#753 id=753 data-nosnippet>753</a>        <span class="kw">let </span><span class="kw-2">mut </span>root_indices = <span class="macro">vec!</span>[];
+<a href=#754 id=754 data-nosnippet>754</a>
+<a href=#755 id=755 data-nosnippet>755</a>        <span class="kw">for </span>field_id <span class="kw">in </span>field_ids.iter() {
+<a href=#756 id=756 data-nosnippet>756</a>            <span class="kw">let </span>parquet_pos = (<span class="kw-2">*</span>field_id - <span class="number">1</span>) <span class="kw">as </span>usize;
 <a href=#757 id=757 data-nosnippet>757</a>
-<a href=#758 id=758 data-nosnippet>758</a>    <span class="kw">fn </span>get_selected_row_group_indices(
-<a href=#759 id=759 data-nosnippet>759</a>        predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#760 id=760 data-nosnippet>760</a>        parquet_metadata: <span class="kw-2">&amp;</span>Arc&lt;ParquetMetaData&gt;,
-<a href=#761 id=761 data-nosnippet>761</a>        field_id_map: <span class="kw-2">&amp;</span>HashMap&lt;i32, usize&gt;,
-<a href=#762 id=762 data-nosnippet>762</a>        snapshot_schema: <span class="kw-2">&amp;</span>Schema,
-<a href=#763 id=763 data-nosnippet>763</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;usize&gt;&gt; {
-<a href=#764 id=764 data-nosnippet>764</a>        <span class="kw">let </span>row_groups_metadata = parquet_metadata.row_groups();
-<a href=#765 id=765 data-nosnippet>765</a>        <span class="kw">let </span><span class="kw-2">mut </span>results = Vec::with_capacity(row_groups_metadata.len());
-<a href=#766 id=766 data-nosnippet>766</a>
-<a href=#767 id=767 data-nosnippet>767</a>        <span class="kw">for </span>(idx, row_group_metadata) <span class="kw">in </span>row_groups_metadata.iter().enumerate() {
-<a href=#768 id=768 data-nosnippet>768</a>            <span class="kw">if </span>RowGroupMetricsEvaluator::eval(
-<a href=#769 id=769 data-nosnippet>769</a>                predicate,
-<a href=#770 id=770 data-nosnippet>770</a>                row_group_metadata,
-<a href=#771 id=771 data-nosnippet>771</a>                field_id_map,
-<a href=#772 id=772 data-nosnippet>772</a>                snapshot_schema,
-<a href=#773 id=773 data-nosnippet>773</a>            )<span class="question-mark">? </span>{
-<a href=#774 id=774 data-nosnippet>774</a>                results.push(idx);
-<a href=#775 id=775 data-nosnippet>775</a>            }
-<a href=#776 id=776 data-nosnippet>776</a>        }
-<a href=#777 id=777 data-nosnippet>777</a>
-<a href=#778 id=778 data-nosnippet>778</a>        <span class="prelude-val">Ok</span>(results)
-<a href=#779 id=779 data-nosnippet>779</a>    }
-<a href=#780 id=780 data-nosnippet>780</a>
-<a href=#781 id=781 data-nosnippet>781</a>    <span class="kw">fn </span>get_row_selection_for_filter_predicate(
-<a href=#782 id=782 data-nosnippet>782</a>        predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#783 id=783 data-nosnippet>783</a>        parquet_metadata: <span class="kw-2">&amp;</span>Arc&lt;ParquetMetaData&gt;,
-<a href=#784 id=784 data-nosnippet>784</a>        selected_row_groups: <span class="kw-2">&amp;</span><span class="prelude-ty">Option</span>&lt;Vec&lt;usize&gt;&gt;,
-<a href=#785 id=785 data-nosnippet>785</a>        field_id_map: <span class="kw-2">&amp;</span>HashMap&lt;i32, usize&gt;,
-<a href=#786 id=786 data-nosnippet>786</a>        snapshot_schema: <span class="kw-2">&amp;</span>Schema,
-<a href=#787 id=787 data-nosnippet>787</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;RowSelection&gt; {
-<a href=#788 id=788 data-nosnippet>788</a>        <span class="kw">let </span><span class="prelude-val">Some</span>(column_index) = parquet_metadata.column_index() <span class="kw">else </span>{
-<a href=#789 id=789 data-nosnippet>789</a>            <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#790 id=790 data-nosnippet>790</a>                ErrorKind::Unexpected,
-<a href=#791 id=791 data-nosnippet>791</a>                <span class="string">"Parquet file metadata does not contain a column index"</span>,
-<a href=#792 id=792 data-nosnippet>792</a>            ));
-<a href=#793 id=793 data-nosnippet>793</a>        };
-<a href=#794 id=794 data-nosnippet>794</a>
-<a href=#795 id=795 data-nosnippet>795</a>        <span class="kw">let </span><span class="prelude-val">Some</span>(offset_index) = parquet_metadata.offset_index() <span class="kw">else </span>{
-<a href=#796 id=796 data-nosnippet>796</a>            <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#797 id=797 data-nosnippet>797</a>                ErrorKind::Unexpected,
-<a href=#798 id=798 data-nosnippet>798</a>                <span class="string">"Parquet file metadata does not contain an offset index"</span>,
-<a href=#799 id=799 data-nosnippet>799</a>            ));
-<a href=#800 id=800 data-nosnippet>800</a>        };
-<a href=#801 id=801 data-nosnippet>801</a>
-<a href=#802 id=802 data-nosnippet>802</a>        <span class="comment">// If all row groups were filtered out, return an empty RowSelection (select no rows)
-<a href=#803 id=803 data-nosnippet>803</a>        </span><span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_groups) = selected_row_groups {
-<a href=#804 id=804 data-nosnippet>804</a>            <span class="kw">if </span>selected_row_groups.is_empty() {
-<a href=#805 id=805 data-nosnippet>805</a>                <span class="kw">return </span><span class="prelude-val">Ok</span>(RowSelection::from(Vec::new()));
-<a href=#806 id=806 data-nosnippet>806</a>            }
-<a href=#807 id=807 data-nosnippet>807</a>        }
+<a href=#758 id=758 data-nosnippet>758</a>            <span class="kw">if </span>parquet_pos &lt; parquet_root_fields.len() {
+<a href=#759 id=759 data-nosnippet>759</a>                root_indices.push(parquet_pos);
+<a href=#760 id=760 data-nosnippet>760</a>            }
+<a href=#761 id=761 data-nosnippet>761</a>            <span class="comment">// RecordBatchTransformer adds missing columns with NULL values
+<a href=#762 id=762 data-nosnippet>762</a>        </span>}
+<a href=#763 id=763 data-nosnippet>763</a>
+<a href=#764 id=764 data-nosnippet>764</a>        <span class="kw">if </span>root_indices.is_empty() {
+<a href=#765 id=765 data-nosnippet>765</a>            <span class="prelude-val">Ok</span>(ProjectionMask::all())
+<a href=#766 id=766 data-nosnippet>766</a>        } <span class="kw">else </span>{
+<a href=#767 id=767 data-nosnippet>767</a>            <span class="prelude-val">Ok</span>(ProjectionMask::roots(parquet_schema, root_indices))
+<a href=#768 id=768 data-nosnippet>768</a>        }
+<a href=#769 id=769 data-nosnippet>769</a>    }
+<a href=#770 id=770 data-nosnippet>770</a>
+<a href=#771 id=771 data-nosnippet>771</a>    <span class="kw">fn </span>get_row_filter(
+<a href=#772 id=772 data-nosnippet>772</a>        predicates: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#773 id=773 data-nosnippet>773</a>        parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor,
+<a href=#774 id=774 data-nosnippet>774</a>        iceberg_field_ids: <span class="kw-2">&amp;</span>HashSet&lt;i32&gt;,
+<a href=#775 id=775 data-nosnippet>775</a>        field_id_map: <span class="kw-2">&amp;</span>HashMap&lt;i32, usize&gt;,
+<a href=#776 id=776 data-nosnippet>776</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;RowFilter&gt; {
+<a href=#777 id=777 data-nosnippet>777</a>        <span class="comment">// Collect Parquet column indices from field ids.
+<a href=#778 id=778 data-nosnippet>778</a>        // If the field id is not found in Parquet schema, it will be ignored due to schema evolution.
+<a href=#779 id=779 data-nosnippet>779</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>column_indices = iceberg_field_ids
+<a href=#780 id=780 data-nosnippet>780</a>            .iter()
+<a href=#781 id=781 data-nosnippet>781</a>            .filter_map(|field_id| field_id_map.get(field_id).cloned())
+<a href=#782 id=782 data-nosnippet>782</a>            .collect::&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;();
+<a href=#783 id=783 data-nosnippet>783</a>        column_indices.sort();
+<a href=#784 id=784 data-nosnippet>784</a>
+<a href=#785 id=785 data-nosnippet>785</a>        <span class="comment">// The converter that converts `BoundPredicates` to `ArrowPredicates`
+<a href=#786 id=786 data-nosnippet>786</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>converter = PredicateConverter {
+<a href=#787 id=787 data-nosnippet>787</a>            parquet_schema,
+<a href=#788 id=788 data-nosnippet>788</a>            column_map: field_id_map,
+<a href=#789 id=789 data-nosnippet>789</a>            column_indices: <span class="kw-2">&amp;</span>column_indices,
+<a href=#790 id=790 data-nosnippet>790</a>        };
+<a href=#791 id=791 data-nosnippet>791</a>
+<a href=#792 id=792 data-nosnippet>792</a>        <span class="comment">// After collecting required leaf column indices used in the predicate,
+<a href=#793 id=793 data-nosnippet>793</a>        // creates the projection mask for the Arrow predicates.
+<a href=#794 id=794 data-nosnippet>794</a>        </span><span class="kw">let </span>projection_mask = ProjectionMask::leaves(parquet_schema, column_indices.clone());
+<a href=#795 id=795 data-nosnippet>795</a>        <span class="kw">let </span>predicate_func = visit(<span class="kw-2">&amp;mut </span>converter, predicates)<span class="question-mark">?</span>;
+<a href=#796 id=796 data-nosnippet>796</a>        <span class="kw">let </span>arrow_predicate = ArrowPredicateFn::new(projection_mask, predicate_func);
+<a href=#797 id=797 data-nosnippet>797</a>        <span class="prelude-val">Ok</span>(RowFilter::new(<span class="macro">vec!</span>[Box::new(arrow_predicate)]))
+<a href=#798 id=798 data-nosnippet>798</a>    }
+<a href=#799 id=799 data-nosnippet>799</a>
+<a href=#800 id=800 data-nosnippet>800</a>    <span class="kw">fn </span>get_selected_row_group_indices(
+<a href=#801 id=801 data-nosnippet>801</a>        predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#802 id=802 data-nosnippet>802</a>        parquet_metadata: <span class="kw-2">&amp;</span>Arc&lt;ParquetMetaData&gt;,
+<a href=#803 id=803 data-nosnippet>803</a>        field_id_map: <span class="kw-2">&amp;</span>HashMap&lt;i32, usize&gt;,
+<a href=#804 id=804 data-nosnippet>804</a>        snapshot_schema: <span class="kw-2">&amp;</span>Schema,
+<a href=#805 id=805 data-nosnippet>805</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;usize&gt;&gt; {
+<a href=#806 id=806 data-nosnippet>806</a>        <span class="kw">let </span>row_groups_metadata = parquet_metadata.row_groups();
+<a href=#807 id=807 data-nosnippet>807</a>        <span class="kw">let </span><span class="kw-2">mut </span>results = Vec::with_capacity(row_groups_metadata.len());
 <a href=#808 id=808 data-nosnippet>808</a>
-<a href=#809 id=809 data-nosnippet>809</a>        <span class="kw">let </span><span class="kw-2">mut </span>selected_row_groups_idx = <span class="number">0</span>;
-<a href=#810 id=810 data-nosnippet>810</a>
-<a href=#811 id=811 data-nosnippet>811</a>        <span class="kw">let </span>page_index = column_index
-<a href=#812 id=812 data-nosnippet>812</a>            .iter()
-<a href=#813 id=813 data-nosnippet>813</a>            .enumerate()
-<a href=#814 id=814 data-nosnippet>814</a>            .zip(offset_index)
-<a href=#815 id=815 data-nosnippet>815</a>            .zip(parquet_metadata.row_groups());
-<a href=#816 id=816 data-nosnippet>816</a>
-<a href=#817 id=817 data-nosnippet>817</a>        <span class="kw">let </span><span class="kw-2">mut </span>results = Vec::new();
-<a href=#818 id=818 data-nosnippet>818</a>        <span class="kw">for </span>(((idx, column_index), offset_index), row_group_metadata) <span class="kw">in </span>page_index {
-<a href=#819 id=819 data-nosnippet>819</a>            <span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_groups) = selected_row_groups {
-<a href=#820 id=820 data-nosnippet>820</a>                <span class="comment">// skip row groups that aren't present in selected_row_groups
-<a href=#821 id=821 data-nosnippet>821</a>                </span><span class="kw">if </span>idx == selected_row_groups[selected_row_groups_idx] {
-<a href=#822 id=822 data-nosnippet>822</a>                    selected_row_groups_idx += <span class="number">1</span>;
-<a href=#823 id=823 data-nosnippet>823</a>                } <span class="kw">else </span>{
-<a href=#824 id=824 data-nosnippet>824</a>                    <span class="kw">continue</span>;
-<a href=#825 id=825 data-nosnippet>825</a>                }
-<a href=#826 id=826 data-nosnippet>826</a>            }
-<a href=#827 id=827 data-nosnippet>827</a>
-<a href=#828 id=828 data-nosnippet>828</a>            <span class="kw">let </span>selections_for_page = PageIndexEvaluator::eval(
-<a href=#829 id=829 data-nosnippet>829</a>                predicate,
-<a href=#830 id=830 data-nosnippet>830</a>                column_index,
-<a href=#831 id=831 data-nosnippet>831</a>                offset_index,
-<a href=#832 id=832 data-nosnippet>832</a>                row_group_metadata,
-<a href=#833 id=833 data-nosnippet>833</a>                field_id_map,
-<a href=#834 id=834 data-nosnippet>834</a>                snapshot_schema,
-<a href=#835 id=835 data-nosnippet>835</a>            )<span class="question-mark">?</span>;
+<a href=#809 id=809 data-nosnippet>809</a>        <span class="kw">for </span>(idx, row_group_metadata) <span class="kw">in </span>row_groups_metadata.iter().enumerate() {
+<a href=#810 id=810 data-nosnippet>810</a>            <span class="kw">if </span>RowGroupMetricsEvaluator::eval(
+<a href=#811 id=811 data-nosnippet>811</a>                predicate,
+<a href=#812 id=812 data-nosnippet>812</a>                row_group_metadata,
+<a href=#813 id=813 data-nosnippet>813</a>                field_id_map,
+<a href=#814 id=814 data-nosnippet>814</a>                snapshot_schema,
+<a href=#815 id=815 data-nosnippet>815</a>            )<span class="question-mark">? </span>{
+<a href=#816 id=816 data-nosnippet>816</a>                results.push(idx);
+<a href=#817 id=817 data-nosnippet>817</a>            }
+<a href=#818 id=818 data-nosnippet>818</a>        }
+<a href=#819 id=819 data-nosnippet>819</a>
+<a href=#820 id=820 data-nosnippet>820</a>        <span class="prelude-val">Ok</span>(results)
+<a href=#821 id=821 data-nosnippet>821</a>    }
+<a href=#822 id=822 data-nosnippet>822</a>
+<a href=#823 id=823 data-nosnippet>823</a>    <span class="kw">fn </span>get_row_selection_for_filter_predicate(
+<a href=#824 id=824 data-nosnippet>824</a>        predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#825 id=825 data-nosnippet>825</a>        parquet_metadata: <span class="kw-2">&amp;</span>Arc&lt;ParquetMetaData&gt;,
+<a href=#826 id=826 data-nosnippet>826</a>        selected_row_groups: <span class="kw-2">&amp;</span><span class="prelude-ty">Option</span>&lt;Vec&lt;usize&gt;&gt;,
+<a href=#827 id=827 data-nosnippet>827</a>        field_id_map: <span class="kw-2">&amp;</span>HashMap&lt;i32, usize&gt;,
+<a href=#828 id=828 data-nosnippet>828</a>        snapshot_schema: <span class="kw-2">&amp;</span>Schema,
+<a href=#829 id=829 data-nosnippet>829</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;RowSelection&gt; {
+<a href=#830 id=830 data-nosnippet>830</a>        <span class="kw">let </span><span class="prelude-val">Some</span>(column_index) = parquet_metadata.column_index() <span class="kw">else </span>{
+<a href=#831 id=831 data-nosnippet>831</a>            <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#832 id=832 data-nosnippet>832</a>                ErrorKind::Unexpected,
+<a href=#833 id=833 data-nosnippet>833</a>                <span class="string">"Parquet file metadata does not contain a column index"</span>,
+<a href=#834 id=834 data-nosnippet>834</a>            ));
+<a href=#835 id=835 data-nosnippet>835</a>        };
 <a href=#836 id=836 data-nosnippet>836</a>
-<a href=#837 id=837 data-nosnippet>837</a>            results.push(selections_for_page);
-<a href=#838 id=838 data-nosnippet>838</a>
-<a href=#839 id=839 data-nosnippet>839</a>            <span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_groups) = selected_row_groups {
-<a href=#840 id=840 data-nosnippet>840</a>                <span class="kw">if </span>selected_row_groups_idx == selected_row_groups.len() {
-<a href=#841 id=841 data-nosnippet>841</a>                    <span class="kw">break</span>;
-<a href=#842 id=842 data-nosnippet>842</a>                }
-<a href=#843 id=843 data-nosnippet>843</a>            }
-<a href=#844 id=844 data-nosnippet>844</a>        }
-<a href=#845 id=845 data-nosnippet>845</a>
-<a href=#846 id=846 data-nosnippet>846</a>        <span class="prelude-val">Ok</span>(results.into_iter().flatten().collect::&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;().into())
-<a href=#847 id=847 data-nosnippet>847</a>    }
-<a href=#848 id=848 data-nosnippet>848</a>
-<a href=#849 id=849 data-nosnippet>849</a>    <span class="doccomment">/// Filters row groups by byte range to support Iceberg's file splitting.
-<a href=#850 id=850 data-nosnippet>850</a>    ///
-<a href=#851 id=851 data-nosnippet>851</a>    /// Iceberg splits large files at row group boundaries, so we only read row groups
-<a href=#852 id=852 data-nosnippet>852</a>    /// whose byte ranges overlap with [start, start+length).
-<a href=#853 id=853 data-nosnippet>853</a>    </span><span class="kw">fn </span>filter_row_groups_by_byte_range(
-<a href=#854 id=854 data-nosnippet>854</a>        parquet_metadata: <span class="kw-2">&amp;</span>Arc&lt;ParquetMetaData&gt;,
-<a href=#855 id=855 data-nosnippet>855</a>        start: u64,
-<a href=#856 id=856 data-nosnippet>856</a>        length: u64,
-<a href=#857 id=857 data-nosnippet>857</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;usize&gt;&gt; {
-<a href=#858 id=858 data-nosnippet>858</a>        <span class="kw">let </span>row_groups = parquet_metadata.row_groups();
-<a href=#859 id=859 data-nosnippet>859</a>        <span class="kw">let </span><span class="kw-2">mut </span>selected = Vec::new();
-<a href=#860 id=860 data-nosnippet>860</a>        <span class="kw">let </span>end = start + length;
-<a href=#861 id=861 data-nosnippet>861</a>
-<a href=#862 id=862 data-nosnippet>862</a>        <span class="comment">// Row groups are stored sequentially after the 4-byte magic header.
-<a href=#863 id=863 data-nosnippet>863</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>current_byte_offset = <span class="number">4u64</span>;
-<a href=#864 id=864 data-nosnippet>864</a>
-<a href=#865 id=865 data-nosnippet>865</a>        <span class="kw">for </span>(idx, row_group) <span class="kw">in </span>row_groups.iter().enumerate() {
-<a href=#866 id=866 data-nosnippet>866</a>            <span class="kw">let </span>row_group_size = row_group.compressed_size() <span class="kw">as </span>u64;
-<a href=#867 id=867 data-nosnippet>867</a>            <span class="kw">let </span>row_group_end = current_byte_offset + row_group_size;
-<a href=#868 id=868 data-nosnippet>868</a>
-<a href=#869 id=869 data-nosnippet>869</a>            <span class="kw">if </span>current_byte_offset &lt; end &amp;&amp; start &lt; row_group_end {
-<a href=#870 id=870 data-nosnippet>870</a>                selected.push(idx);
-<a href=#871 id=871 data-nosnippet>871</a>            }
-<a href=#872 id=872 data-nosnippet>872</a>
-<a href=#873 id=873 data-nosnippet>873</a>            current_byte_offset = row_group_end;
-<a href=#874 id=874 data-nosnippet>874</a>        }
-<a href=#875 id=875 data-nosnippet>875</a>
-<a href=#876 id=876 data-nosnippet>876</a>        <span class="prelude-val">Ok</span>(selected)
-<a href=#877 id=877 data-nosnippet>877</a>    }
-<a href=#878 id=878 data-nosnippet>878</a>}
-<a href=#879 id=879 data-nosnippet>879</a>
-<a href=#880 id=880 data-nosnippet>880</a><span class="doccomment">/// Build the map of parquet field id to Parquet column index in the schema.
-<a href=#881 id=881 data-nosnippet>881</a>/// Returns None if the Parquet file doesn't have field IDs embedded (e.g., migrated tables).
-<a href=#882 id=882 data-nosnippet>882</a></span><span class="kw">fn </span>build_field_id_map(parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor) -&gt; <span class="prelude-ty">Result</span>&lt;<span class="prelude-ty">Option</span>&lt;HashMap&lt;i32, usize&gt;&gt;&gt; {
-<a href=#883 id=883 data-nosnippet>883</a>    <span class="kw">let </span><span class="kw-2">mut </span>column_map = HashMap::new();
-<a href=#884 id=884 data-nosnippet>884</a>
-<a href=#885 id=885 data-nosnippet>885</a>    <span class="kw">for </span>(idx, field) <span class="kw">in </span>parquet_schema.columns().iter().enumerate() {
-<a href=#886 id=886 data-nosnippet>886</a>        <span class="kw">let </span>field_type = field.self_type();
-<a href=#887 id=887 data-nosnippet>887</a>        <span class="kw">match </span>field_type {
-<a href=#888 id=888 data-nosnippet>888</a>            ParquetType::PrimitiveType { basic_info, .. } =&gt; {
-<a href=#889 id=889 data-nosnippet>889</a>                <span class="kw">if </span>!basic_info.has_id() {
-<a href=#890 id=890 data-nosnippet>890</a>                    <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>);
-<a href=#891 id=891 data-nosnippet>891</a>                }
-<a href=#892 id=892 data-nosnippet>892</a>                column_map.insert(basic_info.id(), idx);
-<a href=#893 id=893 data-nosnippet>893</a>            }
-<a href=#894 id=894 data-nosnippet>894</a>            ParquetType::GroupType { .. } =&gt; {
-<a href=#895 id=895 data-nosnippet>895</a>                <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#896 id=896 data-nosnippet>896</a>                    ErrorKind::DataInvalid,
-<a href=#897 id=897 data-nosnippet>897</a>                    <span class="macro">format!</span>(
-<a href=#898 id=898 data-nosnippet>898</a>                        <span class="string">"Leave column in schema should be primitive type but got {field_type:?}"
-<a href=#899 id=899 data-nosnippet>899</a>                    </span>),
-<a href=#900 id=900 data-nosnippet>900</a>                ));
-<a href=#901 id=901 data-nosnippet>901</a>            }
-<a href=#902 id=902 data-nosnippet>902</a>        };
-<a href=#903 id=903 data-nosnippet>903</a>    }
-<a href=#904 id=904 data-nosnippet>904</a>
-<a href=#905 id=905 data-nosnippet>905</a>    <span class="prelude-val">Ok</span>(<span class="prelude-val">Some</span>(column_map))
-<a href=#906 id=906 data-nosnippet>906</a>}
-<a href=#907 id=907 data-nosnippet>907</a>
-<a href=#908 id=908 data-nosnippet>908</a><span class="doccomment">/// Build a fallback field ID map for Parquet files without embedded field IDs.
-<a href=#909 id=909 data-nosnippet>909</a>/// Position-based (1, 2, 3, ...) for compatibility with iceberg-java migrations.
-<a href=#910 id=910 data-nosnippet>910</a></span><span class="kw">fn </span>build_fallback_field_id_map(parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor) -&gt; HashMap&lt;i32, usize&gt; {
-<a href=#911 id=911 data-nosnippet>911</a>    <span class="kw">let </span><span class="kw-2">mut </span>column_map = HashMap::new();
-<a href=#912 id=912 data-nosnippet>912</a>
-<a href=#913 id=913 data-nosnippet>913</a>    <span class="comment">// 1-indexed to match iceberg-java's convention
-<a href=#914 id=914 data-nosnippet>914</a>    </span><span class="kw">for </span>(idx, _field) <span class="kw">in </span>parquet_schema.columns().iter().enumerate() {
-<a href=#915 id=915 data-nosnippet>915</a>        <span class="kw">let </span>field_id = (idx + <span class="number">1</span>) <span class="kw">as </span>i32;
-<a href=#916 id=916 data-nosnippet>916</a>        column_map.insert(field_id, idx);
-<a href=#917 id=917 data-nosnippet>917</a>    }
-<a href=#918 id=918 data-nosnippet>918</a>
-<a href=#919 id=919 data-nosnippet>919</a>    column_map
+<a href=#837 id=837 data-nosnippet>837</a>        <span class="kw">let </span><span class="prelude-val">Some</span>(offset_index) = parquet_metadata.offset_index() <span class="kw">else </span>{
+<a href=#838 id=838 data-nosnippet>838</a>            <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#839 id=839 data-nosnippet>839</a>                ErrorKind::Unexpected,
+<a href=#840 id=840 data-nosnippet>840</a>                <span class="string">"Parquet file metadata does not contain an offset index"</span>,
+<a href=#841 id=841 data-nosnippet>841</a>            ));
+<a href=#842 id=842 data-nosnippet>842</a>        };
+<a href=#843 id=843 data-nosnippet>843</a>
+<a href=#844 id=844 data-nosnippet>844</a>        <span class="comment">// If all row groups were filtered out, return an empty RowSelection (select no rows)
+<a href=#845 id=845 data-nosnippet>845</a>        </span><span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_groups) = selected_row_groups {
+<a href=#846 id=846 data-nosnippet>846</a>            <span class="kw">if </span>selected_row_groups.is_empty() {
+<a href=#847 id=847 data-nosnippet>847</a>                <span class="kw">return </span><span class="prelude-val">Ok</span>(RowSelection::from(Vec::new()));
+<a href=#848 id=848 data-nosnippet>848</a>            }
+<a href=#849 id=849 data-nosnippet>849</a>        }
+<a href=#850 id=850 data-nosnippet>850</a>
+<a href=#851 id=851 data-nosnippet>851</a>        <span class="kw">let </span><span class="kw-2">mut </span>selected_row_groups_idx = <span class="number">0</span>;
+<a href=#852 id=852 data-nosnippet>852</a>
+<a href=#853 id=853 data-nosnippet>853</a>        <span class="kw">let </span>page_index = column_index
+<a href=#854 id=854 data-nosnippet>854</a>            .iter()
+<a href=#855 id=855 data-nosnippet>855</a>            .enumerate()
+<a href=#856 id=856 data-nosnippet>856</a>            .zip(offset_index)
+<a href=#857 id=857 data-nosnippet>857</a>            .zip(parquet_metadata.row_groups());
+<a href=#858 id=858 data-nosnippet>858</a>
+<a href=#859 id=859 data-nosnippet>859</a>        <span class="kw">let </span><span class="kw-2">mut </span>results = Vec::new();
+<a href=#860 id=860 data-nosnippet>860</a>        <span class="kw">for </span>(((idx, column_index), offset_index), row_group_metadata) <span class="kw">in </span>page_index {
+<a href=#861 id=861 data-nosnippet>861</a>            <span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_groups) = selected_row_groups {
+<a href=#862 id=862 data-nosnippet>862</a>                <span class="comment">// skip row groups that aren't present in selected_row_groups
+<a href=#863 id=863 data-nosnippet>863</a>                </span><span class="kw">if </span>idx == selected_row_groups[selected_row_groups_idx] {
+<a href=#864 id=864 data-nosnippet>864</a>                    selected_row_groups_idx += <span class="number">1</span>;
+<a href=#865 id=865 data-nosnippet>865</a>                } <span class="kw">else </span>{
+<a href=#866 id=866 data-nosnippet>866</a>                    <span class="kw">continue</span>;
+<a href=#867 id=867 data-nosnippet>867</a>                }
+<a href=#868 id=868 data-nosnippet>868</a>            }
+<a href=#869 id=869 data-nosnippet>869</a>
+<a href=#870 id=870 data-nosnippet>870</a>            <span class="kw">let </span>selections_for_page = PageIndexEvaluator::eval(
+<a href=#871 id=871 data-nosnippet>871</a>                predicate,
+<a href=#872 id=872 data-nosnippet>872</a>                column_index,
+<a href=#873 id=873 data-nosnippet>873</a>                offset_index,
+<a href=#874 id=874 data-nosnippet>874</a>                row_group_metadata,
+<a href=#875 id=875 data-nosnippet>875</a>                field_id_map,
+<a href=#876 id=876 data-nosnippet>876</a>                snapshot_schema,
+<a href=#877 id=877 data-nosnippet>877</a>            )<span class="question-mark">?</span>;
+<a href=#878 id=878 data-nosnippet>878</a>
+<a href=#879 id=879 data-nosnippet>879</a>            results.push(selections_for_page);
+<a href=#880 id=880 data-nosnippet>880</a>
+<a href=#881 id=881 data-nosnippet>881</a>            <span class="kw">if let </span><span class="prelude-val">Some</span>(selected_row_groups) = selected_row_groups {
+<a href=#882 id=882 data-nosnippet>882</a>                <span class="kw">if </span>selected_row_groups_idx == selected_row_groups.len() {
+<a href=#883 id=883 data-nosnippet>883</a>                    <span class="kw">break</span>;
+<a href=#884 id=884 data-nosnippet>884</a>                }
+<a href=#885 id=885 data-nosnippet>885</a>            }
+<a href=#886 id=886 data-nosnippet>886</a>        }
+<a href=#887 id=887 data-nosnippet>887</a>
+<a href=#888 id=888 data-nosnippet>888</a>        <span class="prelude-val">Ok</span>(results.into_iter().flatten().collect::&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;().into())
+<a href=#889 id=889 data-nosnippet>889</a>    }
+<a href=#890 id=890 data-nosnippet>890</a>
+<a href=#891 id=891 data-nosnippet>891</a>    <span class="doccomment">/// Filters row groups by byte range to support Iceberg's file splitting.
+<a href=#892 id=892 data-nosnippet>892</a>    ///
+<a href=#893 id=893 data-nosnippet>893</a>    /// Iceberg splits large files at row group boundaries, so we only read row groups
+<a href=#894 id=894 data-nosnippet>894</a>    /// whose byte ranges overlap with [start, start+length).
+<a href=#895 id=895 data-nosnippet>895</a>    </span><span class="kw">fn </span>filter_row_groups_by_byte_range(
+<a href=#896 id=896 data-nosnippet>896</a>        parquet_metadata: <span class="kw-2">&amp;</span>Arc&lt;ParquetMetaData&gt;,
+<a href=#897 id=897 data-nosnippet>897</a>        start: u64,
+<a href=#898 id=898 data-nosnippet>898</a>        length: u64,
+<a href=#899 id=899 data-nosnippet>899</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;usize&gt;&gt; {
+<a href=#900 id=900 data-nosnippet>900</a>        <span class="kw">let </span>row_groups = parquet_metadata.row_groups();
+<a href=#901 id=901 data-nosnippet>901</a>        <span class="kw">let </span><span class="kw-2">mut </span>selected = Vec::new();
+<a href=#902 id=902 data-nosnippet>902</a>        <span class="kw">let </span>end = start + length;
+<a href=#903 id=903 data-nosnippet>903</a>
+<a href=#904 id=904 data-nosnippet>904</a>        <span class="comment">// Row groups are stored sequentially after the 4-byte magic header.
+<a href=#905 id=905 data-nosnippet>905</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>current_byte_offset = <span class="number">4u64</span>;
+<a href=#906 id=906 data-nosnippet>906</a>
+<a href=#907 id=907 data-nosnippet>907</a>        <span class="kw">for </span>(idx, row_group) <span class="kw">in </span>row_groups.iter().enumerate() {
+<a href=#908 id=908 data-nosnippet>908</a>            <span class="kw">let </span>row_group_size = row_group.compressed_size() <span class="kw">as </span>u64;
+<a href=#909 id=909 data-nosnippet>909</a>            <span class="kw">let </span>row_group_end = current_byte_offset + row_group_size;
+<a href=#910 id=910 data-nosnippet>910</a>
+<a href=#911 id=911 data-nosnippet>911</a>            <span class="kw">if </span>current_byte_offset &lt; end &amp;&amp; start &lt; row_group_end {
+<a href=#912 id=912 data-nosnippet>912</a>                selected.push(idx);
+<a href=#913 id=913 data-nosnippet>913</a>            }
+<a href=#914 id=914 data-nosnippet>914</a>
+<a href=#915 id=915 data-nosnippet>915</a>            current_byte_offset = row_group_end;
+<a href=#916 id=916 data-nosnippet>916</a>        }
+<a href=#917 id=917 data-nosnippet>917</a>
+<a href=#918 id=918 data-nosnippet>918</a>        <span class="prelude-val">Ok</span>(selected)
+<a href=#919 id=919 data-nosnippet>919</a>    }
 <a href=#920 id=920 data-nosnippet>920</a>}
 <a href=#921 id=921 data-nosnippet>921</a>
-<a href=#922 id=922 data-nosnippet>922</a><span class="doccomment">/// Add position-based fallback field IDs to Arrow schema for Parquet files lacking them.
-<a href=#923 id=923 data-nosnippet>923</a>/// Enables projection on migrated files (e.g., from Hive/Spark).
-<a href=#924 id=924 data-nosnippet>924</a>///
-<a href=#925 id=925 data-nosnippet>925</a>/// Why at schema level (not per-batch): Efficiency - avoids repeated schema modification.
-<a href=#926 id=926 data-nosnippet>926</a>/// Why only top-level: Nested projection uses leaf column indices, not parent struct IDs.
-<a href=#927 id=927 data-nosnippet>927</a>/// Why 1-indexed: Compatibility with iceberg-java's ParquetSchemaUtil.addFallbackIds().
-<a href=#928 id=928 data-nosnippet>928</a></span><span class="kw">fn </span>add_fallback_field_ids_to_arrow_schema(arrow_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef) -&gt; Arc&lt;ArrowSchema&gt; {
-<a href=#929 id=929 data-nosnippet>929</a>    <span class="macro">debug_assert!</span>(
-<a href=#930 id=930 data-nosnippet>930</a>        arrow_schema
-<a href=#931 id=931 data-nosnippet>931</a>            .fields()
-<a href=#932 id=932 data-nosnippet>932</a>            .iter()
-<a href=#933 id=933 data-nosnippet>933</a>            .next()
-<a href=#934 id=934 data-nosnippet>934</a>            .is_none_or(|f| f.metadata().get(PARQUET_FIELD_ID_META_KEY).is_none()),
-<a href=#935 id=935 data-nosnippet>935</a>        <span class="string">"Schema already has field IDs"
-<a href=#936 id=936 data-nosnippet>936</a>    </span>);
-<a href=#937 id=937 data-nosnippet>937</a>
-<a href=#938 id=938 data-nosnippet>938</a>    <span class="kw">use </span>arrow_schema::Field;
-<a href=#939 id=939 data-nosnippet>939</a>
-<a href=#940 id=940 data-nosnippet>940</a>    <span class="kw">let </span>fields_with_fallback_ids: Vec&lt;<span class="kw">_</span>&gt; = arrow_schema
-<a href=#941 id=941 data-nosnippet>941</a>        .fields()
-<a href=#942 id=942 data-nosnippet>942</a>        .iter()
-<a href=#943 id=943 data-nosnippet>943</a>        .enumerate()
-<a href=#944 id=944 data-nosnippet>944</a>        .map(|(pos, field)| {
-<a href=#945 id=945 data-nosnippet>945</a>            <span class="kw">let </span><span class="kw-2">mut </span>metadata = field.metadata().clone();
-<a href=#946 id=946 data-nosnippet>946</a>            <span class="kw">let </span>field_id = (pos + <span class="number">1</span>) <span class="kw">as </span>i32; <span class="comment">// 1-indexed for Java compatibility
-<a href=#947 id=947 data-nosnippet>947</a>            </span>metadata.insert(PARQUET_FIELD_ID_META_KEY.to_string(), field_id.to_string());
-<a href=#948 id=948 data-nosnippet>948</a>
-<a href=#949 id=949 data-nosnippet>949</a>            Field::new(field.name(), field.data_type().clone(), field.is_nullable())
-<a href=#950 id=950 data-nosnippet>950</a>                .with_metadata(metadata)
-<a href=#951 id=951 data-nosnippet>951</a>        })
-<a href=#952 id=952 data-nosnippet>952</a>        .collect();
-<a href=#953 id=953 data-nosnippet>953</a>
-<a href=#954 id=954 data-nosnippet>954</a>    Arc::new(ArrowSchema::new_with_metadata(
-<a href=#955 id=955 data-nosnippet>955</a>        fields_with_fallback_ids,
-<a href=#956 id=956 data-nosnippet>956</a>        arrow_schema.metadata().clone(),
-<a href=#957 id=957 data-nosnippet>957</a>    ))
-<a href=#958 id=958 data-nosnippet>958</a>}
-<a href=#959 id=959 data-nosnippet>959</a>
-<a href=#960 id=960 data-nosnippet>960</a><span class="doccomment">/// A visitor to collect field ids from bound predicates.
-<a href=#961 id=961 data-nosnippet>961</a></span><span class="kw">struct </span>CollectFieldIdVisitor {
-<a href=#962 id=962 data-nosnippet>962</a>    field_ids: HashSet&lt;i32&gt;,
-<a href=#963 id=963 data-nosnippet>963</a>}
-<a href=#964 id=964 data-nosnippet>964</a>
-<a href=#965 id=965 data-nosnippet>965</a><span class="kw">impl </span>CollectFieldIdVisitor {
-<a href=#966 id=966 data-nosnippet>966</a>    <span class="kw">fn </span>field_ids(<span class="self">self</span>) -&gt; HashSet&lt;i32&gt; {
-<a href=#967 id=967 data-nosnippet>967</a>        <span class="self">self</span>.field_ids
-<a href=#968 id=968 data-nosnippet>968</a>    }
-<a href=#969 id=969 data-nosnippet>969</a>}
-<a href=#970 id=970 data-nosnippet>970</a>
-<a href=#971 id=971 data-nosnippet>971</a><span class="kw">impl </span>BoundPredicateVisitor <span class="kw">for </span>CollectFieldIdVisitor {
-<a href=#972 id=972 data-nosnippet>972</a>    <span class="kw">type </span>T = ();
-<a href=#973 id=973 data-nosnippet>973</a>
-<a href=#974 id=974 data-nosnippet>974</a>    <span class="kw">fn </span>always_true(<span class="kw-2">&amp;mut </span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#975 id=975 data-nosnippet>975</a>        <span class="prelude-val">Ok</span>(())
-<a href=#976 id=976 data-nosnippet>976</a>    }
-<a href=#977 id=977 data-nosnippet>977</a>
-<a href=#978 id=978 data-nosnippet>978</a>    <span class="kw">fn </span>always_false(<span class="kw-2">&amp;mut </span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#979 id=979 data-nosnippet>979</a>        <span class="prelude-val">Ok</span>(())
-<a href=#980 id=980 data-nosnippet>980</a>    }
-<a href=#981 id=981 data-nosnippet>981</a>
-<a href=#982 id=982 data-nosnippet>982</a>    <span class="kw">fn </span>and(<span class="kw-2">&amp;mut </span><span class="self">self</span>, _lhs: (), _rhs: ()) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#983 id=983 data-nosnippet>983</a>        <span class="prelude-val">Ok</span>(())
-<a href=#984 id=984 data-nosnippet>984</a>    }
-<a href=#985 id=985 data-nosnippet>985</a>
-<a href=#986 id=986 data-nosnippet>986</a>    <span class="kw">fn </span>or(<span class="kw-2">&amp;mut </span><span class="self">self</span>, _lhs: (), _rhs: ()) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#987 id=987 data-nosnippet>987</a>        <span class="prelude-val">Ok</span>(())
-<a href=#988 id=988 data-nosnippet>988</a>    }
-<a href=#989 id=989 data-nosnippet>989</a>
-<a href=#990 id=990 data-nosnippet>990</a>    <span class="kw">fn </span>not(<span class="kw-2">&amp;mut </span><span class="self">self</span>, _inner: ()) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#991 id=991 data-nosnippet>991</a>        <span class="prelude-val">Ok</span>(())
-<a href=#992 id=992 data-nosnippet>992</a>    }
-<a href=#993 id=993 data-nosnippet>993</a>
-<a href=#994 id=994 data-nosnippet>994</a>    <span class="kw">fn </span>is_null(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference, _predicate: <span class="kw-2">&amp;</span>BoundPredicate) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#995 id=995 data-nosnippet>995</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#996 id=996 data-nosnippet>996</a>        <span class="prelude-val">Ok</span>(())
-<a href=#997 id=997 data-nosnippet>997</a>    }
-<a href=#998 id=998 data-nosnippet>998</a>
-<a href=#999 id=999 data-nosnippet>999</a>    <span class="kw">fn </span>not_null(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference, _predicate: <span class="kw-2">&amp;</span>BoundPredicate) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1000 id=1000 data-nosnippet>1000</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1001 id=1001 data-nosnippet>1001</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1002 id=1002 data-nosnippet>1002</a>    }
-<a href=#1003 id=1003 data-nosnippet>1003</a>
-<a href=#1004 id=1004 data-nosnippet>1004</a>    <span class="kw">fn </span>is_nan(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference, _predicate: <span class="kw-2">&amp;</span>BoundPredicate) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1005 id=1005 data-nosnippet>1005</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1006 id=1006 data-nosnippet>1006</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1007 id=1007 data-nosnippet>1007</a>    }
-<a href=#1008 id=1008 data-nosnippet>1008</a>
-<a href=#1009 id=1009 data-nosnippet>1009</a>    <span class="kw">fn </span>not_nan(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference, _predicate: <span class="kw-2">&amp;</span>BoundPredicate) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1010 id=1010 data-nosnippet>1010</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1011 id=1011 data-nosnippet>1011</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1012 id=1012 data-nosnippet>1012</a>    }
-<a href=#1013 id=1013 data-nosnippet>1013</a>
-<a href=#1014 id=1014 data-nosnippet>1014</a>    <span class="kw">fn </span>less_than(
-<a href=#1015 id=1015 data-nosnippet>1015</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1016 id=1016 data-nosnippet>1016</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1017 id=1017 data-nosnippet>1017</a>        _literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1018 id=1018 data-nosnippet>1018</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1019 id=1019 data-nosnippet>1019</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1020 id=1020 data-nosnippet>1020</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1021 id=1021 data-nosnippet>1021</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1022 id=1022 data-nosnippet>1022</a>    }
+<a href=#922 id=922 data-nosnippet>922</a><span class="doccomment">/// Build the map of parquet field id to Parquet column index in the schema.
+<a href=#923 id=923 data-nosnippet>923</a>/// Returns None if the Parquet file doesn't have field IDs embedded (e.g., migrated tables).
+<a href=#924 id=924 data-nosnippet>924</a></span><span class="kw">fn </span>build_field_id_map(parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor) -&gt; <span class="prelude-ty">Result</span>&lt;<span class="prelude-ty">Option</span>&lt;HashMap&lt;i32, usize&gt;&gt;&gt; {
+<a href=#925 id=925 data-nosnippet>925</a>    <span class="kw">let </span><span class="kw-2">mut </span>column_map = HashMap::new();
+<a href=#926 id=926 data-nosnippet>926</a>
+<a href=#927 id=927 data-nosnippet>927</a>    <span class="kw">for </span>(idx, field) <span class="kw">in </span>parquet_schema.columns().iter().enumerate() {
+<a href=#928 id=928 data-nosnippet>928</a>        <span class="kw">let </span>field_type = field.self_type();
+<a href=#929 id=929 data-nosnippet>929</a>        <span class="kw">match </span>field_type {
+<a href=#930 id=930 data-nosnippet>930</a>            ParquetType::PrimitiveType { basic_info, .. } =&gt; {
+<a href=#931 id=931 data-nosnippet>931</a>                <span class="kw">if </span>!basic_info.has_id() {
+<a href=#932 id=932 data-nosnippet>932</a>                    <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>);
+<a href=#933 id=933 data-nosnippet>933</a>                }
+<a href=#934 id=934 data-nosnippet>934</a>                column_map.insert(basic_info.id(), idx);
+<a href=#935 id=935 data-nosnippet>935</a>            }
+<a href=#936 id=936 data-nosnippet>936</a>            ParquetType::GroupType { .. } =&gt; {
+<a href=#937 id=937 data-nosnippet>937</a>                <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#938 id=938 data-nosnippet>938</a>                    ErrorKind::DataInvalid,
+<a href=#939 id=939 data-nosnippet>939</a>                    <span class="macro">format!</span>(
+<a href=#940 id=940 data-nosnippet>940</a>                        <span class="string">"Leave column in schema should be primitive type but got {field_type:?}"
+<a href=#941 id=941 data-nosnippet>941</a>                    </span>),
+<a href=#942 id=942 data-nosnippet>942</a>                ));
+<a href=#943 id=943 data-nosnippet>943</a>            }
+<a href=#944 id=944 data-nosnippet>944</a>        };
+<a href=#945 id=945 data-nosnippet>945</a>    }
+<a href=#946 id=946 data-nosnippet>946</a>
+<a href=#947 id=947 data-nosnippet>947</a>    <span class="prelude-val">Ok</span>(<span class="prelude-val">Some</span>(column_map))
+<a href=#948 id=948 data-nosnippet>948</a>}
+<a href=#949 id=949 data-nosnippet>949</a>
+<a href=#950 id=950 data-nosnippet>950</a><span class="doccomment">/// Build a fallback field ID map for Parquet files without embedded field IDs.
+<a href=#951 id=951 data-nosnippet>951</a>/// Position-based (1, 2, 3, ...) for compatibility with iceberg-java migrations.
+<a href=#952 id=952 data-nosnippet>952</a></span><span class="kw">fn </span>build_fallback_field_id_map(parquet_schema: <span class="kw-2">&amp;</span>SchemaDescriptor) -&gt; HashMap&lt;i32, usize&gt; {
+<a href=#953 id=953 data-nosnippet>953</a>    <span class="kw">let </span><span class="kw-2">mut </span>column_map = HashMap::new();
+<a href=#954 id=954 data-nosnippet>954</a>
+<a href=#955 id=955 data-nosnippet>955</a>    <span class="comment">// 1-indexed to match iceberg-java's convention
+<a href=#956 id=956 data-nosnippet>956</a>    </span><span class="kw">for </span>(idx, _field) <span class="kw">in </span>parquet_schema.columns().iter().enumerate() {
+<a href=#957 id=957 data-nosnippet>957</a>        <span class="kw">let </span>field_id = (idx + <span class="number">1</span>) <span class="kw">as </span>i32;
+<a href=#958 id=958 data-nosnippet>958</a>        column_map.insert(field_id, idx);
+<a href=#959 id=959 data-nosnippet>959</a>    }
+<a href=#960 id=960 data-nosnippet>960</a>
+<a href=#961 id=961 data-nosnippet>961</a>    column_map
+<a href=#962 id=962 data-nosnippet>962</a>}
+<a href=#963 id=963 data-nosnippet>963</a>
+<a href=#964 id=964 data-nosnippet>964</a><span class="doccomment">/// Apply name mapping to Arrow schema for Parquet files lacking field IDs.
+<a href=#965 id=965 data-nosnippet>965</a>///
+<a href=#966 id=966 data-nosnippet>966</a>/// Assigns Iceberg field IDs based on column names using the name mapping,
+<a href=#967 id=967 data-nosnippet>967</a>/// enabling correct projection on migrated files (e.g., from Hive/Spark via add_files).
+<a href=#968 id=968 data-nosnippet>968</a>///
+<a href=#969 id=969 data-nosnippet>969</a>/// Per Iceberg spec Column Projection rule #2:
+<a href=#970 id=970 data-nosnippet>970</a>/// "Use schema.name-mapping.default metadata to map field id to columns without field id"
+<a href=#971 id=971 data-nosnippet>971</a>/// https://iceberg.apache.org/spec/#column-projection
+<a href=#972 id=972 data-nosnippet>972</a>///
+<a href=#973 id=973 data-nosnippet>973</a>/// Corresponds to Java's ParquetSchemaUtil.applyNameMapping() and ApplyNameMapping visitor.
+<a href=#974 id=974 data-nosnippet>974</a>/// The key difference is Java operates on Parquet MessageType, while we operate on Arrow Schema.
+<a href=#975 id=975 data-nosnippet>975</a>///
+<a href=#976 id=976 data-nosnippet>976</a>/// # Arguments
+<a href=#977 id=977 data-nosnippet>977</a>/// * `arrow_schema` - Arrow schema from Parquet file (without field IDs)
+<a href=#978 id=978 data-nosnippet>978</a>/// * `name_mapping` - Name mapping from table metadata (TableProperties.DEFAULT_NAME_MAPPING)
+<a href=#979 id=979 data-nosnippet>979</a>///
+<a href=#980 id=980 data-nosnippet>980</a>/// # Returns
+<a href=#981 id=981 data-nosnippet>981</a>/// Arrow schema with field IDs assigned based on name mapping
+<a href=#982 id=982 data-nosnippet>982</a></span><span class="kw">fn </span>apply_name_mapping_to_arrow_schema(
+<a href=#983 id=983 data-nosnippet>983</a>    arrow_schema: ArrowSchemaRef,
+<a href=#984 id=984 data-nosnippet>984</a>    name_mapping: <span class="kw-2">&amp;</span>NameMapping,
+<a href=#985 id=985 data-nosnippet>985</a>) -&gt; <span class="prelude-ty">Result</span>&lt;Arc&lt;ArrowSchema&gt;&gt; {
+<a href=#986 id=986 data-nosnippet>986</a>    <span class="macro">debug_assert!</span>(
+<a href=#987 id=987 data-nosnippet>987</a>        arrow_schema
+<a href=#988 id=988 data-nosnippet>988</a>            .fields()
+<a href=#989 id=989 data-nosnippet>989</a>            .iter()
+<a href=#990 id=990 data-nosnippet>990</a>            .next()
+<a href=#991 id=991 data-nosnippet>991</a>            .is_none_or(|f| f.metadata().get(PARQUET_FIELD_ID_META_KEY).is_none()),
+<a href=#992 id=992 data-nosnippet>992</a>        <span class="string">"Schema already has field IDs - name mapping should not be applied"
+<a href=#993 id=993 data-nosnippet>993</a>    </span>);
+<a href=#994 id=994 data-nosnippet>994</a>
+<a href=#995 id=995 data-nosnippet>995</a>    <span class="kw">use </span>arrow_schema::Field;
+<a href=#996 id=996 data-nosnippet>996</a>
+<a href=#997 id=997 data-nosnippet>997</a>    <span class="kw">let </span>fields_with_mapped_ids: Vec&lt;<span class="kw">_</span>&gt; = arrow_schema
+<a href=#998 id=998 data-nosnippet>998</a>        .fields()
+<a href=#999 id=999 data-nosnippet>999</a>        .iter()
+<a href=#1000 id=1000 data-nosnippet>1000</a>        .map(|field| {
+<a href=#1001 id=1001 data-nosnippet>1001</a>            <span class="comment">// Look up this column name in name mapping to get the Iceberg field ID.
+<a href=#1002 id=1002 data-nosnippet>1002</a>            // Corresponds to Java's ApplyNameMapping visitor which calls
+<a href=#1003 id=1003 data-nosnippet>1003</a>            // nameMapping.find(currentPath()) and returns field.withId() if found.
+<a href=#1004 id=1004 data-nosnippet>1004</a>            //
+<a href=#1005 id=1005 data-nosnippet>1005</a>            // If the field isn't in the mapping, leave it WITHOUT assigning an ID
+<a href=#1006 id=1006 data-nosnippet>1006</a>            // (matching Java's behavior of returning the field unchanged).
+<a href=#1007 id=1007 data-nosnippet>1007</a>            // Later, during projection, fields without IDs are filtered out.
+<a href=#1008 id=1008 data-nosnippet>1008</a>            </span><span class="kw">let </span>mapped_field_opt = name_mapping
+<a href=#1009 id=1009 data-nosnippet>1009</a>                .fields()
+<a href=#1010 id=1010 data-nosnippet>1010</a>                .iter()
+<a href=#1011 id=1011 data-nosnippet>1011</a>                .find(|f| f.names().contains(<span class="kw-2">&amp;</span>field.name().to_string()));
+<a href=#1012 id=1012 data-nosnippet>1012</a>
+<a href=#1013 id=1013 data-nosnippet>1013</a>            <span class="kw">let </span><span class="kw-2">mut </span>metadata = field.metadata().clone();
+<a href=#1014 id=1014 data-nosnippet>1014</a>
+<a href=#1015 id=1015 data-nosnippet>1015</a>            <span class="kw">if let </span><span class="prelude-val">Some</span>(mapped_field) = mapped_field_opt {
+<a href=#1016 id=1016 data-nosnippet>1016</a>                <span class="kw">if let </span><span class="prelude-val">Some</span>(field_id) = mapped_field.field_id() {
+<a href=#1017 id=1017 data-nosnippet>1017</a>                    <span class="comment">// Field found in mapping with a field_id → assign it
+<a href=#1018 id=1018 data-nosnippet>1018</a>                    </span>metadata.insert(PARQUET_FIELD_ID_META_KEY.to_string(), field_id.to_string());
+<a href=#1019 id=1019 data-nosnippet>1019</a>                }
+<a href=#1020 id=1020 data-nosnippet>1020</a>                <span class="comment">// If field_id is None, leave the field without an ID (will be filtered by projection)
+<a href=#1021 id=1021 data-nosnippet>1021</a>            </span>}
+<a href=#1022 id=1022 data-nosnippet>1022</a>            <span class="comment">// If field not found in mapping, leave it without an ID (will be filtered by projection)
 <a href=#1023 id=1023 data-nosnippet>1023</a>
-<a href=#1024 id=1024 data-nosnippet>1024</a>    <span class="kw">fn </span>less_than_or_eq(
-<a href=#1025 id=1025 data-nosnippet>1025</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1026 id=1026 data-nosnippet>1026</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1027 id=1027 data-nosnippet>1027</a>        _literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1028 id=1028 data-nosnippet>1028</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1029 id=1029 data-nosnippet>1029</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1030 id=1030 data-nosnippet>1030</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1031 id=1031 data-nosnippet>1031</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1032 id=1032 data-nosnippet>1032</a>    }
-<a href=#1033 id=1033 data-nosnippet>1033</a>
-<a href=#1034 id=1034 data-nosnippet>1034</a>    <span class="kw">fn </span>greater_than(
-<a href=#1035 id=1035 data-nosnippet>1035</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1036 id=1036 data-nosnippet>1036</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1037 id=1037 data-nosnippet>1037</a>        _literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1038 id=1038 data-nosnippet>1038</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1039 id=1039 data-nosnippet>1039</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1040 id=1040 data-nosnippet>1040</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1041 id=1041 data-nosnippet>1041</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1042 id=1042 data-nosnippet>1042</a>    }
-<a href=#1043 id=1043 data-nosnippet>1043</a>
-<a href=#1044 id=1044 data-nosnippet>1044</a>    <span class="kw">fn </span>greater_than_or_eq(
-<a href=#1045 id=1045 data-nosnippet>1045</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1046 id=1046 data-nosnippet>1046</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1047 id=1047 data-nosnippet>1047</a>        _literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1048 id=1048 data-nosnippet>1048</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1049 id=1049 data-nosnippet>1049</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1050 id=1050 data-nosnippet>1050</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1051 id=1051 data-nosnippet>1051</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1052 id=1052 data-nosnippet>1052</a>    }
-<a href=#1053 id=1053 data-nosnippet>1053</a>
-<a href=#1054 id=1054 data-nosnippet>1054</a>    <span class="kw">fn </span>eq(
-<a href=#1055 id=1055 data-nosnippet>1055</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1056 id=1056 data-nosnippet>1056</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1057 id=1057 data-nosnippet>1057</a>        _literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1058 id=1058 data-nosnippet>1058</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1059 id=1059 data-nosnippet>1059</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1060 id=1060 data-nosnippet>1060</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1061 id=1061 data-nosnippet>1061</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1062 id=1062 data-nosnippet>1062</a>    }
-<a href=#1063 id=1063 data-nosnippet>1063</a>
-<a href=#1064 id=1064 data-nosnippet>1064</a>    <span class="kw">fn </span>not_eq(
-<a href=#1065 id=1065 data-nosnippet>1065</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1066 id=1066 data-nosnippet>1066</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1067 id=1067 data-nosnippet>1067</a>        _literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1068 id=1068 data-nosnippet>1068</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1069 id=1069 data-nosnippet>1069</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1070 id=1070 data-nosnippet>1070</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1071 id=1071 data-nosnippet>1071</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1072 id=1072 data-nosnippet>1072</a>    }
-<a href=#1073 id=1073 data-nosnippet>1073</a>
-<a href=#1074 id=1074 data-nosnippet>1074</a>    <span class="kw">fn </span>starts_with(
-<a href=#1075 id=1075 data-nosnippet>1075</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1076 id=1076 data-nosnippet>1076</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1077 id=1077 data-nosnippet>1077</a>        _literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1078 id=1078 data-nosnippet>1078</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1079 id=1079 data-nosnippet>1079</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1080 id=1080 data-nosnippet>1080</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1081 id=1081 data-nosnippet>1081</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1082 id=1082 data-nosnippet>1082</a>    }
+<a href=#1024 id=1024 data-nosnippet>1024</a>            </span>Field::new(field.name(), field.data_type().clone(), field.is_nullable())
+<a href=#1025 id=1025 data-nosnippet>1025</a>                .with_metadata(metadata)
+<a href=#1026 id=1026 data-nosnippet>1026</a>        })
+<a href=#1027 id=1027 data-nosnippet>1027</a>        .collect();
+<a href=#1028 id=1028 data-nosnippet>1028</a>
+<a href=#1029 id=1029 data-nosnippet>1029</a>    <span class="prelude-val">Ok</span>(Arc::new(ArrowSchema::new_with_metadata(
+<a href=#1030 id=1030 data-nosnippet>1030</a>        fields_with_mapped_ids,
+<a href=#1031 id=1031 data-nosnippet>1031</a>        arrow_schema.metadata().clone(),
+<a href=#1032 id=1032 data-nosnippet>1032</a>    )))
+<a href=#1033 id=1033 data-nosnippet>1033</a>}
+<a href=#1034 id=1034 data-nosnippet>1034</a>
+<a href=#1035 id=1035 data-nosnippet>1035</a><span class="doccomment">/// Add position-based fallback field IDs to Arrow schema for Parquet files lacking them.
+<a href=#1036 id=1036 data-nosnippet>1036</a>/// Enables projection on migrated files (e.g., from Hive/Spark).
+<a href=#1037 id=1037 data-nosnippet>1037</a>///
+<a href=#1038 id=1038 data-nosnippet>1038</a>/// Why at schema level (not per-batch): Efficiency - avoids repeated schema modification.
+<a href=#1039 id=1039 data-nosnippet>1039</a>/// Why only top-level: Nested projection uses leaf column indices, not parent struct IDs.
+<a href=#1040 id=1040 data-nosnippet>1040</a>/// Why 1-indexed: Compatibility with iceberg-java's ParquetSchemaUtil.addFallbackIds().
+<a href=#1041 id=1041 data-nosnippet>1041</a></span><span class="kw">fn </span>add_fallback_field_ids_to_arrow_schema(arrow_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef) -&gt; Arc&lt;ArrowSchema&gt; {
+<a href=#1042 id=1042 data-nosnippet>1042</a>    <span class="macro">debug_assert!</span>(
+<a href=#1043 id=1043 data-nosnippet>1043</a>        arrow_schema
+<a href=#1044 id=1044 data-nosnippet>1044</a>            .fields()
+<a href=#1045 id=1045 data-nosnippet>1045</a>            .iter()
+<a href=#1046 id=1046 data-nosnippet>1046</a>            .next()
+<a href=#1047 id=1047 data-nosnippet>1047</a>            .is_none_or(|f| f.metadata().get(PARQUET_FIELD_ID_META_KEY).is_none()),
+<a href=#1048 id=1048 data-nosnippet>1048</a>        <span class="string">"Schema already has field IDs"
+<a href=#1049 id=1049 data-nosnippet>1049</a>    </span>);
+<a href=#1050 id=1050 data-nosnippet>1050</a>
+<a href=#1051 id=1051 data-nosnippet>1051</a>    <span class="kw">use </span>arrow_schema::Field;
+<a href=#1052 id=1052 data-nosnippet>1052</a>
+<a href=#1053 id=1053 data-nosnippet>1053</a>    <span class="kw">let </span>fields_with_fallback_ids: Vec&lt;<span class="kw">_</span>&gt; = arrow_schema
+<a href=#1054 id=1054 data-nosnippet>1054</a>        .fields()
+<a href=#1055 id=1055 data-nosnippet>1055</a>        .iter()
+<a href=#1056 id=1056 data-nosnippet>1056</a>        .enumerate()
+<a href=#1057 id=1057 data-nosnippet>1057</a>        .map(|(pos, field)| {
+<a href=#1058 id=1058 data-nosnippet>1058</a>            <span class="kw">let </span><span class="kw-2">mut </span>metadata = field.metadata().clone();
+<a href=#1059 id=1059 data-nosnippet>1059</a>            <span class="kw">let </span>field_id = (pos + <span class="number">1</span>) <span class="kw">as </span>i32; <span class="comment">// 1-indexed for Java compatibility
+<a href=#1060 id=1060 data-nosnippet>1060</a>            </span>metadata.insert(PARQUET_FIELD_ID_META_KEY.to_string(), field_id.to_string());
+<a href=#1061 id=1061 data-nosnippet>1061</a>
+<a href=#1062 id=1062 data-nosnippet>1062</a>            Field::new(field.name(), field.data_type().clone(), field.is_nullable())
+<a href=#1063 id=1063 data-nosnippet>1063</a>                .with_metadata(metadata)
+<a href=#1064 id=1064 data-nosnippet>1064</a>        })
+<a href=#1065 id=1065 data-nosnippet>1065</a>        .collect();
+<a href=#1066 id=1066 data-nosnippet>1066</a>
+<a href=#1067 id=1067 data-nosnippet>1067</a>    Arc::new(ArrowSchema::new_with_metadata(
+<a href=#1068 id=1068 data-nosnippet>1068</a>        fields_with_fallback_ids,
+<a href=#1069 id=1069 data-nosnippet>1069</a>        arrow_schema.metadata().clone(),
+<a href=#1070 id=1070 data-nosnippet>1070</a>    ))
+<a href=#1071 id=1071 data-nosnippet>1071</a>}
+<a href=#1072 id=1072 data-nosnippet>1072</a>
+<a href=#1073 id=1073 data-nosnippet>1073</a><span class="doccomment">/// A visitor to collect field ids from bound predicates.
+<a href=#1074 id=1074 data-nosnippet>1074</a></span><span class="kw">struct </span>CollectFieldIdVisitor {
+<a href=#1075 id=1075 data-nosnippet>1075</a>    field_ids: HashSet&lt;i32&gt;,
+<a href=#1076 id=1076 data-nosnippet>1076</a>}
+<a href=#1077 id=1077 data-nosnippet>1077</a>
+<a href=#1078 id=1078 data-nosnippet>1078</a><span class="kw">impl </span>CollectFieldIdVisitor {
+<a href=#1079 id=1079 data-nosnippet>1079</a>    <span class="kw">fn </span>field_ids(<span class="self">self</span>) -&gt; HashSet&lt;i32&gt; {
+<a href=#1080 id=1080 data-nosnippet>1080</a>        <span class="self">self</span>.field_ids
+<a href=#1081 id=1081 data-nosnippet>1081</a>    }
+<a href=#1082 id=1082 data-nosnippet>1082</a>}
 <a href=#1083 id=1083 data-nosnippet>1083</a>
-<a href=#1084 id=1084 data-nosnippet>1084</a>    <span class="kw">fn </span>not_starts_with(
-<a href=#1085 id=1085 data-nosnippet>1085</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1086 id=1086 data-nosnippet>1086</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1087 id=1087 data-nosnippet>1087</a>        _literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1088 id=1088 data-nosnippet>1088</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1089 id=1089 data-nosnippet>1089</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1090 id=1090 data-nosnippet>1090</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1091 id=1091 data-nosnippet>1091</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1092 id=1092 data-nosnippet>1092</a>    }
-<a href=#1093 id=1093 data-nosnippet>1093</a>
-<a href=#1094 id=1094 data-nosnippet>1094</a>    <span class="kw">fn </span>r#in(
-<a href=#1095 id=1095 data-nosnippet>1095</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1096 id=1096 data-nosnippet>1096</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1097 id=1097 data-nosnippet>1097</a>        _literals: <span class="kw-2">&amp;</span>FnvHashSet&lt;Datum&gt;,
-<a href=#1098 id=1098 data-nosnippet>1098</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1099 id=1099 data-nosnippet>1099</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1100 id=1100 data-nosnippet>1100</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1101 id=1101 data-nosnippet>1101</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1102 id=1102 data-nosnippet>1102</a>    }
-<a href=#1103 id=1103 data-nosnippet>1103</a>
-<a href=#1104 id=1104 data-nosnippet>1104</a>    <span class="kw">fn </span>not_in(
-<a href=#1105 id=1105 data-nosnippet>1105</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1106 id=1106 data-nosnippet>1106</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1107 id=1107 data-nosnippet>1107</a>        _literals: <span class="kw-2">&amp;</span>FnvHashSet&lt;Datum&gt;,
-<a href=#1108 id=1108 data-nosnippet>1108</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1109 id=1109 data-nosnippet>1109</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
-<a href=#1110 id=1110 data-nosnippet>1110</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
-<a href=#1111 id=1111 data-nosnippet>1111</a>        <span class="prelude-val">Ok</span>(())
-<a href=#1112 id=1112 data-nosnippet>1112</a>    }
-<a href=#1113 id=1113 data-nosnippet>1113</a>}
-<a href=#1114 id=1114 data-nosnippet>1114</a>
-<a href=#1115 id=1115 data-nosnippet>1115</a><span class="doccomment">/// A visitor to convert Iceberg bound predicates to Arrow predicates.
-<a href=#1116 id=1116 data-nosnippet>1116</a></span><span class="kw">struct </span>PredicateConverter&lt;<span class="lifetime">'a</span>&gt; {
-<a href=#1117 id=1117 data-nosnippet>1117</a>    <span class="doccomment">/// The Parquet schema descriptor.
-<a href=#1118 id=1118 data-nosnippet>1118</a>    </span><span class="kw">pub </span>parquet_schema: <span class="kw-2">&amp;</span><span class="lifetime">'a </span>SchemaDescriptor,
-<a href=#1119 id=1119 data-nosnippet>1119</a>    <span class="doccomment">/// The map between field id and leaf column index in Parquet schema.
-<a href=#1120 id=1120 data-nosnippet>1120</a>    </span><span class="kw">pub </span>column_map: <span class="kw-2">&amp;</span><span class="lifetime">'a </span>HashMap&lt;i32, usize&gt;,
-<a href=#1121 id=1121 data-nosnippet>1121</a>    <span class="doccomment">/// The required column indices in Parquet schema for the predicates.
-<a href=#1122 id=1122 data-nosnippet>1122</a>    </span><span class="kw">pub </span>column_indices: <span class="kw-2">&amp;</span><span class="lifetime">'a </span>Vec&lt;usize&gt;,
-<a href=#1123 id=1123 data-nosnippet>1123</a>}
-<a href=#1124 id=1124 data-nosnippet>1124</a>
-<a href=#1125 id=1125 data-nosnippet>1125</a><span class="kw">impl </span>PredicateConverter&lt;<span class="lifetime">'_</span>&gt; {
-<a href=#1126 id=1126 data-nosnippet>1126</a>    <span class="doccomment">/// When visiting a bound reference, we return index of the leaf column in the
-<a href=#1127 id=1127 data-nosnippet>1127</a>    /// required column indices which is used to project the column in the record batch.
-<a href=#1128 id=1128 data-nosnippet>1128</a>    /// Return None if the field id is not found in the column map, which is possible
-<a href=#1129 id=1129 data-nosnippet>1129</a>    /// due to schema evolution.
-<a href=#1130 id=1130 data-nosnippet>1130</a>    </span><span class="kw">fn </span>bound_reference(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference) -&gt; <span class="prelude-ty">Result</span>&lt;<span class="prelude-ty">Option</span>&lt;usize&gt;&gt; {
-<a href=#1131 id=1131 data-nosnippet>1131</a>        <span class="comment">// The leaf column's index in Parquet schema.
-<a href=#1132 id=1132 data-nosnippet>1132</a>        </span><span class="kw">if let </span><span class="prelude-val">Some</span>(column_idx) = <span class="self">self</span>.column_map.get(<span class="kw-2">&amp;</span>reference.field().id) {
-<a href=#1133 id=1133 data-nosnippet>1133</a>            <span class="kw">if </span><span class="self">self</span>.parquet_schema.get_column_root(<span class="kw-2">*</span>column_idx).is_group() {
-<a href=#1134 id=1134 data-nosnippet>1134</a>                <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#1135 id=1135 data-nosnippet>1135</a>                    ErrorKind::DataInvalid,
-<a href=#1136 id=1136 data-nosnippet>1136</a>                    <span class="macro">format!</span>(
-<a href=#1137 id=1137 data-nosnippet>1137</a>                        <span class="string">"Leave column `{}` in predicates isn't a root column in Parquet schema."</span>,
-<a href=#1138 id=1138 data-nosnippet>1138</a>                        reference.field().name
-<a href=#1139 id=1139 data-nosnippet>1139</a>                    ),
-<a href=#1140 id=1140 data-nosnippet>1140</a>                ));
-<a href=#1141 id=1141 data-nosnippet>1141</a>            }
-<a href=#1142 id=1142 data-nosnippet>1142</a>
-<a href=#1143 id=1143 data-nosnippet>1143</a>            <span class="comment">// The leaf column's index in the required column indices.
-<a href=#1144 id=1144 data-nosnippet>1144</a>            </span><span class="kw">let </span>index = <span class="self">self
-<a href=#1145 id=1145 data-nosnippet>1145</a>                </span>.column_indices
-<a href=#1146 id=1146 data-nosnippet>1146</a>                .iter()
-<a href=#1147 id=1147 data-nosnippet>1147</a>                .position(|<span class="kw-2">&amp;</span>idx| idx == <span class="kw-2">*</span>column_idx)
-<a href=#1148 id=1148 data-nosnippet>1148</a>                .ok_or(Error::new(
-<a href=#1149 id=1149 data-nosnippet>1149</a>                    ErrorKind::DataInvalid,
-<a href=#1150 id=1150 data-nosnippet>1150</a>                    <span class="macro">format!</span>(
-<a href=#1151 id=1151 data-nosnippet>1151</a>                <span class="string">"Leave column `{}` in predicates cannot be found in the required column indices."</span>,
-<a href=#1152 id=1152 data-nosnippet>1152</a>                reference.field().name
-<a href=#1153 id=1153 data-nosnippet>1153</a>            ),
-<a href=#1154 id=1154 data-nosnippet>1154</a>                ))<span class="question-mark">?</span>;
-<a href=#1155 id=1155 data-nosnippet>1155</a>
-<a href=#1156 id=1156 data-nosnippet>1156</a>            <span class="prelude-val">Ok</span>(<span class="prelude-val">Some</span>(index))
-<a href=#1157 id=1157 data-nosnippet>1157</a>        } <span class="kw">else </span>{
-<a href=#1158 id=1158 data-nosnippet>1158</a>            <span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>)
-<a href=#1159 id=1159 data-nosnippet>1159</a>        }
-<a href=#1160 id=1160 data-nosnippet>1160</a>    }
-<a href=#1161 id=1161 data-nosnippet>1161</a>
-<a href=#1162 id=1162 data-nosnippet>1162</a>    <span class="doccomment">/// Build an Arrow predicate that always returns true.
-<a href=#1163 id=1163 data-nosnippet>1163</a>    </span><span class="kw">fn </span>build_always_true(<span class="kw-2">&amp;</span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1164 id=1164 data-nosnippet>1164</a>        <span class="prelude-val">Ok</span>(Box::new(|batch| {
-<a href=#1165 id=1165 data-nosnippet>1165</a>            <span class="prelude-val">Ok</span>(BooleanArray::from(<span class="macro">vec!</span>[<span class="bool-val">true</span>; batch.num_rows()]))
-<a href=#1166 id=1166 data-nosnippet>1166</a>        }))
-<a href=#1167 id=1167 data-nosnippet>1167</a>    }
-<a href=#1168 id=1168 data-nosnippet>1168</a>
-<a href=#1169 id=1169 data-nosnippet>1169</a>    <span class="doccomment">/// Build an Arrow predicate that always returns false.
-<a href=#1170 id=1170 data-nosnippet>1170</a>    </span><span class="kw">fn </span>build_always_false(<span class="kw-2">&amp;</span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1171 id=1171 data-nosnippet>1171</a>        <span class="prelude-val">Ok</span>(Box::new(|batch| {
-<a href=#1172 id=1172 data-nosnippet>1172</a>            <span class="prelude-val">Ok</span>(BooleanArray::from(<span class="macro">vec!</span>[<span class="bool-val">false</span>; batch.num_rows()]))
-<a href=#1173 id=1173 data-nosnippet>1173</a>        }))
-<a href=#1174 id=1174 data-nosnippet>1174</a>    }
-<a href=#1175 id=1175 data-nosnippet>1175</a>}
+<a href=#1084 id=1084 data-nosnippet>1084</a><span class="kw">impl </span>BoundPredicateVisitor <span class="kw">for </span>CollectFieldIdVisitor {
+<a href=#1085 id=1085 data-nosnippet>1085</a>    <span class="kw">type </span>T = ();
+<a href=#1086 id=1086 data-nosnippet>1086</a>
+<a href=#1087 id=1087 data-nosnippet>1087</a>    <span class="kw">fn </span>always_true(<span class="kw-2">&amp;mut </span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1088 id=1088 data-nosnippet>1088</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1089 id=1089 data-nosnippet>1089</a>    }
+<a href=#1090 id=1090 data-nosnippet>1090</a>
+<a href=#1091 id=1091 data-nosnippet>1091</a>    <span class="kw">fn </span>always_false(<span class="kw-2">&amp;mut </span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1092 id=1092 data-nosnippet>1092</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1093 id=1093 data-nosnippet>1093</a>    }
+<a href=#1094 id=1094 data-nosnippet>1094</a>
+<a href=#1095 id=1095 data-nosnippet>1095</a>    <span class="kw">fn </span>and(<span class="kw-2">&amp;mut </span><span class="self">self</span>, _lhs: (), _rhs: ()) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1096 id=1096 data-nosnippet>1096</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1097 id=1097 data-nosnippet>1097</a>    }
+<a href=#1098 id=1098 data-nosnippet>1098</a>
+<a href=#1099 id=1099 data-nosnippet>1099</a>    <span class="kw">fn </span>or(<span class="kw-2">&amp;mut </span><span class="self">self</span>, _lhs: (), _rhs: ()) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1100 id=1100 data-nosnippet>1100</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1101 id=1101 data-nosnippet>1101</a>    }
+<a href=#1102 id=1102 data-nosnippet>1102</a>
+<a href=#1103 id=1103 data-nosnippet>1103</a>    <span class="kw">fn </span>not(<span class="kw-2">&amp;mut </span><span class="self">self</span>, _inner: ()) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1104 id=1104 data-nosnippet>1104</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1105 id=1105 data-nosnippet>1105</a>    }
+<a href=#1106 id=1106 data-nosnippet>1106</a>
+<a href=#1107 id=1107 data-nosnippet>1107</a>    <span class="kw">fn </span>is_null(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference, _predicate: <span class="kw-2">&amp;</span>BoundPredicate) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1108 id=1108 data-nosnippet>1108</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1109 id=1109 data-nosnippet>1109</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1110 id=1110 data-nosnippet>1110</a>    }
+<a href=#1111 id=1111 data-nosnippet>1111</a>
+<a href=#1112 id=1112 data-nosnippet>1112</a>    <span class="kw">fn </span>not_null(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference, _predicate: <span class="kw-2">&amp;</span>BoundPredicate) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1113 id=1113 data-nosnippet>1113</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1114 id=1114 data-nosnippet>1114</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1115 id=1115 data-nosnippet>1115</a>    }
+<a href=#1116 id=1116 data-nosnippet>1116</a>
+<a href=#1117 id=1117 data-nosnippet>1117</a>    <span class="kw">fn </span>is_nan(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference, _predicate: <span class="kw-2">&amp;</span>BoundPredicate) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1118 id=1118 data-nosnippet>1118</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1119 id=1119 data-nosnippet>1119</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1120 id=1120 data-nosnippet>1120</a>    }
+<a href=#1121 id=1121 data-nosnippet>1121</a>
+<a href=#1122 id=1122 data-nosnippet>1122</a>    <span class="kw">fn </span>not_nan(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference, _predicate: <span class="kw-2">&amp;</span>BoundPredicate) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1123 id=1123 data-nosnippet>1123</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1124 id=1124 data-nosnippet>1124</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1125 id=1125 data-nosnippet>1125</a>    }
+<a href=#1126 id=1126 data-nosnippet>1126</a>
+<a href=#1127 id=1127 data-nosnippet>1127</a>    <span class="kw">fn </span>less_than(
+<a href=#1128 id=1128 data-nosnippet>1128</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1129 id=1129 data-nosnippet>1129</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1130 id=1130 data-nosnippet>1130</a>        _literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1131 id=1131 data-nosnippet>1131</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1132 id=1132 data-nosnippet>1132</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1133 id=1133 data-nosnippet>1133</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1134 id=1134 data-nosnippet>1134</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1135 id=1135 data-nosnippet>1135</a>    }
+<a href=#1136 id=1136 data-nosnippet>1136</a>
+<a href=#1137 id=1137 data-nosnippet>1137</a>    <span class="kw">fn </span>less_than_or_eq(
+<a href=#1138 id=1138 data-nosnippet>1138</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1139 id=1139 data-nosnippet>1139</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1140 id=1140 data-nosnippet>1140</a>        _literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1141 id=1141 data-nosnippet>1141</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1142 id=1142 data-nosnippet>1142</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1143 id=1143 data-nosnippet>1143</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1144 id=1144 data-nosnippet>1144</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1145 id=1145 data-nosnippet>1145</a>    }
+<a href=#1146 id=1146 data-nosnippet>1146</a>
+<a href=#1147 id=1147 data-nosnippet>1147</a>    <span class="kw">fn </span>greater_than(
+<a href=#1148 id=1148 data-nosnippet>1148</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1149 id=1149 data-nosnippet>1149</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1150 id=1150 data-nosnippet>1150</a>        _literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1151 id=1151 data-nosnippet>1151</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1152 id=1152 data-nosnippet>1152</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1153 id=1153 data-nosnippet>1153</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1154 id=1154 data-nosnippet>1154</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1155 id=1155 data-nosnippet>1155</a>    }
+<a href=#1156 id=1156 data-nosnippet>1156</a>
+<a href=#1157 id=1157 data-nosnippet>1157</a>    <span class="kw">fn </span>greater_than_or_eq(
+<a href=#1158 id=1158 data-nosnippet>1158</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1159 id=1159 data-nosnippet>1159</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1160 id=1160 data-nosnippet>1160</a>        _literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1161 id=1161 data-nosnippet>1161</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1162 id=1162 data-nosnippet>1162</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1163 id=1163 data-nosnippet>1163</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1164 id=1164 data-nosnippet>1164</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1165 id=1165 data-nosnippet>1165</a>    }
+<a href=#1166 id=1166 data-nosnippet>1166</a>
+<a href=#1167 id=1167 data-nosnippet>1167</a>    <span class="kw">fn </span>eq(
+<a href=#1168 id=1168 data-nosnippet>1168</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1169 id=1169 data-nosnippet>1169</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1170 id=1170 data-nosnippet>1170</a>        _literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1171 id=1171 data-nosnippet>1171</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1172 id=1172 data-nosnippet>1172</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1173 id=1173 data-nosnippet>1173</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1174 id=1174 data-nosnippet>1174</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1175 id=1175 data-nosnippet>1175</a>    }
 <a href=#1176 id=1176 data-nosnippet>1176</a>
-<a href=#1177 id=1177 data-nosnippet>1177</a><span class="doccomment">/// Gets the leaf column from the record batch for the required column index. Only
-<a href=#1178 id=1178 data-nosnippet>1178</a>/// supports top-level columns for now.
-<a href=#1179 id=1179 data-nosnippet>1179</a></span><span class="kw">fn </span>project_column(
-<a href=#1180 id=1180 data-nosnippet>1180</a>    batch: <span class="kw-2">&amp;</span>RecordBatch,
-<a href=#1181 id=1181 data-nosnippet>1181</a>    column_idx: usize,
-<a href=#1182 id=1182 data-nosnippet>1182</a>) -&gt; std::result::Result&lt;ArrayRef, ArrowError&gt; {
-<a href=#1183 id=1183 data-nosnippet>1183</a>    <span class="kw">let </span>column = batch.column(column_idx);
-<a href=#1184 id=1184 data-nosnippet>1184</a>
-<a href=#1185 id=1185 data-nosnippet>1185</a>    <span class="kw">match </span>column.data_type() {
-<a href=#1186 id=1186 data-nosnippet>1186</a>        DataType::Struct(<span class="kw">_</span>) =&gt; <span class="prelude-val">Err</span>(ArrowError::SchemaError(
-<a href=#1187 id=1187 data-nosnippet>1187</a>            <span class="string">"Does not support struct column yet."</span>.to_string(),
-<a href=#1188 id=1188 data-nosnippet>1188</a>        )),
-<a href=#1189 id=1189 data-nosnippet>1189</a>        <span class="kw">_ </span>=&gt; <span class="prelude-val">Ok</span>(column.clone()),
-<a href=#1190 id=1190 data-nosnippet>1190</a>    }
-<a href=#1191 id=1191 data-nosnippet>1191</a>}
-<a href=#1192 id=1192 data-nosnippet>1192</a>
-<a href=#1193 id=1193 data-nosnippet>1193</a><span class="kw">type </span>PredicateResult =
-<a href=#1194 id=1194 data-nosnippet>1194</a>    <span class="kw">dyn </span>FnMut(RecordBatch) -&gt; std::result::Result&lt;BooleanArray, ArrowError&gt; + Send + <span class="lifetime">'static</span>;
-<a href=#1195 id=1195 data-nosnippet>1195</a>
-<a href=#1196 id=1196 data-nosnippet>1196</a><span class="kw">impl </span>BoundPredicateVisitor <span class="kw">for </span>PredicateConverter&lt;<span class="lifetime">'_</span>&gt; {
-<a href=#1197 id=1197 data-nosnippet>1197</a>    <span class="kw">type </span>T = Box&lt;PredicateResult&gt;;
-<a href=#1198 id=1198 data-nosnippet>1198</a>
-<a href=#1199 id=1199 data-nosnippet>1199</a>    <span class="kw">fn </span>always_true(<span class="kw-2">&amp;mut </span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1200 id=1200 data-nosnippet>1200</a>        <span class="self">self</span>.build_always_true()
-<a href=#1201 id=1201 data-nosnippet>1201</a>    }
-<a href=#1202 id=1202 data-nosnippet>1202</a>
-<a href=#1203 id=1203 data-nosnippet>1203</a>    <span class="kw">fn </span>always_false(<span class="kw-2">&amp;mut </span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1204 id=1204 data-nosnippet>1204</a>        <span class="self">self</span>.build_always_false()
+<a href=#1177 id=1177 data-nosnippet>1177</a>    <span class="kw">fn </span>not_eq(
+<a href=#1178 id=1178 data-nosnippet>1178</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1179 id=1179 data-nosnippet>1179</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1180 id=1180 data-nosnippet>1180</a>        _literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1181 id=1181 data-nosnippet>1181</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1182 id=1182 data-nosnippet>1182</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1183 id=1183 data-nosnippet>1183</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1184 id=1184 data-nosnippet>1184</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1185 id=1185 data-nosnippet>1185</a>    }
+<a href=#1186 id=1186 data-nosnippet>1186</a>
+<a href=#1187 id=1187 data-nosnippet>1187</a>    <span class="kw">fn </span>starts_with(
+<a href=#1188 id=1188 data-nosnippet>1188</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1189 id=1189 data-nosnippet>1189</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1190 id=1190 data-nosnippet>1190</a>        _literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1191 id=1191 data-nosnippet>1191</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1192 id=1192 data-nosnippet>1192</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1193 id=1193 data-nosnippet>1193</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1194 id=1194 data-nosnippet>1194</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1195 id=1195 data-nosnippet>1195</a>    }
+<a href=#1196 id=1196 data-nosnippet>1196</a>
+<a href=#1197 id=1197 data-nosnippet>1197</a>    <span class="kw">fn </span>not_starts_with(
+<a href=#1198 id=1198 data-nosnippet>1198</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1199 id=1199 data-nosnippet>1199</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1200 id=1200 data-nosnippet>1200</a>        _literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1201 id=1201 data-nosnippet>1201</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1202 id=1202 data-nosnippet>1202</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1203 id=1203 data-nosnippet>1203</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1204 id=1204 data-nosnippet>1204</a>        <span class="prelude-val">Ok</span>(())
 <a href=#1205 id=1205 data-nosnippet>1205</a>    }
 <a href=#1206 id=1206 data-nosnippet>1206</a>
-<a href=#1207 id=1207 data-nosnippet>1207</a>    <span class="kw">fn </span>and(
+<a href=#1207 id=1207 data-nosnippet>1207</a>    <span class="kw">fn </span>r#in(
 <a href=#1208 id=1208 data-nosnippet>1208</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1209 id=1209 data-nosnippet>1209</a>        <span class="kw-2">mut </span>lhs: Box&lt;PredicateResult&gt;,
-<a href=#1210 id=1210 data-nosnippet>1210</a>        <span class="kw-2">mut </span>rhs: Box&lt;PredicateResult&gt;,
-<a href=#1211 id=1211 data-nosnippet>1211</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1212 id=1212 data-nosnippet>1212</a>        <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1213 id=1213 data-nosnippet>1213</a>            <span class="kw">let </span>left = lhs(batch.clone())<span class="question-mark">?</span>;
-<a href=#1214 id=1214 data-nosnippet>1214</a>            <span class="kw">let </span>right = rhs(batch)<span class="question-mark">?</span>;
-<a href=#1215 id=1215 data-nosnippet>1215</a>            and_kleene(<span class="kw-2">&amp;</span>left, <span class="kw-2">&amp;</span>right)
-<a href=#1216 id=1216 data-nosnippet>1216</a>        }))
-<a href=#1217 id=1217 data-nosnippet>1217</a>    }
-<a href=#1218 id=1218 data-nosnippet>1218</a>
-<a href=#1219 id=1219 data-nosnippet>1219</a>    <span class="kw">fn </span>or(
-<a href=#1220 id=1220 data-nosnippet>1220</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1221 id=1221 data-nosnippet>1221</a>        <span class="kw-2">mut </span>lhs: Box&lt;PredicateResult&gt;,
-<a href=#1222 id=1222 data-nosnippet>1222</a>        <span class="kw-2">mut </span>rhs: Box&lt;PredicateResult&gt;,
-<a href=#1223 id=1223 data-nosnippet>1223</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1224 id=1224 data-nosnippet>1224</a>        <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1225 id=1225 data-nosnippet>1225</a>            <span class="kw">let </span>left = lhs(batch.clone())<span class="question-mark">?</span>;
-<a href=#1226 id=1226 data-nosnippet>1226</a>            <span class="kw">let </span>right = rhs(batch)<span class="question-mark">?</span>;
-<a href=#1227 id=1227 data-nosnippet>1227</a>            or_kleene(<span class="kw-2">&amp;</span>left, <span class="kw-2">&amp;</span>right)
-<a href=#1228 id=1228 data-nosnippet>1228</a>        }))
-<a href=#1229 id=1229 data-nosnippet>1229</a>    }
-<a href=#1230 id=1230 data-nosnippet>1230</a>
-<a href=#1231 id=1231 data-nosnippet>1231</a>    <span class="kw">fn </span>not(<span class="kw-2">&amp;mut </span><span class="self">self</span>, <span class="kw-2">mut </span>inner: Box&lt;PredicateResult&gt;) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1232 id=1232 data-nosnippet>1232</a>        <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1233 id=1233 data-nosnippet>1233</a>            <span class="kw">let </span>pred_ret = inner(batch)<span class="question-mark">?</span>;
-<a href=#1234 id=1234 data-nosnippet>1234</a>            not(<span class="kw-2">&amp;</span>pred_ret)
-<a href=#1235 id=1235 data-nosnippet>1235</a>        }))
-<a href=#1236 id=1236 data-nosnippet>1236</a>    }
+<a href=#1209 id=1209 data-nosnippet>1209</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1210 id=1210 data-nosnippet>1210</a>        _literals: <span class="kw-2">&amp;</span>FnvHashSet&lt;Datum&gt;,
+<a href=#1211 id=1211 data-nosnippet>1211</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1212 id=1212 data-nosnippet>1212</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1213 id=1213 data-nosnippet>1213</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1214 id=1214 data-nosnippet>1214</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1215 id=1215 data-nosnippet>1215</a>    }
+<a href=#1216 id=1216 data-nosnippet>1216</a>
+<a href=#1217 id=1217 data-nosnippet>1217</a>    <span class="kw">fn </span>not_in(
+<a href=#1218 id=1218 data-nosnippet>1218</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1219 id=1219 data-nosnippet>1219</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1220 id=1220 data-nosnippet>1220</a>        _literals: <span class="kw-2">&amp;</span>FnvHashSet&lt;Datum&gt;,
+<a href=#1221 id=1221 data-nosnippet>1221</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1222 id=1222 data-nosnippet>1222</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;()&gt; {
+<a href=#1223 id=1223 data-nosnippet>1223</a>        <span class="self">self</span>.field_ids.insert(reference.field().id);
+<a href=#1224 id=1224 data-nosnippet>1224</a>        <span class="prelude-val">Ok</span>(())
+<a href=#1225 id=1225 data-nosnippet>1225</a>    }
+<a href=#1226 id=1226 data-nosnippet>1226</a>}
+<a href=#1227 id=1227 data-nosnippet>1227</a>
+<a href=#1228 id=1228 data-nosnippet>1228</a><span class="doccomment">/// A visitor to convert Iceberg bound predicates to Arrow predicates.
+<a href=#1229 id=1229 data-nosnippet>1229</a></span><span class="kw">struct </span>PredicateConverter&lt;<span class="lifetime">'a</span>&gt; {
+<a href=#1230 id=1230 data-nosnippet>1230</a>    <span class="doccomment">/// The Parquet schema descriptor.
+<a href=#1231 id=1231 data-nosnippet>1231</a>    </span><span class="kw">pub </span>parquet_schema: <span class="kw-2">&amp;</span><span class="lifetime">'a </span>SchemaDescriptor,
+<a href=#1232 id=1232 data-nosnippet>1232</a>    <span class="doccomment">/// The map between field id and leaf column index in Parquet schema.
+<a href=#1233 id=1233 data-nosnippet>1233</a>    </span><span class="kw">pub </span>column_map: <span class="kw-2">&amp;</span><span class="lifetime">'a </span>HashMap&lt;i32, usize&gt;,
+<a href=#1234 id=1234 data-nosnippet>1234</a>    <span class="doccomment">/// The required column indices in Parquet schema for the predicates.
+<a href=#1235 id=1235 data-nosnippet>1235</a>    </span><span class="kw">pub </span>column_indices: <span class="kw-2">&amp;</span><span class="lifetime">'a </span>Vec&lt;usize&gt;,
+<a href=#1236 id=1236 data-nosnippet>1236</a>}
 <a href=#1237 id=1237 data-nosnippet>1237</a>
-<a href=#1238 id=1238 data-nosnippet>1238</a>    <span class="kw">fn </span>is_null(
-<a href=#1239 id=1239 data-nosnippet>1239</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1240 id=1240 data-nosnippet>1240</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1241 id=1241 data-nosnippet>1241</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1242 id=1242 data-nosnippet>1242</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1243 id=1243 data-nosnippet>1243</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1244 id=1244 data-nosnippet>1244</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1245 id=1245 data-nosnippet>1245</a>                <span class="kw">let </span>column = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1246 id=1246 data-nosnippet>1246</a>                is_null(<span class="kw-2">&amp;</span>column)
-<a href=#1247 id=1247 data-nosnippet>1247</a>            }))
-<a href=#1248 id=1248 data-nosnippet>1248</a>        } <span class="kw">else </span>{
-<a href=#1249 id=1249 data-nosnippet>1249</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1250 id=1250 data-nosnippet>1250</a>            </span><span class="self">self</span>.build_always_true()
-<a href=#1251 id=1251 data-nosnippet>1251</a>        }
-<a href=#1252 id=1252 data-nosnippet>1252</a>    }
-<a href=#1253 id=1253 data-nosnippet>1253</a>
-<a href=#1254 id=1254 data-nosnippet>1254</a>    <span class="kw">fn </span>not_null(
-<a href=#1255 id=1255 data-nosnippet>1255</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1256 id=1256 data-nosnippet>1256</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1257 id=1257 data-nosnippet>1257</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1258 id=1258 data-nosnippet>1258</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1259 id=1259 data-nosnippet>1259</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1260 id=1260 data-nosnippet>1260</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1261 id=1261 data-nosnippet>1261</a>                <span class="kw">let </span>column = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1262 id=1262 data-nosnippet>1262</a>                is_not_null(<span class="kw-2">&amp;</span>column)
-<a href=#1263 id=1263 data-nosnippet>1263</a>            }))
-<a href=#1264 id=1264 data-nosnippet>1264</a>        } <span class="kw">else </span>{
-<a href=#1265 id=1265 data-nosnippet>1265</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1266 id=1266 data-nosnippet>1266</a>            </span><span class="self">self</span>.build_always_false()
-<a href=#1267 id=1267 data-nosnippet>1267</a>        }
-<a href=#1268 id=1268 data-nosnippet>1268</a>    }
-<a href=#1269 id=1269 data-nosnippet>1269</a>
-<a href=#1270 id=1270 data-nosnippet>1270</a>    <span class="kw">fn </span>is_nan(
-<a href=#1271 id=1271 data-nosnippet>1271</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1272 id=1272 data-nosnippet>1272</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1273 id=1273 data-nosnippet>1273</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1274 id=1274 data-nosnippet>1274</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1275 id=1275 data-nosnippet>1275</a>        <span class="kw">if </span><span class="self">self</span>.bound_reference(reference)<span class="question-mark">?</span>.is_some() {
-<a href=#1276 id=1276 data-nosnippet>1276</a>            <span class="self">self</span>.build_always_true()
-<a href=#1277 id=1277 data-nosnippet>1277</a>        } <span class="kw">else </span>{
-<a href=#1278 id=1278 data-nosnippet>1278</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1279 id=1279 data-nosnippet>1279</a>            </span><span class="self">self</span>.build_always_false()
-<a href=#1280 id=1280 data-nosnippet>1280</a>        }
-<a href=#1281 id=1281 data-nosnippet>1281</a>    }
-<a href=#1282 id=1282 data-nosnippet>1282</a>
-<a href=#1283 id=1283 data-nosnippet>1283</a>    <span class="kw">fn </span>not_nan(
-<a href=#1284 id=1284 data-nosnippet>1284</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1285 id=1285 data-nosnippet>1285</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1286 id=1286 data-nosnippet>1286</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1287 id=1287 data-nosnippet>1287</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1288 id=1288 data-nosnippet>1288</a>        <span class="kw">if </span><span class="self">self</span>.bound_reference(reference)<span class="question-mark">?</span>.is_some() {
-<a href=#1289 id=1289 data-nosnippet>1289</a>            <span class="self">self</span>.build_always_false()
-<a href=#1290 id=1290 data-nosnippet>1290</a>        } <span class="kw">else </span>{
-<a href=#1291 id=1291 data-nosnippet>1291</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1292 id=1292 data-nosnippet>1292</a>            </span><span class="self">self</span>.build_always_true()
-<a href=#1293 id=1293 data-nosnippet>1293</a>        }
-<a href=#1294 id=1294 data-nosnippet>1294</a>    }
-<a href=#1295 id=1295 data-nosnippet>1295</a>
-<a href=#1296 id=1296 data-nosnippet>1296</a>    <span class="kw">fn </span>less_than(
-<a href=#1297 id=1297 data-nosnippet>1297</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1298 id=1298 data-nosnippet>1298</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1299 id=1299 data-nosnippet>1299</a>        literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1300 id=1300 data-nosnippet>1300</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1301 id=1301 data-nosnippet>1301</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1302 id=1302 data-nosnippet>1302</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1303 id=1303 data-nosnippet>1303</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
-<a href=#1304 id=1304 data-nosnippet>1304</a>
-<a href=#1305 id=1305 data-nosnippet>1305</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1306 id=1306 data-nosnippet>1306</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1307 id=1307 data-nosnippet>1307</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1308 id=1308 data-nosnippet>1308</a>                lt(<span class="kw-2">&amp;</span>left, literal.as_ref())
-<a href=#1309 id=1309 data-nosnippet>1309</a>            }))
-<a href=#1310 id=1310 data-nosnippet>1310</a>        } <span class="kw">else </span>{
-<a href=#1311 id=1311 data-nosnippet>1311</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1312 id=1312 data-nosnippet>1312</a>            </span><span class="self">self</span>.build_always_true()
-<a href=#1313 id=1313 data-nosnippet>1313</a>        }
+<a href=#1238 id=1238 data-nosnippet>1238</a><span class="kw">impl </span>PredicateConverter&lt;<span class="lifetime">'_</span>&gt; {
+<a href=#1239 id=1239 data-nosnippet>1239</a>    <span class="doccomment">/// When visiting a bound reference, we return index of the leaf column in the
+<a href=#1240 id=1240 data-nosnippet>1240</a>    /// required column indices which is used to project the column in the record batch.
+<a href=#1241 id=1241 data-nosnippet>1241</a>    /// Return None if the field id is not found in the column map, which is possible
+<a href=#1242 id=1242 data-nosnippet>1242</a>    /// due to schema evolution.
+<a href=#1243 id=1243 data-nosnippet>1243</a>    </span><span class="kw">fn </span>bound_reference(<span class="kw-2">&amp;mut </span><span class="self">self</span>, reference: <span class="kw-2">&amp;</span>BoundReference) -&gt; <span class="prelude-ty">Result</span>&lt;<span class="prelude-ty">Option</span>&lt;usize&gt;&gt; {
+<a href=#1244 id=1244 data-nosnippet>1244</a>        <span class="comment">// The leaf column's index in Parquet schema.
+<a href=#1245 id=1245 data-nosnippet>1245</a>        </span><span class="kw">if let </span><span class="prelude-val">Some</span>(column_idx) = <span class="self">self</span>.column_map.get(<span class="kw-2">&amp;</span>reference.field().id) {
+<a href=#1246 id=1246 data-nosnippet>1246</a>            <span class="kw">if </span><span class="self">self</span>.parquet_schema.get_column_root(<span class="kw-2">*</span>column_idx).is_group() {
+<a href=#1247 id=1247 data-nosnippet>1247</a>                <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#1248 id=1248 data-nosnippet>1248</a>                    ErrorKind::DataInvalid,
+<a href=#1249 id=1249 data-nosnippet>1249</a>                    <span class="macro">format!</span>(
+<a href=#1250 id=1250 data-nosnippet>1250</a>                        <span class="string">"Leave column `{}` in predicates isn't a root column in Parquet schema."</span>,
+<a href=#1251 id=1251 data-nosnippet>1251</a>                        reference.field().name
+<a href=#1252 id=1252 data-nosnippet>1252</a>                    ),
+<a href=#1253 id=1253 data-nosnippet>1253</a>                ));
+<a href=#1254 id=1254 data-nosnippet>1254</a>            }
+<a href=#1255 id=1255 data-nosnippet>1255</a>
+<a href=#1256 id=1256 data-nosnippet>1256</a>            <span class="comment">// The leaf column's index in the required column indices.
+<a href=#1257 id=1257 data-nosnippet>1257</a>            </span><span class="kw">let </span>index = <span class="self">self
+<a href=#1258 id=1258 data-nosnippet>1258</a>                </span>.column_indices
+<a href=#1259 id=1259 data-nosnippet>1259</a>                .iter()
+<a href=#1260 id=1260 data-nosnippet>1260</a>                .position(|<span class="kw-2">&amp;</span>idx| idx == <span class="kw-2">*</span>column_idx)
+<a href=#1261 id=1261 data-nosnippet>1261</a>                .ok_or(Error::new(
+<a href=#1262 id=1262 data-nosnippet>1262</a>                    ErrorKind::DataInvalid,
+<a href=#1263 id=1263 data-nosnippet>1263</a>                    <span class="macro">format!</span>(
+<a href=#1264 id=1264 data-nosnippet>1264</a>                <span class="string">"Leave column `{}` in predicates cannot be found in the required column indices."</span>,
+<a href=#1265 id=1265 data-nosnippet>1265</a>                reference.field().name
+<a href=#1266 id=1266 data-nosnippet>1266</a>            ),
+<a href=#1267 id=1267 data-nosnippet>1267</a>                ))<span class="question-mark">?</span>;
+<a href=#1268 id=1268 data-nosnippet>1268</a>
+<a href=#1269 id=1269 data-nosnippet>1269</a>            <span class="prelude-val">Ok</span>(<span class="prelude-val">Some</span>(index))
+<a href=#1270 id=1270 data-nosnippet>1270</a>        } <span class="kw">else </span>{
+<a href=#1271 id=1271 data-nosnippet>1271</a>            <span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>)
+<a href=#1272 id=1272 data-nosnippet>1272</a>        }
+<a href=#1273 id=1273 data-nosnippet>1273</a>    }
+<a href=#1274 id=1274 data-nosnippet>1274</a>
+<a href=#1275 id=1275 data-nosnippet>1275</a>    <span class="doccomment">/// Build an Arrow predicate that always returns true.
+<a href=#1276 id=1276 data-nosnippet>1276</a>    </span><span class="kw">fn </span>build_always_true(<span class="kw-2">&amp;</span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1277 id=1277 data-nosnippet>1277</a>        <span class="prelude-val">Ok</span>(Box::new(|batch| {
+<a href=#1278 id=1278 data-nosnippet>1278</a>            <span class="prelude-val">Ok</span>(BooleanArray::from(<span class="macro">vec!</span>[<span class="bool-val">true</span>; batch.num_rows()]))
+<a href=#1279 id=1279 data-nosnippet>1279</a>        }))
+<a href=#1280 id=1280 data-nosnippet>1280</a>    }
+<a href=#1281 id=1281 data-nosnippet>1281</a>
+<a href=#1282 id=1282 data-nosnippet>1282</a>    <span class="doccomment">/// Build an Arrow predicate that always returns false.
+<a href=#1283 id=1283 data-nosnippet>1283</a>    </span><span class="kw">fn </span>build_always_false(<span class="kw-2">&amp;</span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1284 id=1284 data-nosnippet>1284</a>        <span class="prelude-val">Ok</span>(Box::new(|batch| {
+<a href=#1285 id=1285 data-nosnippet>1285</a>            <span class="prelude-val">Ok</span>(BooleanArray::from(<span class="macro">vec!</span>[<span class="bool-val">false</span>; batch.num_rows()]))
+<a href=#1286 id=1286 data-nosnippet>1286</a>        }))
+<a href=#1287 id=1287 data-nosnippet>1287</a>    }
+<a href=#1288 id=1288 data-nosnippet>1288</a>}
+<a href=#1289 id=1289 data-nosnippet>1289</a>
+<a href=#1290 id=1290 data-nosnippet>1290</a><span class="doccomment">/// Gets the leaf column from the record batch for the required column index. Only
+<a href=#1291 id=1291 data-nosnippet>1291</a>/// supports top-level columns for now.
+<a href=#1292 id=1292 data-nosnippet>1292</a></span><span class="kw">fn </span>project_column(
+<a href=#1293 id=1293 data-nosnippet>1293</a>    batch: <span class="kw-2">&amp;</span>RecordBatch,
+<a href=#1294 id=1294 data-nosnippet>1294</a>    column_idx: usize,
+<a href=#1295 id=1295 data-nosnippet>1295</a>) -&gt; std::result::Result&lt;ArrayRef, ArrowError&gt; {
+<a href=#1296 id=1296 data-nosnippet>1296</a>    <span class="kw">let </span>column = batch.column(column_idx);
+<a href=#1297 id=1297 data-nosnippet>1297</a>
+<a href=#1298 id=1298 data-nosnippet>1298</a>    <span class="kw">match </span>column.data_type() {
+<a href=#1299 id=1299 data-nosnippet>1299</a>        DataType::Struct(<span class="kw">_</span>) =&gt; <span class="prelude-val">Err</span>(ArrowError::SchemaError(
+<a href=#1300 id=1300 data-nosnippet>1300</a>            <span class="string">"Does not support struct column yet."</span>.to_string(),
+<a href=#1301 id=1301 data-nosnippet>1301</a>        )),
+<a href=#1302 id=1302 data-nosnippet>1302</a>        <span class="kw">_ </span>=&gt; <span class="prelude-val">Ok</span>(column.clone()),
+<a href=#1303 id=1303 data-nosnippet>1303</a>    }
+<a href=#1304 id=1304 data-nosnippet>1304</a>}
+<a href=#1305 id=1305 data-nosnippet>1305</a>
+<a href=#1306 id=1306 data-nosnippet>1306</a><span class="kw">type </span>PredicateResult =
+<a href=#1307 id=1307 data-nosnippet>1307</a>    <span class="kw">dyn </span>FnMut(RecordBatch) -&gt; std::result::Result&lt;BooleanArray, ArrowError&gt; + Send + <span class="lifetime">'static</span>;
+<a href=#1308 id=1308 data-nosnippet>1308</a>
+<a href=#1309 id=1309 data-nosnippet>1309</a><span class="kw">impl </span>BoundPredicateVisitor <span class="kw">for </span>PredicateConverter&lt;<span class="lifetime">'_</span>&gt; {
+<a href=#1310 id=1310 data-nosnippet>1310</a>    <span class="kw">type </span>T = Box&lt;PredicateResult&gt;;
+<a href=#1311 id=1311 data-nosnippet>1311</a>
+<a href=#1312 id=1312 data-nosnippet>1312</a>    <span class="kw">fn </span>always_true(<span class="kw-2">&amp;mut </span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1313 id=1313 data-nosnippet>1313</a>        <span class="self">self</span>.build_always_true()
 <a href=#1314 id=1314 data-nosnippet>1314</a>    }
 <a href=#1315 id=1315 data-nosnippet>1315</a>
-<a href=#1316 id=1316 data-nosnippet>1316</a>    <span class="kw">fn </span>less_than_or_eq(
-<a href=#1317 id=1317 data-nosnippet>1317</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1318 id=1318 data-nosnippet>1318</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1319 id=1319 data-nosnippet>1319</a>        literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1320 id=1320 data-nosnippet>1320</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1321 id=1321 data-nosnippet>1321</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1322 id=1322 data-nosnippet>1322</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1323 id=1323 data-nosnippet>1323</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
-<a href=#1324 id=1324 data-nosnippet>1324</a>
-<a href=#1325 id=1325 data-nosnippet>1325</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1326 id=1326 data-nosnippet>1326</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1327 id=1327 data-nosnippet>1327</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1328 id=1328 data-nosnippet>1328</a>                lt_eq(<span class="kw-2">&amp;</span>left, literal.as_ref())
-<a href=#1329 id=1329 data-nosnippet>1329</a>            }))
-<a href=#1330 id=1330 data-nosnippet>1330</a>        } <span class="kw">else </span>{
-<a href=#1331 id=1331 data-nosnippet>1331</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1332 id=1332 data-nosnippet>1332</a>            </span><span class="self">self</span>.build_always_true()
-<a href=#1333 id=1333 data-nosnippet>1333</a>        }
-<a href=#1334 id=1334 data-nosnippet>1334</a>    }
-<a href=#1335 id=1335 data-nosnippet>1335</a>
-<a href=#1336 id=1336 data-nosnippet>1336</a>    <span class="kw">fn </span>greater_than(
-<a href=#1337 id=1337 data-nosnippet>1337</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1338 id=1338 data-nosnippet>1338</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1339 id=1339 data-nosnippet>1339</a>        literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1340 id=1340 data-nosnippet>1340</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1341 id=1341 data-nosnippet>1341</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1342 id=1342 data-nosnippet>1342</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1343 id=1343 data-nosnippet>1343</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
-<a href=#1344 id=1344 data-nosnippet>1344</a>
-<a href=#1345 id=1345 data-nosnippet>1345</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1346 id=1346 data-nosnippet>1346</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1347 id=1347 data-nosnippet>1347</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1348 id=1348 data-nosnippet>1348</a>                gt(<span class="kw-2">&amp;</span>left, literal.as_ref())
-<a href=#1349 id=1349 data-nosnippet>1349</a>            }))
-<a href=#1350 id=1350 data-nosnippet>1350</a>        } <span class="kw">else </span>{
-<a href=#1351 id=1351 data-nosnippet>1351</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1352 id=1352 data-nosnippet>1352</a>            </span><span class="self">self</span>.build_always_false()
-<a href=#1353 id=1353 data-nosnippet>1353</a>        }
-<a href=#1354 id=1354 data-nosnippet>1354</a>    }
-<a href=#1355 id=1355 data-nosnippet>1355</a>
-<a href=#1356 id=1356 data-nosnippet>1356</a>    <span class="kw">fn </span>greater_than_or_eq(
-<a href=#1357 id=1357 data-nosnippet>1357</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1358 id=1358 data-nosnippet>1358</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1359 id=1359 data-nosnippet>1359</a>        literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1360 id=1360 data-nosnippet>1360</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1361 id=1361 data-nosnippet>1361</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1362 id=1362 data-nosnippet>1362</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1363 id=1363 data-nosnippet>1363</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
-<a href=#1364 id=1364 data-nosnippet>1364</a>
-<a href=#1365 id=1365 data-nosnippet>1365</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1366 id=1366 data-nosnippet>1366</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1367 id=1367 data-nosnippet>1367</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1368 id=1368 data-nosnippet>1368</a>                gt_eq(<span class="kw-2">&amp;</span>left, literal.as_ref())
-<a href=#1369 id=1369 data-nosnippet>1369</a>            }))
-<a href=#1370 id=1370 data-nosnippet>1370</a>        } <span class="kw">else </span>{
-<a href=#1371 id=1371 data-nosnippet>1371</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1372 id=1372 data-nosnippet>1372</a>            </span><span class="self">self</span>.build_always_false()
-<a href=#1373 id=1373 data-nosnippet>1373</a>        }
-<a href=#1374 id=1374 data-nosnippet>1374</a>    }
-<a href=#1375 id=1375 data-nosnippet>1375</a>
-<a href=#1376 id=1376 data-nosnippet>1376</a>    <span class="kw">fn </span>eq(
-<a href=#1377 id=1377 data-nosnippet>1377</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1378 id=1378 data-nosnippet>1378</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1379 id=1379 data-nosnippet>1379</a>        literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1380 id=1380 data-nosnippet>1380</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1381 id=1381 data-nosnippet>1381</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1382 id=1382 data-nosnippet>1382</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1383 id=1383 data-nosnippet>1383</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
-<a href=#1384 id=1384 data-nosnippet>1384</a>
-<a href=#1385 id=1385 data-nosnippet>1385</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1386 id=1386 data-nosnippet>1386</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1387 id=1387 data-nosnippet>1387</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1388 id=1388 data-nosnippet>1388</a>                eq(<span class="kw-2">&amp;</span>left, literal.as_ref())
-<a href=#1389 id=1389 data-nosnippet>1389</a>            }))
+<a href=#1316 id=1316 data-nosnippet>1316</a>    <span class="kw">fn </span>always_false(<span class="kw-2">&amp;mut </span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1317 id=1317 data-nosnippet>1317</a>        <span class="self">self</span>.build_always_false()
+<a href=#1318 id=1318 data-nosnippet>1318</a>    }
+<a href=#1319 id=1319 data-nosnippet>1319</a>
+<a href=#1320 id=1320 data-nosnippet>1320</a>    <span class="kw">fn </span>and(
+<a href=#1321 id=1321 data-nosnippet>1321</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1322 id=1322 data-nosnippet>1322</a>        <span class="kw-2">mut </span>lhs: Box&lt;PredicateResult&gt;,
+<a href=#1323 id=1323 data-nosnippet>1323</a>        <span class="kw-2">mut </span>rhs: Box&lt;PredicateResult&gt;,
+<a href=#1324 id=1324 data-nosnippet>1324</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1325 id=1325 data-nosnippet>1325</a>        <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1326 id=1326 data-nosnippet>1326</a>            <span class="kw">let </span>left = lhs(batch.clone())<span class="question-mark">?</span>;
+<a href=#1327 id=1327 data-nosnippet>1327</a>            <span class="kw">let </span>right = rhs(batch)<span class="question-mark">?</span>;
+<a href=#1328 id=1328 data-nosnippet>1328</a>            and_kleene(<span class="kw-2">&amp;</span>left, <span class="kw-2">&amp;</span>right)
+<a href=#1329 id=1329 data-nosnippet>1329</a>        }))
+<a href=#1330 id=1330 data-nosnippet>1330</a>    }
+<a href=#1331 id=1331 data-nosnippet>1331</a>
+<a href=#1332 id=1332 data-nosnippet>1332</a>    <span class="kw">fn </span>or(
+<a href=#1333 id=1333 data-nosnippet>1333</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1334 id=1334 data-nosnippet>1334</a>        <span class="kw-2">mut </span>lhs: Box&lt;PredicateResult&gt;,
+<a href=#1335 id=1335 data-nosnippet>1335</a>        <span class="kw-2">mut </span>rhs: Box&lt;PredicateResult&gt;,
+<a href=#1336 id=1336 data-nosnippet>1336</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1337 id=1337 data-nosnippet>1337</a>        <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1338 id=1338 data-nosnippet>1338</a>            <span class="kw">let </span>left = lhs(batch.clone())<span class="question-mark">?</span>;
+<a href=#1339 id=1339 data-nosnippet>1339</a>            <span class="kw">let </span>right = rhs(batch)<span class="question-mark">?</span>;
+<a href=#1340 id=1340 data-nosnippet>1340</a>            or_kleene(<span class="kw-2">&amp;</span>left, <span class="kw-2">&amp;</span>right)
+<a href=#1341 id=1341 data-nosnippet>1341</a>        }))
+<a href=#1342 id=1342 data-nosnippet>1342</a>    }
+<a href=#1343 id=1343 data-nosnippet>1343</a>
+<a href=#1344 id=1344 data-nosnippet>1344</a>    <span class="kw">fn </span>not(<span class="kw-2">&amp;mut </span><span class="self">self</span>, <span class="kw-2">mut </span>inner: Box&lt;PredicateResult&gt;) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1345 id=1345 data-nosnippet>1345</a>        <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1346 id=1346 data-nosnippet>1346</a>            <span class="kw">let </span>pred_ret = inner(batch)<span class="question-mark">?</span>;
+<a href=#1347 id=1347 data-nosnippet>1347</a>            not(<span class="kw-2">&amp;</span>pred_ret)
+<a href=#1348 id=1348 data-nosnippet>1348</a>        }))
+<a href=#1349 id=1349 data-nosnippet>1349</a>    }
+<a href=#1350 id=1350 data-nosnippet>1350</a>
+<a href=#1351 id=1351 data-nosnippet>1351</a>    <span class="kw">fn </span>is_null(
+<a href=#1352 id=1352 data-nosnippet>1352</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1353 id=1353 data-nosnippet>1353</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1354 id=1354 data-nosnippet>1354</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1355 id=1355 data-nosnippet>1355</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1356 id=1356 data-nosnippet>1356</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1357 id=1357 data-nosnippet>1357</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1358 id=1358 data-nosnippet>1358</a>                <span class="kw">let </span>column = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1359 id=1359 data-nosnippet>1359</a>                is_null(<span class="kw-2">&amp;</span>column)
+<a href=#1360 id=1360 data-nosnippet>1360</a>            }))
+<a href=#1361 id=1361 data-nosnippet>1361</a>        } <span class="kw">else </span>{
+<a href=#1362 id=1362 data-nosnippet>1362</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1363 id=1363 data-nosnippet>1363</a>            </span><span class="self">self</span>.build_always_true()
+<a href=#1364 id=1364 data-nosnippet>1364</a>        }
+<a href=#1365 id=1365 data-nosnippet>1365</a>    }
+<a href=#1366 id=1366 data-nosnippet>1366</a>
+<a href=#1367 id=1367 data-nosnippet>1367</a>    <span class="kw">fn </span>not_null(
+<a href=#1368 id=1368 data-nosnippet>1368</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1369 id=1369 data-nosnippet>1369</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1370 id=1370 data-nosnippet>1370</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1371 id=1371 data-nosnippet>1371</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1372 id=1372 data-nosnippet>1372</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1373 id=1373 data-nosnippet>1373</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1374 id=1374 data-nosnippet>1374</a>                <span class="kw">let </span>column = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1375 id=1375 data-nosnippet>1375</a>                is_not_null(<span class="kw-2">&amp;</span>column)
+<a href=#1376 id=1376 data-nosnippet>1376</a>            }))
+<a href=#1377 id=1377 data-nosnippet>1377</a>        } <span class="kw">else </span>{
+<a href=#1378 id=1378 data-nosnippet>1378</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1379 id=1379 data-nosnippet>1379</a>            </span><span class="self">self</span>.build_always_false()
+<a href=#1380 id=1380 data-nosnippet>1380</a>        }
+<a href=#1381 id=1381 data-nosnippet>1381</a>    }
+<a href=#1382 id=1382 data-nosnippet>1382</a>
+<a href=#1383 id=1383 data-nosnippet>1383</a>    <span class="kw">fn </span>is_nan(
+<a href=#1384 id=1384 data-nosnippet>1384</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1385 id=1385 data-nosnippet>1385</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1386 id=1386 data-nosnippet>1386</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1387 id=1387 data-nosnippet>1387</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1388 id=1388 data-nosnippet>1388</a>        <span class="kw">if </span><span class="self">self</span>.bound_reference(reference)<span class="question-mark">?</span>.is_some() {
+<a href=#1389 id=1389 data-nosnippet>1389</a>            <span class="self">self</span>.build_always_true()
 <a href=#1390 id=1390 data-nosnippet>1390</a>        } <span class="kw">else </span>{
 <a href=#1391 id=1391 data-nosnippet>1391</a>            <span class="comment">// A missing column, treating it as null.
 <a href=#1392 id=1392 data-nosnippet>1392</a>            </span><span class="self">self</span>.build_always_false()
 <a href=#1393 id=1393 data-nosnippet>1393</a>        }
 <a href=#1394 id=1394 data-nosnippet>1394</a>    }
 <a href=#1395 id=1395 data-nosnippet>1395</a>
-<a href=#1396 id=1396 data-nosnippet>1396</a>    <span class="kw">fn </span>not_eq(
+<a href=#1396 id=1396 data-nosnippet>1396</a>    <span class="kw">fn </span>not_nan(
 <a href=#1397 id=1397 data-nosnippet>1397</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
 <a href=#1398 id=1398 data-nosnippet>1398</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1399 id=1399 data-nosnippet>1399</a>        literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1400 id=1400 data-nosnippet>1400</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1401 id=1401 data-nosnippet>1401</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1402 id=1402 data-nosnippet>1402</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1403 id=1403 data-nosnippet>1403</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
-<a href=#1404 id=1404 data-nosnippet>1404</a>
-<a href=#1405 id=1405 data-nosnippet>1405</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1406 id=1406 data-nosnippet>1406</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1407 id=1407 data-nosnippet>1407</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1408 id=1408 data-nosnippet>1408</a>                neq(<span class="kw-2">&amp;</span>left, literal.as_ref())
-<a href=#1409 id=1409 data-nosnippet>1409</a>            }))
-<a href=#1410 id=1410 data-nosnippet>1410</a>        } <span class="kw">else </span>{
-<a href=#1411 id=1411 data-nosnippet>1411</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1412 id=1412 data-nosnippet>1412</a>            </span><span class="self">self</span>.build_always_false()
-<a href=#1413 id=1413 data-nosnippet>1413</a>        }
-<a href=#1414 id=1414 data-nosnippet>1414</a>    }
-<a href=#1415 id=1415 data-nosnippet>1415</a>
-<a href=#1416 id=1416 data-nosnippet>1416</a>    <span class="kw">fn </span>starts_with(
-<a href=#1417 id=1417 data-nosnippet>1417</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1418 id=1418 data-nosnippet>1418</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1419 id=1419 data-nosnippet>1419</a>        literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1420 id=1420 data-nosnippet>1420</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1421 id=1421 data-nosnippet>1421</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1422 id=1422 data-nosnippet>1422</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1423 id=1423 data-nosnippet>1423</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
-<a href=#1424 id=1424 data-nosnippet>1424</a>
-<a href=#1425 id=1425 data-nosnippet>1425</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1426 id=1426 data-nosnippet>1426</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1427 id=1427 data-nosnippet>1427</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1428 id=1428 data-nosnippet>1428</a>                starts_with(<span class="kw-2">&amp;</span>left, literal.as_ref())
-<a href=#1429 id=1429 data-nosnippet>1429</a>            }))
-<a href=#1430 id=1430 data-nosnippet>1430</a>        } <span class="kw">else </span>{
-<a href=#1431 id=1431 data-nosnippet>1431</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1432 id=1432 data-nosnippet>1432</a>            </span><span class="self">self</span>.build_always_false()
-<a href=#1433 id=1433 data-nosnippet>1433</a>        }
-<a href=#1434 id=1434 data-nosnippet>1434</a>    }
-<a href=#1435 id=1435 data-nosnippet>1435</a>
-<a href=#1436 id=1436 data-nosnippet>1436</a>    <span class="kw">fn </span>not_starts_with(
-<a href=#1437 id=1437 data-nosnippet>1437</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1438 id=1438 data-nosnippet>1438</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1439 id=1439 data-nosnippet>1439</a>        literal: <span class="kw-2">&amp;</span>Datum,
-<a href=#1440 id=1440 data-nosnippet>1440</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1441 id=1441 data-nosnippet>1441</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1442 id=1442 data-nosnippet>1442</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1443 id=1443 data-nosnippet>1443</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
-<a href=#1444 id=1444 data-nosnippet>1444</a>
-<a href=#1445 id=1445 data-nosnippet>1445</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1446 id=1446 data-nosnippet>1446</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1447 id=1447 data-nosnippet>1447</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1448 id=1448 data-nosnippet>1448</a>                <span class="comment">// update here if arrow ever adds a native not_starts_with
-<a href=#1449 id=1449 data-nosnippet>1449</a>                </span>not(<span class="kw-2">&amp;</span>starts_with(<span class="kw-2">&amp;</span>left, literal.as_ref())<span class="question-mark">?</span>)
-<a href=#1450 id=1450 data-nosnippet>1450</a>            }))
-<a href=#1451 id=1451 data-nosnippet>1451</a>        } <span class="kw">else </span>{
-<a href=#1452 id=1452 data-nosnippet>1452</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1453 id=1453 data-nosnippet>1453</a>            </span><span class="self">self</span>.build_always_true()
-<a href=#1454 id=1454 data-nosnippet>1454</a>        }
-<a href=#1455 id=1455 data-nosnippet>1455</a>    }
-<a href=#1456 id=1456 data-nosnippet>1456</a>
-<a href=#1457 id=1457 data-nosnippet>1457</a>    <span class="kw">fn </span>r#in(
-<a href=#1458 id=1458 data-nosnippet>1458</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1459 id=1459 data-nosnippet>1459</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1460 id=1460 data-nosnippet>1460</a>        literals: <span class="kw-2">&amp;</span>FnvHashSet&lt;Datum&gt;,
-<a href=#1461 id=1461 data-nosnippet>1461</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1462 id=1462 data-nosnippet>1462</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1463 id=1463 data-nosnippet>1463</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1464 id=1464 data-nosnippet>1464</a>            <span class="kw">let </span>literals: Vec&lt;<span class="kw">_</span>&gt; = literals
-<a href=#1465 id=1465 data-nosnippet>1465</a>                .iter()
-<a href=#1466 id=1466 data-nosnippet>1466</a>                .map(|lit| get_arrow_datum(lit).unwrap())
-<a href=#1467 id=1467 data-nosnippet>1467</a>                .collect();
+<a href=#1399 id=1399 data-nosnippet>1399</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1400 id=1400 data-nosnippet>1400</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1401 id=1401 data-nosnippet>1401</a>        <span class="kw">if </span><span class="self">self</span>.bound_reference(reference)<span class="question-mark">?</span>.is_some() {
+<a href=#1402 id=1402 data-nosnippet>1402</a>            <span class="self">self</span>.build_always_false()
+<a href=#1403 id=1403 data-nosnippet>1403</a>        } <span class="kw">else </span>{
+<a href=#1404 id=1404 data-nosnippet>1404</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1405 id=1405 data-nosnippet>1405</a>            </span><span class="self">self</span>.build_always_true()
+<a href=#1406 id=1406 data-nosnippet>1406</a>        }
+<a href=#1407 id=1407 data-nosnippet>1407</a>    }
+<a href=#1408 id=1408 data-nosnippet>1408</a>
+<a href=#1409 id=1409 data-nosnippet>1409</a>    <span class="kw">fn </span>less_than(
+<a href=#1410 id=1410 data-nosnippet>1410</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1411 id=1411 data-nosnippet>1411</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1412 id=1412 data-nosnippet>1412</a>        literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1413 id=1413 data-nosnippet>1413</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1414 id=1414 data-nosnippet>1414</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1415 id=1415 data-nosnippet>1415</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1416 id=1416 data-nosnippet>1416</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
+<a href=#1417 id=1417 data-nosnippet>1417</a>
+<a href=#1418 id=1418 data-nosnippet>1418</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1419 id=1419 data-nosnippet>1419</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1420 id=1420 data-nosnippet>1420</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1421 id=1421 data-nosnippet>1421</a>                lt(<span class="kw-2">&amp;</span>left, literal.as_ref())
+<a href=#1422 id=1422 data-nosnippet>1422</a>            }))
+<a href=#1423 id=1423 data-nosnippet>1423</a>        } <span class="kw">else </span>{
+<a href=#1424 id=1424 data-nosnippet>1424</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1425 id=1425 data-nosnippet>1425</a>            </span><span class="self">self</span>.build_always_true()
+<a href=#1426 id=1426 data-nosnippet>1426</a>        }
+<a href=#1427 id=1427 data-nosnippet>1427</a>    }
+<a href=#1428 id=1428 data-nosnippet>1428</a>
+<a href=#1429 id=1429 data-nosnippet>1429</a>    <span class="kw">fn </span>less_than_or_eq(
+<a href=#1430 id=1430 data-nosnippet>1430</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1431 id=1431 data-nosnippet>1431</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1432 id=1432 data-nosnippet>1432</a>        literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1433 id=1433 data-nosnippet>1433</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1434 id=1434 data-nosnippet>1434</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1435 id=1435 data-nosnippet>1435</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1436 id=1436 data-nosnippet>1436</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
+<a href=#1437 id=1437 data-nosnippet>1437</a>
+<a href=#1438 id=1438 data-nosnippet>1438</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1439 id=1439 data-nosnippet>1439</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1440 id=1440 data-nosnippet>1440</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1441 id=1441 data-nosnippet>1441</a>                lt_eq(<span class="kw-2">&amp;</span>left, literal.as_ref())
+<a href=#1442 id=1442 data-nosnippet>1442</a>            }))
+<a href=#1443 id=1443 data-nosnippet>1443</a>        } <span class="kw">else </span>{
+<a href=#1444 id=1444 data-nosnippet>1444</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1445 id=1445 data-nosnippet>1445</a>            </span><span class="self">self</span>.build_always_true()
+<a href=#1446 id=1446 data-nosnippet>1446</a>        }
+<a href=#1447 id=1447 data-nosnippet>1447</a>    }
+<a href=#1448 id=1448 data-nosnippet>1448</a>
+<a href=#1449 id=1449 data-nosnippet>1449</a>    <span class="kw">fn </span>greater_than(
+<a href=#1450 id=1450 data-nosnippet>1450</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1451 id=1451 data-nosnippet>1451</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1452 id=1452 data-nosnippet>1452</a>        literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1453 id=1453 data-nosnippet>1453</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1454 id=1454 data-nosnippet>1454</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1455 id=1455 data-nosnippet>1455</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1456 id=1456 data-nosnippet>1456</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
+<a href=#1457 id=1457 data-nosnippet>1457</a>
+<a href=#1458 id=1458 data-nosnippet>1458</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1459 id=1459 data-nosnippet>1459</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1460 id=1460 data-nosnippet>1460</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1461 id=1461 data-nosnippet>1461</a>                gt(<span class="kw-2">&amp;</span>left, literal.as_ref())
+<a href=#1462 id=1462 data-nosnippet>1462</a>            }))
+<a href=#1463 id=1463 data-nosnippet>1463</a>        } <span class="kw">else </span>{
+<a href=#1464 id=1464 data-nosnippet>1464</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1465 id=1465 data-nosnippet>1465</a>            </span><span class="self">self</span>.build_always_false()
+<a href=#1466 id=1466 data-nosnippet>1466</a>        }
+<a href=#1467 id=1467 data-nosnippet>1467</a>    }
 <a href=#1468 id=1468 data-nosnippet>1468</a>
-<a href=#1469 id=1469 data-nosnippet>1469</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1470 id=1470 data-nosnippet>1470</a>                <span class="comment">// update this if arrow ever adds a native is_in kernel
-<a href=#1471 id=1471 data-nosnippet>1471</a>                </span><span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1472 id=1472 data-nosnippet>1472</a>
-<a href=#1473 id=1473 data-nosnippet>1473</a>                <span class="kw">let </span><span class="kw-2">mut </span>acc = BooleanArray::from(<span class="macro">vec!</span>[<span class="bool-val">false</span>; batch.num_rows()]);
-<a href=#1474 id=1474 data-nosnippet>1474</a>                <span class="kw">for </span>literal <span class="kw">in </span><span class="kw-2">&amp;</span>literals {
-<a href=#1475 id=1475 data-nosnippet>1475</a>                    <span class="kw">let </span>literal = try_cast_literal(literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1476 id=1476 data-nosnippet>1476</a>                    acc = or(<span class="kw-2">&amp;</span>acc, <span class="kw-2">&amp;</span>eq(<span class="kw-2">&amp;</span>left, literal.as_ref())<span class="question-mark">?</span>)<span class="question-mark">?
-<a href=#1477 id=1477 data-nosnippet>1477</a>                </span>}
-<a href=#1478 id=1478 data-nosnippet>1478</a>
-<a href=#1479 id=1479 data-nosnippet>1479</a>                <span class="prelude-val">Ok</span>(acc)
-<a href=#1480 id=1480 data-nosnippet>1480</a>            }))
-<a href=#1481 id=1481 data-nosnippet>1481</a>        } <span class="kw">else </span>{
-<a href=#1482 id=1482 data-nosnippet>1482</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1483 id=1483 data-nosnippet>1483</a>            </span><span class="self">self</span>.build_always_false()
-<a href=#1484 id=1484 data-nosnippet>1484</a>        }
-<a href=#1485 id=1485 data-nosnippet>1485</a>    }
-<a href=#1486 id=1486 data-nosnippet>1486</a>
-<a href=#1487 id=1487 data-nosnippet>1487</a>    <span class="kw">fn </span>not_in(
-<a href=#1488 id=1488 data-nosnippet>1488</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1489 id=1489 data-nosnippet>1489</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
-<a href=#1490 id=1490 data-nosnippet>1490</a>        literals: <span class="kw-2">&amp;</span>FnvHashSet&lt;Datum&gt;,
-<a href=#1491 id=1491 data-nosnippet>1491</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
-<a href=#1492 id=1492 data-nosnippet>1492</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
-<a href=#1493 id=1493 data-nosnippet>1493</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
-<a href=#1494 id=1494 data-nosnippet>1494</a>            <span class="kw">let </span>literals: Vec&lt;<span class="kw">_</span>&gt; = literals
-<a href=#1495 id=1495 data-nosnippet>1495</a>                .iter()
-<a href=#1496 id=1496 data-nosnippet>1496</a>                .map(|lit| get_arrow_datum(lit).unwrap())
-<a href=#1497 id=1497 data-nosnippet>1497</a>                .collect();
-<a href=#1498 id=1498 data-nosnippet>1498</a>
-<a href=#1499 id=1499 data-nosnippet>1499</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
-<a href=#1500 id=1500 data-nosnippet>1500</a>                <span class="comment">// update this if arrow ever adds a native not_in kernel
-<a href=#1501 id=1501 data-nosnippet>1501</a>                </span><span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
-<a href=#1502 id=1502 data-nosnippet>1502</a>                <span class="kw">let </span><span class="kw-2">mut </span>acc = BooleanArray::from(<span class="macro">vec!</span>[<span class="bool-val">true</span>; batch.num_rows()]);
-<a href=#1503 id=1503 data-nosnippet>1503</a>                <span class="kw">for </span>literal <span class="kw">in </span><span class="kw-2">&amp;</span>literals {
-<a href=#1504 id=1504 data-nosnippet>1504</a>                    <span class="kw">let </span>literal = try_cast_literal(literal, left.data_type())<span class="question-mark">?</span>;
-<a href=#1505 id=1505 data-nosnippet>1505</a>                    acc = and(<span class="kw-2">&amp;</span>acc, <span class="kw-2">&amp;</span>neq(<span class="kw-2">&amp;</span>left, literal.as_ref())<span class="question-mark">?</span>)<span class="question-mark">?
-<a href=#1506 id=1506 data-nosnippet>1506</a>                </span>}
-<a href=#1507 id=1507 data-nosnippet>1507</a>
-<a href=#1508 id=1508 data-nosnippet>1508</a>                <span class="prelude-val">Ok</span>(acc)
-<a href=#1509 id=1509 data-nosnippet>1509</a>            }))
-<a href=#1510 id=1510 data-nosnippet>1510</a>        } <span class="kw">else </span>{
-<a href=#1511 id=1511 data-nosnippet>1511</a>            <span class="comment">// A missing column, treating it as null.
-<a href=#1512 id=1512 data-nosnippet>1512</a>            </span><span class="self">self</span>.build_always_true()
-<a href=#1513 id=1513 data-nosnippet>1513</a>        }
-<a href=#1514 id=1514 data-nosnippet>1514</a>    }
-<a href=#1515 id=1515 data-nosnippet>1515</a>}
-<a href=#1516 id=1516 data-nosnippet>1516</a>
-<a href=#1517 id=1517 data-nosnippet>1517</a><span class="doccomment">/// ArrowFileReader is a wrapper around a FileRead that impls parquets AsyncFileReader.
-<a href=#1518 id=1518 data-nosnippet>1518</a></span><span class="kw">pub struct </span>ArrowFileReader&lt;R: FileRead&gt; {
-<a href=#1519 id=1519 data-nosnippet>1519</a>    meta: FileMetadata,
-<a href=#1520 id=1520 data-nosnippet>1520</a>    preload_column_index: bool,
-<a href=#1521 id=1521 data-nosnippet>1521</a>    preload_offset_index: bool,
-<a href=#1522 id=1522 data-nosnippet>1522</a>    preload_page_index: bool,
-<a href=#1523 id=1523 data-nosnippet>1523</a>    metadata_size_hint: <span class="prelude-ty">Option</span>&lt;usize&gt;,
-<a href=#1524 id=1524 data-nosnippet>1524</a>    r: R,
-<a href=#1525 id=1525 data-nosnippet>1525</a>}
-<a href=#1526 id=1526 data-nosnippet>1526</a>
-<a href=#1527 id=1527 data-nosnippet>1527</a><span class="kw">impl</span>&lt;R: FileRead&gt; ArrowFileReader&lt;R&gt; {
-<a href=#1528 id=1528 data-nosnippet>1528</a>    <span class="doccomment">/// Create a new ArrowFileReader
-<a href=#1529 id=1529 data-nosnippet>1529</a>    </span><span class="kw">pub fn </span>new(meta: FileMetadata, r: R) -&gt; <span class="self">Self </span>{
-<a href=#1530 id=1530 data-nosnippet>1530</a>        <span class="self">Self </span>{
-<a href=#1531 id=1531 data-nosnippet>1531</a>            meta,
-<a href=#1532 id=1532 data-nosnippet>1532</a>            preload_column_index: <span class="bool-val">false</span>,
-<a href=#1533 id=1533 data-nosnippet>1533</a>            preload_offset_index: <span class="bool-val">false</span>,
-<a href=#1534 id=1534 data-nosnippet>1534</a>            preload_page_index: <span class="bool-val">false</span>,
-<a href=#1535 id=1535 data-nosnippet>1535</a>            metadata_size_hint: <span class="prelude-val">None</span>,
-<a href=#1536 id=1536 data-nosnippet>1536</a>            r,
-<a href=#1537 id=1537 data-nosnippet>1537</a>        }
-<a href=#1538 id=1538 data-nosnippet>1538</a>    }
-<a href=#1539 id=1539 data-nosnippet>1539</a>
-<a href=#1540 id=1540 data-nosnippet>1540</a>    <span class="doccomment">/// Enable or disable preloading of the column index
-<a href=#1541 id=1541 data-nosnippet>1541</a>    </span><span class="kw">pub fn </span>with_preload_column_index(<span class="kw-2">mut </span><span class="self">self</span>, preload: bool) -&gt; <span class="self">Self </span>{
-<a href=#1542 id=1542 data-nosnippet>1542</a>        <span class="self">self</span>.preload_column_index = preload;
-<a href=#1543 id=1543 data-nosnippet>1543</a>        <span class="self">self
-<a href=#1544 id=1544 data-nosnippet>1544</a>    </span>}
-<a href=#1545 id=1545 data-nosnippet>1545</a>
-<a href=#1546 id=1546 data-nosnippet>1546</a>    <span class="doccomment">/// Enable or disable preloading of the offset index
-<a href=#1547 id=1547 data-nosnippet>1547</a>    </span><span class="kw">pub fn </span>with_preload_offset_index(<span class="kw-2">mut </span><span class="self">self</span>, preload: bool) -&gt; <span class="self">Self </span>{
-<a href=#1548 id=1548 data-nosnippet>1548</a>        <span class="self">self</span>.preload_offset_index = preload;
-<a href=#1549 id=1549 data-nosnippet>1549</a>        <span class="self">self
-<a href=#1550 id=1550 data-nosnippet>1550</a>    </span>}
-<a href=#1551 id=1551 data-nosnippet>1551</a>
-<a href=#1552 id=1552 data-nosnippet>1552</a>    <span class="doccomment">/// Enable or disable preloading of the page index
-<a href=#1553 id=1553 data-nosnippet>1553</a>    </span><span class="kw">pub fn </span>with_preload_page_index(<span class="kw-2">mut </span><span class="self">self</span>, preload: bool) -&gt; <span class="self">Self </span>{
-<a href=#1554 id=1554 data-nosnippet>1554</a>        <span class="self">self</span>.preload_page_index = preload;
-<a href=#1555 id=1555 data-nosnippet>1555</a>        <span class="self">self
-<a href=#1556 id=1556 data-nosnippet>1556</a>    </span>}
+<a href=#1469 id=1469 data-nosnippet>1469</a>    <span class="kw">fn </span>greater_than_or_eq(
+<a href=#1470 id=1470 data-nosnippet>1470</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1471 id=1471 data-nosnippet>1471</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1472 id=1472 data-nosnippet>1472</a>        literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1473 id=1473 data-nosnippet>1473</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1474 id=1474 data-nosnippet>1474</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1475 id=1475 data-nosnippet>1475</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1476 id=1476 data-nosnippet>1476</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
+<a href=#1477 id=1477 data-nosnippet>1477</a>
+<a href=#1478 id=1478 data-nosnippet>1478</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1479 id=1479 data-nosnippet>1479</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1480 id=1480 data-nosnippet>1480</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1481 id=1481 data-nosnippet>1481</a>                gt_eq(<span class="kw-2">&amp;</span>left, literal.as_ref())
+<a href=#1482 id=1482 data-nosnippet>1482</a>            }))
+<a href=#1483 id=1483 data-nosnippet>1483</a>        } <span class="kw">else </span>{
+<a href=#1484 id=1484 data-nosnippet>1484</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1485 id=1485 data-nosnippet>1485</a>            </span><span class="self">self</span>.build_always_false()
+<a href=#1486 id=1486 data-nosnippet>1486</a>        }
+<a href=#1487 id=1487 data-nosnippet>1487</a>    }
+<a href=#1488 id=1488 data-nosnippet>1488</a>
+<a href=#1489 id=1489 data-nosnippet>1489</a>    <span class="kw">fn </span>eq(
+<a href=#1490 id=1490 data-nosnippet>1490</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1491 id=1491 data-nosnippet>1491</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1492 id=1492 data-nosnippet>1492</a>        literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1493 id=1493 data-nosnippet>1493</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1494 id=1494 data-nosnippet>1494</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1495 id=1495 data-nosnippet>1495</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1496 id=1496 data-nosnippet>1496</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
+<a href=#1497 id=1497 data-nosnippet>1497</a>
+<a href=#1498 id=1498 data-nosnippet>1498</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1499 id=1499 data-nosnippet>1499</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1500 id=1500 data-nosnippet>1500</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1501 id=1501 data-nosnippet>1501</a>                eq(<span class="kw-2">&amp;</span>left, literal.as_ref())
+<a href=#1502 id=1502 data-nosnippet>1502</a>            }))
+<a href=#1503 id=1503 data-nosnippet>1503</a>        } <span class="kw">else </span>{
+<a href=#1504 id=1504 data-nosnippet>1504</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1505 id=1505 data-nosnippet>1505</a>            </span><span class="self">self</span>.build_always_false()
+<a href=#1506 id=1506 data-nosnippet>1506</a>        }
+<a href=#1507 id=1507 data-nosnippet>1507</a>    }
+<a href=#1508 id=1508 data-nosnippet>1508</a>
+<a href=#1509 id=1509 data-nosnippet>1509</a>    <span class="kw">fn </span>not_eq(
+<a href=#1510 id=1510 data-nosnippet>1510</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1511 id=1511 data-nosnippet>1511</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1512 id=1512 data-nosnippet>1512</a>        literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1513 id=1513 data-nosnippet>1513</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1514 id=1514 data-nosnippet>1514</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1515 id=1515 data-nosnippet>1515</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1516 id=1516 data-nosnippet>1516</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
+<a href=#1517 id=1517 data-nosnippet>1517</a>
+<a href=#1518 id=1518 data-nosnippet>1518</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1519 id=1519 data-nosnippet>1519</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1520 id=1520 data-nosnippet>1520</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1521 id=1521 data-nosnippet>1521</a>                neq(<span class="kw-2">&amp;</span>left, literal.as_ref())
+<a href=#1522 id=1522 data-nosnippet>1522</a>            }))
+<a href=#1523 id=1523 data-nosnippet>1523</a>        } <span class="kw">else </span>{
+<a href=#1524 id=1524 data-nosnippet>1524</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1525 id=1525 data-nosnippet>1525</a>            </span><span class="self">self</span>.build_always_false()
+<a href=#1526 id=1526 data-nosnippet>1526</a>        }
+<a href=#1527 id=1527 data-nosnippet>1527</a>    }
+<a href=#1528 id=1528 data-nosnippet>1528</a>
+<a href=#1529 id=1529 data-nosnippet>1529</a>    <span class="kw">fn </span>starts_with(
+<a href=#1530 id=1530 data-nosnippet>1530</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1531 id=1531 data-nosnippet>1531</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1532 id=1532 data-nosnippet>1532</a>        literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1533 id=1533 data-nosnippet>1533</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1534 id=1534 data-nosnippet>1534</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1535 id=1535 data-nosnippet>1535</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1536 id=1536 data-nosnippet>1536</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
+<a href=#1537 id=1537 data-nosnippet>1537</a>
+<a href=#1538 id=1538 data-nosnippet>1538</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1539 id=1539 data-nosnippet>1539</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1540 id=1540 data-nosnippet>1540</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1541 id=1541 data-nosnippet>1541</a>                starts_with(<span class="kw-2">&amp;</span>left, literal.as_ref())
+<a href=#1542 id=1542 data-nosnippet>1542</a>            }))
+<a href=#1543 id=1543 data-nosnippet>1543</a>        } <span class="kw">else </span>{
+<a href=#1544 id=1544 data-nosnippet>1544</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1545 id=1545 data-nosnippet>1545</a>            </span><span class="self">self</span>.build_always_false()
+<a href=#1546 id=1546 data-nosnippet>1546</a>        }
+<a href=#1547 id=1547 data-nosnippet>1547</a>    }
+<a href=#1548 id=1548 data-nosnippet>1548</a>
+<a href=#1549 id=1549 data-nosnippet>1549</a>    <span class="kw">fn </span>not_starts_with(
+<a href=#1550 id=1550 data-nosnippet>1550</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1551 id=1551 data-nosnippet>1551</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1552 id=1552 data-nosnippet>1552</a>        literal: <span class="kw-2">&amp;</span>Datum,
+<a href=#1553 id=1553 data-nosnippet>1553</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1554 id=1554 data-nosnippet>1554</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1555 id=1555 data-nosnippet>1555</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1556 id=1556 data-nosnippet>1556</a>            <span class="kw">let </span>literal = get_arrow_datum(literal)<span class="question-mark">?</span>;
 <a href=#1557 id=1557 data-nosnippet>1557</a>
-<a href=#1558 id=1558 data-nosnippet>1558</a>    <span class="doccomment">/// Provide a hint as to the number of bytes to prefetch for parsing the Parquet metadata
-<a href=#1559 id=1559 data-nosnippet>1559</a>    ///
-<a href=#1560 id=1560 data-nosnippet>1560</a>    /// This hint can help reduce the number of fetch requests. For more details see the
-<a href=#1561 id=1561 data-nosnippet>1561</a>    /// [ParquetMetaDataReader documentation](https://docs.rs/parquet/latest/parquet/file/metadata/struct.ParquetMetaDataReader.html#method.with_prefetch_hint).
-<a href=#1562 id=1562 data-nosnippet>1562</a>    </span><span class="kw">pub fn </span>with_metadata_size_hint(<span class="kw-2">mut </span><span class="self">self</span>, hint: usize) -&gt; <span class="self">Self </span>{
-<a href=#1563 id=1563 data-nosnippet>1563</a>        <span class="self">self</span>.metadata_size_hint = <span class="prelude-val">Some</span>(hint);
-<a href=#1564 id=1564 data-nosnippet>1564</a>        <span class="self">self
-<a href=#1565 id=1565 data-nosnippet>1565</a>    </span>}
-<a href=#1566 id=1566 data-nosnippet>1566</a>}
-<a href=#1567 id=1567 data-nosnippet>1567</a>
-<a href=#1568 id=1568 data-nosnippet>1568</a><span class="kw">impl</span>&lt;R: FileRead&gt; AsyncFileReader <span class="kw">for </span>ArrowFileReader&lt;R&gt; {
-<a href=#1569 id=1569 data-nosnippet>1569</a>    <span class="kw">fn </span>get_bytes(<span class="kw-2">&amp;mut </span><span class="self">self</span>, range: Range&lt;u64&gt;) -&gt; BoxFuture&lt;<span class="lifetime">'_</span>, parquet::errors::Result&lt;Bytes&gt;&gt; {
-<a href=#1570 id=1570 data-nosnippet>1570</a>        Box::pin(
-<a href=#1571 id=1571 data-nosnippet>1571</a>            <span class="self">self</span>.r
-<a href=#1572 id=1572 data-nosnippet>1572</a>                .read(range.start..range.end)
-<a href=#1573 id=1573 data-nosnippet>1573</a>                .map_err(|err| parquet::errors::ParquetError::External(Box::new(err))),
-<a href=#1574 id=1574 data-nosnippet>1574</a>        )
-<a href=#1575 id=1575 data-nosnippet>1575</a>    }
-<a href=#1576 id=1576 data-nosnippet>1576</a>
-<a href=#1577 id=1577 data-nosnippet>1577</a>    <span class="comment">// TODO: currently we don't respect `ArrowReaderOptions` cause it don't expose any method to access the option field
-<a href=#1578 id=1578 data-nosnippet>1578</a>    // we will fix it after `v55.1.0` is released in https://github.com/apache/arrow-rs/issues/7393
-<a href=#1579 id=1579 data-nosnippet>1579</a>    </span><span class="kw">fn </span>get_metadata(
-<a href=#1580 id=1580 data-nosnippet>1580</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#1581 id=1581 data-nosnippet>1581</a>        _options: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span><span class="lifetime">'_ </span>ArrowReaderOptions&gt;,
-<a href=#1582 id=1582 data-nosnippet>1582</a>    ) -&gt; BoxFuture&lt;<span class="lifetime">'_</span>, parquet::errors::Result&lt;Arc&lt;ParquetMetaData&gt;&gt;&gt; {
-<a href=#1583 id=1583 data-nosnippet>1583</a>        <span class="kw">async move </span>{
-<a href=#1584 id=1584 data-nosnippet>1584</a>            <span class="kw">let </span>reader = ParquetMetaDataReader::new()
-<a href=#1585 id=1585 data-nosnippet>1585</a>                .with_prefetch_hint(<span class="self">self</span>.metadata_size_hint)
-<a href=#1586 id=1586 data-nosnippet>1586</a>                <span class="comment">// Set the page policy first because it updates both column and offset policies.
-<a href=#1587 id=1587 data-nosnippet>1587</a>                </span>.with_page_index_policy(PageIndexPolicy::from(<span class="self">self</span>.preload_page_index))
-<a href=#1588 id=1588 data-nosnippet>1588</a>                .with_column_index_policy(PageIndexPolicy::from(<span class="self">self</span>.preload_column_index))
-<a href=#1589 id=1589 data-nosnippet>1589</a>                .with_offset_index_policy(PageIndexPolicy::from(<span class="self">self</span>.preload_offset_index));
-<a href=#1590 id=1590 data-nosnippet>1590</a>            <span class="kw">let </span>size = <span class="self">self</span>.meta.size;
-<a href=#1591 id=1591 data-nosnippet>1591</a>            <span class="kw">let </span>meta = reader.load_and_finish(<span class="self">self</span>, size).<span class="kw">await</span><span class="question-mark">?</span>;
-<a href=#1592 id=1592 data-nosnippet>1592</a>
-<a href=#1593 id=1593 data-nosnippet>1593</a>            <span class="prelude-val">Ok</span>(Arc::new(meta))
-<a href=#1594 id=1594 data-nosnippet>1594</a>        }
-<a href=#1595 id=1595 data-nosnippet>1595</a>        .boxed()
-<a href=#1596 id=1596 data-nosnippet>1596</a>    }
-<a href=#1597 id=1597 data-nosnippet>1597</a>}
-<a href=#1598 id=1598 data-nosnippet>1598</a>
-<a href=#1599 id=1599 data-nosnippet>1599</a><span class="doccomment">/// The Arrow type of an array that the Parquet reader reads may not match the exact Arrow type
-<a href=#1600 id=1600 data-nosnippet>1600</a>/// that Iceberg uses for literals - but they are effectively the same logical type,
-<a href=#1601 id=1601 data-nosnippet>1601</a>/// i.e. LargeUtf8 and Utf8 or Utf8View and Utf8 or Utf8View and LargeUtf8.
-<a href=#1602 id=1602 data-nosnippet>1602</a>///
-<a href=#1603 id=1603 data-nosnippet>1603</a>/// The Arrow compute kernels that we use must match the type exactly, so first cast the literal
-<a href=#1604 id=1604 data-nosnippet>1604</a>/// into the type of the batch we read from Parquet before sending it to the compute kernel.
-<a href=#1605 id=1605 data-nosnippet>1605</a></span><span class="kw">fn </span>try_cast_literal(
-<a href=#1606 id=1606 data-nosnippet>1606</a>    literal: <span class="kw-2">&amp;</span>Arc&lt;<span class="kw">dyn </span>ArrowDatum + Send + Sync&gt;,
-<a href=#1607 id=1607 data-nosnippet>1607</a>    column_type: <span class="kw-2">&amp;</span>DataType,
-<a href=#1608 id=1608 data-nosnippet>1608</a>) -&gt; std::result::Result&lt;Arc&lt;<span class="kw">dyn </span>ArrowDatum + Send + Sync&gt;, ArrowError&gt; {
-<a href=#1609 id=1609 data-nosnippet>1609</a>    <span class="kw">let </span>literal_array = literal.get().<span class="number">0</span>;
-<a href=#1610 id=1610 data-nosnippet>1610</a>
-<a href=#1611 id=1611 data-nosnippet>1611</a>    <span class="comment">// No cast required
-<a href=#1612 id=1612 data-nosnippet>1612</a>    </span><span class="kw">if </span>literal_array.data_type() == column_type {
-<a href=#1613 id=1613 data-nosnippet>1613</a>        <span class="kw">return </span><span class="prelude-val">Ok</span>(Arc::clone(literal));
-<a href=#1614 id=1614 data-nosnippet>1614</a>    }
-<a href=#1615 id=1615 data-nosnippet>1615</a>
-<a href=#1616 id=1616 data-nosnippet>1616</a>    <span class="kw">let </span>literal_array = cast(literal_array, column_type)<span class="question-mark">?</span>;
-<a href=#1617 id=1617 data-nosnippet>1617</a>    <span class="prelude-val">Ok</span>(Arc::new(Scalar::new(literal_array)))
-<a href=#1618 id=1618 data-nosnippet>1618</a>}
-<a href=#1619 id=1619 data-nosnippet>1619</a>
-<a href=#1620 id=1620 data-nosnippet>1620</a><span class="attr">#[cfg(test)]
-<a href=#1621 id=1621 data-nosnippet>1621</a></span><span class="kw">mod </span>tests {
-<a href=#1622 id=1622 data-nosnippet>1622</a>    <span class="kw">use </span>std::collections::{HashMap, HashSet};
-<a href=#1623 id=1623 data-nosnippet>1623</a>    <span class="kw">use </span>std::fs::File;
-<a href=#1624 id=1624 data-nosnippet>1624</a>    <span class="kw">use </span>std::sync::Arc;
-<a href=#1625 id=1625 data-nosnippet>1625</a>
-<a href=#1626 id=1626 data-nosnippet>1626</a>    <span class="kw">use </span>arrow_array::cast::AsArray;
-<a href=#1627 id=1627 data-nosnippet>1627</a>    <span class="kw">use </span>arrow_array::{ArrayRef, LargeStringArray, RecordBatch, StringArray};
-<a href=#1628 id=1628 data-nosnippet>1628</a>    <span class="kw">use </span>arrow_schema::{DataType, Field, Schema <span class="kw">as </span>ArrowSchema, TimeUnit};
-<a href=#1629 id=1629 data-nosnippet>1629</a>    <span class="kw">use </span>futures::TryStreamExt;
-<a href=#1630 id=1630 data-nosnippet>1630</a>    <span class="kw">use </span>parquet::arrow::arrow_reader::{RowSelection, RowSelector};
-<a href=#1631 id=1631 data-nosnippet>1631</a>    <span class="kw">use </span>parquet::arrow::{ArrowWriter, ProjectionMask};
-<a href=#1632 id=1632 data-nosnippet>1632</a>    <span class="kw">use </span>parquet::basic::Compression;
-<a href=#1633 id=1633 data-nosnippet>1633</a>    <span class="kw">use </span>parquet::file::metadata::{ColumnChunkMetaData, RowGroupMetaData};
-<a href=#1634 id=1634 data-nosnippet>1634</a>    <span class="kw">use </span>parquet::file::properties::WriterProperties;
-<a href=#1635 id=1635 data-nosnippet>1635</a>    <span class="kw">use </span>parquet::schema::parser::parse_message_type;
-<a href=#1636 id=1636 data-nosnippet>1636</a>    <span class="kw">use </span>parquet::schema::types::{SchemaDescPtr, SchemaDescriptor};
-<a href=#1637 id=1637 data-nosnippet>1637</a>    <span class="kw">use </span>roaring::RoaringTreemap;
-<a href=#1638 id=1638 data-nosnippet>1638</a>    <span class="kw">use </span>tempfile::TempDir;
+<a href=#1558 id=1558 data-nosnippet>1558</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1559 id=1559 data-nosnippet>1559</a>                <span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1560 id=1560 data-nosnippet>1560</a>                <span class="kw">let </span>literal = try_cast_literal(<span class="kw-2">&amp;</span>literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1561 id=1561 data-nosnippet>1561</a>                <span class="comment">// update here if arrow ever adds a native not_starts_with
+<a href=#1562 id=1562 data-nosnippet>1562</a>                </span>not(<span class="kw-2">&amp;</span>starts_with(<span class="kw-2">&amp;</span>left, literal.as_ref())<span class="question-mark">?</span>)
+<a href=#1563 id=1563 data-nosnippet>1563</a>            }))
+<a href=#1564 id=1564 data-nosnippet>1564</a>        } <span class="kw">else </span>{
+<a href=#1565 id=1565 data-nosnippet>1565</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1566 id=1566 data-nosnippet>1566</a>            </span><span class="self">self</span>.build_always_true()
+<a href=#1567 id=1567 data-nosnippet>1567</a>        }
+<a href=#1568 id=1568 data-nosnippet>1568</a>    }
+<a href=#1569 id=1569 data-nosnippet>1569</a>
+<a href=#1570 id=1570 data-nosnippet>1570</a>    <span class="kw">fn </span>r#in(
+<a href=#1571 id=1571 data-nosnippet>1571</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1572 id=1572 data-nosnippet>1572</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1573 id=1573 data-nosnippet>1573</a>        literals: <span class="kw-2">&amp;</span>FnvHashSet&lt;Datum&gt;,
+<a href=#1574 id=1574 data-nosnippet>1574</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1575 id=1575 data-nosnippet>1575</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1576 id=1576 data-nosnippet>1576</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1577 id=1577 data-nosnippet>1577</a>            <span class="kw">let </span>literals: Vec&lt;<span class="kw">_</span>&gt; = literals
+<a href=#1578 id=1578 data-nosnippet>1578</a>                .iter()
+<a href=#1579 id=1579 data-nosnippet>1579</a>                .map(|lit| get_arrow_datum(lit).unwrap())
+<a href=#1580 id=1580 data-nosnippet>1580</a>                .collect();
+<a href=#1581 id=1581 data-nosnippet>1581</a>
+<a href=#1582 id=1582 data-nosnippet>1582</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1583 id=1583 data-nosnippet>1583</a>                <span class="comment">// update this if arrow ever adds a native is_in kernel
+<a href=#1584 id=1584 data-nosnippet>1584</a>                </span><span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1585 id=1585 data-nosnippet>1585</a>
+<a href=#1586 id=1586 data-nosnippet>1586</a>                <span class="kw">let </span><span class="kw-2">mut </span>acc = BooleanArray::from(<span class="macro">vec!</span>[<span class="bool-val">false</span>; batch.num_rows()]);
+<a href=#1587 id=1587 data-nosnippet>1587</a>                <span class="kw">for </span>literal <span class="kw">in </span><span class="kw-2">&amp;</span>literals {
+<a href=#1588 id=1588 data-nosnippet>1588</a>                    <span class="kw">let </span>literal = try_cast_literal(literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1589 id=1589 data-nosnippet>1589</a>                    acc = or(<span class="kw-2">&amp;</span>acc, <span class="kw-2">&amp;</span>eq(<span class="kw-2">&amp;</span>left, literal.as_ref())<span class="question-mark">?</span>)<span class="question-mark">?
+<a href=#1590 id=1590 data-nosnippet>1590</a>                </span>}
+<a href=#1591 id=1591 data-nosnippet>1591</a>
+<a href=#1592 id=1592 data-nosnippet>1592</a>                <span class="prelude-val">Ok</span>(acc)
+<a href=#1593 id=1593 data-nosnippet>1593</a>            }))
+<a href=#1594 id=1594 data-nosnippet>1594</a>        } <span class="kw">else </span>{
+<a href=#1595 id=1595 data-nosnippet>1595</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1596 id=1596 data-nosnippet>1596</a>            </span><span class="self">self</span>.build_always_false()
+<a href=#1597 id=1597 data-nosnippet>1597</a>        }
+<a href=#1598 id=1598 data-nosnippet>1598</a>    }
+<a href=#1599 id=1599 data-nosnippet>1599</a>
+<a href=#1600 id=1600 data-nosnippet>1600</a>    <span class="kw">fn </span>not_in(
+<a href=#1601 id=1601 data-nosnippet>1601</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1602 id=1602 data-nosnippet>1602</a>        reference: <span class="kw-2">&amp;</span>BoundReference,
+<a href=#1603 id=1603 data-nosnippet>1603</a>        literals: <span class="kw-2">&amp;</span>FnvHashSet&lt;Datum&gt;,
+<a href=#1604 id=1604 data-nosnippet>1604</a>        _predicate: <span class="kw-2">&amp;</span>BoundPredicate,
+<a href=#1605 id=1605 data-nosnippet>1605</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;PredicateResult&gt;&gt; {
+<a href=#1606 id=1606 data-nosnippet>1606</a>        <span class="kw">if let </span><span class="prelude-val">Some</span>(idx) = <span class="self">self</span>.bound_reference(reference)<span class="question-mark">? </span>{
+<a href=#1607 id=1607 data-nosnippet>1607</a>            <span class="kw">let </span>literals: Vec&lt;<span class="kw">_</span>&gt; = literals
+<a href=#1608 id=1608 data-nosnippet>1608</a>                .iter()
+<a href=#1609 id=1609 data-nosnippet>1609</a>                .map(|lit| get_arrow_datum(lit).unwrap())
+<a href=#1610 id=1610 data-nosnippet>1610</a>                .collect();
+<a href=#1611 id=1611 data-nosnippet>1611</a>
+<a href=#1612 id=1612 data-nosnippet>1612</a>            <span class="prelude-val">Ok</span>(Box::new(<span class="kw">move </span>|batch| {
+<a href=#1613 id=1613 data-nosnippet>1613</a>                <span class="comment">// update this if arrow ever adds a native not_in kernel
+<a href=#1614 id=1614 data-nosnippet>1614</a>                </span><span class="kw">let </span>left = project_column(<span class="kw-2">&amp;</span>batch, idx)<span class="question-mark">?</span>;
+<a href=#1615 id=1615 data-nosnippet>1615</a>                <span class="kw">let </span><span class="kw-2">mut </span>acc = BooleanArray::from(<span class="macro">vec!</span>[<span class="bool-val">true</span>; batch.num_rows()]);
+<a href=#1616 id=1616 data-nosnippet>1616</a>                <span class="kw">for </span>literal <span class="kw">in </span><span class="kw-2">&amp;</span>literals {
+<a href=#1617 id=1617 data-nosnippet>1617</a>                    <span class="kw">let </span>literal = try_cast_literal(literal, left.data_type())<span class="question-mark">?</span>;
+<a href=#1618 id=1618 data-nosnippet>1618</a>                    acc = and(<span class="kw-2">&amp;</span>acc, <span class="kw-2">&amp;</span>neq(<span class="kw-2">&amp;</span>left, literal.as_ref())<span class="question-mark">?</span>)<span class="question-mark">?
+<a href=#1619 id=1619 data-nosnippet>1619</a>                </span>}
+<a href=#1620 id=1620 data-nosnippet>1620</a>
+<a href=#1621 id=1621 data-nosnippet>1621</a>                <span class="prelude-val">Ok</span>(acc)
+<a href=#1622 id=1622 data-nosnippet>1622</a>            }))
+<a href=#1623 id=1623 data-nosnippet>1623</a>        } <span class="kw">else </span>{
+<a href=#1624 id=1624 data-nosnippet>1624</a>            <span class="comment">// A missing column, treating it as null.
+<a href=#1625 id=1625 data-nosnippet>1625</a>            </span><span class="self">self</span>.build_always_true()
+<a href=#1626 id=1626 data-nosnippet>1626</a>        }
+<a href=#1627 id=1627 data-nosnippet>1627</a>    }
+<a href=#1628 id=1628 data-nosnippet>1628</a>}
+<a href=#1629 id=1629 data-nosnippet>1629</a>
+<a href=#1630 id=1630 data-nosnippet>1630</a><span class="doccomment">/// ArrowFileReader is a wrapper around a FileRead that impls parquets AsyncFileReader.
+<a href=#1631 id=1631 data-nosnippet>1631</a></span><span class="kw">pub struct </span>ArrowFileReader&lt;R: FileRead&gt; {
+<a href=#1632 id=1632 data-nosnippet>1632</a>    meta: FileMetadata,
+<a href=#1633 id=1633 data-nosnippet>1633</a>    preload_column_index: bool,
+<a href=#1634 id=1634 data-nosnippet>1634</a>    preload_offset_index: bool,
+<a href=#1635 id=1635 data-nosnippet>1635</a>    preload_page_index: bool,
+<a href=#1636 id=1636 data-nosnippet>1636</a>    metadata_size_hint: <span class="prelude-ty">Option</span>&lt;usize&gt;,
+<a href=#1637 id=1637 data-nosnippet>1637</a>    r: R,
+<a href=#1638 id=1638 data-nosnippet>1638</a>}
 <a href=#1639 id=1639 data-nosnippet>1639</a>
-<a href=#1640 id=1640 data-nosnippet>1640</a>    <span class="kw">use </span><span class="kw">crate</span>::ErrorKind;
-<a href=#1641 id=1641 data-nosnippet>1641</a>    <span class="kw">use </span><span class="kw">crate</span>::arrow::reader::{CollectFieldIdVisitor, PARQUET_FIELD_ID_META_KEY};
-<a href=#1642 id=1642 data-nosnippet>1642</a>    <span class="kw">use </span><span class="kw">crate</span>::arrow::{ArrowReader, ArrowReaderBuilder};
-<a href=#1643 id=1643 data-nosnippet>1643</a>    <span class="kw">use </span><span class="kw">crate</span>::delete_vector::DeleteVector;
-<a href=#1644 id=1644 data-nosnippet>1644</a>    <span class="kw">use </span><span class="kw">crate</span>::expr::visitors::bound_predicate_visitor::visit;
-<a href=#1645 id=1645 data-nosnippet>1645</a>    <span class="kw">use </span><span class="kw">crate</span>::expr::{Bind, Predicate, Reference};
-<a href=#1646 id=1646 data-nosnippet>1646</a>    <span class="kw">use </span><span class="kw">crate</span>::io::FileIO;
-<a href=#1647 id=1647 data-nosnippet>1647</a>    <span class="kw">use </span><span class="kw">crate</span>::scan::{FileScanTask, FileScanTaskDeleteFile, FileScanTaskStream};
-<a href=#1648 id=1648 data-nosnippet>1648</a>    <span class="kw">use </span><span class="kw">crate</span>::spec::{
-<a href=#1649 id=1649 data-nosnippet>1649</a>        DataContentType, DataFileFormat, Datum, NestedField, PrimitiveType, Schema, SchemaRef, Type,
-<a href=#1650 id=1650 data-nosnippet>1650</a>    };
-<a href=#1651 id=1651 data-nosnippet>1651</a>
-<a href=#1652 id=1652 data-nosnippet>1652</a>    <span class="kw">fn </span>table_schema_simple() -&gt; SchemaRef {
-<a href=#1653 id=1653 data-nosnippet>1653</a>        Arc::new(
-<a href=#1654 id=1654 data-nosnippet>1654</a>            Schema::builder()
-<a href=#1655 id=1655 data-nosnippet>1655</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#1656 id=1656 data-nosnippet>1656</a>                .with_identifier_field_ids(<span class="macro">vec!</span>[<span class="number">2</span>])
-<a href=#1657 id=1657 data-nosnippet>1657</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#1658 id=1658 data-nosnippet>1658</a>                    NestedField::optional(<span class="number">1</span>, <span class="string">"foo"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#1659 id=1659 data-nosnippet>1659</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"bar"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#1660 id=1660 data-nosnippet>1660</a>                    NestedField::optional(<span class="number">3</span>, <span class="string">"baz"</span>, Type::Primitive(PrimitiveType::Boolean)).into(),
-<a href=#1661 id=1661 data-nosnippet>1661</a>                    NestedField::optional(<span class="number">4</span>, <span class="string">"qux"</span>, Type::Primitive(PrimitiveType::Float)).into(),
-<a href=#1662 id=1662 data-nosnippet>1662</a>                ])
-<a href=#1663 id=1663 data-nosnippet>1663</a>                .build()
-<a href=#1664 id=1664 data-nosnippet>1664</a>                .unwrap(),
-<a href=#1665 id=1665 data-nosnippet>1665</a>        )
-<a href=#1666 id=1666 data-nosnippet>1666</a>    }
-<a href=#1667 id=1667 data-nosnippet>1667</a>
-<a href=#1668 id=1668 data-nosnippet>1668</a>    <span class="attr">#[test]
-<a href=#1669 id=1669 data-nosnippet>1669</a>    </span><span class="kw">fn </span>test_collect_field_id() {
-<a href=#1670 id=1670 data-nosnippet>1670</a>        <span class="kw">let </span>schema = table_schema_simple();
-<a href=#1671 id=1671 data-nosnippet>1671</a>        <span class="kw">let </span>expr = Reference::new(<span class="string">"qux"</span>).is_null();
-<a href=#1672 id=1672 data-nosnippet>1672</a>        <span class="kw">let </span>bound_expr = expr.bind(schema, <span class="bool-val">true</span>).unwrap();
-<a href=#1673 id=1673 data-nosnippet>1673</a>
-<a href=#1674 id=1674 data-nosnippet>1674</a>        <span class="kw">let </span><span class="kw-2">mut </span>visitor = CollectFieldIdVisitor {
-<a href=#1675 id=1675 data-nosnippet>1675</a>            field_ids: HashSet::default(),
-<a href=#1676 id=1676 data-nosnippet>1676</a>        };
-<a href=#1677 id=1677 data-nosnippet>1677</a>        visit(<span class="kw-2">&amp;mut </span>visitor, <span class="kw-2">&amp;</span>bound_expr).unwrap();
-<a href=#1678 id=1678 data-nosnippet>1678</a>
-<a href=#1679 id=1679 data-nosnippet>1679</a>        <span class="kw">let </span><span class="kw-2">mut </span>expected = HashSet::default();
-<a href=#1680 id=1680 data-nosnippet>1680</a>        expected.insert(<span class="number">4_i32</span>);
-<a href=#1681 id=1681 data-nosnippet>1681</a>
-<a href=#1682 id=1682 data-nosnippet>1682</a>        <span class="macro">assert_eq!</span>(visitor.field_ids, expected);
-<a href=#1683 id=1683 data-nosnippet>1683</a>    }
-<a href=#1684 id=1684 data-nosnippet>1684</a>
-<a href=#1685 id=1685 data-nosnippet>1685</a>    <span class="attr">#[test]
-<a href=#1686 id=1686 data-nosnippet>1686</a>    </span><span class="kw">fn </span>test_collect_field_id_with_and() {
-<a href=#1687 id=1687 data-nosnippet>1687</a>        <span class="kw">let </span>schema = table_schema_simple();
-<a href=#1688 id=1688 data-nosnippet>1688</a>        <span class="kw">let </span>expr = Reference::new(<span class="string">"qux"</span>)
-<a href=#1689 id=1689 data-nosnippet>1689</a>            .is_null()
-<a href=#1690 id=1690 data-nosnippet>1690</a>            .and(Reference::new(<span class="string">"baz"</span>).is_null());
-<a href=#1691 id=1691 data-nosnippet>1691</a>        <span class="kw">let </span>bound_expr = expr.bind(schema, <span class="bool-val">true</span>).unwrap();
-<a href=#1692 id=1692 data-nosnippet>1692</a>
-<a href=#1693 id=1693 data-nosnippet>1693</a>        <span class="kw">let </span><span class="kw-2">mut </span>visitor = CollectFieldIdVisitor {
-<a href=#1694 id=1694 data-nosnippet>1694</a>            field_ids: HashSet::default(),
-<a href=#1695 id=1695 data-nosnippet>1695</a>        };
-<a href=#1696 id=1696 data-nosnippet>1696</a>        visit(<span class="kw-2">&amp;mut </span>visitor, <span class="kw-2">&amp;</span>bound_expr).unwrap();
-<a href=#1697 id=1697 data-nosnippet>1697</a>
-<a href=#1698 id=1698 data-nosnippet>1698</a>        <span class="kw">let </span><span class="kw-2">mut </span>expected = HashSet::default();
-<a href=#1699 id=1699 data-nosnippet>1699</a>        expected.insert(<span class="number">4_i32</span>);
-<a href=#1700 id=1700 data-nosnippet>1700</a>        expected.insert(<span class="number">3</span>);
-<a href=#1701 id=1701 data-nosnippet>1701</a>
-<a href=#1702 id=1702 data-nosnippet>1702</a>        <span class="macro">assert_eq!</span>(visitor.field_ids, expected);
-<a href=#1703 id=1703 data-nosnippet>1703</a>    }
-<a href=#1704 id=1704 data-nosnippet>1704</a>
-<a href=#1705 id=1705 data-nosnippet>1705</a>    <span class="attr">#[test]
-<a href=#1706 id=1706 data-nosnippet>1706</a>    </span><span class="kw">fn </span>test_collect_field_id_with_or() {
-<a href=#1707 id=1707 data-nosnippet>1707</a>        <span class="kw">let </span>schema = table_schema_simple();
-<a href=#1708 id=1708 data-nosnippet>1708</a>        <span class="kw">let </span>expr = Reference::new(<span class="string">"qux"</span>)
-<a href=#1709 id=1709 data-nosnippet>1709</a>            .is_null()
-<a href=#1710 id=1710 data-nosnippet>1710</a>            .or(Reference::new(<span class="string">"baz"</span>).is_null());
-<a href=#1711 id=1711 data-nosnippet>1711</a>        <span class="kw">let </span>bound_expr = expr.bind(schema, <span class="bool-val">true</span>).unwrap();
-<a href=#1712 id=1712 data-nosnippet>1712</a>
-<a href=#1713 id=1713 data-nosnippet>1713</a>        <span class="kw">let </span><span class="kw-2">mut </span>visitor = CollectFieldIdVisitor {
-<a href=#1714 id=1714 data-nosnippet>1714</a>            field_ids: HashSet::default(),
-<a href=#1715 id=1715 data-nosnippet>1715</a>        };
-<a href=#1716 id=1716 data-nosnippet>1716</a>        visit(<span class="kw-2">&amp;mut </span>visitor, <span class="kw-2">&amp;</span>bound_expr).unwrap();
-<a href=#1717 id=1717 data-nosnippet>1717</a>
-<a href=#1718 id=1718 data-nosnippet>1718</a>        <span class="kw">let </span><span class="kw-2">mut </span>expected = HashSet::default();
-<a href=#1719 id=1719 data-nosnippet>1719</a>        expected.insert(<span class="number">4_i32</span>);
-<a href=#1720 id=1720 data-nosnippet>1720</a>        expected.insert(<span class="number">3</span>);
-<a href=#1721 id=1721 data-nosnippet>1721</a>
-<a href=#1722 id=1722 data-nosnippet>1722</a>        <span class="macro">assert_eq!</span>(visitor.field_ids, expected);
-<a href=#1723 id=1723 data-nosnippet>1723</a>    }
-<a href=#1724 id=1724 data-nosnippet>1724</a>
-<a href=#1725 id=1725 data-nosnippet>1725</a>    <span class="attr">#[test]
-<a href=#1726 id=1726 data-nosnippet>1726</a>    </span><span class="kw">fn </span>test_arrow_projection_mask() {
-<a href=#1727 id=1727 data-nosnippet>1727</a>        <span class="kw">let </span>schema = Arc::new(
-<a href=#1728 id=1728 data-nosnippet>1728</a>            Schema::builder()
-<a href=#1729 id=1729 data-nosnippet>1729</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#1730 id=1730 data-nosnippet>1730</a>                .with_identifier_field_ids(<span class="macro">vec!</span>[<span class="number">1</span>])
-<a href=#1731 id=1731 data-nosnippet>1731</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#1732 id=1732 data-nosnippet>1732</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"c1"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#1733 id=1733 data-nosnippet>1733</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"c2"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#1734 id=1734 data-nosnippet>1734</a>                    NestedField::optional(
-<a href=#1735 id=1735 data-nosnippet>1735</a>                        <span class="number">3</span>,
-<a href=#1736 id=1736 data-nosnippet>1736</a>                        <span class="string">"c3"</span>,
-<a href=#1737 id=1737 data-nosnippet>1737</a>                        Type::Primitive(PrimitiveType::Decimal {
-<a href=#1738 id=1738 data-nosnippet>1738</a>                            precision: <span class="number">38</span>,
-<a href=#1739 id=1739 data-nosnippet>1739</a>                            scale: <span class="number">3</span>,
-<a href=#1740 id=1740 data-nosnippet>1740</a>                        }),
-<a href=#1741 id=1741 data-nosnippet>1741</a>                    )
-<a href=#1742 id=1742 data-nosnippet>1742</a>                    .into(),
-<a href=#1743 id=1743 data-nosnippet>1743</a>                ])
-<a href=#1744 id=1744 data-nosnippet>1744</a>                .build()
-<a href=#1745 id=1745 data-nosnippet>1745</a>                .unwrap(),
-<a href=#1746 id=1746 data-nosnippet>1746</a>        );
-<a href=#1747 id=1747 data-nosnippet>1747</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#1748 id=1748 data-nosnippet>1748</a>            Field::new(<span class="string">"c1"</span>, DataType::Utf8, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#1749 id=1749 data-nosnippet>1749</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#1750 id=1750 data-nosnippet>1750</a>                <span class="string">"1"</span>.to_string(),
-<a href=#1751 id=1751 data-nosnippet>1751</a>            )])),
-<a href=#1752 id=1752 data-nosnippet>1752</a>            <span class="comment">// Type not supported
-<a href=#1753 id=1753 data-nosnippet>1753</a>            </span>Field::new(<span class="string">"c2"</span>, DataType::Duration(TimeUnit::Microsecond), <span class="bool-val">true</span>).with_metadata(
-<a href=#1754 id=1754 data-nosnippet>1754</a>                HashMap::from([(PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"2"</span>.to_string())]),
-<a href=#1755 id=1755 data-nosnippet>1755</a>            ),
-<a href=#1756 id=1756 data-nosnippet>1756</a>            <span class="comment">// Precision is beyond the supported range
-<a href=#1757 id=1757 data-nosnippet>1757</a>            </span>Field::new(<span class="string">"c3"</span>, DataType::Decimal128(<span class="number">39</span>, <span class="number">3</span>), <span class="bool-val">true</span>).with_metadata(HashMap::from([(
-<a href=#1758 id=1758 data-nosnippet>1758</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#1759 id=1759 data-nosnippet>1759</a>                <span class="string">"3"</span>.to_string(),
-<a href=#1760 id=1760 data-nosnippet>1760</a>            )])),
-<a href=#1761 id=1761 data-nosnippet>1761</a>        ]));
-<a href=#1762 id=1762 data-nosnippet>1762</a>
-<a href=#1763 id=1763 data-nosnippet>1763</a>        <span class="kw">let </span>message_type = <span class="string">"
-<a href=#1764 id=1764 data-nosnippet>1764</a>message schema {
-<a href=#1765 id=1765 data-nosnippet>1765</a>  required binary c1 (STRING) = 1;
-<a href=#1766 id=1766 data-nosnippet>1766</a>  optional int32 c2 (INTEGER(8,true)) = 2;
-<a href=#1767 id=1767 data-nosnippet>1767</a>  optional fixed_len_byte_array(17) c3 (DECIMAL(39,3)) = 3;
-<a href=#1768 id=1768 data-nosnippet>1768</a>}
-<a href=#1769 id=1769 data-nosnippet>1769</a>    "</span>;
-<a href=#1770 id=1770 data-nosnippet>1770</a>        <span class="kw">let </span>parquet_type = parse_message_type(message_type).expect(<span class="string">"should parse schema"</span>);
-<a href=#1771 id=1771 data-nosnippet>1771</a>        <span class="kw">let </span>parquet_schema = SchemaDescriptor::new(Arc::new(parquet_type));
-<a href=#1772 id=1772 data-nosnippet>1772</a>
-<a href=#1773 id=1773 data-nosnippet>1773</a>        <span class="comment">// Try projecting the fields c2 and c3 with the unsupported data types
-<a href=#1774 id=1774 data-nosnippet>1774</a>        </span><span class="kw">let </span>err = ArrowReader::get_arrow_projection_mask(
-<a href=#1775 id=1775 data-nosnippet>1775</a>            <span class="kw-2">&amp;</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>],
-<a href=#1776 id=1776 data-nosnippet>1776</a>            <span class="kw-2">&amp;</span>schema,
-<a href=#1777 id=1777 data-nosnippet>1777</a>            <span class="kw-2">&amp;</span>parquet_schema,
-<a href=#1778 id=1778 data-nosnippet>1778</a>            <span class="kw-2">&amp;</span>arrow_schema,
-<a href=#1779 id=1779 data-nosnippet>1779</a>            <span class="bool-val">false</span>,
-<a href=#1780 id=1780 data-nosnippet>1780</a>        )
-<a href=#1781 id=1781 data-nosnippet>1781</a>        .unwrap_err();
-<a href=#1782 id=1782 data-nosnippet>1782</a>
-<a href=#1783 id=1783 data-nosnippet>1783</a>        <span class="macro">assert_eq!</span>(err.kind(), ErrorKind::DataInvalid);
-<a href=#1784 id=1784 data-nosnippet>1784</a>        <span class="macro">assert_eq!</span>(
-<a href=#1785 id=1785 data-nosnippet>1785</a>            err.to_string(),
-<a href=#1786 id=1786 data-nosnippet>1786</a>            <span class="string">"DataInvalid =&gt; Unsupported Arrow data type: Duration(Microsecond)"</span>.to_string()
-<a href=#1787 id=1787 data-nosnippet>1787</a>        );
-<a href=#1788 id=1788 data-nosnippet>1788</a>
-<a href=#1789 id=1789 data-nosnippet>1789</a>        <span class="comment">// Omitting field c2, we still get an error due to c3 being selected
-<a href=#1790 id=1790 data-nosnippet>1790</a>        </span><span class="kw">let </span>err = ArrowReader::get_arrow_projection_mask(
-<a href=#1791 id=1791 data-nosnippet>1791</a>            <span class="kw-2">&amp;</span>[<span class="number">1</span>, <span class="number">3</span>],
-<a href=#1792 id=1792 data-nosnippet>1792</a>            <span class="kw-2">&amp;</span>schema,
-<a href=#1793 id=1793 data-nosnippet>1793</a>            <span class="kw-2">&amp;</span>parquet_schema,
-<a href=#1794 id=1794 data-nosnippet>1794</a>            <span class="kw-2">&amp;</span>arrow_schema,
-<a href=#1795 id=1795 data-nosnippet>1795</a>            <span class="bool-val">false</span>,
-<a href=#1796 id=1796 data-nosnippet>1796</a>        )
-<a href=#1797 id=1797 data-nosnippet>1797</a>        .unwrap_err();
-<a href=#1798 id=1798 data-nosnippet>1798</a>
-<a href=#1799 id=1799 data-nosnippet>1799</a>        <span class="macro">assert_eq!</span>(err.kind(), ErrorKind::DataInvalid);
-<a href=#1800 id=1800 data-nosnippet>1800</a>        <span class="macro">assert_eq!</span>(
-<a href=#1801 id=1801 data-nosnippet>1801</a>            err.to_string(),
-<a href=#1802 id=1802 data-nosnippet>1802</a>            <span class="string">"DataInvalid =&gt; Failed to create decimal type, source: DataInvalid =&gt; Decimals with precision larger than 38 are not supported: 39"</span>.to_string()
-<a href=#1803 id=1803 data-nosnippet>1803</a>        );
-<a href=#1804 id=1804 data-nosnippet>1804</a>
-<a href=#1805 id=1805 data-nosnippet>1805</a>        <span class="comment">// Finally avoid selecting fields with unsupported data types
-<a href=#1806 id=1806 data-nosnippet>1806</a>        </span><span class="kw">let </span>mask = ArrowReader::get_arrow_projection_mask(
-<a href=#1807 id=1807 data-nosnippet>1807</a>            <span class="kw-2">&amp;</span>[<span class="number">1</span>],
-<a href=#1808 id=1808 data-nosnippet>1808</a>            <span class="kw-2">&amp;</span>schema,
-<a href=#1809 id=1809 data-nosnippet>1809</a>            <span class="kw-2">&amp;</span>parquet_schema,
-<a href=#1810 id=1810 data-nosnippet>1810</a>            <span class="kw-2">&amp;</span>arrow_schema,
-<a href=#1811 id=1811 data-nosnippet>1811</a>            <span class="bool-val">false</span>,
-<a href=#1812 id=1812 data-nosnippet>1812</a>        )
-<a href=#1813 id=1813 data-nosnippet>1813</a>        .expect(<span class="string">"Some ProjectionMask"</span>);
-<a href=#1814 id=1814 data-nosnippet>1814</a>        <span class="macro">assert_eq!</span>(mask, ProjectionMask::leaves(<span class="kw-2">&amp;</span>parquet_schema, <span class="macro">vec!</span>[<span class="number">0</span>]));
-<a href=#1815 id=1815 data-nosnippet>1815</a>    }
-<a href=#1816 id=1816 data-nosnippet>1816</a>
-<a href=#1817 id=1817 data-nosnippet>1817</a>    <span class="attr">#[tokio::test]
-<a href=#1818 id=1818 data-nosnippet>1818</a>    </span><span class="kw">async fn </span>test_kleene_logic_or_behaviour() {
-<a href=#1819 id=1819 data-nosnippet>1819</a>        <span class="comment">// a IS NULL OR a = 'foo'
-<a href=#1820 id=1820 data-nosnippet>1820</a>        </span><span class="kw">let </span>predicate = Reference::new(<span class="string">"a"</span>)
-<a href=#1821 id=1821 data-nosnippet>1821</a>            .is_null()
-<a href=#1822 id=1822 data-nosnippet>1822</a>            .or(Reference::new(<span class="string">"a"</span>).equal_to(Datum::string(<span class="string">"foo"</span>)));
-<a href=#1823 id=1823 data-nosnippet>1823</a>
-<a href=#1824 id=1824 data-nosnippet>1824</a>        <span class="comment">// Table data: [NULL, "foo", "bar"]
-<a href=#1825 id=1825 data-nosnippet>1825</a>        </span><span class="kw">let </span>data_for_col_a = <span class="macro">vec!</span>[<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()), <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())];
-<a href=#1826 id=1826 data-nosnippet>1826</a>
-<a href=#1827 id=1827 data-nosnippet>1827</a>        <span class="comment">// Expected: [NULL, "foo"].
-<a href=#1828 id=1828 data-nosnippet>1828</a>        </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string())];
-<a href=#1829 id=1829 data-nosnippet>1829</a>
-<a href=#1830 id=1830 data-nosnippet>1830</a>        <span class="kw">let </span>(file_io, schema, table_location, _temp_dir) =
-<a href=#1831 id=1831 data-nosnippet>1831</a>            setup_kleene_logic(data_for_col_a, DataType::Utf8);
-<a href=#1832 id=1832 data-nosnippet>1832</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#1833 id=1833 data-nosnippet>1833</a>
-<a href=#1834 id=1834 data-nosnippet>1834</a>        <span class="kw">let </span>result_data = test_perform_read(predicate, schema, table_location, reader).<span class="kw">await</span>;
-<a href=#1835 id=1835 data-nosnippet>1835</a>
-<a href=#1836 id=1836 data-nosnippet>1836</a>        <span class="macro">assert_eq!</span>(result_data, expected);
-<a href=#1837 id=1837 data-nosnippet>1837</a>    }
-<a href=#1838 id=1838 data-nosnippet>1838</a>
-<a href=#1839 id=1839 data-nosnippet>1839</a>    <span class="attr">#[tokio::test]
-<a href=#1840 id=1840 data-nosnippet>1840</a>    </span><span class="kw">async fn </span>test_kleene_logic_and_behaviour() {
-<a href=#1841 id=1841 data-nosnippet>1841</a>        <span class="comment">// a IS NOT NULL AND a != 'foo'
-<a href=#1842 id=1842 data-nosnippet>1842</a>        </span><span class="kw">let </span>predicate = Reference::new(<span class="string">"a"</span>)
-<a href=#1843 id=1843 data-nosnippet>1843</a>            .is_not_null()
-<a href=#1844 id=1844 data-nosnippet>1844</a>            .and(Reference::new(<span class="string">"a"</span>).not_equal_to(Datum::string(<span class="string">"foo"</span>)));
-<a href=#1845 id=1845 data-nosnippet>1845</a>
-<a href=#1846 id=1846 data-nosnippet>1846</a>        <span class="comment">// Table data: [NULL, "foo", "bar"]
-<a href=#1847 id=1847 data-nosnippet>1847</a>        </span><span class="kw">let </span>data_for_col_a = <span class="macro">vec!</span>[<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()), <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())];
-<a href=#1848 id=1848 data-nosnippet>1848</a>
-<a href=#1849 id=1849 data-nosnippet>1849</a>        <span class="comment">// Expected: ["bar"].
-<a href=#1850 id=1850 data-nosnippet>1850</a>        </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())];
-<a href=#1851 id=1851 data-nosnippet>1851</a>
-<a href=#1852 id=1852 data-nosnippet>1852</a>        <span class="kw">let </span>(file_io, schema, table_location, _temp_dir) =
-<a href=#1853 id=1853 data-nosnippet>1853</a>            setup_kleene_logic(data_for_col_a, DataType::Utf8);
-<a href=#1854 id=1854 data-nosnippet>1854</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#1855 id=1855 data-nosnippet>1855</a>
-<a href=#1856 id=1856 data-nosnippet>1856</a>        <span class="kw">let </span>result_data = test_perform_read(predicate, schema, table_location, reader).<span class="kw">await</span>;
-<a href=#1857 id=1857 data-nosnippet>1857</a>
-<a href=#1858 id=1858 data-nosnippet>1858</a>        <span class="macro">assert_eq!</span>(result_data, expected);
-<a href=#1859 id=1859 data-nosnippet>1859</a>    }
-<a href=#1860 id=1860 data-nosnippet>1860</a>
-<a href=#1861 id=1861 data-nosnippet>1861</a>    <span class="attr">#[tokio::test]
-<a href=#1862 id=1862 data-nosnippet>1862</a>    </span><span class="kw">async fn </span>test_predicate_cast_literal() {
-<a href=#1863 id=1863 data-nosnippet>1863</a>        <span class="kw">let </span>predicates = <span class="macro">vec!</span>[
-<a href=#1864 id=1864 data-nosnippet>1864</a>            <span class="comment">// a == 'foo'
-<a href=#1865 id=1865 data-nosnippet>1865</a>            </span>(Reference::new(<span class="string">"a"</span>).equal_to(Datum::string(<span class="string">"foo"</span>)), <span class="macro">vec!</span>[
-<a href=#1866 id=1866 data-nosnippet>1866</a>                <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()),
-<a href=#1867 id=1867 data-nosnippet>1867</a>            ]),
-<a href=#1868 id=1868 data-nosnippet>1868</a>            <span class="comment">// a != 'foo'
-<a href=#1869 id=1869 data-nosnippet>1869</a>            </span>(
-<a href=#1870 id=1870 data-nosnippet>1870</a>                Reference::new(<span class="string">"a"</span>).not_equal_to(Datum::string(<span class="string">"foo"</span>)),
-<a href=#1871 id=1871 data-nosnippet>1871</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())],
-<a href=#1872 id=1872 data-nosnippet>1872</a>            ),
-<a href=#1873 id=1873 data-nosnippet>1873</a>            <span class="comment">// STARTS_WITH(a, 'foo')
-<a href=#1874 id=1874 data-nosnippet>1874</a>            </span>(Reference::new(<span class="string">"a"</span>).starts_with(Datum::string(<span class="string">"f"</span>)), <span class="macro">vec!</span>[
-<a href=#1875 id=1875 data-nosnippet>1875</a>                <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()),
-<a href=#1876 id=1876 data-nosnippet>1876</a>            ]),
-<a href=#1877 id=1877 data-nosnippet>1877</a>            <span class="comment">// NOT STARTS_WITH(a, 'foo')
-<a href=#1878 id=1878 data-nosnippet>1878</a>            </span>(
-<a href=#1879 id=1879 data-nosnippet>1879</a>                Reference::new(<span class="string">"a"</span>).not_starts_with(Datum::string(<span class="string">"f"</span>)),
-<a href=#1880 id=1880 data-nosnippet>1880</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())],
-<a href=#1881 id=1881 data-nosnippet>1881</a>            ),
-<a href=#1882 id=1882 data-nosnippet>1882</a>            <span class="comment">// a &lt; 'foo'
-<a href=#1883 id=1883 data-nosnippet>1883</a>            </span>(Reference::new(<span class="string">"a"</span>).less_than(Datum::string(<span class="string">"foo"</span>)), <span class="macro">vec!</span>[
-<a href=#1884 id=1884 data-nosnippet>1884</a>                <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string()),
-<a href=#1885 id=1885 data-nosnippet>1885</a>            ]),
-<a href=#1886 id=1886 data-nosnippet>1886</a>            <span class="comment">// a &lt;= 'foo'
-<a href=#1887 id=1887 data-nosnippet>1887</a>            </span>(
-<a href=#1888 id=1888 data-nosnippet>1888</a>                Reference::new(<span class="string">"a"</span>).less_than_or_equal_to(Datum::string(<span class="string">"foo"</span>)),
-<a href=#1889 id=1889 data-nosnippet>1889</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()), <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())],
-<a href=#1890 id=1890 data-nosnippet>1890</a>            ),
-<a href=#1891 id=1891 data-nosnippet>1891</a>            <span class="comment">// a &gt; 'foo'
-<a href=#1892 id=1892 data-nosnippet>1892</a>            </span>(
-<a href=#1893 id=1893 data-nosnippet>1893</a>                Reference::new(<span class="string">"a"</span>).greater_than(Datum::string(<span class="string">"bar"</span>)),
-<a href=#1894 id=1894 data-nosnippet>1894</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string())],
-<a href=#1895 id=1895 data-nosnippet>1895</a>            ),
-<a href=#1896 id=1896 data-nosnippet>1896</a>            <span class="comment">// a &gt;= 'foo'
-<a href=#1897 id=1897 data-nosnippet>1897</a>            </span>(
-<a href=#1898 id=1898 data-nosnippet>1898</a>                Reference::new(<span class="string">"a"</span>).greater_than_or_equal_to(Datum::string(<span class="string">"foo"</span>)),
-<a href=#1899 id=1899 data-nosnippet>1899</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string())],
-<a href=#1900 id=1900 data-nosnippet>1900</a>            ),
-<a href=#1901 id=1901 data-nosnippet>1901</a>            <span class="comment">// a IN ('foo', 'bar')
-<a href=#1902 id=1902 data-nosnippet>1902</a>            </span>(
-<a href=#1903 id=1903 data-nosnippet>1903</a>                Reference::new(<span class="string">"a"</span>).is_in([Datum::string(<span class="string">"foo"</span>), Datum::string(<span class="string">"baz"</span>)]),
-<a href=#1904 id=1904 data-nosnippet>1904</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string())],
-<a href=#1905 id=1905 data-nosnippet>1905</a>            ),
-<a href=#1906 id=1906 data-nosnippet>1906</a>            <span class="comment">// a NOT IN ('foo', 'bar')
-<a href=#1907 id=1907 data-nosnippet>1907</a>            </span>(
-<a href=#1908 id=1908 data-nosnippet>1908</a>                Reference::new(<span class="string">"a"</span>).is_not_in([Datum::string(<span class="string">"foo"</span>), Datum::string(<span class="string">"baz"</span>)]),
-<a href=#1909 id=1909 data-nosnippet>1909</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())],
-<a href=#1910 id=1910 data-nosnippet>1910</a>            ),
-<a href=#1911 id=1911 data-nosnippet>1911</a>        ];
-<a href=#1912 id=1912 data-nosnippet>1912</a>
-<a href=#1913 id=1913 data-nosnippet>1913</a>        <span class="comment">// Table data: ["foo", "bar"]
-<a href=#1914 id=1914 data-nosnippet>1914</a>        </span><span class="kw">let </span>data_for_col_a = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()), <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())];
-<a href=#1915 id=1915 data-nosnippet>1915</a>
-<a href=#1916 id=1916 data-nosnippet>1916</a>        <span class="kw">let </span>(file_io, schema, table_location, _temp_dir) =
-<a href=#1917 id=1917 data-nosnippet>1917</a>            setup_kleene_logic(data_for_col_a, DataType::LargeUtf8);
-<a href=#1918 id=1918 data-nosnippet>1918</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#1919 id=1919 data-nosnippet>1919</a>
-<a href=#1920 id=1920 data-nosnippet>1920</a>        <span class="kw">for </span>(predicate, expected) <span class="kw">in </span>predicates {
-<a href=#1921 id=1921 data-nosnippet>1921</a>            <span class="macro">println!</span>(<span class="string">"testing predicate {predicate}"</span>);
-<a href=#1922 id=1922 data-nosnippet>1922</a>            <span class="kw">let </span>result_data = test_perform_read(
-<a href=#1923 id=1923 data-nosnippet>1923</a>                predicate.clone(),
-<a href=#1924 id=1924 data-nosnippet>1924</a>                schema.clone(),
-<a href=#1925 id=1925 data-nosnippet>1925</a>                table_location.clone(),
-<a href=#1926 id=1926 data-nosnippet>1926</a>                reader.clone(),
-<a href=#1927 id=1927 data-nosnippet>1927</a>            )
-<a href=#1928 id=1928 data-nosnippet>1928</a>            .<span class="kw">await</span>;
+<a href=#1640 id=1640 data-nosnippet>1640</a><span class="kw">impl</span>&lt;R: FileRead&gt; ArrowFileReader&lt;R&gt; {
+<a href=#1641 id=1641 data-nosnippet>1641</a>    <span class="doccomment">/// Create a new ArrowFileReader
+<a href=#1642 id=1642 data-nosnippet>1642</a>    </span><span class="kw">pub fn </span>new(meta: FileMetadata, r: R) -&gt; <span class="self">Self </span>{
+<a href=#1643 id=1643 data-nosnippet>1643</a>        <span class="self">Self </span>{
+<a href=#1644 id=1644 data-nosnippet>1644</a>            meta,
+<a href=#1645 id=1645 data-nosnippet>1645</a>            preload_column_index: <span class="bool-val">false</span>,
+<a href=#1646 id=1646 data-nosnippet>1646</a>            preload_offset_index: <span class="bool-val">false</span>,
+<a href=#1647 id=1647 data-nosnippet>1647</a>            preload_page_index: <span class="bool-val">false</span>,
+<a href=#1648 id=1648 data-nosnippet>1648</a>            metadata_size_hint: <span class="prelude-val">None</span>,
+<a href=#1649 id=1649 data-nosnippet>1649</a>            r,
+<a href=#1650 id=1650 data-nosnippet>1650</a>        }
+<a href=#1651 id=1651 data-nosnippet>1651</a>    }
+<a href=#1652 id=1652 data-nosnippet>1652</a>
+<a href=#1653 id=1653 data-nosnippet>1653</a>    <span class="doccomment">/// Enable or disable preloading of the column index
+<a href=#1654 id=1654 data-nosnippet>1654</a>    </span><span class="kw">pub fn </span>with_preload_column_index(<span class="kw-2">mut </span><span class="self">self</span>, preload: bool) -&gt; <span class="self">Self </span>{
+<a href=#1655 id=1655 data-nosnippet>1655</a>        <span class="self">self</span>.preload_column_index = preload;
+<a href=#1656 id=1656 data-nosnippet>1656</a>        <span class="self">self
+<a href=#1657 id=1657 data-nosnippet>1657</a>    </span>}
+<a href=#1658 id=1658 data-nosnippet>1658</a>
+<a href=#1659 id=1659 data-nosnippet>1659</a>    <span class="doccomment">/// Enable or disable preloading of the offset index
+<a href=#1660 id=1660 data-nosnippet>1660</a>    </span><span class="kw">pub fn </span>with_preload_offset_index(<span class="kw-2">mut </span><span class="self">self</span>, preload: bool) -&gt; <span class="self">Self </span>{
+<a href=#1661 id=1661 data-nosnippet>1661</a>        <span class="self">self</span>.preload_offset_index = preload;
+<a href=#1662 id=1662 data-nosnippet>1662</a>        <span class="self">self
+<a href=#1663 id=1663 data-nosnippet>1663</a>    </span>}
+<a href=#1664 id=1664 data-nosnippet>1664</a>
+<a href=#1665 id=1665 data-nosnippet>1665</a>    <span class="doccomment">/// Enable or disable preloading of the page index
+<a href=#1666 id=1666 data-nosnippet>1666</a>    </span><span class="kw">pub fn </span>with_preload_page_index(<span class="kw-2">mut </span><span class="self">self</span>, preload: bool) -&gt; <span class="self">Self </span>{
+<a href=#1667 id=1667 data-nosnippet>1667</a>        <span class="self">self</span>.preload_page_index = preload;
+<a href=#1668 id=1668 data-nosnippet>1668</a>        <span class="self">self
+<a href=#1669 id=1669 data-nosnippet>1669</a>    </span>}
+<a href=#1670 id=1670 data-nosnippet>1670</a>
+<a href=#1671 id=1671 data-nosnippet>1671</a>    <span class="doccomment">/// Provide a hint as to the number of bytes to prefetch for parsing the Parquet metadata
+<a href=#1672 id=1672 data-nosnippet>1672</a>    ///
+<a href=#1673 id=1673 data-nosnippet>1673</a>    /// This hint can help reduce the number of fetch requests. For more details see the
+<a href=#1674 id=1674 data-nosnippet>1674</a>    /// [ParquetMetaDataReader documentation](https://docs.rs/parquet/latest/parquet/file/metadata/struct.ParquetMetaDataReader.html#method.with_prefetch_hint).
+<a href=#1675 id=1675 data-nosnippet>1675</a>    </span><span class="kw">pub fn </span>with_metadata_size_hint(<span class="kw-2">mut </span><span class="self">self</span>, hint: usize) -&gt; <span class="self">Self </span>{
+<a href=#1676 id=1676 data-nosnippet>1676</a>        <span class="self">self</span>.metadata_size_hint = <span class="prelude-val">Some</span>(hint);
+<a href=#1677 id=1677 data-nosnippet>1677</a>        <span class="self">self
+<a href=#1678 id=1678 data-nosnippet>1678</a>    </span>}
+<a href=#1679 id=1679 data-nosnippet>1679</a>}
+<a href=#1680 id=1680 data-nosnippet>1680</a>
+<a href=#1681 id=1681 data-nosnippet>1681</a><span class="kw">impl</span>&lt;R: FileRead&gt; AsyncFileReader <span class="kw">for </span>ArrowFileReader&lt;R&gt; {
+<a href=#1682 id=1682 data-nosnippet>1682</a>    <span class="kw">fn </span>get_bytes(<span class="kw-2">&amp;mut </span><span class="self">self</span>, range: Range&lt;u64&gt;) -&gt; BoxFuture&lt;<span class="lifetime">'_</span>, parquet::errors::Result&lt;Bytes&gt;&gt; {
+<a href=#1683 id=1683 data-nosnippet>1683</a>        Box::pin(
+<a href=#1684 id=1684 data-nosnippet>1684</a>            <span class="self">self</span>.r
+<a href=#1685 id=1685 data-nosnippet>1685</a>                .read(range.start..range.end)
+<a href=#1686 id=1686 data-nosnippet>1686</a>                .map_err(|err| parquet::errors::ParquetError::External(Box::new(err))),
+<a href=#1687 id=1687 data-nosnippet>1687</a>        )
+<a href=#1688 id=1688 data-nosnippet>1688</a>    }
+<a href=#1689 id=1689 data-nosnippet>1689</a>
+<a href=#1690 id=1690 data-nosnippet>1690</a>    <span class="comment">// TODO: currently we don't respect `ArrowReaderOptions` cause it don't expose any method to access the option field
+<a href=#1691 id=1691 data-nosnippet>1691</a>    // we will fix it after `v55.1.0` is released in https://github.com/apache/arrow-rs/issues/7393
+<a href=#1692 id=1692 data-nosnippet>1692</a>    </span><span class="kw">fn </span>get_metadata(
+<a href=#1693 id=1693 data-nosnippet>1693</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#1694 id=1694 data-nosnippet>1694</a>        _options: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span><span class="lifetime">'_ </span>ArrowReaderOptions&gt;,
+<a href=#1695 id=1695 data-nosnippet>1695</a>    ) -&gt; BoxFuture&lt;<span class="lifetime">'_</span>, parquet::errors::Result&lt;Arc&lt;ParquetMetaData&gt;&gt;&gt; {
+<a href=#1696 id=1696 data-nosnippet>1696</a>        <span class="kw">async move </span>{
+<a href=#1697 id=1697 data-nosnippet>1697</a>            <span class="kw">let </span>reader = ParquetMetaDataReader::new()
+<a href=#1698 id=1698 data-nosnippet>1698</a>                .with_prefetch_hint(<span class="self">self</span>.metadata_size_hint)
+<a href=#1699 id=1699 data-nosnippet>1699</a>                <span class="comment">// Set the page policy first because it updates both column and offset policies.
+<a href=#1700 id=1700 data-nosnippet>1700</a>                </span>.with_page_index_policy(PageIndexPolicy::from(<span class="self">self</span>.preload_page_index))
+<a href=#1701 id=1701 data-nosnippet>1701</a>                .with_column_index_policy(PageIndexPolicy::from(<span class="self">self</span>.preload_column_index))
+<a href=#1702 id=1702 data-nosnippet>1702</a>                .with_offset_index_policy(PageIndexPolicy::from(<span class="self">self</span>.preload_offset_index));
+<a href=#1703 id=1703 data-nosnippet>1703</a>            <span class="kw">let </span>size = <span class="self">self</span>.meta.size;
+<a href=#1704 id=1704 data-nosnippet>1704</a>            <span class="kw">let </span>meta = reader.load_and_finish(<span class="self">self</span>, size).<span class="kw">await</span><span class="question-mark">?</span>;
+<a href=#1705 id=1705 data-nosnippet>1705</a>
+<a href=#1706 id=1706 data-nosnippet>1706</a>            <span class="prelude-val">Ok</span>(Arc::new(meta))
+<a href=#1707 id=1707 data-nosnippet>1707</a>        }
+<a href=#1708 id=1708 data-nosnippet>1708</a>        .boxed()
+<a href=#1709 id=1709 data-nosnippet>1709</a>    }
+<a href=#1710 id=1710 data-nosnippet>1710</a>}
+<a href=#1711 id=1711 data-nosnippet>1711</a>
+<a href=#1712 id=1712 data-nosnippet>1712</a><span class="doccomment">/// The Arrow type of an array that the Parquet reader reads may not match the exact Arrow type
+<a href=#1713 id=1713 data-nosnippet>1713</a>/// that Iceberg uses for literals - but they are effectively the same logical type,
+<a href=#1714 id=1714 data-nosnippet>1714</a>/// i.e. LargeUtf8 and Utf8 or Utf8View and Utf8 or Utf8View and LargeUtf8.
+<a href=#1715 id=1715 data-nosnippet>1715</a>///
+<a href=#1716 id=1716 data-nosnippet>1716</a>/// The Arrow compute kernels that we use must match the type exactly, so first cast the literal
+<a href=#1717 id=1717 data-nosnippet>1717</a>/// into the type of the batch we read from Parquet before sending it to the compute kernel.
+<a href=#1718 id=1718 data-nosnippet>1718</a></span><span class="kw">fn </span>try_cast_literal(
+<a href=#1719 id=1719 data-nosnippet>1719</a>    literal: <span class="kw-2">&amp;</span>Arc&lt;<span class="kw">dyn </span>ArrowDatum + Send + Sync&gt;,
+<a href=#1720 id=1720 data-nosnippet>1720</a>    column_type: <span class="kw-2">&amp;</span>DataType,
+<a href=#1721 id=1721 data-nosnippet>1721</a>) -&gt; std::result::Result&lt;Arc&lt;<span class="kw">dyn </span>ArrowDatum + Send + Sync&gt;, ArrowError&gt; {
+<a href=#1722 id=1722 data-nosnippet>1722</a>    <span class="kw">let </span>literal_array = literal.get().<span class="number">0</span>;
+<a href=#1723 id=1723 data-nosnippet>1723</a>
+<a href=#1724 id=1724 data-nosnippet>1724</a>    <span class="comment">// No cast required
+<a href=#1725 id=1725 data-nosnippet>1725</a>    </span><span class="kw">if </span>literal_array.data_type() == column_type {
+<a href=#1726 id=1726 data-nosnippet>1726</a>        <span class="kw">return </span><span class="prelude-val">Ok</span>(Arc::clone(literal));
+<a href=#1727 id=1727 data-nosnippet>1727</a>    }
+<a href=#1728 id=1728 data-nosnippet>1728</a>
+<a href=#1729 id=1729 data-nosnippet>1729</a>    <span class="kw">let </span>literal_array = cast(literal_array, column_type)<span class="question-mark">?</span>;
+<a href=#1730 id=1730 data-nosnippet>1730</a>    <span class="prelude-val">Ok</span>(Arc::new(Scalar::new(literal_array)))
+<a href=#1731 id=1731 data-nosnippet>1731</a>}
+<a href=#1732 id=1732 data-nosnippet>1732</a>
+<a href=#1733 id=1733 data-nosnippet>1733</a><span class="attr">#[cfg(test)]
+<a href=#1734 id=1734 data-nosnippet>1734</a></span><span class="kw">mod </span>tests {
+<a href=#1735 id=1735 data-nosnippet>1735</a>    <span class="kw">use </span>std::collections::{HashMap, HashSet};
+<a href=#1736 id=1736 data-nosnippet>1736</a>    <span class="kw">use </span>std::fs::File;
+<a href=#1737 id=1737 data-nosnippet>1737</a>    <span class="kw">use </span>std::sync::Arc;
+<a href=#1738 id=1738 data-nosnippet>1738</a>
+<a href=#1739 id=1739 data-nosnippet>1739</a>    <span class="kw">use </span>arrow_array::cast::AsArray;
+<a href=#1740 id=1740 data-nosnippet>1740</a>    <span class="kw">use </span>arrow_array::{ArrayRef, LargeStringArray, RecordBatch, StringArray};
+<a href=#1741 id=1741 data-nosnippet>1741</a>    <span class="kw">use </span>arrow_schema::{DataType, Field, Schema <span class="kw">as </span>ArrowSchema, TimeUnit};
+<a href=#1742 id=1742 data-nosnippet>1742</a>    <span class="kw">use </span>futures::TryStreamExt;
+<a href=#1743 id=1743 data-nosnippet>1743</a>    <span class="kw">use </span>parquet::arrow::arrow_reader::{RowSelection, RowSelector};
+<a href=#1744 id=1744 data-nosnippet>1744</a>    <span class="kw">use </span>parquet::arrow::{ArrowWriter, ProjectionMask};
+<a href=#1745 id=1745 data-nosnippet>1745</a>    <span class="kw">use </span>parquet::basic::Compression;
+<a href=#1746 id=1746 data-nosnippet>1746</a>    <span class="kw">use </span>parquet::file::metadata::{ColumnChunkMetaData, RowGroupMetaData};
+<a href=#1747 id=1747 data-nosnippet>1747</a>    <span class="kw">use </span>parquet::file::properties::WriterProperties;
+<a href=#1748 id=1748 data-nosnippet>1748</a>    <span class="kw">use </span>parquet::schema::parser::parse_message_type;
+<a href=#1749 id=1749 data-nosnippet>1749</a>    <span class="kw">use </span>parquet::schema::types::{SchemaDescPtr, SchemaDescriptor};
+<a href=#1750 id=1750 data-nosnippet>1750</a>    <span class="kw">use </span>roaring::RoaringTreemap;
+<a href=#1751 id=1751 data-nosnippet>1751</a>    <span class="kw">use </span>tempfile::TempDir;
+<a href=#1752 id=1752 data-nosnippet>1752</a>
+<a href=#1753 id=1753 data-nosnippet>1753</a>    <span class="kw">use </span><span class="kw">crate</span>::ErrorKind;
+<a href=#1754 id=1754 data-nosnippet>1754</a>    <span class="kw">use </span><span class="kw">crate</span>::arrow::reader::{CollectFieldIdVisitor, PARQUET_FIELD_ID_META_KEY};
+<a href=#1755 id=1755 data-nosnippet>1755</a>    <span class="kw">use </span><span class="kw">crate</span>::arrow::{ArrowReader, ArrowReaderBuilder};
+<a href=#1756 id=1756 data-nosnippet>1756</a>    <span class="kw">use </span><span class="kw">crate</span>::delete_vector::DeleteVector;
+<a href=#1757 id=1757 data-nosnippet>1757</a>    <span class="kw">use </span><span class="kw">crate</span>::expr::visitors::bound_predicate_visitor::visit;
+<a href=#1758 id=1758 data-nosnippet>1758</a>    <span class="kw">use </span><span class="kw">crate</span>::expr::{Bind, Predicate, Reference};
+<a href=#1759 id=1759 data-nosnippet>1759</a>    <span class="kw">use </span><span class="kw">crate</span>::io::FileIO;
+<a href=#1760 id=1760 data-nosnippet>1760</a>    <span class="kw">use </span><span class="kw">crate</span>::scan::{FileScanTask, FileScanTaskDeleteFile, FileScanTaskStream};
+<a href=#1761 id=1761 data-nosnippet>1761</a>    <span class="kw">use </span><span class="kw">crate</span>::spec::{
+<a href=#1762 id=1762 data-nosnippet>1762</a>        DataContentType, DataFileFormat, Datum, NestedField, PrimitiveType, Schema, SchemaRef, Type,
+<a href=#1763 id=1763 data-nosnippet>1763</a>    };
+<a href=#1764 id=1764 data-nosnippet>1764</a>
+<a href=#1765 id=1765 data-nosnippet>1765</a>    <span class="kw">fn </span>table_schema_simple() -&gt; SchemaRef {
+<a href=#1766 id=1766 data-nosnippet>1766</a>        Arc::new(
+<a href=#1767 id=1767 data-nosnippet>1767</a>            Schema::builder()
+<a href=#1768 id=1768 data-nosnippet>1768</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#1769 id=1769 data-nosnippet>1769</a>                .with_identifier_field_ids(<span class="macro">vec!</span>[<span class="number">2</span>])
+<a href=#1770 id=1770 data-nosnippet>1770</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#1771 id=1771 data-nosnippet>1771</a>                    NestedField::optional(<span class="number">1</span>, <span class="string">"foo"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#1772 id=1772 data-nosnippet>1772</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"bar"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#1773 id=1773 data-nosnippet>1773</a>                    NestedField::optional(<span class="number">3</span>, <span class="string">"baz"</span>, Type::Primitive(PrimitiveType::Boolean)).into(),
+<a href=#1774 id=1774 data-nosnippet>1774</a>                    NestedField::optional(<span class="number">4</span>, <span class="string">"qux"</span>, Type::Primitive(PrimitiveType::Float)).into(),
+<a href=#1775 id=1775 data-nosnippet>1775</a>                ])
+<a href=#1776 id=1776 data-nosnippet>1776</a>                .build()
+<a href=#1777 id=1777 data-nosnippet>1777</a>                .unwrap(),
+<a href=#1778 id=1778 data-nosnippet>1778</a>        )
+<a href=#1779 id=1779 data-nosnippet>1779</a>    }
+<a href=#1780 id=1780 data-nosnippet>1780</a>
+<a href=#1781 id=1781 data-nosnippet>1781</a>    <span class="attr">#[test]
+<a href=#1782 id=1782 data-nosnippet>1782</a>    </span><span class="kw">fn </span>test_collect_field_id() {
+<a href=#1783 id=1783 data-nosnippet>1783</a>        <span class="kw">let </span>schema = table_schema_simple();
+<a href=#1784 id=1784 data-nosnippet>1784</a>        <span class="kw">let </span>expr = Reference::new(<span class="string">"qux"</span>).is_null();
+<a href=#1785 id=1785 data-nosnippet>1785</a>        <span class="kw">let </span>bound_expr = expr.bind(schema, <span class="bool-val">true</span>).unwrap();
+<a href=#1786 id=1786 data-nosnippet>1786</a>
+<a href=#1787 id=1787 data-nosnippet>1787</a>        <span class="kw">let </span><span class="kw-2">mut </span>visitor = CollectFieldIdVisitor {
+<a href=#1788 id=1788 data-nosnippet>1788</a>            field_ids: HashSet::default(),
+<a href=#1789 id=1789 data-nosnippet>1789</a>        };
+<a href=#1790 id=1790 data-nosnippet>1790</a>        visit(<span class="kw-2">&amp;mut </span>visitor, <span class="kw-2">&amp;</span>bound_expr).unwrap();
+<a href=#1791 id=1791 data-nosnippet>1791</a>
+<a href=#1792 id=1792 data-nosnippet>1792</a>        <span class="kw">let </span><span class="kw-2">mut </span>expected = HashSet::default();
+<a href=#1793 id=1793 data-nosnippet>1793</a>        expected.insert(<span class="number">4_i32</span>);
+<a href=#1794 id=1794 data-nosnippet>1794</a>
+<a href=#1795 id=1795 data-nosnippet>1795</a>        <span class="macro">assert_eq!</span>(visitor.field_ids, expected);
+<a href=#1796 id=1796 data-nosnippet>1796</a>    }
+<a href=#1797 id=1797 data-nosnippet>1797</a>
+<a href=#1798 id=1798 data-nosnippet>1798</a>    <span class="attr">#[test]
+<a href=#1799 id=1799 data-nosnippet>1799</a>    </span><span class="kw">fn </span>test_collect_field_id_with_and() {
+<a href=#1800 id=1800 data-nosnippet>1800</a>        <span class="kw">let </span>schema = table_schema_simple();
+<a href=#1801 id=1801 data-nosnippet>1801</a>        <span class="kw">let </span>expr = Reference::new(<span class="string">"qux"</span>)
+<a href=#1802 id=1802 data-nosnippet>1802</a>            .is_null()
+<a href=#1803 id=1803 data-nosnippet>1803</a>            .and(Reference::new(<span class="string">"baz"</span>).is_null());
+<a href=#1804 id=1804 data-nosnippet>1804</a>        <span class="kw">let </span>bound_expr = expr.bind(schema, <span class="bool-val">true</span>).unwrap();
+<a href=#1805 id=1805 data-nosnippet>1805</a>
+<a href=#1806 id=1806 data-nosnippet>1806</a>        <span class="kw">let </span><span class="kw-2">mut </span>visitor = CollectFieldIdVisitor {
+<a href=#1807 id=1807 data-nosnippet>1807</a>            field_ids: HashSet::default(),
+<a href=#1808 id=1808 data-nosnippet>1808</a>        };
+<a href=#1809 id=1809 data-nosnippet>1809</a>        visit(<span class="kw-2">&amp;mut </span>visitor, <span class="kw-2">&amp;</span>bound_expr).unwrap();
+<a href=#1810 id=1810 data-nosnippet>1810</a>
+<a href=#1811 id=1811 data-nosnippet>1811</a>        <span class="kw">let </span><span class="kw-2">mut </span>expected = HashSet::default();
+<a href=#1812 id=1812 data-nosnippet>1812</a>        expected.insert(<span class="number">4_i32</span>);
+<a href=#1813 id=1813 data-nosnippet>1813</a>        expected.insert(<span class="number">3</span>);
+<a href=#1814 id=1814 data-nosnippet>1814</a>
+<a href=#1815 id=1815 data-nosnippet>1815</a>        <span class="macro">assert_eq!</span>(visitor.field_ids, expected);
+<a href=#1816 id=1816 data-nosnippet>1816</a>    }
+<a href=#1817 id=1817 data-nosnippet>1817</a>
+<a href=#1818 id=1818 data-nosnippet>1818</a>    <span class="attr">#[test]
+<a href=#1819 id=1819 data-nosnippet>1819</a>    </span><span class="kw">fn </span>test_collect_field_id_with_or() {
+<a href=#1820 id=1820 data-nosnippet>1820</a>        <span class="kw">let </span>schema = table_schema_simple();
+<a href=#1821 id=1821 data-nosnippet>1821</a>        <span class="kw">let </span>expr = Reference::new(<span class="string">"qux"</span>)
+<a href=#1822 id=1822 data-nosnippet>1822</a>            .is_null()
+<a href=#1823 id=1823 data-nosnippet>1823</a>            .or(Reference::new(<span class="string">"baz"</span>).is_null());
+<a href=#1824 id=1824 data-nosnippet>1824</a>        <span class="kw">let </span>bound_expr = expr.bind(schema, <span class="bool-val">true</span>).unwrap();
+<a href=#1825 id=1825 data-nosnippet>1825</a>
+<a href=#1826 id=1826 data-nosnippet>1826</a>        <span class="kw">let </span><span class="kw-2">mut </span>visitor = CollectFieldIdVisitor {
+<a href=#1827 id=1827 data-nosnippet>1827</a>            field_ids: HashSet::default(),
+<a href=#1828 id=1828 data-nosnippet>1828</a>        };
+<a href=#1829 id=1829 data-nosnippet>1829</a>        visit(<span class="kw-2">&amp;mut </span>visitor, <span class="kw-2">&amp;</span>bound_expr).unwrap();
+<a href=#1830 id=1830 data-nosnippet>1830</a>
+<a href=#1831 id=1831 data-nosnippet>1831</a>        <span class="kw">let </span><span class="kw-2">mut </span>expected = HashSet::default();
+<a href=#1832 id=1832 data-nosnippet>1832</a>        expected.insert(<span class="number">4_i32</span>);
+<a href=#1833 id=1833 data-nosnippet>1833</a>        expected.insert(<span class="number">3</span>);
+<a href=#1834 id=1834 data-nosnippet>1834</a>
+<a href=#1835 id=1835 data-nosnippet>1835</a>        <span class="macro">assert_eq!</span>(visitor.field_ids, expected);
+<a href=#1836 id=1836 data-nosnippet>1836</a>    }
+<a href=#1837 id=1837 data-nosnippet>1837</a>
+<a href=#1838 id=1838 data-nosnippet>1838</a>    <span class="attr">#[test]
+<a href=#1839 id=1839 data-nosnippet>1839</a>    </span><span class="kw">fn </span>test_arrow_projection_mask() {
+<a href=#1840 id=1840 data-nosnippet>1840</a>        <span class="kw">let </span>schema = Arc::new(
+<a href=#1841 id=1841 data-nosnippet>1841</a>            Schema::builder()
+<a href=#1842 id=1842 data-nosnippet>1842</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#1843 id=1843 data-nosnippet>1843</a>                .with_identifier_field_ids(<span class="macro">vec!</span>[<span class="number">1</span>])
+<a href=#1844 id=1844 data-nosnippet>1844</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#1845 id=1845 data-nosnippet>1845</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"c1"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#1846 id=1846 data-nosnippet>1846</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"c2"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#1847 id=1847 data-nosnippet>1847</a>                    NestedField::optional(
+<a href=#1848 id=1848 data-nosnippet>1848</a>                        <span class="number">3</span>,
+<a href=#1849 id=1849 data-nosnippet>1849</a>                        <span class="string">"c3"</span>,
+<a href=#1850 id=1850 data-nosnippet>1850</a>                        Type::Primitive(PrimitiveType::Decimal {
+<a href=#1851 id=1851 data-nosnippet>1851</a>                            precision: <span class="number">38</span>,
+<a href=#1852 id=1852 data-nosnippet>1852</a>                            scale: <span class="number">3</span>,
+<a href=#1853 id=1853 data-nosnippet>1853</a>                        }),
+<a href=#1854 id=1854 data-nosnippet>1854</a>                    )
+<a href=#1855 id=1855 data-nosnippet>1855</a>                    .into(),
+<a href=#1856 id=1856 data-nosnippet>1856</a>                ])
+<a href=#1857 id=1857 data-nosnippet>1857</a>                .build()
+<a href=#1858 id=1858 data-nosnippet>1858</a>                .unwrap(),
+<a href=#1859 id=1859 data-nosnippet>1859</a>        );
+<a href=#1860 id=1860 data-nosnippet>1860</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#1861 id=1861 data-nosnippet>1861</a>            Field::new(<span class="string">"c1"</span>, DataType::Utf8, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#1862 id=1862 data-nosnippet>1862</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#1863 id=1863 data-nosnippet>1863</a>                <span class="string">"1"</span>.to_string(),
+<a href=#1864 id=1864 data-nosnippet>1864</a>            )])),
+<a href=#1865 id=1865 data-nosnippet>1865</a>            <span class="comment">// Type not supported
+<a href=#1866 id=1866 data-nosnippet>1866</a>            </span>Field::new(<span class="string">"c2"</span>, DataType::Duration(TimeUnit::Microsecond), <span class="bool-val">true</span>).with_metadata(
+<a href=#1867 id=1867 data-nosnippet>1867</a>                HashMap::from([(PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"2"</span>.to_string())]),
+<a href=#1868 id=1868 data-nosnippet>1868</a>            ),
+<a href=#1869 id=1869 data-nosnippet>1869</a>            <span class="comment">// Precision is beyond the supported range
+<a href=#1870 id=1870 data-nosnippet>1870</a>            </span>Field::new(<span class="string">"c3"</span>, DataType::Decimal128(<span class="number">39</span>, <span class="number">3</span>), <span class="bool-val">true</span>).with_metadata(HashMap::from([(
+<a href=#1871 id=1871 data-nosnippet>1871</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#1872 id=1872 data-nosnippet>1872</a>                <span class="string">"3"</span>.to_string(),
+<a href=#1873 id=1873 data-nosnippet>1873</a>            )])),
+<a href=#1874 id=1874 data-nosnippet>1874</a>        ]));
+<a href=#1875 id=1875 data-nosnippet>1875</a>
+<a href=#1876 id=1876 data-nosnippet>1876</a>        <span class="kw">let </span>message_type = <span class="string">"
+<a href=#1877 id=1877 data-nosnippet>1877</a>message schema {
+<a href=#1878 id=1878 data-nosnippet>1878</a>  required binary c1 (STRING) = 1;
+<a href=#1879 id=1879 data-nosnippet>1879</a>  optional int32 c2 (INTEGER(8,true)) = 2;
+<a href=#1880 id=1880 data-nosnippet>1880</a>  optional fixed_len_byte_array(17) c3 (DECIMAL(39,3)) = 3;
+<a href=#1881 id=1881 data-nosnippet>1881</a>}
+<a href=#1882 id=1882 data-nosnippet>1882</a>    "</span>;
+<a href=#1883 id=1883 data-nosnippet>1883</a>        <span class="kw">let </span>parquet_type = parse_message_type(message_type).expect(<span class="string">"should parse schema"</span>);
+<a href=#1884 id=1884 data-nosnippet>1884</a>        <span class="kw">let </span>parquet_schema = SchemaDescriptor::new(Arc::new(parquet_type));
+<a href=#1885 id=1885 data-nosnippet>1885</a>
+<a href=#1886 id=1886 data-nosnippet>1886</a>        <span class="comment">// Try projecting the fields c2 and c3 with the unsupported data types
+<a href=#1887 id=1887 data-nosnippet>1887</a>        </span><span class="kw">let </span>err = ArrowReader::get_arrow_projection_mask(
+<a href=#1888 id=1888 data-nosnippet>1888</a>            <span class="kw-2">&amp;</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>],
+<a href=#1889 id=1889 data-nosnippet>1889</a>            <span class="kw-2">&amp;</span>schema,
+<a href=#1890 id=1890 data-nosnippet>1890</a>            <span class="kw-2">&amp;</span>parquet_schema,
+<a href=#1891 id=1891 data-nosnippet>1891</a>            <span class="kw-2">&amp;</span>arrow_schema,
+<a href=#1892 id=1892 data-nosnippet>1892</a>            <span class="bool-val">false</span>,
+<a href=#1893 id=1893 data-nosnippet>1893</a>        )
+<a href=#1894 id=1894 data-nosnippet>1894</a>        .unwrap_err();
+<a href=#1895 id=1895 data-nosnippet>1895</a>
+<a href=#1896 id=1896 data-nosnippet>1896</a>        <span class="macro">assert_eq!</span>(err.kind(), ErrorKind::DataInvalid);
+<a href=#1897 id=1897 data-nosnippet>1897</a>        <span class="macro">assert_eq!</span>(
+<a href=#1898 id=1898 data-nosnippet>1898</a>            err.to_string(),
+<a href=#1899 id=1899 data-nosnippet>1899</a>            <span class="string">"DataInvalid =&gt; Unsupported Arrow data type: Duration(Microsecond)"</span>.to_string()
+<a href=#1900 id=1900 data-nosnippet>1900</a>        );
+<a href=#1901 id=1901 data-nosnippet>1901</a>
+<a href=#1902 id=1902 data-nosnippet>1902</a>        <span class="comment">// Omitting field c2, we still get an error due to c3 being selected
+<a href=#1903 id=1903 data-nosnippet>1903</a>        </span><span class="kw">let </span>err = ArrowReader::get_arrow_projection_mask(
+<a href=#1904 id=1904 data-nosnippet>1904</a>            <span class="kw-2">&amp;</span>[<span class="number">1</span>, <span class="number">3</span>],
+<a href=#1905 id=1905 data-nosnippet>1905</a>            <span class="kw-2">&amp;</span>schema,
+<a href=#1906 id=1906 data-nosnippet>1906</a>            <span class="kw-2">&amp;</span>parquet_schema,
+<a href=#1907 id=1907 data-nosnippet>1907</a>            <span class="kw-2">&amp;</span>arrow_schema,
+<a href=#1908 id=1908 data-nosnippet>1908</a>            <span class="bool-val">false</span>,
+<a href=#1909 id=1909 data-nosnippet>1909</a>        )
+<a href=#1910 id=1910 data-nosnippet>1910</a>        .unwrap_err();
+<a href=#1911 id=1911 data-nosnippet>1911</a>
+<a href=#1912 id=1912 data-nosnippet>1912</a>        <span class="macro">assert_eq!</span>(err.kind(), ErrorKind::DataInvalid);
+<a href=#1913 id=1913 data-nosnippet>1913</a>        <span class="macro">assert_eq!</span>(
+<a href=#1914 id=1914 data-nosnippet>1914</a>            err.to_string(),
+<a href=#1915 id=1915 data-nosnippet>1915</a>            <span class="string">"DataInvalid =&gt; Failed to create decimal type, source: DataInvalid =&gt; Decimals with precision larger than 38 are not supported: 39"</span>.to_string()
+<a href=#1916 id=1916 data-nosnippet>1916</a>        );
+<a href=#1917 id=1917 data-nosnippet>1917</a>
+<a href=#1918 id=1918 data-nosnippet>1918</a>        <span class="comment">// Finally avoid selecting fields with unsupported data types
+<a href=#1919 id=1919 data-nosnippet>1919</a>        </span><span class="kw">let </span>mask = ArrowReader::get_arrow_projection_mask(
+<a href=#1920 id=1920 data-nosnippet>1920</a>            <span class="kw-2">&amp;</span>[<span class="number">1</span>],
+<a href=#1921 id=1921 data-nosnippet>1921</a>            <span class="kw-2">&amp;</span>schema,
+<a href=#1922 id=1922 data-nosnippet>1922</a>            <span class="kw-2">&amp;</span>parquet_schema,
+<a href=#1923 id=1923 data-nosnippet>1923</a>            <span class="kw-2">&amp;</span>arrow_schema,
+<a href=#1924 id=1924 data-nosnippet>1924</a>            <span class="bool-val">false</span>,
+<a href=#1925 id=1925 data-nosnippet>1925</a>        )
+<a href=#1926 id=1926 data-nosnippet>1926</a>        .expect(<span class="string">"Some ProjectionMask"</span>);
+<a href=#1927 id=1927 data-nosnippet>1927</a>        <span class="macro">assert_eq!</span>(mask, ProjectionMask::leaves(<span class="kw-2">&amp;</span>parquet_schema, <span class="macro">vec!</span>[<span class="number">0</span>]));
+<a href=#1928 id=1928 data-nosnippet>1928</a>    }
 <a href=#1929 id=1929 data-nosnippet>1929</a>
-<a href=#1930 id=1930 data-nosnippet>1930</a>            <span class="macro">assert_eq!</span>(result_data, expected, <span class="string">"predicate={predicate}"</span>);
-<a href=#1931 id=1931 data-nosnippet>1931</a>        }
-<a href=#1932 id=1932 data-nosnippet>1932</a>    }
-<a href=#1933 id=1933 data-nosnippet>1933</a>
-<a href=#1934 id=1934 data-nosnippet>1934</a>    <span class="kw">async fn </span>test_perform_read(
-<a href=#1935 id=1935 data-nosnippet>1935</a>        predicate: Predicate,
-<a href=#1936 id=1936 data-nosnippet>1936</a>        schema: SchemaRef,
-<a href=#1937 id=1937 data-nosnippet>1937</a>        table_location: String,
-<a href=#1938 id=1938 data-nosnippet>1938</a>        reader: ArrowReader,
-<a href=#1939 id=1939 data-nosnippet>1939</a>    ) -&gt; Vec&lt;<span class="prelude-ty">Option</span>&lt;String&gt;&gt; {
-<a href=#1940 id=1940 data-nosnippet>1940</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#1941 id=1941 data-nosnippet>1941</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#1942 id=1942 data-nosnippet>1942</a>                start: <span class="number">0</span>,
-<a href=#1943 id=1943 data-nosnippet>1943</a>                length: <span class="number">0</span>,
-<a href=#1944 id=1944 data-nosnippet>1944</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#1945 id=1945 data-nosnippet>1945</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{table_location}/1.parquet"</span>),
-<a href=#1946 id=1946 data-nosnippet>1946</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#1947 id=1947 data-nosnippet>1947</a>                schema: schema.clone(),
-<a href=#1948 id=1948 data-nosnippet>1948</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
-<a href=#1949 id=1949 data-nosnippet>1949</a>                predicate: <span class="prelude-val">Some</span>(predicate.bind(schema, <span class="bool-val">true</span>).unwrap()),
-<a href=#1950 id=1950 data-nosnippet>1950</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#1951 id=1951 data-nosnippet>1951</a>            })]
-<a href=#1952 id=1952 data-nosnippet>1952</a>            .into_iter(),
-<a href=#1953 id=1953 data-nosnippet>1953</a>        )) <span class="kw">as </span>FileScanTaskStream;
-<a href=#1954 id=1954 data-nosnippet>1954</a>
-<a href=#1955 id=1955 data-nosnippet>1955</a>        <span class="kw">let </span>result = reader
-<a href=#1956 id=1956 data-nosnippet>1956</a>            .read(tasks)
-<a href=#1957 id=1957 data-nosnippet>1957</a>            .unwrap()
-<a href=#1958 id=1958 data-nosnippet>1958</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#1959 id=1959 data-nosnippet>1959</a>            .<span class="kw">await
-<a href=#1960 id=1960 data-nosnippet>1960</a>            </span>.unwrap();
+<a href=#1930 id=1930 data-nosnippet>1930</a>    <span class="attr">#[tokio::test]
+<a href=#1931 id=1931 data-nosnippet>1931</a>    </span><span class="kw">async fn </span>test_kleene_logic_or_behaviour() {
+<a href=#1932 id=1932 data-nosnippet>1932</a>        <span class="comment">// a IS NULL OR a = 'foo'
+<a href=#1933 id=1933 data-nosnippet>1933</a>        </span><span class="kw">let </span>predicate = Reference::new(<span class="string">"a"</span>)
+<a href=#1934 id=1934 data-nosnippet>1934</a>            .is_null()
+<a href=#1935 id=1935 data-nosnippet>1935</a>            .or(Reference::new(<span class="string">"a"</span>).equal_to(Datum::string(<span class="string">"foo"</span>)));
+<a href=#1936 id=1936 data-nosnippet>1936</a>
+<a href=#1937 id=1937 data-nosnippet>1937</a>        <span class="comment">// Table data: [NULL, "foo", "bar"]
+<a href=#1938 id=1938 data-nosnippet>1938</a>        </span><span class="kw">let </span>data_for_col_a = <span class="macro">vec!</span>[<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()), <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())];
+<a href=#1939 id=1939 data-nosnippet>1939</a>
+<a href=#1940 id=1940 data-nosnippet>1940</a>        <span class="comment">// Expected: [NULL, "foo"].
+<a href=#1941 id=1941 data-nosnippet>1941</a>        </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string())];
+<a href=#1942 id=1942 data-nosnippet>1942</a>
+<a href=#1943 id=1943 data-nosnippet>1943</a>        <span class="kw">let </span>(file_io, schema, table_location, _temp_dir) =
+<a href=#1944 id=1944 data-nosnippet>1944</a>            setup_kleene_logic(data_for_col_a, DataType::Utf8);
+<a href=#1945 id=1945 data-nosnippet>1945</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#1946 id=1946 data-nosnippet>1946</a>
+<a href=#1947 id=1947 data-nosnippet>1947</a>        <span class="kw">let </span>result_data = test_perform_read(predicate, schema, table_location, reader).<span class="kw">await</span>;
+<a href=#1948 id=1948 data-nosnippet>1948</a>
+<a href=#1949 id=1949 data-nosnippet>1949</a>        <span class="macro">assert_eq!</span>(result_data, expected);
+<a href=#1950 id=1950 data-nosnippet>1950</a>    }
+<a href=#1951 id=1951 data-nosnippet>1951</a>
+<a href=#1952 id=1952 data-nosnippet>1952</a>    <span class="attr">#[tokio::test]
+<a href=#1953 id=1953 data-nosnippet>1953</a>    </span><span class="kw">async fn </span>test_kleene_logic_and_behaviour() {
+<a href=#1954 id=1954 data-nosnippet>1954</a>        <span class="comment">// a IS NOT NULL AND a != 'foo'
+<a href=#1955 id=1955 data-nosnippet>1955</a>        </span><span class="kw">let </span>predicate = Reference::new(<span class="string">"a"</span>)
+<a href=#1956 id=1956 data-nosnippet>1956</a>            .is_not_null()
+<a href=#1957 id=1957 data-nosnippet>1957</a>            .and(Reference::new(<span class="string">"a"</span>).not_equal_to(Datum::string(<span class="string">"foo"</span>)));
+<a href=#1958 id=1958 data-nosnippet>1958</a>
+<a href=#1959 id=1959 data-nosnippet>1959</a>        <span class="comment">// Table data: [NULL, "foo", "bar"]
+<a href=#1960 id=1960 data-nosnippet>1960</a>        </span><span class="kw">let </span>data_for_col_a = <span class="macro">vec!</span>[<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()), <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())];
 <a href=#1961 id=1961 data-nosnippet>1961</a>
-<a href=#1962 id=1962 data-nosnippet>1962</a>        result[<span class="number">0</span>].columns()[<span class="number">0</span>]
-<a href=#1963 id=1963 data-nosnippet>1963</a>            .as_string_opt::&lt;i32&gt;()
-<a href=#1964 id=1964 data-nosnippet>1964</a>            .unwrap()
-<a href=#1965 id=1965 data-nosnippet>1965</a>            .iter()
-<a href=#1966 id=1966 data-nosnippet>1966</a>            .map(|v| v.map(ToOwned::to_owned))
-<a href=#1967 id=1967 data-nosnippet>1967</a>            .collect::&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;()
-<a href=#1968 id=1968 data-nosnippet>1968</a>    }
-<a href=#1969 id=1969 data-nosnippet>1969</a>
-<a href=#1970 id=1970 data-nosnippet>1970</a>    <span class="kw">fn </span>setup_kleene_logic(
-<a href=#1971 id=1971 data-nosnippet>1971</a>        data_for_col_a: Vec&lt;<span class="prelude-ty">Option</span>&lt;String&gt;&gt;,
-<a href=#1972 id=1972 data-nosnippet>1972</a>        col_a_type: DataType,
-<a href=#1973 id=1973 data-nosnippet>1973</a>    ) -&gt; (FileIO, SchemaRef, String, TempDir) {
-<a href=#1974 id=1974 data-nosnippet>1974</a>        <span class="kw">let </span>schema = Arc::new(
-<a href=#1975 id=1975 data-nosnippet>1975</a>            Schema::builder()
-<a href=#1976 id=1976 data-nosnippet>1976</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#1977 id=1977 data-nosnippet>1977</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#1978 id=1978 data-nosnippet>1978</a>                    NestedField::optional(<span class="number">1</span>, <span class="string">"a"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#1979 id=1979 data-nosnippet>1979</a>                ])
-<a href=#1980 id=1980 data-nosnippet>1980</a>                .build()
-<a href=#1981 id=1981 data-nosnippet>1981</a>                .unwrap(),
-<a href=#1982 id=1982 data-nosnippet>1982</a>        );
-<a href=#1983 id=1983 data-nosnippet>1983</a>
-<a href=#1984 id=1984 data-nosnippet>1984</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#1985 id=1985 data-nosnippet>1985</a>            Field::new(<span class="string">"a"</span>, col_a_type.clone(), <span class="bool-val">true</span>).with_metadata(HashMap::from([(
-<a href=#1986 id=1986 data-nosnippet>1986</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#1987 id=1987 data-nosnippet>1987</a>                <span class="string">"1"</span>.to_string(),
-<a href=#1988 id=1988 data-nosnippet>1988</a>            )])),
-<a href=#1989 id=1989 data-nosnippet>1989</a>        ]));
-<a href=#1990 id=1990 data-nosnippet>1990</a>
-<a href=#1991 id=1991 data-nosnippet>1991</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#1992 id=1992 data-nosnippet>1992</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#1993 id=1993 data-nosnippet>1993</a>
-<a href=#1994 id=1994 data-nosnippet>1994</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#1995 id=1995 data-nosnippet>1995</a>
-<a href=#1996 id=1996 data-nosnippet>1996</a>        <span class="kw">let </span>col = <span class="kw">match </span>col_a_type {
-<a href=#1997 id=1997 data-nosnippet>1997</a>            DataType::Utf8 =&gt; Arc::new(StringArray::from(data_for_col_a)) <span class="kw">as </span>ArrayRef,
-<a href=#1998 id=1998 data-nosnippet>1998</a>            DataType::LargeUtf8 =&gt; Arc::new(LargeStringArray::from(data_for_col_a)) <span class="kw">as </span>ArrayRef,
-<a href=#1999 id=1999 data-nosnippet>1999</a>            <span class="kw">_ </span>=&gt; <span class="macro">panic!</span>(<span class="string">"unexpected col_a_type"</span>),
-<a href=#2000 id=2000 data-nosnippet>2000</a>        };
-<a href=#2001 id=2001 data-nosnippet>2001</a>
-<a href=#2002 id=2002 data-nosnippet>2002</a>        <span class="kw">let </span>to_write = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[col]).unwrap();
-<a href=#2003 id=2003 data-nosnippet>2003</a>
-<a href=#2004 id=2004 data-nosnippet>2004</a>        <span class="comment">// Write the Parquet files
-<a href=#2005 id=2005 data-nosnippet>2005</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
-<a href=#2006 id=2006 data-nosnippet>2006</a>            .set_compression(Compression::SNAPPY)
-<a href=#2007 id=2007 data-nosnippet>2007</a>            .build();
-<a href=#2008 id=2008 data-nosnippet>2008</a>
-<a href=#2009 id=2009 data-nosnippet>2009</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#2010 id=2010 data-nosnippet>2010</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer =
-<a href=#2011 id=2011 data-nosnippet>2011</a>            ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props.clone())).unwrap();
-<a href=#2012 id=2012 data-nosnippet>2012</a>
-<a href=#2013 id=2013 data-nosnippet>2013</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
-<a href=#2014 id=2014 data-nosnippet>2014</a>
-<a href=#2015 id=2015 data-nosnippet>2015</a>        <span class="comment">// writer must be closed to write footer
-<a href=#2016 id=2016 data-nosnippet>2016</a>        </span>writer.close().unwrap();
-<a href=#2017 id=2017 data-nosnippet>2017</a>
-<a href=#2018 id=2018 data-nosnippet>2018</a>        (file_io, schema, table_location, tmp_dir)
-<a href=#2019 id=2019 data-nosnippet>2019</a>    }
-<a href=#2020 id=2020 data-nosnippet>2020</a>
-<a href=#2021 id=2021 data-nosnippet>2021</a>    <span class="attr">#[test]
-<a href=#2022 id=2022 data-nosnippet>2022</a>    </span><span class="kw">fn </span>test_build_deletes_row_selection() {
-<a href=#2023 id=2023 data-nosnippet>2023</a>        <span class="kw">let </span>schema_descr = get_test_schema_descr();
-<a href=#2024 id=2024 data-nosnippet>2024</a>
-<a href=#2025 id=2025 data-nosnippet>2025</a>        <span class="kw">let </span><span class="kw-2">mut </span>columns = <span class="macro">vec!</span>[];
-<a href=#2026 id=2026 data-nosnippet>2026</a>        <span class="kw">for </span>ptr <span class="kw">in </span>schema_descr.columns() {
-<a href=#2027 id=2027 data-nosnippet>2027</a>            <span class="kw">let </span>column = ColumnChunkMetaData::builder(ptr.clone()).build().unwrap();
-<a href=#2028 id=2028 data-nosnippet>2028</a>            columns.push(column);
-<a href=#2029 id=2029 data-nosnippet>2029</a>        }
-<a href=#2030 id=2030 data-nosnippet>2030</a>
-<a href=#2031 id=2031 data-nosnippet>2031</a>        <span class="kw">let </span>row_groups_metadata = <span class="macro">vec!</span>[
-<a href=#2032 id=2032 data-nosnippet>2032</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">1000</span>, <span class="number">0</span>),
-<a href=#2033 id=2033 data-nosnippet>2033</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">500</span>, <span class="number">1</span>),
-<a href=#2034 id=2034 data-nosnippet>2034</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">500</span>, <span class="number">2</span>),
-<a href=#2035 id=2035 data-nosnippet>2035</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">1000</span>, <span class="number">3</span>),
-<a href=#2036 id=2036 data-nosnippet>2036</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">500</span>, <span class="number">4</span>),
-<a href=#2037 id=2037 data-nosnippet>2037</a>        ];
-<a href=#2038 id=2038 data-nosnippet>2038</a>
-<a href=#2039 id=2039 data-nosnippet>2039</a>        <span class="kw">let </span>selected_row_groups = <span class="prelude-val">Some</span>(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">3</span>]);
-<a href=#2040 id=2040 data-nosnippet>2040</a>
-<a href=#2041 id=2041 data-nosnippet>2041</a>        <span class="comment">/* cases to cover:
-<a href=#2042 id=2042 data-nosnippet>2042</a>           * {skip|select} {first|intermediate|last} {one row|multiple rows} in
-<a href=#2043 id=2043 data-nosnippet>2043</a>             {first|intermediate|last} {skipped|selected} row group
-<a href=#2044 id=2044 data-nosnippet>2044</a>           * row group selection disabled
-<a href=#2045 id=2045 data-nosnippet>2045</a>        */
+<a href=#1962 id=1962 data-nosnippet>1962</a>        <span class="comment">// Expected: ["bar"].
+<a href=#1963 id=1963 data-nosnippet>1963</a>        </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())];
+<a href=#1964 id=1964 data-nosnippet>1964</a>
+<a href=#1965 id=1965 data-nosnippet>1965</a>        <span class="kw">let </span>(file_io, schema, table_location, _temp_dir) =
+<a href=#1966 id=1966 data-nosnippet>1966</a>            setup_kleene_logic(data_for_col_a, DataType::Utf8);
+<a href=#1967 id=1967 data-nosnippet>1967</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#1968 id=1968 data-nosnippet>1968</a>
+<a href=#1969 id=1969 data-nosnippet>1969</a>        <span class="kw">let </span>result_data = test_perform_read(predicate, schema, table_location, reader).<span class="kw">await</span>;
+<a href=#1970 id=1970 data-nosnippet>1970</a>
+<a href=#1971 id=1971 data-nosnippet>1971</a>        <span class="macro">assert_eq!</span>(result_data, expected);
+<a href=#1972 id=1972 data-nosnippet>1972</a>    }
+<a href=#1973 id=1973 data-nosnippet>1973</a>
+<a href=#1974 id=1974 data-nosnippet>1974</a>    <span class="attr">#[tokio::test]
+<a href=#1975 id=1975 data-nosnippet>1975</a>    </span><span class="kw">async fn </span>test_predicate_cast_literal() {
+<a href=#1976 id=1976 data-nosnippet>1976</a>        <span class="kw">let </span>predicates = <span class="macro">vec!</span>[
+<a href=#1977 id=1977 data-nosnippet>1977</a>            <span class="comment">// a == 'foo'
+<a href=#1978 id=1978 data-nosnippet>1978</a>            </span>(Reference::new(<span class="string">"a"</span>).equal_to(Datum::string(<span class="string">"foo"</span>)), <span class="macro">vec!</span>[
+<a href=#1979 id=1979 data-nosnippet>1979</a>                <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()),
+<a href=#1980 id=1980 data-nosnippet>1980</a>            ]),
+<a href=#1981 id=1981 data-nosnippet>1981</a>            <span class="comment">// a != 'foo'
+<a href=#1982 id=1982 data-nosnippet>1982</a>            </span>(
+<a href=#1983 id=1983 data-nosnippet>1983</a>                Reference::new(<span class="string">"a"</span>).not_equal_to(Datum::string(<span class="string">"foo"</span>)),
+<a href=#1984 id=1984 data-nosnippet>1984</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())],
+<a href=#1985 id=1985 data-nosnippet>1985</a>            ),
+<a href=#1986 id=1986 data-nosnippet>1986</a>            <span class="comment">// STARTS_WITH(a, 'foo')
+<a href=#1987 id=1987 data-nosnippet>1987</a>            </span>(Reference::new(<span class="string">"a"</span>).starts_with(Datum::string(<span class="string">"f"</span>)), <span class="macro">vec!</span>[
+<a href=#1988 id=1988 data-nosnippet>1988</a>                <span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()),
+<a href=#1989 id=1989 data-nosnippet>1989</a>            ]),
+<a href=#1990 id=1990 data-nosnippet>1990</a>            <span class="comment">// NOT STARTS_WITH(a, 'foo')
+<a href=#1991 id=1991 data-nosnippet>1991</a>            </span>(
+<a href=#1992 id=1992 data-nosnippet>1992</a>                Reference::new(<span class="string">"a"</span>).not_starts_with(Datum::string(<span class="string">"f"</span>)),
+<a href=#1993 id=1993 data-nosnippet>1993</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())],
+<a href=#1994 id=1994 data-nosnippet>1994</a>            ),
+<a href=#1995 id=1995 data-nosnippet>1995</a>            <span class="comment">// a &lt; 'foo'
+<a href=#1996 id=1996 data-nosnippet>1996</a>            </span>(Reference::new(<span class="string">"a"</span>).less_than(Datum::string(<span class="string">"foo"</span>)), <span class="macro">vec!</span>[
+<a href=#1997 id=1997 data-nosnippet>1997</a>                <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string()),
+<a href=#1998 id=1998 data-nosnippet>1998</a>            ]),
+<a href=#1999 id=1999 data-nosnippet>1999</a>            <span class="comment">// a &lt;= 'foo'
+<a href=#2000 id=2000 data-nosnippet>2000</a>            </span>(
+<a href=#2001 id=2001 data-nosnippet>2001</a>                Reference::new(<span class="string">"a"</span>).less_than_or_equal_to(Datum::string(<span class="string">"foo"</span>)),
+<a href=#2002 id=2002 data-nosnippet>2002</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()), <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())],
+<a href=#2003 id=2003 data-nosnippet>2003</a>            ),
+<a href=#2004 id=2004 data-nosnippet>2004</a>            <span class="comment">// a &gt; 'foo'
+<a href=#2005 id=2005 data-nosnippet>2005</a>            </span>(
+<a href=#2006 id=2006 data-nosnippet>2006</a>                Reference::new(<span class="string">"a"</span>).greater_than(Datum::string(<span class="string">"bar"</span>)),
+<a href=#2007 id=2007 data-nosnippet>2007</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string())],
+<a href=#2008 id=2008 data-nosnippet>2008</a>            ),
+<a href=#2009 id=2009 data-nosnippet>2009</a>            <span class="comment">// a &gt;= 'foo'
+<a href=#2010 id=2010 data-nosnippet>2010</a>            </span>(
+<a href=#2011 id=2011 data-nosnippet>2011</a>                Reference::new(<span class="string">"a"</span>).greater_than_or_equal_to(Datum::string(<span class="string">"foo"</span>)),
+<a href=#2012 id=2012 data-nosnippet>2012</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string())],
+<a href=#2013 id=2013 data-nosnippet>2013</a>            ),
+<a href=#2014 id=2014 data-nosnippet>2014</a>            <span class="comment">// a IN ('foo', 'bar')
+<a href=#2015 id=2015 data-nosnippet>2015</a>            </span>(
+<a href=#2016 id=2016 data-nosnippet>2016</a>                Reference::new(<span class="string">"a"</span>).is_in([Datum::string(<span class="string">"foo"</span>), Datum::string(<span class="string">"baz"</span>)]),
+<a href=#2017 id=2017 data-nosnippet>2017</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string())],
+<a href=#2018 id=2018 data-nosnippet>2018</a>            ),
+<a href=#2019 id=2019 data-nosnippet>2019</a>            <span class="comment">// a NOT IN ('foo', 'bar')
+<a href=#2020 id=2020 data-nosnippet>2020</a>            </span>(
+<a href=#2021 id=2021 data-nosnippet>2021</a>                Reference::new(<span class="string">"a"</span>).is_not_in([Datum::string(<span class="string">"foo"</span>), Datum::string(<span class="string">"baz"</span>)]),
+<a href=#2022 id=2022 data-nosnippet>2022</a>                <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())],
+<a href=#2023 id=2023 data-nosnippet>2023</a>            ),
+<a href=#2024 id=2024 data-nosnippet>2024</a>        ];
+<a href=#2025 id=2025 data-nosnippet>2025</a>
+<a href=#2026 id=2026 data-nosnippet>2026</a>        <span class="comment">// Table data: ["foo", "bar"]
+<a href=#2027 id=2027 data-nosnippet>2027</a>        </span><span class="kw">let </span>data_for_col_a = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"foo"</span>.to_string()), <span class="prelude-val">Some</span>(<span class="string">"bar"</span>.to_string())];
+<a href=#2028 id=2028 data-nosnippet>2028</a>
+<a href=#2029 id=2029 data-nosnippet>2029</a>        <span class="kw">let </span>(file_io, schema, table_location, _temp_dir) =
+<a href=#2030 id=2030 data-nosnippet>2030</a>            setup_kleene_logic(data_for_col_a, DataType::LargeUtf8);
+<a href=#2031 id=2031 data-nosnippet>2031</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#2032 id=2032 data-nosnippet>2032</a>
+<a href=#2033 id=2033 data-nosnippet>2033</a>        <span class="kw">for </span>(predicate, expected) <span class="kw">in </span>predicates {
+<a href=#2034 id=2034 data-nosnippet>2034</a>            <span class="macro">println!</span>(<span class="string">"testing predicate {predicate}"</span>);
+<a href=#2035 id=2035 data-nosnippet>2035</a>            <span class="kw">let </span>result_data = test_perform_read(
+<a href=#2036 id=2036 data-nosnippet>2036</a>                predicate.clone(),
+<a href=#2037 id=2037 data-nosnippet>2037</a>                schema.clone(),
+<a href=#2038 id=2038 data-nosnippet>2038</a>                table_location.clone(),
+<a href=#2039 id=2039 data-nosnippet>2039</a>                reader.clone(),
+<a href=#2040 id=2040 data-nosnippet>2040</a>            )
+<a href=#2041 id=2041 data-nosnippet>2041</a>            .<span class="kw">await</span>;
+<a href=#2042 id=2042 data-nosnippet>2042</a>
+<a href=#2043 id=2043 data-nosnippet>2043</a>            <span class="macro">assert_eq!</span>(result_data, expected, <span class="string">"predicate={predicate}"</span>);
+<a href=#2044 id=2044 data-nosnippet>2044</a>        }
+<a href=#2045 id=2045 data-nosnippet>2045</a>    }
 <a href=#2046 id=2046 data-nosnippet>2046</a>
-<a href=#2047 id=2047 data-nosnippet>2047</a>        </span><span class="kw">let </span>positional_deletes = RoaringTreemap::from_iter(<span class="kw-2">&amp;</span>[
-<a href=#2048 id=2048 data-nosnippet>2048</a>            <span class="number">1</span>, <span class="comment">// in skipped rg 0, should be ignored
-<a href=#2049 id=2049 data-nosnippet>2049</a>            </span><span class="number">3</span>, <span class="comment">// run of three consecutive items in skipped rg0
-<a href=#2050 id=2050 data-nosnippet>2050</a>            </span><span class="number">4</span>, <span class="number">5</span>, <span class="number">998</span>, <span class="comment">// two consecutive items at end of skipped rg0
-<a href=#2051 id=2051 data-nosnippet>2051</a>            </span><span class="number">999</span>, <span class="number">1000</span>, <span class="comment">// solitary row at start of selected rg1 (1, 9)
-<a href=#2052 id=2052 data-nosnippet>2052</a>            </span><span class="number">1010</span>, <span class="comment">// run of 3 rows in selected rg1
-<a href=#2053 id=2053 data-nosnippet>2053</a>            </span><span class="number">1011</span>, <span class="number">1012</span>, <span class="comment">// (3, 485)
-<a href=#2054 id=2054 data-nosnippet>2054</a>            </span><span class="number">1498</span>, <span class="comment">// run of two items at end of selected rg1
-<a href=#2055 id=2055 data-nosnippet>2055</a>            </span><span class="number">1499</span>, <span class="number">1500</span>, <span class="comment">// run of two items at start of skipped rg2
-<a href=#2056 id=2056 data-nosnippet>2056</a>            </span><span class="number">1501</span>, <span class="number">1600</span>, <span class="comment">// should ignore, in skipped rg2
-<a href=#2057 id=2057 data-nosnippet>2057</a>            </span><span class="number">1999</span>, <span class="comment">// single row at end of skipped rg2
-<a href=#2058 id=2058 data-nosnippet>2058</a>            </span><span class="number">2000</span>, <span class="comment">// run of two items at start of selected rg3
-<a href=#2059 id=2059 data-nosnippet>2059</a>            </span><span class="number">2001</span>, <span class="comment">// (4, 98)
-<a href=#2060 id=2060 data-nosnippet>2060</a>            </span><span class="number">2100</span>, <span class="comment">// single row in selected row group 3 (1, 99)
-<a href=#2061 id=2061 data-nosnippet>2061</a>            </span><span class="number">2200</span>, <span class="comment">// run of 3 consecutive rows in selected row group 3
-<a href=#2062 id=2062 data-nosnippet>2062</a>            </span><span class="number">2201</span>, <span class="number">2202</span>, <span class="comment">// (3, 796)
-<a href=#2063 id=2063 data-nosnippet>2063</a>            </span><span class="number">2999</span>, <span class="comment">// single item at end of selected rg3 (1)
-<a href=#2064 id=2064 data-nosnippet>2064</a>            </span><span class="number">3000</span>, <span class="comment">// single item at start of skipped rg4
-<a href=#2065 id=2065 data-nosnippet>2065</a>        </span>]);
-<a href=#2066 id=2066 data-nosnippet>2066</a>
-<a href=#2067 id=2067 data-nosnippet>2067</a>        <span class="kw">let </span>positional_deletes = DeleteVector::new(positional_deletes);
-<a href=#2068 id=2068 data-nosnippet>2068</a>
-<a href=#2069 id=2069 data-nosnippet>2069</a>        <span class="comment">// using selected row groups 1 and 3
-<a href=#2070 id=2070 data-nosnippet>2070</a>        </span><span class="kw">let </span>result = ArrowReader::build_deletes_row_selection(
-<a href=#2071 id=2071 data-nosnippet>2071</a>            <span class="kw-2">&amp;</span>row_groups_metadata,
-<a href=#2072 id=2072 data-nosnippet>2072</a>            <span class="kw-2">&amp;</span>selected_row_groups,
-<a href=#2073 id=2073 data-nosnippet>2073</a>            <span class="kw-2">&amp;</span>positional_deletes,
-<a href=#2074 id=2074 data-nosnippet>2074</a>        )
-<a href=#2075 id=2075 data-nosnippet>2075</a>        .unwrap();
-<a href=#2076 id=2076 data-nosnippet>2076</a>
-<a href=#2077 id=2077 data-nosnippet>2077</a>        <span class="kw">let </span>expected = RowSelection::from(<span class="macro">vec!</span>[
-<a href=#2078 id=2078 data-nosnippet>2078</a>            RowSelector::skip(<span class="number">1</span>),
-<a href=#2079 id=2079 data-nosnippet>2079</a>            RowSelector::select(<span class="number">9</span>),
-<a href=#2080 id=2080 data-nosnippet>2080</a>            RowSelector::skip(<span class="number">3</span>),
-<a href=#2081 id=2081 data-nosnippet>2081</a>            RowSelector::select(<span class="number">485</span>),
-<a href=#2082 id=2082 data-nosnippet>2082</a>            RowSelector::skip(<span class="number">4</span>),
-<a href=#2083 id=2083 data-nosnippet>2083</a>            RowSelector::select(<span class="number">98</span>),
-<a href=#2084 id=2084 data-nosnippet>2084</a>            RowSelector::skip(<span class="number">1</span>),
-<a href=#2085 id=2085 data-nosnippet>2085</a>            RowSelector::select(<span class="number">99</span>),
-<a href=#2086 id=2086 data-nosnippet>2086</a>            RowSelector::skip(<span class="number">3</span>),
-<a href=#2087 id=2087 data-nosnippet>2087</a>            RowSelector::select(<span class="number">796</span>),
-<a href=#2088 id=2088 data-nosnippet>2088</a>            RowSelector::skip(<span class="number">1</span>),
-<a href=#2089 id=2089 data-nosnippet>2089</a>        ]);
-<a href=#2090 id=2090 data-nosnippet>2090</a>
-<a href=#2091 id=2091 data-nosnippet>2091</a>        <span class="macro">assert_eq!</span>(result, expected);
-<a href=#2092 id=2092 data-nosnippet>2092</a>
-<a href=#2093 id=2093 data-nosnippet>2093</a>        <span class="comment">// selecting all row groups
-<a href=#2094 id=2094 data-nosnippet>2094</a>        </span><span class="kw">let </span>result = ArrowReader::build_deletes_row_selection(
-<a href=#2095 id=2095 data-nosnippet>2095</a>            <span class="kw-2">&amp;</span>row_groups_metadata,
-<a href=#2096 id=2096 data-nosnippet>2096</a>            <span class="kw-2">&amp;</span><span class="prelude-val">None</span>,
-<a href=#2097 id=2097 data-nosnippet>2097</a>            <span class="kw-2">&amp;</span>positional_deletes,
-<a href=#2098 id=2098 data-nosnippet>2098</a>        )
-<a href=#2099 id=2099 data-nosnippet>2099</a>        .unwrap();
-<a href=#2100 id=2100 data-nosnippet>2100</a>
-<a href=#2101 id=2101 data-nosnippet>2101</a>        <span class="kw">let </span>expected = RowSelection::from(<span class="macro">vec!</span>[
-<a href=#2102 id=2102 data-nosnippet>2102</a>            RowSelector::select(<span class="number">1</span>),
-<a href=#2103 id=2103 data-nosnippet>2103</a>            RowSelector::skip(<span class="number">1</span>),
-<a href=#2104 id=2104 data-nosnippet>2104</a>            RowSelector::select(<span class="number">1</span>),
-<a href=#2105 id=2105 data-nosnippet>2105</a>            RowSelector::skip(<span class="number">3</span>),
-<a href=#2106 id=2106 data-nosnippet>2106</a>            RowSelector::select(<span class="number">992</span>),
-<a href=#2107 id=2107 data-nosnippet>2107</a>            RowSelector::skip(<span class="number">3</span>),
-<a href=#2108 id=2108 data-nosnippet>2108</a>            RowSelector::select(<span class="number">9</span>),
-<a href=#2109 id=2109 data-nosnippet>2109</a>            RowSelector::skip(<span class="number">3</span>),
-<a href=#2110 id=2110 data-nosnippet>2110</a>            RowSelector::select(<span class="number">485</span>),
-<a href=#2111 id=2111 data-nosnippet>2111</a>            RowSelector::skip(<span class="number">4</span>),
-<a href=#2112 id=2112 data-nosnippet>2112</a>            RowSelector::select(<span class="number">98</span>),
-<a href=#2113 id=2113 data-nosnippet>2113</a>            RowSelector::skip(<span class="number">1</span>),
-<a href=#2114 id=2114 data-nosnippet>2114</a>            RowSelector::select(<span class="number">398</span>),
-<a href=#2115 id=2115 data-nosnippet>2115</a>            RowSelector::skip(<span class="number">3</span>),
-<a href=#2116 id=2116 data-nosnippet>2116</a>            RowSelector::select(<span class="number">98</span>),
-<a href=#2117 id=2117 data-nosnippet>2117</a>            RowSelector::skip(<span class="number">1</span>),
-<a href=#2118 id=2118 data-nosnippet>2118</a>            RowSelector::select(<span class="number">99</span>),
-<a href=#2119 id=2119 data-nosnippet>2119</a>            RowSelector::skip(<span class="number">3</span>),
-<a href=#2120 id=2120 data-nosnippet>2120</a>            RowSelector::select(<span class="number">796</span>),
-<a href=#2121 id=2121 data-nosnippet>2121</a>            RowSelector::skip(<span class="number">2</span>),
-<a href=#2122 id=2122 data-nosnippet>2122</a>            RowSelector::select(<span class="number">499</span>),
-<a href=#2123 id=2123 data-nosnippet>2123</a>        ]);
+<a href=#2047 id=2047 data-nosnippet>2047</a>    <span class="kw">async fn </span>test_perform_read(
+<a href=#2048 id=2048 data-nosnippet>2048</a>        predicate: Predicate,
+<a href=#2049 id=2049 data-nosnippet>2049</a>        schema: SchemaRef,
+<a href=#2050 id=2050 data-nosnippet>2050</a>        table_location: String,
+<a href=#2051 id=2051 data-nosnippet>2051</a>        reader: ArrowReader,
+<a href=#2052 id=2052 data-nosnippet>2052</a>    ) -&gt; Vec&lt;<span class="prelude-ty">Option</span>&lt;String&gt;&gt; {
+<a href=#2053 id=2053 data-nosnippet>2053</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#2054 id=2054 data-nosnippet>2054</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#2055 id=2055 data-nosnippet>2055</a>                start: <span class="number">0</span>,
+<a href=#2056 id=2056 data-nosnippet>2056</a>                length: <span class="number">0</span>,
+<a href=#2057 id=2057 data-nosnippet>2057</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#2058 id=2058 data-nosnippet>2058</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{table_location}/1.parquet"</span>),
+<a href=#2059 id=2059 data-nosnippet>2059</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#2060 id=2060 data-nosnippet>2060</a>                schema: schema.clone(),
+<a href=#2061 id=2061 data-nosnippet>2061</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
+<a href=#2062 id=2062 data-nosnippet>2062</a>                predicate: <span class="prelude-val">Some</span>(predicate.bind(schema, <span class="bool-val">true</span>).unwrap()),
+<a href=#2063 id=2063 data-nosnippet>2063</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#2064 id=2064 data-nosnippet>2064</a>                partition: <span class="prelude-val">None</span>,
+<a href=#2065 id=2065 data-nosnippet>2065</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#2066 id=2066 data-nosnippet>2066</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#2067 id=2067 data-nosnippet>2067</a>            })]
+<a href=#2068 id=2068 data-nosnippet>2068</a>            .into_iter(),
+<a href=#2069 id=2069 data-nosnippet>2069</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#2070 id=2070 data-nosnippet>2070</a>
+<a href=#2071 id=2071 data-nosnippet>2071</a>        <span class="kw">let </span>result = reader
+<a href=#2072 id=2072 data-nosnippet>2072</a>            .read(tasks)
+<a href=#2073 id=2073 data-nosnippet>2073</a>            .unwrap()
+<a href=#2074 id=2074 data-nosnippet>2074</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#2075 id=2075 data-nosnippet>2075</a>            .<span class="kw">await
+<a href=#2076 id=2076 data-nosnippet>2076</a>            </span>.unwrap();
+<a href=#2077 id=2077 data-nosnippet>2077</a>
+<a href=#2078 id=2078 data-nosnippet>2078</a>        result[<span class="number">0</span>].columns()[<span class="number">0</span>]
+<a href=#2079 id=2079 data-nosnippet>2079</a>            .as_string_opt::&lt;i32&gt;()
+<a href=#2080 id=2080 data-nosnippet>2080</a>            .unwrap()
+<a href=#2081 id=2081 data-nosnippet>2081</a>            .iter()
+<a href=#2082 id=2082 data-nosnippet>2082</a>            .map(|v| v.map(ToOwned::to_owned))
+<a href=#2083 id=2083 data-nosnippet>2083</a>            .collect::&lt;Vec&lt;<span class="kw">_</span>&gt;&gt;()
+<a href=#2084 id=2084 data-nosnippet>2084</a>    }
+<a href=#2085 id=2085 data-nosnippet>2085</a>
+<a href=#2086 id=2086 data-nosnippet>2086</a>    <span class="kw">fn </span>setup_kleene_logic(
+<a href=#2087 id=2087 data-nosnippet>2087</a>        data_for_col_a: Vec&lt;<span class="prelude-ty">Option</span>&lt;String&gt;&gt;,
+<a href=#2088 id=2088 data-nosnippet>2088</a>        col_a_type: DataType,
+<a href=#2089 id=2089 data-nosnippet>2089</a>    ) -&gt; (FileIO, SchemaRef, String, TempDir) {
+<a href=#2090 id=2090 data-nosnippet>2090</a>        <span class="kw">let </span>schema = Arc::new(
+<a href=#2091 id=2091 data-nosnippet>2091</a>            Schema::builder()
+<a href=#2092 id=2092 data-nosnippet>2092</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#2093 id=2093 data-nosnippet>2093</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#2094 id=2094 data-nosnippet>2094</a>                    NestedField::optional(<span class="number">1</span>, <span class="string">"a"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#2095 id=2095 data-nosnippet>2095</a>                ])
+<a href=#2096 id=2096 data-nosnippet>2096</a>                .build()
+<a href=#2097 id=2097 data-nosnippet>2097</a>                .unwrap(),
+<a href=#2098 id=2098 data-nosnippet>2098</a>        );
+<a href=#2099 id=2099 data-nosnippet>2099</a>
+<a href=#2100 id=2100 data-nosnippet>2100</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#2101 id=2101 data-nosnippet>2101</a>            Field::new(<span class="string">"a"</span>, col_a_type.clone(), <span class="bool-val">true</span>).with_metadata(HashMap::from([(
+<a href=#2102 id=2102 data-nosnippet>2102</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2103 id=2103 data-nosnippet>2103</a>                <span class="string">"1"</span>.to_string(),
+<a href=#2104 id=2104 data-nosnippet>2104</a>            )])),
+<a href=#2105 id=2105 data-nosnippet>2105</a>        ]));
+<a href=#2106 id=2106 data-nosnippet>2106</a>
+<a href=#2107 id=2107 data-nosnippet>2107</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#2108 id=2108 data-nosnippet>2108</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#2109 id=2109 data-nosnippet>2109</a>
+<a href=#2110 id=2110 data-nosnippet>2110</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#2111 id=2111 data-nosnippet>2111</a>
+<a href=#2112 id=2112 data-nosnippet>2112</a>        <span class="kw">let </span>col = <span class="kw">match </span>col_a_type {
+<a href=#2113 id=2113 data-nosnippet>2113</a>            DataType::Utf8 =&gt; Arc::new(StringArray::from(data_for_col_a)) <span class="kw">as </span>ArrayRef,
+<a href=#2114 id=2114 data-nosnippet>2114</a>            DataType::LargeUtf8 =&gt; Arc::new(LargeStringArray::from(data_for_col_a)) <span class="kw">as </span>ArrayRef,
+<a href=#2115 id=2115 data-nosnippet>2115</a>            <span class="kw">_ </span>=&gt; <span class="macro">panic!</span>(<span class="string">"unexpected col_a_type"</span>),
+<a href=#2116 id=2116 data-nosnippet>2116</a>        };
+<a href=#2117 id=2117 data-nosnippet>2117</a>
+<a href=#2118 id=2118 data-nosnippet>2118</a>        <span class="kw">let </span>to_write = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[col]).unwrap();
+<a href=#2119 id=2119 data-nosnippet>2119</a>
+<a href=#2120 id=2120 data-nosnippet>2120</a>        <span class="comment">// Write the Parquet files
+<a href=#2121 id=2121 data-nosnippet>2121</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
+<a href=#2122 id=2122 data-nosnippet>2122</a>            .set_compression(Compression::SNAPPY)
+<a href=#2123 id=2123 data-nosnippet>2123</a>            .build();
 <a href=#2124 id=2124 data-nosnippet>2124</a>
-<a href=#2125 id=2125 data-nosnippet>2125</a>        <span class="macro">assert_eq!</span>(result, expected);
-<a href=#2126 id=2126 data-nosnippet>2126</a>    }
-<a href=#2127 id=2127 data-nosnippet>2127</a>
-<a href=#2128 id=2128 data-nosnippet>2128</a>    <span class="kw">fn </span>build_test_row_group_meta(
-<a href=#2129 id=2129 data-nosnippet>2129</a>        schema_descr: SchemaDescPtr,
-<a href=#2130 id=2130 data-nosnippet>2130</a>        columns: Vec&lt;ColumnChunkMetaData&gt;,
-<a href=#2131 id=2131 data-nosnippet>2131</a>        num_rows: i64,
-<a href=#2132 id=2132 data-nosnippet>2132</a>        ordinal: i16,
-<a href=#2133 id=2133 data-nosnippet>2133</a>    ) -&gt; RowGroupMetaData {
-<a href=#2134 id=2134 data-nosnippet>2134</a>        RowGroupMetaData::builder(schema_descr.clone())
-<a href=#2135 id=2135 data-nosnippet>2135</a>            .set_num_rows(num_rows)
-<a href=#2136 id=2136 data-nosnippet>2136</a>            .set_total_byte_size(<span class="number">2000</span>)
-<a href=#2137 id=2137 data-nosnippet>2137</a>            .set_column_metadata(columns)
-<a href=#2138 id=2138 data-nosnippet>2138</a>            .set_ordinal(ordinal)
-<a href=#2139 id=2139 data-nosnippet>2139</a>            .build()
-<a href=#2140 id=2140 data-nosnippet>2140</a>            .unwrap()
-<a href=#2141 id=2141 data-nosnippet>2141</a>    }
-<a href=#2142 id=2142 data-nosnippet>2142</a>
-<a href=#2143 id=2143 data-nosnippet>2143</a>    <span class="kw">fn </span>get_test_schema_descr() -&gt; SchemaDescPtr {
-<a href=#2144 id=2144 data-nosnippet>2144</a>        <span class="kw">use </span>parquet::schema::types::Type <span class="kw">as </span>SchemaType;
-<a href=#2145 id=2145 data-nosnippet>2145</a>
-<a href=#2146 id=2146 data-nosnippet>2146</a>        <span class="kw">let </span>schema = SchemaType::group_type_builder(<span class="string">"schema"</span>)
-<a href=#2147 id=2147 data-nosnippet>2147</a>            .with_fields(<span class="macro">vec!</span>[
-<a href=#2148 id=2148 data-nosnippet>2148</a>                Arc::new(
-<a href=#2149 id=2149 data-nosnippet>2149</a>                    SchemaType::primitive_type_builder(<span class="string">"a"</span>, parquet::basic::Type::INT32)
-<a href=#2150 id=2150 data-nosnippet>2150</a>                        .build()
-<a href=#2151 id=2151 data-nosnippet>2151</a>                        .unwrap(),
-<a href=#2152 id=2152 data-nosnippet>2152</a>                ),
-<a href=#2153 id=2153 data-nosnippet>2153</a>                Arc::new(
-<a href=#2154 id=2154 data-nosnippet>2154</a>                    SchemaType::primitive_type_builder(<span class="string">"b"</span>, parquet::basic::Type::INT32)
-<a href=#2155 id=2155 data-nosnippet>2155</a>                        .build()
-<a href=#2156 id=2156 data-nosnippet>2156</a>                        .unwrap(),
-<a href=#2157 id=2157 data-nosnippet>2157</a>                ),
-<a href=#2158 id=2158 data-nosnippet>2158</a>            ])
-<a href=#2159 id=2159 data-nosnippet>2159</a>            .build()
-<a href=#2160 id=2160 data-nosnippet>2160</a>            .unwrap();
-<a href=#2161 id=2161 data-nosnippet>2161</a>
-<a href=#2162 id=2162 data-nosnippet>2162</a>        Arc::new(SchemaDescriptor::new(Arc::new(schema)))
-<a href=#2163 id=2163 data-nosnippet>2163</a>    }
-<a href=#2164 id=2164 data-nosnippet>2164</a>
-<a href=#2165 id=2165 data-nosnippet>2165</a>    <span class="doccomment">/// Verifies that file splits respect byte ranges and only read specific row groups.
-<a href=#2166 id=2166 data-nosnippet>2166</a>    </span><span class="attr">#[tokio::test]
-<a href=#2167 id=2167 data-nosnippet>2167</a>    </span><span class="kw">async fn </span>test_file_splits_respect_byte_ranges() {
-<a href=#2168 id=2168 data-nosnippet>2168</a>        <span class="kw">use </span>arrow_array::Int32Array;
-<a href=#2169 id=2169 data-nosnippet>2169</a>        <span class="kw">use </span>parquet::file::reader::{FileReader, SerializedFileReader};
-<a href=#2170 id=2170 data-nosnippet>2170</a>
-<a href=#2171 id=2171 data-nosnippet>2171</a>        <span class="kw">let </span>schema = Arc::new(
-<a href=#2172 id=2172 data-nosnippet>2172</a>            Schema::builder()
-<a href=#2173 id=2173 data-nosnippet>2173</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#2174 id=2174 data-nosnippet>2174</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#2175 id=2175 data-nosnippet>2175</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#2176 id=2176 data-nosnippet>2176</a>                ])
-<a href=#2177 id=2177 data-nosnippet>2177</a>                .build()
-<a href=#2178 id=2178 data-nosnippet>2178</a>                .unwrap(),
-<a href=#2179 id=2179 data-nosnippet>2179</a>        );
-<a href=#2180 id=2180 data-nosnippet>2180</a>
-<a href=#2181 id=2181 data-nosnippet>2181</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#2182 id=2182 data-nosnippet>2182</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2183 id=2183 data-nosnippet>2183</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2184 id=2184 data-nosnippet>2184</a>                <span class="string">"1"</span>.to_string(),
-<a href=#2185 id=2185 data-nosnippet>2185</a>            )])),
-<a href=#2186 id=2186 data-nosnippet>2186</a>        ]));
-<a href=#2187 id=2187 data-nosnippet>2187</a>
-<a href=#2188 id=2188 data-nosnippet>2188</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#2189 id=2189 data-nosnippet>2189</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#2190 id=2190 data-nosnippet>2190</a>        <span class="kw">let </span>file_path = <span class="macro">format!</span>(<span class="string">"{}/multi_row_group.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
-<a href=#2191 id=2191 data-nosnippet>2191</a>
-<a href=#2192 id=2192 data-nosnippet>2192</a>        <span class="comment">// Force each batch into its own row group for testing byte range filtering.
-<a href=#2193 id=2193 data-nosnippet>2193</a>        </span><span class="kw">let </span>batch1 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(Int32Array::from(
-<a href=#2194 id=2194 data-nosnippet>2194</a>            (<span class="number">0</span>..<span class="number">100</span>).collect::&lt;Vec&lt;i32&gt;&gt;(),
-<a href=#2195 id=2195 data-nosnippet>2195</a>        ))])
-<a href=#2196 id=2196 data-nosnippet>2196</a>        .unwrap();
-<a href=#2197 id=2197 data-nosnippet>2197</a>        <span class="kw">let </span>batch2 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(Int32Array::from(
-<a href=#2198 id=2198 data-nosnippet>2198</a>            (<span class="number">100</span>..<span class="number">200</span>).collect::&lt;Vec&lt;i32&gt;&gt;(),
-<a href=#2199 id=2199 data-nosnippet>2199</a>        ))])
-<a href=#2200 id=2200 data-nosnippet>2200</a>        .unwrap();
-<a href=#2201 id=2201 data-nosnippet>2201</a>        <span class="kw">let </span>batch3 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(Int32Array::from(
-<a href=#2202 id=2202 data-nosnippet>2202</a>            (<span class="number">200</span>..<span class="number">300</span>).collect::&lt;Vec&lt;i32&gt;&gt;(),
-<a href=#2203 id=2203 data-nosnippet>2203</a>        ))])
-<a href=#2204 id=2204 data-nosnippet>2204</a>        .unwrap();
-<a href=#2205 id=2205 data-nosnippet>2205</a>
-<a href=#2206 id=2206 data-nosnippet>2206</a>        <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#2207 id=2207 data-nosnippet>2207</a>            .set_compression(Compression::SNAPPY)
-<a href=#2208 id=2208 data-nosnippet>2208</a>            .set_max_row_group_size(<span class="number">100</span>)
-<a href=#2209 id=2209 data-nosnippet>2209</a>            .build();
-<a href=#2210 id=2210 data-nosnippet>2210</a>
-<a href=#2211 id=2211 data-nosnippet>2211</a>        <span class="kw">let </span>file = File::create(<span class="kw-2">&amp;</span>file_path).unwrap();
-<a href=#2212 id=2212 data-nosnippet>2212</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#2213 id=2213 data-nosnippet>2213</a>        writer.write(<span class="kw-2">&amp;</span>batch1).expect(<span class="string">"Writing batch 1"</span>);
-<a href=#2214 id=2214 data-nosnippet>2214</a>        writer.write(<span class="kw-2">&amp;</span>batch2).expect(<span class="string">"Writing batch 2"</span>);
-<a href=#2215 id=2215 data-nosnippet>2215</a>        writer.write(<span class="kw-2">&amp;</span>batch3).expect(<span class="string">"Writing batch 3"</span>);
-<a href=#2216 id=2216 data-nosnippet>2216</a>        writer.close().unwrap();
-<a href=#2217 id=2217 data-nosnippet>2217</a>
-<a href=#2218 id=2218 data-nosnippet>2218</a>        <span class="comment">// Read the file metadata to get row group byte positions
-<a href=#2219 id=2219 data-nosnippet>2219</a>        </span><span class="kw">let </span>file = File::open(<span class="kw-2">&amp;</span>file_path).unwrap();
-<a href=#2220 id=2220 data-nosnippet>2220</a>        <span class="kw">let </span>reader = SerializedFileReader::new(file).unwrap();
-<a href=#2221 id=2221 data-nosnippet>2221</a>        <span class="kw">let </span>metadata = reader.metadata();
-<a href=#2222 id=2222 data-nosnippet>2222</a>
-<a href=#2223 id=2223 data-nosnippet>2223</a>        <span class="macro">println!</span>(<span class="string">"File has {} row groups"</span>, metadata.num_row_groups());
-<a href=#2224 id=2224 data-nosnippet>2224</a>        <span class="macro">assert_eq!</span>(metadata.num_row_groups(), <span class="number">3</span>, <span class="string">"Expected 3 row groups"</span>);
-<a href=#2225 id=2225 data-nosnippet>2225</a>
-<a href=#2226 id=2226 data-nosnippet>2226</a>        <span class="comment">// Get byte positions for each row group
-<a href=#2227 id=2227 data-nosnippet>2227</a>        </span><span class="kw">let </span>row_group_0 = metadata.row_group(<span class="number">0</span>);
-<a href=#2228 id=2228 data-nosnippet>2228</a>        <span class="kw">let </span>row_group_1 = metadata.row_group(<span class="number">1</span>);
-<a href=#2229 id=2229 data-nosnippet>2229</a>        <span class="kw">let </span>row_group_2 = metadata.row_group(<span class="number">2</span>);
-<a href=#2230 id=2230 data-nosnippet>2230</a>
-<a href=#2231 id=2231 data-nosnippet>2231</a>        <span class="kw">let </span>rg0_start = <span class="number">4u64</span>; <span class="comment">// Parquet files start with 4-byte magic "PAR1"
-<a href=#2232 id=2232 data-nosnippet>2232</a>        </span><span class="kw">let </span>rg1_start = rg0_start + row_group_0.compressed_size() <span class="kw">as </span>u64;
-<a href=#2233 id=2233 data-nosnippet>2233</a>        <span class="kw">let </span>rg2_start = rg1_start + row_group_1.compressed_size() <span class="kw">as </span>u64;
-<a href=#2234 id=2234 data-nosnippet>2234</a>        <span class="kw">let </span>file_end = rg2_start + row_group_2.compressed_size() <span class="kw">as </span>u64;
-<a href=#2235 id=2235 data-nosnippet>2235</a>
-<a href=#2236 id=2236 data-nosnippet>2236</a>        <span class="macro">println!</span>(
-<a href=#2237 id=2237 data-nosnippet>2237</a>            <span class="string">"Row group 0: {} rows, starts at byte {}, {} bytes compressed"</span>,
-<a href=#2238 id=2238 data-nosnippet>2238</a>            row_group_0.num_rows(),
-<a href=#2239 id=2239 data-nosnippet>2239</a>            rg0_start,
-<a href=#2240 id=2240 data-nosnippet>2240</a>            row_group_0.compressed_size()
-<a href=#2241 id=2241 data-nosnippet>2241</a>        );
-<a href=#2242 id=2242 data-nosnippet>2242</a>        <span class="macro">println!</span>(
-<a href=#2243 id=2243 data-nosnippet>2243</a>            <span class="string">"Row group 1: {} rows, starts at byte {}, {} bytes compressed"</span>,
-<a href=#2244 id=2244 data-nosnippet>2244</a>            row_group_1.num_rows(),
-<a href=#2245 id=2245 data-nosnippet>2245</a>            rg1_start,
-<a href=#2246 id=2246 data-nosnippet>2246</a>            row_group_1.compressed_size()
-<a href=#2247 id=2247 data-nosnippet>2247</a>        );
-<a href=#2248 id=2248 data-nosnippet>2248</a>        <span class="macro">println!</span>(
-<a href=#2249 id=2249 data-nosnippet>2249</a>            <span class="string">"Row group 2: {} rows, starts at byte {}, {} bytes compressed"</span>,
-<a href=#2250 id=2250 data-nosnippet>2250</a>            row_group_2.num_rows(),
-<a href=#2251 id=2251 data-nosnippet>2251</a>            rg2_start,
-<a href=#2252 id=2252 data-nosnippet>2252</a>            row_group_2.compressed_size()
-<a href=#2253 id=2253 data-nosnippet>2253</a>        );
-<a href=#2254 id=2254 data-nosnippet>2254</a>
-<a href=#2255 id=2255 data-nosnippet>2255</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#2256 id=2256 data-nosnippet>2256</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#2257 id=2257 data-nosnippet>2257</a>
-<a href=#2258 id=2258 data-nosnippet>2258</a>        <span class="comment">// Task 1: read only the first row group
-<a href=#2259 id=2259 data-nosnippet>2259</a>        </span><span class="kw">let </span>task1 = FileScanTask {
-<a href=#2260 id=2260 data-nosnippet>2260</a>            start: rg0_start,
-<a href=#2261 id=2261 data-nosnippet>2261</a>            length: row_group_0.compressed_size() <span class="kw">as </span>u64,
-<a href=#2262 id=2262 data-nosnippet>2262</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">100</span>),
-<a href=#2263 id=2263 data-nosnippet>2263</a>            data_file_path: file_path.clone(),
-<a href=#2264 id=2264 data-nosnippet>2264</a>            data_file_format: DataFileFormat::Parquet,
-<a href=#2265 id=2265 data-nosnippet>2265</a>            schema: schema.clone(),
-<a href=#2266 id=2266 data-nosnippet>2266</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
-<a href=#2267 id=2267 data-nosnippet>2267</a>            predicate: <span class="prelude-val">None</span>,
-<a href=#2268 id=2268 data-nosnippet>2268</a>            deletes: <span class="macro">vec!</span>[],
-<a href=#2269 id=2269 data-nosnippet>2269</a>        };
-<a href=#2270 id=2270 data-nosnippet>2270</a>
-<a href=#2271 id=2271 data-nosnippet>2271</a>        <span class="comment">// Task 2: read the second and third row groups
-<a href=#2272 id=2272 data-nosnippet>2272</a>        </span><span class="kw">let </span>task2 = FileScanTask {
-<a href=#2273 id=2273 data-nosnippet>2273</a>            start: rg1_start,
-<a href=#2274 id=2274 data-nosnippet>2274</a>            length: file_end - rg1_start,
-<a href=#2275 id=2275 data-nosnippet>2275</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">200</span>),
-<a href=#2276 id=2276 data-nosnippet>2276</a>            data_file_path: file_path.clone(),
-<a href=#2277 id=2277 data-nosnippet>2277</a>            data_file_format: DataFileFormat::Parquet,
-<a href=#2278 id=2278 data-nosnippet>2278</a>            schema: schema.clone(),
-<a href=#2279 id=2279 data-nosnippet>2279</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
-<a href=#2280 id=2280 data-nosnippet>2280</a>            predicate: <span class="prelude-val">None</span>,
-<a href=#2281 id=2281 data-nosnippet>2281</a>            deletes: <span class="macro">vec!</span>[],
-<a href=#2282 id=2282 data-nosnippet>2282</a>        };
-<a href=#2283 id=2283 data-nosnippet>2283</a>
-<a href=#2284 id=2284 data-nosnippet>2284</a>        <span class="kw">let </span>tasks1 = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task1)])) <span class="kw">as </span>FileScanTaskStream;
-<a href=#2285 id=2285 data-nosnippet>2285</a>        <span class="kw">let </span>result1 = reader
-<a href=#2286 id=2286 data-nosnippet>2286</a>            .clone()
-<a href=#2287 id=2287 data-nosnippet>2287</a>            .read(tasks1)
-<a href=#2288 id=2288 data-nosnippet>2288</a>            .unwrap()
-<a href=#2289 id=2289 data-nosnippet>2289</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#2290 id=2290 data-nosnippet>2290</a>            .<span class="kw">await
-<a href=#2291 id=2291 data-nosnippet>2291</a>            </span>.unwrap();
-<a href=#2292 id=2292 data-nosnippet>2292</a>
-<a href=#2293 id=2293 data-nosnippet>2293</a>        <span class="kw">let </span>total_rows_task1: usize = result1.iter().map(|b| b.num_rows()).sum();
-<a href=#2294 id=2294 data-nosnippet>2294</a>        <span class="macro">println!</span>(
-<a href=#2295 id=2295 data-nosnippet>2295</a>            <span class="string">"Task 1 (bytes {}-{}) returned {} rows"</span>,
-<a href=#2296 id=2296 data-nosnippet>2296</a>            rg0_start,
-<a href=#2297 id=2297 data-nosnippet>2297</a>            rg0_start + row_group_0.compressed_size() <span class="kw">as </span>u64,
-<a href=#2298 id=2298 data-nosnippet>2298</a>            total_rows_task1
-<a href=#2299 id=2299 data-nosnippet>2299</a>        );
-<a href=#2300 id=2300 data-nosnippet>2300</a>
-<a href=#2301 id=2301 data-nosnippet>2301</a>        <span class="kw">let </span>tasks2 = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task2)])) <span class="kw">as </span>FileScanTaskStream;
-<a href=#2302 id=2302 data-nosnippet>2302</a>        <span class="kw">let </span>result2 = reader
-<a href=#2303 id=2303 data-nosnippet>2303</a>            .read(tasks2)
-<a href=#2304 id=2304 data-nosnippet>2304</a>            .unwrap()
-<a href=#2305 id=2305 data-nosnippet>2305</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#2306 id=2306 data-nosnippet>2306</a>            .<span class="kw">await
-<a href=#2307 id=2307 data-nosnippet>2307</a>            </span>.unwrap();
-<a href=#2308 id=2308 data-nosnippet>2308</a>
-<a href=#2309 id=2309 data-nosnippet>2309</a>        <span class="kw">let </span>total_rows_task2: usize = result2.iter().map(|b| b.num_rows()).sum();
-<a href=#2310 id=2310 data-nosnippet>2310</a>        <span class="macro">println!</span>(<span class="string">"Task 2 (bytes {rg1_start}-{file_end}) returned {total_rows_task2} rows"</span>);
-<a href=#2311 id=2311 data-nosnippet>2311</a>
-<a href=#2312 id=2312 data-nosnippet>2312</a>        <span class="macro">assert_eq!</span>(
-<a href=#2313 id=2313 data-nosnippet>2313</a>            total_rows_task1, <span class="number">100</span>,
-<a href=#2314 id=2314 data-nosnippet>2314</a>            <span class="string">"Task 1 should read only the first row group (100 rows), but got {total_rows_task1} rows"
-<a href=#2315 id=2315 data-nosnippet>2315</a>        </span>);
-<a href=#2316 id=2316 data-nosnippet>2316</a>
-<a href=#2317 id=2317 data-nosnippet>2317</a>        <span class="macro">assert_eq!</span>(
-<a href=#2318 id=2318 data-nosnippet>2318</a>            total_rows_task2, <span class="number">200</span>,
-<a href=#2319 id=2319 data-nosnippet>2319</a>            <span class="string">"Task 2 should read only the second+third row groups (200 rows), but got {total_rows_task2} rows"
-<a href=#2320 id=2320 data-nosnippet>2320</a>        </span>);
+<a href=#2125 id=2125 data-nosnippet>2125</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#2126 id=2126 data-nosnippet>2126</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer =
+<a href=#2127 id=2127 data-nosnippet>2127</a>            ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props.clone())).unwrap();
+<a href=#2128 id=2128 data-nosnippet>2128</a>
+<a href=#2129 id=2129 data-nosnippet>2129</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
+<a href=#2130 id=2130 data-nosnippet>2130</a>
+<a href=#2131 id=2131 data-nosnippet>2131</a>        <span class="comment">// writer must be closed to write footer
+<a href=#2132 id=2132 data-nosnippet>2132</a>        </span>writer.close().unwrap();
+<a href=#2133 id=2133 data-nosnippet>2133</a>
+<a href=#2134 id=2134 data-nosnippet>2134</a>        (file_io, schema, table_location, tmp_dir)
+<a href=#2135 id=2135 data-nosnippet>2135</a>    }
+<a href=#2136 id=2136 data-nosnippet>2136</a>
+<a href=#2137 id=2137 data-nosnippet>2137</a>    <span class="attr">#[test]
+<a href=#2138 id=2138 data-nosnippet>2138</a>    </span><span class="kw">fn </span>test_build_deletes_row_selection() {
+<a href=#2139 id=2139 data-nosnippet>2139</a>        <span class="kw">let </span>schema_descr = get_test_schema_descr();
+<a href=#2140 id=2140 data-nosnippet>2140</a>
+<a href=#2141 id=2141 data-nosnippet>2141</a>        <span class="kw">let </span><span class="kw-2">mut </span>columns = <span class="macro">vec!</span>[];
+<a href=#2142 id=2142 data-nosnippet>2142</a>        <span class="kw">for </span>ptr <span class="kw">in </span>schema_descr.columns() {
+<a href=#2143 id=2143 data-nosnippet>2143</a>            <span class="kw">let </span>column = ColumnChunkMetaData::builder(ptr.clone()).build().unwrap();
+<a href=#2144 id=2144 data-nosnippet>2144</a>            columns.push(column);
+<a href=#2145 id=2145 data-nosnippet>2145</a>        }
+<a href=#2146 id=2146 data-nosnippet>2146</a>
+<a href=#2147 id=2147 data-nosnippet>2147</a>        <span class="kw">let </span>row_groups_metadata = <span class="macro">vec!</span>[
+<a href=#2148 id=2148 data-nosnippet>2148</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">1000</span>, <span class="number">0</span>),
+<a href=#2149 id=2149 data-nosnippet>2149</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">500</span>, <span class="number">1</span>),
+<a href=#2150 id=2150 data-nosnippet>2150</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">500</span>, <span class="number">2</span>),
+<a href=#2151 id=2151 data-nosnippet>2151</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">1000</span>, <span class="number">3</span>),
+<a href=#2152 id=2152 data-nosnippet>2152</a>            build_test_row_group_meta(schema_descr.clone(), columns.clone(), <span class="number">500</span>, <span class="number">4</span>),
+<a href=#2153 id=2153 data-nosnippet>2153</a>        ];
+<a href=#2154 id=2154 data-nosnippet>2154</a>
+<a href=#2155 id=2155 data-nosnippet>2155</a>        <span class="kw">let </span>selected_row_groups = <span class="prelude-val">Some</span>(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">3</span>]);
+<a href=#2156 id=2156 data-nosnippet>2156</a>
+<a href=#2157 id=2157 data-nosnippet>2157</a>        <span class="comment">/* cases to cover:
+<a href=#2158 id=2158 data-nosnippet>2158</a>           * {skip|select} {first|intermediate|last} {one row|multiple rows} in
+<a href=#2159 id=2159 data-nosnippet>2159</a>             {first|intermediate|last} {skipped|selected} row group
+<a href=#2160 id=2160 data-nosnippet>2160</a>           * row group selection disabled
+<a href=#2161 id=2161 data-nosnippet>2161</a>        */
+<a href=#2162 id=2162 data-nosnippet>2162</a>
+<a href=#2163 id=2163 data-nosnippet>2163</a>        </span><span class="kw">let </span>positional_deletes = RoaringTreemap::from_iter(<span class="kw-2">&amp;</span>[
+<a href=#2164 id=2164 data-nosnippet>2164</a>            <span class="number">1</span>, <span class="comment">// in skipped rg 0, should be ignored
+<a href=#2165 id=2165 data-nosnippet>2165</a>            </span><span class="number">3</span>, <span class="comment">// run of three consecutive items in skipped rg0
+<a href=#2166 id=2166 data-nosnippet>2166</a>            </span><span class="number">4</span>, <span class="number">5</span>, <span class="number">998</span>, <span class="comment">// two consecutive items at end of skipped rg0
+<a href=#2167 id=2167 data-nosnippet>2167</a>            </span><span class="number">999</span>, <span class="number">1000</span>, <span class="comment">// solitary row at start of selected rg1 (1, 9)
+<a href=#2168 id=2168 data-nosnippet>2168</a>            </span><span class="number">1010</span>, <span class="comment">// run of 3 rows in selected rg1
+<a href=#2169 id=2169 data-nosnippet>2169</a>            </span><span class="number">1011</span>, <span class="number">1012</span>, <span class="comment">// (3, 485)
+<a href=#2170 id=2170 data-nosnippet>2170</a>            </span><span class="number">1498</span>, <span class="comment">// run of two items at end of selected rg1
+<a href=#2171 id=2171 data-nosnippet>2171</a>            </span><span class="number">1499</span>, <span class="number">1500</span>, <span class="comment">// run of two items at start of skipped rg2
+<a href=#2172 id=2172 data-nosnippet>2172</a>            </span><span class="number">1501</span>, <span class="number">1600</span>, <span class="comment">// should ignore, in skipped rg2
+<a href=#2173 id=2173 data-nosnippet>2173</a>            </span><span class="number">1999</span>, <span class="comment">// single row at end of skipped rg2
+<a href=#2174 id=2174 data-nosnippet>2174</a>            </span><span class="number">2000</span>, <span class="comment">// run of two items at start of selected rg3
+<a href=#2175 id=2175 data-nosnippet>2175</a>            </span><span class="number">2001</span>, <span class="comment">// (4, 98)
+<a href=#2176 id=2176 data-nosnippet>2176</a>            </span><span class="number">2100</span>, <span class="comment">// single row in selected row group 3 (1, 99)
+<a href=#2177 id=2177 data-nosnippet>2177</a>            </span><span class="number">2200</span>, <span class="comment">// run of 3 consecutive rows in selected row group 3
+<a href=#2178 id=2178 data-nosnippet>2178</a>            </span><span class="number">2201</span>, <span class="number">2202</span>, <span class="comment">// (3, 796)
+<a href=#2179 id=2179 data-nosnippet>2179</a>            </span><span class="number">2999</span>, <span class="comment">// single item at end of selected rg3 (1)
+<a href=#2180 id=2180 data-nosnippet>2180</a>            </span><span class="number">3000</span>, <span class="comment">// single item at start of skipped rg4
+<a href=#2181 id=2181 data-nosnippet>2181</a>        </span>]);
+<a href=#2182 id=2182 data-nosnippet>2182</a>
+<a href=#2183 id=2183 data-nosnippet>2183</a>        <span class="kw">let </span>positional_deletes = DeleteVector::new(positional_deletes);
+<a href=#2184 id=2184 data-nosnippet>2184</a>
+<a href=#2185 id=2185 data-nosnippet>2185</a>        <span class="comment">// using selected row groups 1 and 3
+<a href=#2186 id=2186 data-nosnippet>2186</a>        </span><span class="kw">let </span>result = ArrowReader::build_deletes_row_selection(
+<a href=#2187 id=2187 data-nosnippet>2187</a>            <span class="kw-2">&amp;</span>row_groups_metadata,
+<a href=#2188 id=2188 data-nosnippet>2188</a>            <span class="kw-2">&amp;</span>selected_row_groups,
+<a href=#2189 id=2189 data-nosnippet>2189</a>            <span class="kw-2">&amp;</span>positional_deletes,
+<a href=#2190 id=2190 data-nosnippet>2190</a>        )
+<a href=#2191 id=2191 data-nosnippet>2191</a>        .unwrap();
+<a href=#2192 id=2192 data-nosnippet>2192</a>
+<a href=#2193 id=2193 data-nosnippet>2193</a>        <span class="kw">let </span>expected = RowSelection::from(<span class="macro">vec!</span>[
+<a href=#2194 id=2194 data-nosnippet>2194</a>            RowSelector::skip(<span class="number">1</span>),
+<a href=#2195 id=2195 data-nosnippet>2195</a>            RowSelector::select(<span class="number">9</span>),
+<a href=#2196 id=2196 data-nosnippet>2196</a>            RowSelector::skip(<span class="number">3</span>),
+<a href=#2197 id=2197 data-nosnippet>2197</a>            RowSelector::select(<span class="number">485</span>),
+<a href=#2198 id=2198 data-nosnippet>2198</a>            RowSelector::skip(<span class="number">4</span>),
+<a href=#2199 id=2199 data-nosnippet>2199</a>            RowSelector::select(<span class="number">98</span>),
+<a href=#2200 id=2200 data-nosnippet>2200</a>            RowSelector::skip(<span class="number">1</span>),
+<a href=#2201 id=2201 data-nosnippet>2201</a>            RowSelector::select(<span class="number">99</span>),
+<a href=#2202 id=2202 data-nosnippet>2202</a>            RowSelector::skip(<span class="number">3</span>),
+<a href=#2203 id=2203 data-nosnippet>2203</a>            RowSelector::select(<span class="number">796</span>),
+<a href=#2204 id=2204 data-nosnippet>2204</a>            RowSelector::skip(<span class="number">1</span>),
+<a href=#2205 id=2205 data-nosnippet>2205</a>        ]);
+<a href=#2206 id=2206 data-nosnippet>2206</a>
+<a href=#2207 id=2207 data-nosnippet>2207</a>        <span class="macro">assert_eq!</span>(result, expected);
+<a href=#2208 id=2208 data-nosnippet>2208</a>
+<a href=#2209 id=2209 data-nosnippet>2209</a>        <span class="comment">// selecting all row groups
+<a href=#2210 id=2210 data-nosnippet>2210</a>        </span><span class="kw">let </span>result = ArrowReader::build_deletes_row_selection(
+<a href=#2211 id=2211 data-nosnippet>2211</a>            <span class="kw-2">&amp;</span>row_groups_metadata,
+<a href=#2212 id=2212 data-nosnippet>2212</a>            <span class="kw-2">&amp;</span><span class="prelude-val">None</span>,
+<a href=#2213 id=2213 data-nosnippet>2213</a>            <span class="kw-2">&amp;</span>positional_deletes,
+<a href=#2214 id=2214 data-nosnippet>2214</a>        )
+<a href=#2215 id=2215 data-nosnippet>2215</a>        .unwrap();
+<a href=#2216 id=2216 data-nosnippet>2216</a>
+<a href=#2217 id=2217 data-nosnippet>2217</a>        <span class="kw">let </span>expected = RowSelection::from(<span class="macro">vec!</span>[
+<a href=#2218 id=2218 data-nosnippet>2218</a>            RowSelector::select(<span class="number">1</span>),
+<a href=#2219 id=2219 data-nosnippet>2219</a>            RowSelector::skip(<span class="number">1</span>),
+<a href=#2220 id=2220 data-nosnippet>2220</a>            RowSelector::select(<span class="number">1</span>),
+<a href=#2221 id=2221 data-nosnippet>2221</a>            RowSelector::skip(<span class="number">3</span>),
+<a href=#2222 id=2222 data-nosnippet>2222</a>            RowSelector::select(<span class="number">992</span>),
+<a href=#2223 id=2223 data-nosnippet>2223</a>            RowSelector::skip(<span class="number">3</span>),
+<a href=#2224 id=2224 data-nosnippet>2224</a>            RowSelector::select(<span class="number">9</span>),
+<a href=#2225 id=2225 data-nosnippet>2225</a>            RowSelector::skip(<span class="number">3</span>),
+<a href=#2226 id=2226 data-nosnippet>2226</a>            RowSelector::select(<span class="number">485</span>),
+<a href=#2227 id=2227 data-nosnippet>2227</a>            RowSelector::skip(<span class="number">4</span>),
+<a href=#2228 id=2228 data-nosnippet>2228</a>            RowSelector::select(<span class="number">98</span>),
+<a href=#2229 id=2229 data-nosnippet>2229</a>            RowSelector::skip(<span class="number">1</span>),
+<a href=#2230 id=2230 data-nosnippet>2230</a>            RowSelector::select(<span class="number">398</span>),
+<a href=#2231 id=2231 data-nosnippet>2231</a>            RowSelector::skip(<span class="number">3</span>),
+<a href=#2232 id=2232 data-nosnippet>2232</a>            RowSelector::select(<span class="number">98</span>),
+<a href=#2233 id=2233 data-nosnippet>2233</a>            RowSelector::skip(<span class="number">1</span>),
+<a href=#2234 id=2234 data-nosnippet>2234</a>            RowSelector::select(<span class="number">99</span>),
+<a href=#2235 id=2235 data-nosnippet>2235</a>            RowSelector::skip(<span class="number">3</span>),
+<a href=#2236 id=2236 data-nosnippet>2236</a>            RowSelector::select(<span class="number">796</span>),
+<a href=#2237 id=2237 data-nosnippet>2237</a>            RowSelector::skip(<span class="number">2</span>),
+<a href=#2238 id=2238 data-nosnippet>2238</a>            RowSelector::select(<span class="number">499</span>),
+<a href=#2239 id=2239 data-nosnippet>2239</a>        ]);
+<a href=#2240 id=2240 data-nosnippet>2240</a>
+<a href=#2241 id=2241 data-nosnippet>2241</a>        <span class="macro">assert_eq!</span>(result, expected);
+<a href=#2242 id=2242 data-nosnippet>2242</a>    }
+<a href=#2243 id=2243 data-nosnippet>2243</a>
+<a href=#2244 id=2244 data-nosnippet>2244</a>    <span class="kw">fn </span>build_test_row_group_meta(
+<a href=#2245 id=2245 data-nosnippet>2245</a>        schema_descr: SchemaDescPtr,
+<a href=#2246 id=2246 data-nosnippet>2246</a>        columns: Vec&lt;ColumnChunkMetaData&gt;,
+<a href=#2247 id=2247 data-nosnippet>2247</a>        num_rows: i64,
+<a href=#2248 id=2248 data-nosnippet>2248</a>        ordinal: i16,
+<a href=#2249 id=2249 data-nosnippet>2249</a>    ) -&gt; RowGroupMetaData {
+<a href=#2250 id=2250 data-nosnippet>2250</a>        RowGroupMetaData::builder(schema_descr.clone())
+<a href=#2251 id=2251 data-nosnippet>2251</a>            .set_num_rows(num_rows)
+<a href=#2252 id=2252 data-nosnippet>2252</a>            .set_total_byte_size(<span class="number">2000</span>)
+<a href=#2253 id=2253 data-nosnippet>2253</a>            .set_column_metadata(columns)
+<a href=#2254 id=2254 data-nosnippet>2254</a>            .set_ordinal(ordinal)
+<a href=#2255 id=2255 data-nosnippet>2255</a>            .build()
+<a href=#2256 id=2256 data-nosnippet>2256</a>            .unwrap()
+<a href=#2257 id=2257 data-nosnippet>2257</a>    }
+<a href=#2258 id=2258 data-nosnippet>2258</a>
+<a href=#2259 id=2259 data-nosnippet>2259</a>    <span class="kw">fn </span>get_test_schema_descr() -&gt; SchemaDescPtr {
+<a href=#2260 id=2260 data-nosnippet>2260</a>        <span class="kw">use </span>parquet::schema::types::Type <span class="kw">as </span>SchemaType;
+<a href=#2261 id=2261 data-nosnippet>2261</a>
+<a href=#2262 id=2262 data-nosnippet>2262</a>        <span class="kw">let </span>schema = SchemaType::group_type_builder(<span class="string">"schema"</span>)
+<a href=#2263 id=2263 data-nosnippet>2263</a>            .with_fields(<span class="macro">vec!</span>[
+<a href=#2264 id=2264 data-nosnippet>2264</a>                Arc::new(
+<a href=#2265 id=2265 data-nosnippet>2265</a>                    SchemaType::primitive_type_builder(<span class="string">"a"</span>, parquet::basic::Type::INT32)
+<a href=#2266 id=2266 data-nosnippet>2266</a>                        .build()
+<a href=#2267 id=2267 data-nosnippet>2267</a>                        .unwrap(),
+<a href=#2268 id=2268 data-nosnippet>2268</a>                ),
+<a href=#2269 id=2269 data-nosnippet>2269</a>                Arc::new(
+<a href=#2270 id=2270 data-nosnippet>2270</a>                    SchemaType::primitive_type_builder(<span class="string">"b"</span>, parquet::basic::Type::INT32)
+<a href=#2271 id=2271 data-nosnippet>2271</a>                        .build()
+<a href=#2272 id=2272 data-nosnippet>2272</a>                        .unwrap(),
+<a href=#2273 id=2273 data-nosnippet>2273</a>                ),
+<a href=#2274 id=2274 data-nosnippet>2274</a>            ])
+<a href=#2275 id=2275 data-nosnippet>2275</a>            .build()
+<a href=#2276 id=2276 data-nosnippet>2276</a>            .unwrap();
+<a href=#2277 id=2277 data-nosnippet>2277</a>
+<a href=#2278 id=2278 data-nosnippet>2278</a>        Arc::new(SchemaDescriptor::new(Arc::new(schema)))
+<a href=#2279 id=2279 data-nosnippet>2279</a>    }
+<a href=#2280 id=2280 data-nosnippet>2280</a>
+<a href=#2281 id=2281 data-nosnippet>2281</a>    <span class="doccomment">/// Verifies that file splits respect byte ranges and only read specific row groups.
+<a href=#2282 id=2282 data-nosnippet>2282</a>    </span><span class="attr">#[tokio::test]
+<a href=#2283 id=2283 data-nosnippet>2283</a>    </span><span class="kw">async fn </span>test_file_splits_respect_byte_ranges() {
+<a href=#2284 id=2284 data-nosnippet>2284</a>        <span class="kw">use </span>arrow_array::Int32Array;
+<a href=#2285 id=2285 data-nosnippet>2285</a>        <span class="kw">use </span>parquet::file::reader::{FileReader, SerializedFileReader};
+<a href=#2286 id=2286 data-nosnippet>2286</a>
+<a href=#2287 id=2287 data-nosnippet>2287</a>        <span class="kw">let </span>schema = Arc::new(
+<a href=#2288 id=2288 data-nosnippet>2288</a>            Schema::builder()
+<a href=#2289 id=2289 data-nosnippet>2289</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#2290 id=2290 data-nosnippet>2290</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#2291 id=2291 data-nosnippet>2291</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#2292 id=2292 data-nosnippet>2292</a>                ])
+<a href=#2293 id=2293 data-nosnippet>2293</a>                .build()
+<a href=#2294 id=2294 data-nosnippet>2294</a>                .unwrap(),
+<a href=#2295 id=2295 data-nosnippet>2295</a>        );
+<a href=#2296 id=2296 data-nosnippet>2296</a>
+<a href=#2297 id=2297 data-nosnippet>2297</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#2298 id=2298 data-nosnippet>2298</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#2299 id=2299 data-nosnippet>2299</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2300 id=2300 data-nosnippet>2300</a>                <span class="string">"1"</span>.to_string(),
+<a href=#2301 id=2301 data-nosnippet>2301</a>            )])),
+<a href=#2302 id=2302 data-nosnippet>2302</a>        ]));
+<a href=#2303 id=2303 data-nosnippet>2303</a>
+<a href=#2304 id=2304 data-nosnippet>2304</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#2305 id=2305 data-nosnippet>2305</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#2306 id=2306 data-nosnippet>2306</a>        <span class="kw">let </span>file_path = <span class="macro">format!</span>(<span class="string">"{}/multi_row_group.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#2307 id=2307 data-nosnippet>2307</a>
+<a href=#2308 id=2308 data-nosnippet>2308</a>        <span class="comment">// Force each batch into its own row group for testing byte range filtering.
+<a href=#2309 id=2309 data-nosnippet>2309</a>        </span><span class="kw">let </span>batch1 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(Int32Array::from(
+<a href=#2310 id=2310 data-nosnippet>2310</a>            (<span class="number">0</span>..<span class="number">100</span>).collect::&lt;Vec&lt;i32&gt;&gt;(),
+<a href=#2311 id=2311 data-nosnippet>2311</a>        ))])
+<a href=#2312 id=2312 data-nosnippet>2312</a>        .unwrap();
+<a href=#2313 id=2313 data-nosnippet>2313</a>        <span class="kw">let </span>batch2 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(Int32Array::from(
+<a href=#2314 id=2314 data-nosnippet>2314</a>            (<span class="number">100</span>..<span class="number">200</span>).collect::&lt;Vec&lt;i32&gt;&gt;(),
+<a href=#2315 id=2315 data-nosnippet>2315</a>        ))])
+<a href=#2316 id=2316 data-nosnippet>2316</a>        .unwrap();
+<a href=#2317 id=2317 data-nosnippet>2317</a>        <span class="kw">let </span>batch3 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(Int32Array::from(
+<a href=#2318 id=2318 data-nosnippet>2318</a>            (<span class="number">200</span>..<span class="number">300</span>).collect::&lt;Vec&lt;i32&gt;&gt;(),
+<a href=#2319 id=2319 data-nosnippet>2319</a>        ))])
+<a href=#2320 id=2320 data-nosnippet>2320</a>        .unwrap();
 <a href=#2321 id=2321 data-nosnippet>2321</a>
-<a href=#2322 id=2322 data-nosnippet>2322</a>        <span class="comment">// Verify the actual data values are correct (not just the row count)
-<a href=#2323 id=2323 data-nosnippet>2323</a>        </span><span class="kw">if </span>total_rows_task1 &gt; <span class="number">0 </span>{
-<a href=#2324 id=2324 data-nosnippet>2324</a>            <span class="kw">let </span>first_batch = <span class="kw-2">&amp;</span>result1[<span class="number">0</span>];
-<a href=#2325 id=2325 data-nosnippet>2325</a>            <span class="kw">let </span>id_col = first_batch
-<a href=#2326 id=2326 data-nosnippet>2326</a>                .column(<span class="number">0</span>)
-<a href=#2327 id=2327 data-nosnippet>2327</a>                .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#2328 id=2328 data-nosnippet>2328</a>            <span class="kw">let </span>first_val = id_col.value(<span class="number">0</span>);
-<a href=#2329 id=2329 data-nosnippet>2329</a>            <span class="kw">let </span>last_val = id_col.value(id_col.len() - <span class="number">1</span>);
-<a href=#2330 id=2330 data-nosnippet>2330</a>            <span class="macro">println!</span>(<span class="string">"Task 1 data range: {first_val} to {last_val}"</span>);
-<a href=#2331 id=2331 data-nosnippet>2331</a>
-<a href=#2332 id=2332 data-nosnippet>2332</a>            <span class="macro">assert_eq!</span>(first_val, <span class="number">0</span>, <span class="string">"Task 1 should start with id=0"</span>);
-<a href=#2333 id=2333 data-nosnippet>2333</a>            <span class="macro">assert_eq!</span>(last_val, <span class="number">99</span>, <span class="string">"Task 1 should end with id=99"</span>);
-<a href=#2334 id=2334 data-nosnippet>2334</a>        }
-<a href=#2335 id=2335 data-nosnippet>2335</a>
-<a href=#2336 id=2336 data-nosnippet>2336</a>        <span class="kw">if </span>total_rows_task2 &gt; <span class="number">0 </span>{
-<a href=#2337 id=2337 data-nosnippet>2337</a>            <span class="kw">let </span>first_batch = <span class="kw-2">&amp;</span>result2[<span class="number">0</span>];
-<a href=#2338 id=2338 data-nosnippet>2338</a>            <span class="kw">let </span>id_col = first_batch
-<a href=#2339 id=2339 data-nosnippet>2339</a>                .column(<span class="number">0</span>)
-<a href=#2340 id=2340 data-nosnippet>2340</a>                .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#2341 id=2341 data-nosnippet>2341</a>            <span class="kw">let </span>first_val = id_col.value(<span class="number">0</span>);
-<a href=#2342 id=2342 data-nosnippet>2342</a>            <span class="macro">println!</span>(<span class="string">"Task 2 first value: {first_val}"</span>);
-<a href=#2343 id=2343 data-nosnippet>2343</a>
-<a href=#2344 id=2344 data-nosnippet>2344</a>            <span class="macro">assert_eq!</span>(first_val, <span class="number">100</span>, <span class="string">"Task 2 should start with id=100, not id=0"</span>);
-<a href=#2345 id=2345 data-nosnippet>2345</a>        }
-<a href=#2346 id=2346 data-nosnippet>2346</a>    }
-<a href=#2347 id=2347 data-nosnippet>2347</a>
-<a href=#2348 id=2348 data-nosnippet>2348</a>    <span class="doccomment">/// Test schema evolution: reading old Parquet file (with only column 'a')
-<a href=#2349 id=2349 data-nosnippet>2349</a>    /// using a newer table schema (with columns 'a' and 'b').
-<a href=#2350 id=2350 data-nosnippet>2350</a>    /// This tests that:
-<a href=#2351 id=2351 data-nosnippet>2351</a>    /// 1. get_arrow_projection_mask allows missing columns
-<a href=#2352 id=2352 data-nosnippet>2352</a>    /// 2. RecordBatchTransformer adds missing column 'b' with NULL values
-<a href=#2353 id=2353 data-nosnippet>2353</a>    </span><span class="attr">#[tokio::test]
-<a href=#2354 id=2354 data-nosnippet>2354</a>    </span><span class="kw">async fn </span>test_schema_evolution_add_column() {
-<a href=#2355 id=2355 data-nosnippet>2355</a>        <span class="kw">use </span>arrow_array::{Array, Int32Array};
-<a href=#2356 id=2356 data-nosnippet>2356</a>
-<a href=#2357 id=2357 data-nosnippet>2357</a>        <span class="comment">// New table schema: columns 'a' and 'b' (b was added later, file only has 'a')
-<a href=#2358 id=2358 data-nosnippet>2358</a>        </span><span class="kw">let </span>new_schema = Arc::new(
-<a href=#2359 id=2359 data-nosnippet>2359</a>            Schema::builder()
-<a href=#2360 id=2360 data-nosnippet>2360</a>                .with_schema_id(<span class="number">2</span>)
-<a href=#2361 id=2361 data-nosnippet>2361</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#2362 id=2362 data-nosnippet>2362</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"a"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#2363 id=2363 data-nosnippet>2363</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"b"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#2364 id=2364 data-nosnippet>2364</a>                ])
-<a href=#2365 id=2365 data-nosnippet>2365</a>                .build()
-<a href=#2366 id=2366 data-nosnippet>2366</a>                .unwrap(),
-<a href=#2367 id=2367 data-nosnippet>2367</a>        );
-<a href=#2368 id=2368 data-nosnippet>2368</a>
-<a href=#2369 id=2369 data-nosnippet>2369</a>        <span class="comment">// Create Arrow schema for old Parquet file (only has column 'a')
-<a href=#2370 id=2370 data-nosnippet>2370</a>        </span><span class="kw">let </span>arrow_schema_old = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#2371 id=2371 data-nosnippet>2371</a>            Field::new(<span class="string">"a"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2372 id=2372 data-nosnippet>2372</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2373 id=2373 data-nosnippet>2373</a>                <span class="string">"1"</span>.to_string(),
-<a href=#2374 id=2374 data-nosnippet>2374</a>            )])),
-<a href=#2375 id=2375 data-nosnippet>2375</a>        ]));
-<a href=#2376 id=2376 data-nosnippet>2376</a>
-<a href=#2377 id=2377 data-nosnippet>2377</a>        <span class="comment">// Write old Parquet file with only column 'a'
-<a href=#2378 id=2378 data-nosnippet>2378</a>        </span><span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#2379 id=2379 data-nosnippet>2379</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#2380 id=2380 data-nosnippet>2380</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#2381 id=2381 data-nosnippet>2381</a>
-<a href=#2382 id=2382 data-nosnippet>2382</a>        <span class="kw">let </span>data_a = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#2383 id=2383 data-nosnippet>2383</a>        <span class="kw">let </span>to_write = RecordBatch::try_new(arrow_schema_old.clone(), <span class="macro">vec!</span>[data_a]).unwrap();
-<a href=#2384 id=2384 data-nosnippet>2384</a>
-<a href=#2385 id=2385 data-nosnippet>2385</a>        <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#2386 id=2386 data-nosnippet>2386</a>            .set_compression(Compression::SNAPPY)
-<a href=#2387 id=2387 data-nosnippet>2387</a>            .build();
-<a href=#2388 id=2388 data-nosnippet>2388</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/old_file.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#2389 id=2389 data-nosnippet>2389</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#2390 id=2390 data-nosnippet>2390</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
-<a href=#2391 id=2391 data-nosnippet>2391</a>        writer.close().unwrap();
-<a href=#2392 id=2392 data-nosnippet>2392</a>
-<a href=#2393 id=2393 data-nosnippet>2393</a>        <span class="comment">// Read the old Parquet file using the NEW schema (with column 'b')
-<a href=#2394 id=2394 data-nosnippet>2394</a>        </span><span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#2395 id=2395 data-nosnippet>2395</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#2396 id=2396 data-nosnippet>2396</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#2397 id=2397 data-nosnippet>2397</a>                start: <span class="number">0</span>,
-<a href=#2398 id=2398 data-nosnippet>2398</a>                length: <span class="number">0</span>,
-<a href=#2399 id=2399 data-nosnippet>2399</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#2400 id=2400 data-nosnippet>2400</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{table_location}/old_file.parquet"</span>),
-<a href=#2401 id=2401 data-nosnippet>2401</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#2402 id=2402 data-nosnippet>2402</a>                schema: new_schema.clone(),
-<a href=#2403 id=2403 data-nosnippet>2403</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>], <span class="comment">// Request both columns 'a' and 'b'
-<a href=#2404 id=2404 data-nosnippet>2404</a>                </span>predicate: <span class="prelude-val">None</span>,
-<a href=#2405 id=2405 data-nosnippet>2405</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#2406 id=2406 data-nosnippet>2406</a>            })]
-<a href=#2407 id=2407 data-nosnippet>2407</a>            .into_iter(),
-<a href=#2408 id=2408 data-nosnippet>2408</a>        )) <span class="kw">as </span>FileScanTaskStream;
-<a href=#2409 id=2409 data-nosnippet>2409</a>
-<a href=#2410 id=2410 data-nosnippet>2410</a>        <span class="kw">let </span>result = reader
-<a href=#2411 id=2411 data-nosnippet>2411</a>            .read(tasks)
-<a href=#2412 id=2412 data-nosnippet>2412</a>            .unwrap()
-<a href=#2413 id=2413 data-nosnippet>2413</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#2414 id=2414 data-nosnippet>2414</a>            .<span class="kw">await
-<a href=#2415 id=2415 data-nosnippet>2415</a>            </span>.unwrap();
-<a href=#2416 id=2416 data-nosnippet>2416</a>
-<a href=#2417 id=2417 data-nosnippet>2417</a>        <span class="comment">// Verify we got the correct data
-<a href=#2418 id=2418 data-nosnippet>2418</a>        </span><span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
-<a href=#2419 id=2419 data-nosnippet>2419</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
-<a href=#2420 id=2420 data-nosnippet>2420</a>
-<a href=#2421 id=2421 data-nosnippet>2421</a>        <span class="comment">// Should have 2 columns now
-<a href=#2422 id=2422 data-nosnippet>2422</a>        </span><span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
-<a href=#2423 id=2423 data-nosnippet>2423</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">3</span>);
-<a href=#2424 id=2424 data-nosnippet>2424</a>
-<a href=#2425 id=2425 data-nosnippet>2425</a>        <span class="comment">// Column 'a' should have the original data
-<a href=#2426 id=2426 data-nosnippet>2426</a>        </span><span class="kw">let </span>col_a = batch
-<a href=#2427 id=2427 data-nosnippet>2427</a>            .column(<span class="number">0</span>)
-<a href=#2428 id=2428 data-nosnippet>2428</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#2429 id=2429 data-nosnippet>2429</a>        <span class="macro">assert_eq!</span>(col_a.values(), <span class="kw-2">&amp;</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>]);
+<a href=#2322 id=2322 data-nosnippet>2322</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#2323 id=2323 data-nosnippet>2323</a>            .set_compression(Compression::SNAPPY)
+<a href=#2324 id=2324 data-nosnippet>2324</a>            .set_max_row_group_size(<span class="number">100</span>)
+<a href=#2325 id=2325 data-nosnippet>2325</a>            .build();
+<a href=#2326 id=2326 data-nosnippet>2326</a>
+<a href=#2327 id=2327 data-nosnippet>2327</a>        <span class="kw">let </span>file = File::create(<span class="kw-2">&amp;</span>file_path).unwrap();
+<a href=#2328 id=2328 data-nosnippet>2328</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#2329 id=2329 data-nosnippet>2329</a>        writer.write(<span class="kw-2">&amp;</span>batch1).expect(<span class="string">"Writing batch 1"</span>);
+<a href=#2330 id=2330 data-nosnippet>2330</a>        writer.write(<span class="kw-2">&amp;</span>batch2).expect(<span class="string">"Writing batch 2"</span>);
+<a href=#2331 id=2331 data-nosnippet>2331</a>        writer.write(<span class="kw-2">&amp;</span>batch3).expect(<span class="string">"Writing batch 3"</span>);
+<a href=#2332 id=2332 data-nosnippet>2332</a>        writer.close().unwrap();
+<a href=#2333 id=2333 data-nosnippet>2333</a>
+<a href=#2334 id=2334 data-nosnippet>2334</a>        <span class="comment">// Read the file metadata to get row group byte positions
+<a href=#2335 id=2335 data-nosnippet>2335</a>        </span><span class="kw">let </span>file = File::open(<span class="kw-2">&amp;</span>file_path).unwrap();
+<a href=#2336 id=2336 data-nosnippet>2336</a>        <span class="kw">let </span>reader = SerializedFileReader::new(file).unwrap();
+<a href=#2337 id=2337 data-nosnippet>2337</a>        <span class="kw">let </span>metadata = reader.metadata();
+<a href=#2338 id=2338 data-nosnippet>2338</a>
+<a href=#2339 id=2339 data-nosnippet>2339</a>        <span class="macro">println!</span>(<span class="string">"File has {} row groups"</span>, metadata.num_row_groups());
+<a href=#2340 id=2340 data-nosnippet>2340</a>        <span class="macro">assert_eq!</span>(metadata.num_row_groups(), <span class="number">3</span>, <span class="string">"Expected 3 row groups"</span>);
+<a href=#2341 id=2341 data-nosnippet>2341</a>
+<a href=#2342 id=2342 data-nosnippet>2342</a>        <span class="comment">// Get byte positions for each row group
+<a href=#2343 id=2343 data-nosnippet>2343</a>        </span><span class="kw">let </span>row_group_0 = metadata.row_group(<span class="number">0</span>);
+<a href=#2344 id=2344 data-nosnippet>2344</a>        <span class="kw">let </span>row_group_1 = metadata.row_group(<span class="number">1</span>);
+<a href=#2345 id=2345 data-nosnippet>2345</a>        <span class="kw">let </span>row_group_2 = metadata.row_group(<span class="number">2</span>);
+<a href=#2346 id=2346 data-nosnippet>2346</a>
+<a href=#2347 id=2347 data-nosnippet>2347</a>        <span class="kw">let </span>rg0_start = <span class="number">4u64</span>; <span class="comment">// Parquet files start with 4-byte magic "PAR1"
+<a href=#2348 id=2348 data-nosnippet>2348</a>        </span><span class="kw">let </span>rg1_start = rg0_start + row_group_0.compressed_size() <span class="kw">as </span>u64;
+<a href=#2349 id=2349 data-nosnippet>2349</a>        <span class="kw">let </span>rg2_start = rg1_start + row_group_1.compressed_size() <span class="kw">as </span>u64;
+<a href=#2350 id=2350 data-nosnippet>2350</a>        <span class="kw">let </span>file_end = rg2_start + row_group_2.compressed_size() <span class="kw">as </span>u64;
+<a href=#2351 id=2351 data-nosnippet>2351</a>
+<a href=#2352 id=2352 data-nosnippet>2352</a>        <span class="macro">println!</span>(
+<a href=#2353 id=2353 data-nosnippet>2353</a>            <span class="string">"Row group 0: {} rows, starts at byte {}, {} bytes compressed"</span>,
+<a href=#2354 id=2354 data-nosnippet>2354</a>            row_group_0.num_rows(),
+<a href=#2355 id=2355 data-nosnippet>2355</a>            rg0_start,
+<a href=#2356 id=2356 data-nosnippet>2356</a>            row_group_0.compressed_size()
+<a href=#2357 id=2357 data-nosnippet>2357</a>        );
+<a href=#2358 id=2358 data-nosnippet>2358</a>        <span class="macro">println!</span>(
+<a href=#2359 id=2359 data-nosnippet>2359</a>            <span class="string">"Row group 1: {} rows, starts at byte {}, {} bytes compressed"</span>,
+<a href=#2360 id=2360 data-nosnippet>2360</a>            row_group_1.num_rows(),
+<a href=#2361 id=2361 data-nosnippet>2361</a>            rg1_start,
+<a href=#2362 id=2362 data-nosnippet>2362</a>            row_group_1.compressed_size()
+<a href=#2363 id=2363 data-nosnippet>2363</a>        );
+<a href=#2364 id=2364 data-nosnippet>2364</a>        <span class="macro">println!</span>(
+<a href=#2365 id=2365 data-nosnippet>2365</a>            <span class="string">"Row group 2: {} rows, starts at byte {}, {} bytes compressed"</span>,
+<a href=#2366 id=2366 data-nosnippet>2366</a>            row_group_2.num_rows(),
+<a href=#2367 id=2367 data-nosnippet>2367</a>            rg2_start,
+<a href=#2368 id=2368 data-nosnippet>2368</a>            row_group_2.compressed_size()
+<a href=#2369 id=2369 data-nosnippet>2369</a>        );
+<a href=#2370 id=2370 data-nosnippet>2370</a>
+<a href=#2371 id=2371 data-nosnippet>2371</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#2372 id=2372 data-nosnippet>2372</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#2373 id=2373 data-nosnippet>2373</a>
+<a href=#2374 id=2374 data-nosnippet>2374</a>        <span class="comment">// Task 1: read only the first row group
+<a href=#2375 id=2375 data-nosnippet>2375</a>        </span><span class="kw">let </span>task1 = FileScanTask {
+<a href=#2376 id=2376 data-nosnippet>2376</a>            start: rg0_start,
+<a href=#2377 id=2377 data-nosnippet>2377</a>            length: row_group_0.compressed_size() <span class="kw">as </span>u64,
+<a href=#2378 id=2378 data-nosnippet>2378</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">100</span>),
+<a href=#2379 id=2379 data-nosnippet>2379</a>            data_file_path: file_path.clone(),
+<a href=#2380 id=2380 data-nosnippet>2380</a>            data_file_format: DataFileFormat::Parquet,
+<a href=#2381 id=2381 data-nosnippet>2381</a>            schema: schema.clone(),
+<a href=#2382 id=2382 data-nosnippet>2382</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
+<a href=#2383 id=2383 data-nosnippet>2383</a>            predicate: <span class="prelude-val">None</span>,
+<a href=#2384 id=2384 data-nosnippet>2384</a>            deletes: <span class="macro">vec!</span>[],
+<a href=#2385 id=2385 data-nosnippet>2385</a>            partition: <span class="prelude-val">None</span>,
+<a href=#2386 id=2386 data-nosnippet>2386</a>            partition_spec: <span class="prelude-val">None</span>,
+<a href=#2387 id=2387 data-nosnippet>2387</a>            name_mapping: <span class="prelude-val">None</span>,
+<a href=#2388 id=2388 data-nosnippet>2388</a>        };
+<a href=#2389 id=2389 data-nosnippet>2389</a>
+<a href=#2390 id=2390 data-nosnippet>2390</a>        <span class="comment">// Task 2: read the second and third row groups
+<a href=#2391 id=2391 data-nosnippet>2391</a>        </span><span class="kw">let </span>task2 = FileScanTask {
+<a href=#2392 id=2392 data-nosnippet>2392</a>            start: rg1_start,
+<a href=#2393 id=2393 data-nosnippet>2393</a>            length: file_end - rg1_start,
+<a href=#2394 id=2394 data-nosnippet>2394</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">200</span>),
+<a href=#2395 id=2395 data-nosnippet>2395</a>            data_file_path: file_path.clone(),
+<a href=#2396 id=2396 data-nosnippet>2396</a>            data_file_format: DataFileFormat::Parquet,
+<a href=#2397 id=2397 data-nosnippet>2397</a>            schema: schema.clone(),
+<a href=#2398 id=2398 data-nosnippet>2398</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
+<a href=#2399 id=2399 data-nosnippet>2399</a>            predicate: <span class="prelude-val">None</span>,
+<a href=#2400 id=2400 data-nosnippet>2400</a>            deletes: <span class="macro">vec!</span>[],
+<a href=#2401 id=2401 data-nosnippet>2401</a>            partition: <span class="prelude-val">None</span>,
+<a href=#2402 id=2402 data-nosnippet>2402</a>            partition_spec: <span class="prelude-val">None</span>,
+<a href=#2403 id=2403 data-nosnippet>2403</a>            name_mapping: <span class="prelude-val">None</span>,
+<a href=#2404 id=2404 data-nosnippet>2404</a>        };
+<a href=#2405 id=2405 data-nosnippet>2405</a>
+<a href=#2406 id=2406 data-nosnippet>2406</a>        <span class="kw">let </span>tasks1 = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task1)])) <span class="kw">as </span>FileScanTaskStream;
+<a href=#2407 id=2407 data-nosnippet>2407</a>        <span class="kw">let </span>result1 = reader
+<a href=#2408 id=2408 data-nosnippet>2408</a>            .clone()
+<a href=#2409 id=2409 data-nosnippet>2409</a>            .read(tasks1)
+<a href=#2410 id=2410 data-nosnippet>2410</a>            .unwrap()
+<a href=#2411 id=2411 data-nosnippet>2411</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#2412 id=2412 data-nosnippet>2412</a>            .<span class="kw">await
+<a href=#2413 id=2413 data-nosnippet>2413</a>            </span>.unwrap();
+<a href=#2414 id=2414 data-nosnippet>2414</a>
+<a href=#2415 id=2415 data-nosnippet>2415</a>        <span class="kw">let </span>total_rows_task1: usize = result1.iter().map(|b| b.num_rows()).sum();
+<a href=#2416 id=2416 data-nosnippet>2416</a>        <span class="macro">println!</span>(
+<a href=#2417 id=2417 data-nosnippet>2417</a>            <span class="string">"Task 1 (bytes {}-{}) returned {} rows"</span>,
+<a href=#2418 id=2418 data-nosnippet>2418</a>            rg0_start,
+<a href=#2419 id=2419 data-nosnippet>2419</a>            rg0_start + row_group_0.compressed_size() <span class="kw">as </span>u64,
+<a href=#2420 id=2420 data-nosnippet>2420</a>            total_rows_task1
+<a href=#2421 id=2421 data-nosnippet>2421</a>        );
+<a href=#2422 id=2422 data-nosnippet>2422</a>
+<a href=#2423 id=2423 data-nosnippet>2423</a>        <span class="kw">let </span>tasks2 = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task2)])) <span class="kw">as </span>FileScanTaskStream;
+<a href=#2424 id=2424 data-nosnippet>2424</a>        <span class="kw">let </span>result2 = reader
+<a href=#2425 id=2425 data-nosnippet>2425</a>            .read(tasks2)
+<a href=#2426 id=2426 data-nosnippet>2426</a>            .unwrap()
+<a href=#2427 id=2427 data-nosnippet>2427</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#2428 id=2428 data-nosnippet>2428</a>            .<span class="kw">await
+<a href=#2429 id=2429 data-nosnippet>2429</a>            </span>.unwrap();
 <a href=#2430 id=2430 data-nosnippet>2430</a>
-<a href=#2431 id=2431 data-nosnippet>2431</a>        <span class="comment">// Column 'b' should be all NULLs (it didn't exist in the old file)
-<a href=#2432 id=2432 data-nosnippet>2432</a>        </span><span class="kw">let </span>col_b = batch
-<a href=#2433 id=2433 data-nosnippet>2433</a>            .column(<span class="number">1</span>)
-<a href=#2434 id=2434 data-nosnippet>2434</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#2435 id=2435 data-nosnippet>2435</a>        <span class="macro">assert_eq!</span>(col_b.null_count(), <span class="number">3</span>);
-<a href=#2436 id=2436 data-nosnippet>2436</a>        <span class="macro">assert!</span>(col_b.is_null(<span class="number">0</span>));
-<a href=#2437 id=2437 data-nosnippet>2437</a>        <span class="macro">assert!</span>(col_b.is_null(<span class="number">1</span>));
-<a href=#2438 id=2438 data-nosnippet>2438</a>        <span class="macro">assert!</span>(col_b.is_null(<span class="number">2</span>));
-<a href=#2439 id=2439 data-nosnippet>2439</a>    }
-<a href=#2440 id=2440 data-nosnippet>2440</a>
-<a href=#2441 id=2441 data-nosnippet>2441</a>    <span class="doccomment">/// Test for bug where position deletes in later row groups are not applied correctly.
-<a href=#2442 id=2442 data-nosnippet>2442</a>    ///
-<a href=#2443 id=2443 data-nosnippet>2443</a>    /// When a file has multiple row groups and a position delete targets a row in a later
-<a href=#2444 id=2444 data-nosnippet>2444</a>    /// row group, the `build_deletes_row_selection` function had a bug where it would
-<a href=#2445 id=2445 data-nosnippet>2445</a>    /// fail to increment `current_row_group_base_idx` when skipping row groups.
-<a href=#2446 id=2446 data-nosnippet>2446</a>    ///
-<a href=#2447 id=2447 data-nosnippet>2447</a>    /// This test creates:
-<a href=#2448 id=2448 data-nosnippet>2448</a>    /// - A data file with 200 rows split into 2 row groups (0-99, 100-199)
-<a href=#2449 id=2449 data-nosnippet>2449</a>    /// - A position delete file that deletes row 199 (last row in second row group)
-<a href=#2450 id=2450 data-nosnippet>2450</a>    ///
-<a href=#2451 id=2451 data-nosnippet>2451</a>    /// Expected behavior: Should return 199 rows (with id=200 deleted)
-<a href=#2452 id=2452 data-nosnippet>2452</a>    /// Bug behavior: Returns 200 rows (delete is not applied)
-<a href=#2453 id=2453 data-nosnippet>2453</a>    ///
-<a href=#2454 id=2454 data-nosnippet>2454</a>    /// This bug was discovered while running Apache Spark + Apache Iceberg integration tests
-<a href=#2455 id=2455 data-nosnippet>2455</a>    /// through DataFusion Comet. The following Iceberg Java tests failed due to this bug:
-<a href=#2456 id=2456 data-nosnippet>2456</a>    /// - `org.apache.iceberg.spark.extensions.TestMergeOnReadDelete::testDeleteWithMultipleRowGroupsParquet`
-<a href=#2457 id=2457 data-nosnippet>2457</a>    /// - `org.apache.iceberg.spark.extensions.TestMergeOnReadUpdate::testUpdateWithMultipleRowGroupsParquet`
-<a href=#2458 id=2458 data-nosnippet>2458</a>    </span><span class="attr">#[tokio::test]
-<a href=#2459 id=2459 data-nosnippet>2459</a>    </span><span class="kw">async fn </span>test_position_delete_across_multiple_row_groups() {
-<a href=#2460 id=2460 data-nosnippet>2460</a>        <span class="kw">use </span>arrow_array::{Int32Array, Int64Array};
-<a href=#2461 id=2461 data-nosnippet>2461</a>        <span class="kw">use </span>parquet::file::reader::{FileReader, SerializedFileReader};
-<a href=#2462 id=2462 data-nosnippet>2462</a>
-<a href=#2463 id=2463 data-nosnippet>2463</a>        <span class="comment">// Field IDs for positional delete schema
-<a href=#2464 id=2464 data-nosnippet>2464</a>        </span><span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_FILE_PATH: u64 = <span class="number">2147483546</span>;
-<a href=#2465 id=2465 data-nosnippet>2465</a>        <span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_POS: u64 = <span class="number">2147483545</span>;
-<a href=#2466 id=2466 data-nosnippet>2466</a>
-<a href=#2467 id=2467 data-nosnippet>2467</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#2468 id=2468 data-nosnippet>2468</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#2431 id=2431 data-nosnippet>2431</a>        <span class="kw">let </span>total_rows_task2: usize = result2.iter().map(|b| b.num_rows()).sum();
+<a href=#2432 id=2432 data-nosnippet>2432</a>        <span class="macro">println!</span>(<span class="string">"Task 2 (bytes {rg1_start}-{file_end}) returned {total_rows_task2} rows"</span>);
+<a href=#2433 id=2433 data-nosnippet>2433</a>
+<a href=#2434 id=2434 data-nosnippet>2434</a>        <span class="macro">assert_eq!</span>(
+<a href=#2435 id=2435 data-nosnippet>2435</a>            total_rows_task1, <span class="number">100</span>,
+<a href=#2436 id=2436 data-nosnippet>2436</a>            <span class="string">"Task 1 should read only the first row group (100 rows), but got {total_rows_task1} rows"
+<a href=#2437 id=2437 data-nosnippet>2437</a>        </span>);
+<a href=#2438 id=2438 data-nosnippet>2438</a>
+<a href=#2439 id=2439 data-nosnippet>2439</a>        <span class="macro">assert_eq!</span>(
+<a href=#2440 id=2440 data-nosnippet>2440</a>            total_rows_task2, <span class="number">200</span>,
+<a href=#2441 id=2441 data-nosnippet>2441</a>            <span class="string">"Task 2 should read only the second+third row groups (200 rows), but got {total_rows_task2} rows"
+<a href=#2442 id=2442 data-nosnippet>2442</a>        </span>);
+<a href=#2443 id=2443 data-nosnippet>2443</a>
+<a href=#2444 id=2444 data-nosnippet>2444</a>        <span class="comment">// Verify the actual data values are correct (not just the row count)
+<a href=#2445 id=2445 data-nosnippet>2445</a>        </span><span class="kw">if </span>total_rows_task1 &gt; <span class="number">0 </span>{
+<a href=#2446 id=2446 data-nosnippet>2446</a>            <span class="kw">let </span>first_batch = <span class="kw-2">&amp;</span>result1[<span class="number">0</span>];
+<a href=#2447 id=2447 data-nosnippet>2447</a>            <span class="kw">let </span>id_col = first_batch
+<a href=#2448 id=2448 data-nosnippet>2448</a>                .column(<span class="number">0</span>)
+<a href=#2449 id=2449 data-nosnippet>2449</a>                .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#2450 id=2450 data-nosnippet>2450</a>            <span class="kw">let </span>first_val = id_col.value(<span class="number">0</span>);
+<a href=#2451 id=2451 data-nosnippet>2451</a>            <span class="kw">let </span>last_val = id_col.value(id_col.len() - <span class="number">1</span>);
+<a href=#2452 id=2452 data-nosnippet>2452</a>            <span class="macro">println!</span>(<span class="string">"Task 1 data range: {first_val} to {last_val}"</span>);
+<a href=#2453 id=2453 data-nosnippet>2453</a>
+<a href=#2454 id=2454 data-nosnippet>2454</a>            <span class="macro">assert_eq!</span>(first_val, <span class="number">0</span>, <span class="string">"Task 1 should start with id=0"</span>);
+<a href=#2455 id=2455 data-nosnippet>2455</a>            <span class="macro">assert_eq!</span>(last_val, <span class="number">99</span>, <span class="string">"Task 1 should end with id=99"</span>);
+<a href=#2456 id=2456 data-nosnippet>2456</a>        }
+<a href=#2457 id=2457 data-nosnippet>2457</a>
+<a href=#2458 id=2458 data-nosnippet>2458</a>        <span class="kw">if </span>total_rows_task2 &gt; <span class="number">0 </span>{
+<a href=#2459 id=2459 data-nosnippet>2459</a>            <span class="kw">let </span>first_batch = <span class="kw-2">&amp;</span>result2[<span class="number">0</span>];
+<a href=#2460 id=2460 data-nosnippet>2460</a>            <span class="kw">let </span>id_col = first_batch
+<a href=#2461 id=2461 data-nosnippet>2461</a>                .column(<span class="number">0</span>)
+<a href=#2462 id=2462 data-nosnippet>2462</a>                .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#2463 id=2463 data-nosnippet>2463</a>            <span class="kw">let </span>first_val = id_col.value(<span class="number">0</span>);
+<a href=#2464 id=2464 data-nosnippet>2464</a>            <span class="macro">println!</span>(<span class="string">"Task 2 first value: {first_val}"</span>);
+<a href=#2465 id=2465 data-nosnippet>2465</a>
+<a href=#2466 id=2466 data-nosnippet>2466</a>            <span class="macro">assert_eq!</span>(first_val, <span class="number">100</span>, <span class="string">"Task 2 should start with id=100, not id=0"</span>);
+<a href=#2467 id=2467 data-nosnippet>2467</a>        }
+<a href=#2468 id=2468 data-nosnippet>2468</a>    }
 <a href=#2469 id=2469 data-nosnippet>2469</a>
-<a href=#2470 id=2470 data-nosnippet>2470</a>        <span class="comment">// Create table schema with a single 'id' column
-<a href=#2471 id=2471 data-nosnippet>2471</a>        </span><span class="kw">let </span>table_schema = Arc::new(
-<a href=#2472 id=2472 data-nosnippet>2472</a>            Schema::builder()
-<a href=#2473 id=2473 data-nosnippet>2473</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#2474 id=2474 data-nosnippet>2474</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#2475 id=2475 data-nosnippet>2475</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#2476 id=2476 data-nosnippet>2476</a>                ])
-<a href=#2477 id=2477 data-nosnippet>2477</a>                .build()
-<a href=#2478 id=2478 data-nosnippet>2478</a>                .unwrap(),
-<a href=#2479 id=2479 data-nosnippet>2479</a>        );
-<a href=#2480 id=2480 data-nosnippet>2480</a>
-<a href=#2481 id=2481 data-nosnippet>2481</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#2482 id=2482 data-nosnippet>2482</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2483 id=2483 data-nosnippet>2483</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2484 id=2484 data-nosnippet>2484</a>                <span class="string">"1"</span>.to_string(),
-<a href=#2485 id=2485 data-nosnippet>2485</a>            )])),
-<a href=#2486 id=2486 data-nosnippet>2486</a>        ]));
-<a href=#2487 id=2487 data-nosnippet>2487</a>
-<a href=#2488 id=2488 data-nosnippet>2488</a>        <span class="comment">// Step 1: Create data file with 200 rows in 2 row groups
-<a href=#2489 id=2489 data-nosnippet>2489</a>        // Row group 0: rows 0-99 (ids 1-100)
-<a href=#2490 id=2490 data-nosnippet>2490</a>        // Row group 1: rows 100-199 (ids 101-200)
-<a href=#2491 id=2491 data-nosnippet>2491</a>        </span><span class="kw">let </span>data_file_path = <span class="macro">format!</span>(<span class="string">"{}/data.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
-<a href=#2492 id=2492 data-nosnippet>2492</a>
-<a href=#2493 id=2493 data-nosnippet>2493</a>        <span class="kw">let </span>batch1 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
-<a href=#2494 id=2494 data-nosnippet>2494</a>            Int32Array::from_iter_values(<span class="number">1</span>..=<span class="number">100</span>),
-<a href=#2495 id=2495 data-nosnippet>2495</a>        )])
-<a href=#2496 id=2496 data-nosnippet>2496</a>        .unwrap();
-<a href=#2497 id=2497 data-nosnippet>2497</a>
-<a href=#2498 id=2498 data-nosnippet>2498</a>        <span class="kw">let </span>batch2 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
-<a href=#2499 id=2499 data-nosnippet>2499</a>            Int32Array::from_iter_values(<span class="number">101</span>..=<span class="number">200</span>),
-<a href=#2500 id=2500 data-nosnippet>2500</a>        )])
-<a href=#2501 id=2501 data-nosnippet>2501</a>        .unwrap();
-<a href=#2502 id=2502 data-nosnippet>2502</a>
-<a href=#2503 id=2503 data-nosnippet>2503</a>        <span class="comment">// Force each batch into its own row group
-<a href=#2504 id=2504 data-nosnippet>2504</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
-<a href=#2505 id=2505 data-nosnippet>2505</a>            .set_compression(Compression::SNAPPY)
-<a href=#2506 id=2506 data-nosnippet>2506</a>            .set_max_row_group_size(<span class="number">100</span>)
-<a href=#2507 id=2507 data-nosnippet>2507</a>            .build();
-<a href=#2508 id=2508 data-nosnippet>2508</a>
-<a href=#2509 id=2509 data-nosnippet>2509</a>        <span class="kw">let </span>file = File::create(<span class="kw-2">&amp;</span>data_file_path).unwrap();
-<a href=#2510 id=2510 data-nosnippet>2510</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#2511 id=2511 data-nosnippet>2511</a>        writer.write(<span class="kw-2">&amp;</span>batch1).expect(<span class="string">"Writing batch 1"</span>);
-<a href=#2512 id=2512 data-nosnippet>2512</a>        writer.write(<span class="kw-2">&amp;</span>batch2).expect(<span class="string">"Writing batch 2"</span>);
+<a href=#2470 id=2470 data-nosnippet>2470</a>    <span class="doccomment">/// Test schema evolution: reading old Parquet file (with only column 'a')
+<a href=#2471 id=2471 data-nosnippet>2471</a>    /// using a newer table schema (with columns 'a' and 'b').
+<a href=#2472 id=2472 data-nosnippet>2472</a>    /// This tests that:
+<a href=#2473 id=2473 data-nosnippet>2473</a>    /// 1. get_arrow_projection_mask allows missing columns
+<a href=#2474 id=2474 data-nosnippet>2474</a>    /// 2. RecordBatchTransformer adds missing column 'b' with NULL values
+<a href=#2475 id=2475 data-nosnippet>2475</a>    </span><span class="attr">#[tokio::test]
+<a href=#2476 id=2476 data-nosnippet>2476</a>    </span><span class="kw">async fn </span>test_schema_evolution_add_column() {
+<a href=#2477 id=2477 data-nosnippet>2477</a>        <span class="kw">use </span>arrow_array::{Array, Int32Array};
+<a href=#2478 id=2478 data-nosnippet>2478</a>
+<a href=#2479 id=2479 data-nosnippet>2479</a>        <span class="comment">// New table schema: columns 'a' and 'b' (b was added later, file only has 'a')
+<a href=#2480 id=2480 data-nosnippet>2480</a>        </span><span class="kw">let </span>new_schema = Arc::new(
+<a href=#2481 id=2481 data-nosnippet>2481</a>            Schema::builder()
+<a href=#2482 id=2482 data-nosnippet>2482</a>                .with_schema_id(<span class="number">2</span>)
+<a href=#2483 id=2483 data-nosnippet>2483</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#2484 id=2484 data-nosnippet>2484</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"a"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#2485 id=2485 data-nosnippet>2485</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"b"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#2486 id=2486 data-nosnippet>2486</a>                ])
+<a href=#2487 id=2487 data-nosnippet>2487</a>                .build()
+<a href=#2488 id=2488 data-nosnippet>2488</a>                .unwrap(),
+<a href=#2489 id=2489 data-nosnippet>2489</a>        );
+<a href=#2490 id=2490 data-nosnippet>2490</a>
+<a href=#2491 id=2491 data-nosnippet>2491</a>        <span class="comment">// Create Arrow schema for old Parquet file (only has column 'a')
+<a href=#2492 id=2492 data-nosnippet>2492</a>        </span><span class="kw">let </span>arrow_schema_old = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#2493 id=2493 data-nosnippet>2493</a>            Field::new(<span class="string">"a"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#2494 id=2494 data-nosnippet>2494</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2495 id=2495 data-nosnippet>2495</a>                <span class="string">"1"</span>.to_string(),
+<a href=#2496 id=2496 data-nosnippet>2496</a>            )])),
+<a href=#2497 id=2497 data-nosnippet>2497</a>        ]));
+<a href=#2498 id=2498 data-nosnippet>2498</a>
+<a href=#2499 id=2499 data-nosnippet>2499</a>        <span class="comment">// Write old Parquet file with only column 'a'
+<a href=#2500 id=2500 data-nosnippet>2500</a>        </span><span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#2501 id=2501 data-nosnippet>2501</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#2502 id=2502 data-nosnippet>2502</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#2503 id=2503 data-nosnippet>2503</a>
+<a href=#2504 id=2504 data-nosnippet>2504</a>        <span class="kw">let </span>data_a = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#2505 id=2505 data-nosnippet>2505</a>        <span class="kw">let </span>to_write = RecordBatch::try_new(arrow_schema_old.clone(), <span class="macro">vec!</span>[data_a]).unwrap();
+<a href=#2506 id=2506 data-nosnippet>2506</a>
+<a href=#2507 id=2507 data-nosnippet>2507</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#2508 id=2508 data-nosnippet>2508</a>            .set_compression(Compression::SNAPPY)
+<a href=#2509 id=2509 data-nosnippet>2509</a>            .build();
+<a href=#2510 id=2510 data-nosnippet>2510</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/old_file.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#2511 id=2511 data-nosnippet>2511</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#2512 id=2512 data-nosnippet>2512</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
 <a href=#2513 id=2513 data-nosnippet>2513</a>        writer.close().unwrap();
 <a href=#2514 id=2514 data-nosnippet>2514</a>
-<a href=#2515 id=2515 data-nosnippet>2515</a>        <span class="comment">// Verify we created 2 row groups
-<a href=#2516 id=2516 data-nosnippet>2516</a>        </span><span class="kw">let </span>verify_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
-<a href=#2517 id=2517 data-nosnippet>2517</a>        <span class="kw">let </span>verify_reader = SerializedFileReader::new(verify_file).unwrap();
-<a href=#2518 id=2518 data-nosnippet>2518</a>        <span class="macro">assert_eq!</span>(
-<a href=#2519 id=2519 data-nosnippet>2519</a>            verify_reader.metadata().num_row_groups(),
-<a href=#2520 id=2520 data-nosnippet>2520</a>            <span class="number">2</span>,
-<a href=#2521 id=2521 data-nosnippet>2521</a>            <span class="string">"Should have 2 row groups"
-<a href=#2522 id=2522 data-nosnippet>2522</a>        </span>);
-<a href=#2523 id=2523 data-nosnippet>2523</a>
-<a href=#2524 id=2524 data-nosnippet>2524</a>        <span class="comment">// Step 2: Create position delete file that deletes row 199 (id=200, last row in row group 1)
-<a href=#2525 id=2525 data-nosnippet>2525</a>        </span><span class="kw">let </span>delete_file_path = <span class="macro">format!</span>(<span class="string">"{}/deletes.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
-<a href=#2526 id=2526 data-nosnippet>2526</a>
-<a href=#2527 id=2527 data-nosnippet>2527</a>        <span class="kw">let </span>delete_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#2528 id=2528 data-nosnippet>2528</a>            Field::new(<span class="string">"file_path"</span>, DataType::Utf8, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2529 id=2529 data-nosnippet>2529</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2530 id=2530 data-nosnippet>2530</a>                FIELD_ID_POSITIONAL_DELETE_FILE_PATH.to_string(),
-<a href=#2531 id=2531 data-nosnippet>2531</a>            )])),
-<a href=#2532 id=2532 data-nosnippet>2532</a>            Field::new(<span class="string">"pos"</span>, DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2533 id=2533 data-nosnippet>2533</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2534 id=2534 data-nosnippet>2534</a>                FIELD_ID_POSITIONAL_DELETE_POS.to_string(),
-<a href=#2535 id=2535 data-nosnippet>2535</a>            )])),
-<a href=#2536 id=2536 data-nosnippet>2536</a>        ]));
-<a href=#2537 id=2537 data-nosnippet>2537</a>
-<a href=#2538 id=2538 data-nosnippet>2538</a>        <span class="comment">// Delete row at position 199 (0-indexed, so it's the last row: id=200)
-<a href=#2539 id=2539 data-nosnippet>2539</a>        </span><span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[
-<a href=#2540 id=2540 data-nosnippet>2540</a>            Arc::new(StringArray::from_iter_values(<span class="macro">vec!</span>[data_file_path.clone()])),
-<a href=#2541 id=2541 data-nosnippet>2541</a>            Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[<span class="number">199i64</span>])),
-<a href=#2542 id=2542 data-nosnippet>2542</a>        ])
-<a href=#2543 id=2543 data-nosnippet>2543</a>        .unwrap();
-<a href=#2544 id=2544 data-nosnippet>2544</a>
-<a href=#2545 id=2545 data-nosnippet>2545</a>        <span class="kw">let </span>delete_props = WriterProperties::builder()
-<a href=#2546 id=2546 data-nosnippet>2546</a>            .set_compression(Compression::SNAPPY)
-<a href=#2547 id=2547 data-nosnippet>2547</a>            .build();
-<a href=#2548 id=2548 data-nosnippet>2548</a>
-<a href=#2549 id=2549 data-nosnippet>2549</a>        <span class="kw">let </span>delete_file = File::create(<span class="kw-2">&amp;</span>delete_file_path).unwrap();
-<a href=#2550 id=2550 data-nosnippet>2550</a>        <span class="kw">let </span><span class="kw-2">mut </span>delete_writer =
-<a href=#2551 id=2551 data-nosnippet>2551</a>            ArrowWriter::try_new(delete_file, delete_schema, <span class="prelude-val">Some</span>(delete_props)).unwrap();
-<a href=#2552 id=2552 data-nosnippet>2552</a>        delete_writer.write(<span class="kw-2">&amp;</span>delete_batch).unwrap();
-<a href=#2553 id=2553 data-nosnippet>2553</a>        delete_writer.close().unwrap();
-<a href=#2554 id=2554 data-nosnippet>2554</a>
-<a href=#2555 id=2555 data-nosnippet>2555</a>        <span class="comment">// Step 3: Read the data file with the delete applied
-<a href=#2556 id=2556 data-nosnippet>2556</a>        </span><span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#2557 id=2557 data-nosnippet>2557</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#2558 id=2558 data-nosnippet>2558</a>
-<a href=#2559 id=2559 data-nosnippet>2559</a>        <span class="kw">let </span>task = FileScanTask {
-<a href=#2560 id=2560 data-nosnippet>2560</a>            start: <span class="number">0</span>,
-<a href=#2561 id=2561 data-nosnippet>2561</a>            length: <span class="number">0</span>,
-<a href=#2562 id=2562 data-nosnippet>2562</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">200</span>),
-<a href=#2563 id=2563 data-nosnippet>2563</a>            data_file_path: data_file_path.clone(),
-<a href=#2564 id=2564 data-nosnippet>2564</a>            data_file_format: DataFileFormat::Parquet,
-<a href=#2565 id=2565 data-nosnippet>2565</a>            schema: table_schema.clone(),
-<a href=#2566 id=2566 data-nosnippet>2566</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
-<a href=#2567 id=2567 data-nosnippet>2567</a>            predicate: <span class="prelude-val">None</span>,
-<a href=#2568 id=2568 data-nosnippet>2568</a>            deletes: <span class="macro">vec!</span>[FileScanTaskDeleteFile {
-<a href=#2569 id=2569 data-nosnippet>2569</a>                file_path: delete_file_path,
-<a href=#2570 id=2570 data-nosnippet>2570</a>                file_type: DataContentType::PositionDeletes,
-<a href=#2571 id=2571 data-nosnippet>2571</a>                partition_spec_id: <span class="number">0</span>,
-<a href=#2572 id=2572 data-nosnippet>2572</a>                equality_ids: <span class="prelude-val">None</span>,
-<a href=#2573 id=2573 data-nosnippet>2573</a>            }],
-<a href=#2574 id=2574 data-nosnippet>2574</a>        };
-<a href=#2575 id=2575 data-nosnippet>2575</a>
-<a href=#2576 id=2576 data-nosnippet>2576</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task)])) <span class="kw">as </span>FileScanTaskStream;
-<a href=#2577 id=2577 data-nosnippet>2577</a>        <span class="kw">let </span>result = reader
-<a href=#2578 id=2578 data-nosnippet>2578</a>            .read(tasks)
-<a href=#2579 id=2579 data-nosnippet>2579</a>            .unwrap()
-<a href=#2580 id=2580 data-nosnippet>2580</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#2581 id=2581 data-nosnippet>2581</a>            .<span class="kw">await
-<a href=#2582 id=2582 data-nosnippet>2582</a>            </span>.unwrap();
-<a href=#2583 id=2583 data-nosnippet>2583</a>
-<a href=#2584 id=2584 data-nosnippet>2584</a>        <span class="comment">// Step 4: Verify we got 199 rows (not 200)
-<a href=#2585 id=2585 data-nosnippet>2585</a>        </span><span class="kw">let </span>total_rows: usize = result.iter().map(|b| b.num_rows()).sum();
-<a href=#2586 id=2586 data-nosnippet>2586</a>
-<a href=#2587 id=2587 data-nosnippet>2587</a>        <span class="macro">println!</span>(<span class="string">"Total rows read: {}"</span>, total_rows);
-<a href=#2588 id=2588 data-nosnippet>2588</a>        <span class="macro">println!</span>(<span class="string">"Expected: 199 rows (deleted row 199 which had id=200)"</span>);
-<a href=#2589 id=2589 data-nosnippet>2589</a>
-<a href=#2590 id=2590 data-nosnippet>2590</a>        <span class="comment">// This assertion will FAIL before the fix and PASS after the fix
-<a href=#2591 id=2591 data-nosnippet>2591</a>        </span><span class="macro">assert_eq!</span>(
-<a href=#2592 id=2592 data-nosnippet>2592</a>            total_rows, <span class="number">199</span>,
-<a href=#2593 id=2593 data-nosnippet>2593</a>            <span class="string">"Expected 199 rows after deleting row 199, but got {} rows. \
-<a href=#2594 id=2594 data-nosnippet>2594</a>             The bug causes position deletes in later row groups to be ignored."</span>,
-<a href=#2595 id=2595 data-nosnippet>2595</a>            total_rows
-<a href=#2596 id=2596 data-nosnippet>2596</a>        );
-<a href=#2597 id=2597 data-nosnippet>2597</a>
-<a href=#2598 id=2598 data-nosnippet>2598</a>        <span class="comment">// Verify the deleted row (id=200) is not present
-<a href=#2599 id=2599 data-nosnippet>2599</a>        </span><span class="kw">let </span>all_ids: Vec&lt;i32&gt; = result
-<a href=#2600 id=2600 data-nosnippet>2600</a>            .iter()
-<a href=#2601 id=2601 data-nosnippet>2601</a>            .flat_map(|batch| {
-<a href=#2602 id=2602 data-nosnippet>2602</a>                batch
-<a href=#2603 id=2603 data-nosnippet>2603</a>                    .column(<span class="number">0</span>)
-<a href=#2604 id=2604 data-nosnippet>2604</a>                    .as_primitive::&lt;arrow_array::types::Int32Type&gt;()
-<a href=#2605 id=2605 data-nosnippet>2605</a>                    .values()
-<a href=#2606 id=2606 data-nosnippet>2606</a>                    .iter()
-<a href=#2607 id=2607 data-nosnippet>2607</a>                    .copied()
-<a href=#2608 id=2608 data-nosnippet>2608</a>            })
-<a href=#2609 id=2609 data-nosnippet>2609</a>            .collect();
-<a href=#2610 id=2610 data-nosnippet>2610</a>
-<a href=#2611 id=2611 data-nosnippet>2611</a>        <span class="macro">assert!</span>(
-<a href=#2612 id=2612 data-nosnippet>2612</a>            !all_ids.contains(<span class="kw-2">&amp;</span><span class="number">200</span>),
-<a href=#2613 id=2613 data-nosnippet>2613</a>            <span class="string">"Row with id=200 should be deleted but was found in results"
-<a href=#2614 id=2614 data-nosnippet>2614</a>        </span>);
-<a href=#2615 id=2615 data-nosnippet>2615</a>
-<a href=#2616 id=2616 data-nosnippet>2616</a>        <span class="comment">// Verify we have all other ids (1-199)
-<a href=#2617 id=2617 data-nosnippet>2617</a>        </span><span class="kw">let </span>expected_ids: Vec&lt;i32&gt; = (<span class="number">1</span>..=<span class="number">199</span>).collect();
-<a href=#2618 id=2618 data-nosnippet>2618</a>        <span class="macro">assert_eq!</span>(
-<a href=#2619 id=2619 data-nosnippet>2619</a>            all_ids, expected_ids,
-<a href=#2620 id=2620 data-nosnippet>2620</a>            <span class="string">"Should have ids 1-199 but got different values"
-<a href=#2621 id=2621 data-nosnippet>2621</a>        </span>);
-<a href=#2622 id=2622 data-nosnippet>2622</a>    }
-<a href=#2623 id=2623 data-nosnippet>2623</a>
-<a href=#2624 id=2624 data-nosnippet>2624</a>    <span class="doccomment">/// Test for bug where position deletes are lost when skipping unselected row groups.
-<a href=#2625 id=2625 data-nosnippet>2625</a>    ///
-<a href=#2626 id=2626 data-nosnippet>2626</a>    /// This is a variant of `test_position_delete_across_multiple_row_groups` that exercises
-<a href=#2627 id=2627 data-nosnippet>2627</a>    /// the row group selection code path (`selected_row_groups: Some([...])`).
-<a href=#2628 id=2628 data-nosnippet>2628</a>    ///
-<a href=#2629 id=2629 data-nosnippet>2629</a>    /// When a file has multiple row groups and only some are selected for reading,
-<a href=#2630 id=2630 data-nosnippet>2630</a>    /// the `build_deletes_row_selection` function must correctly skip over deletes in
-<a href=#2631 id=2631 data-nosnippet>2631</a>    /// unselected row groups WITHOUT consuming deletes that belong to selected row groups.
-<a href=#2632 id=2632 data-nosnippet>2632</a>    ///
-<a href=#2633 id=2633 data-nosnippet>2633</a>    /// This test creates:
-<a href=#2634 id=2634 data-nosnippet>2634</a>    /// - A data file with 200 rows split into 2 row groups (0-99, 100-199)
-<a href=#2635 id=2635 data-nosnippet>2635</a>    /// - A position delete file that deletes row 199 (last row in second row group)
-<a href=#2636 id=2636 data-nosnippet>2636</a>    /// - Row group selection that reads ONLY row group 1 (rows 100-199)
-<a href=#2637 id=2637 data-nosnippet>2637</a>    ///
-<a href=#2638 id=2638 data-nosnippet>2638</a>    /// Expected behavior: Should return 99 rows (with row 199 deleted)
-<a href=#2639 id=2639 data-nosnippet>2639</a>    /// Bug behavior: Returns 100 rows (delete is lost when skipping row group 0)
-<a href=#2640 id=2640 data-nosnippet>2640</a>    ///
-<a href=#2641 id=2641 data-nosnippet>2641</a>    /// The bug occurs when processing row group 0 (unselected):
-<a href=#2642 id=2642 data-nosnippet>2642</a>    /// ```rust
-<a href=#2643 id=2643 data-nosnippet>2643</a>    /// delete_vector_iter.advance_to(next_row_group_base_idx); // Position at first delete &gt;= 100
-<a href=#2644 id=2644 data-nosnippet>2644</a>    /// next_deleted_row_idx_opt = delete_vector_iter.next(); // BUG: Consumes delete at 199!
-<a href=#2645 id=2645 data-nosnippet>2645</a>    /// ```
-<a href=#2646 id=2646 data-nosnippet>2646</a>    ///
-<a href=#2647 id=2647 data-nosnippet>2647</a>    /// The fix is to NOT call `next()` after `advance_to()` when skipping unselected row groups,
-<a href=#2648 id=2648 data-nosnippet>2648</a>    /// because `advance_to()` already positions the iterator correctly without consuming elements.
-<a href=#2649 id=2649 data-nosnippet>2649</a>    </span><span class="attr">#[tokio::test]
-<a href=#2650 id=2650 data-nosnippet>2650</a>    </span><span class="kw">async fn </span>test_position_delete_with_row_group_selection() {
-<a href=#2651 id=2651 data-nosnippet>2651</a>        <span class="kw">use </span>arrow_array::{Int32Array, Int64Array};
-<a href=#2652 id=2652 data-nosnippet>2652</a>        <span class="kw">use </span>parquet::file::reader::{FileReader, SerializedFileReader};
-<a href=#2653 id=2653 data-nosnippet>2653</a>
-<a href=#2654 id=2654 data-nosnippet>2654</a>        <span class="comment">// Field IDs for positional delete schema
-<a href=#2655 id=2655 data-nosnippet>2655</a>        </span><span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_FILE_PATH: u64 = <span class="number">2147483546</span>;
-<a href=#2656 id=2656 data-nosnippet>2656</a>        <span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_POS: u64 = <span class="number">2147483545</span>;
-<a href=#2657 id=2657 data-nosnippet>2657</a>
-<a href=#2658 id=2658 data-nosnippet>2658</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#2659 id=2659 data-nosnippet>2659</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#2660 id=2660 data-nosnippet>2660</a>
-<a href=#2661 id=2661 data-nosnippet>2661</a>        <span class="comment">// Create table schema with a single 'id' column
-<a href=#2662 id=2662 data-nosnippet>2662</a>        </span><span class="kw">let </span>table_schema = Arc::new(
-<a href=#2663 id=2663 data-nosnippet>2663</a>            Schema::builder()
-<a href=#2664 id=2664 data-nosnippet>2664</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#2665 id=2665 data-nosnippet>2665</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#2666 id=2666 data-nosnippet>2666</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#2667 id=2667 data-nosnippet>2667</a>                ])
-<a href=#2668 id=2668 data-nosnippet>2668</a>                .build()
-<a href=#2669 id=2669 data-nosnippet>2669</a>                .unwrap(),
-<a href=#2670 id=2670 data-nosnippet>2670</a>        );
-<a href=#2671 id=2671 data-nosnippet>2671</a>
-<a href=#2672 id=2672 data-nosnippet>2672</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#2673 id=2673 data-nosnippet>2673</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2674 id=2674 data-nosnippet>2674</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2675 id=2675 data-nosnippet>2675</a>                <span class="string">"1"</span>.to_string(),
-<a href=#2676 id=2676 data-nosnippet>2676</a>            )])),
-<a href=#2677 id=2677 data-nosnippet>2677</a>        ]));
-<a href=#2678 id=2678 data-nosnippet>2678</a>
-<a href=#2679 id=2679 data-nosnippet>2679</a>        <span class="comment">// Step 1: Create data file with 200 rows in 2 row groups
-<a href=#2680 id=2680 data-nosnippet>2680</a>        // Row group 0: rows 0-99 (ids 1-100)
-<a href=#2681 id=2681 data-nosnippet>2681</a>        // Row group 1: rows 100-199 (ids 101-200)
-<a href=#2682 id=2682 data-nosnippet>2682</a>        </span><span class="kw">let </span>data_file_path = <span class="macro">format!</span>(<span class="string">"{}/data.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#2515 id=2515 data-nosnippet>2515</a>        <span class="comment">// Read the old Parquet file using the NEW schema (with column 'b')
+<a href=#2516 id=2516 data-nosnippet>2516</a>        </span><span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#2517 id=2517 data-nosnippet>2517</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#2518 id=2518 data-nosnippet>2518</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#2519 id=2519 data-nosnippet>2519</a>                start: <span class="number">0</span>,
+<a href=#2520 id=2520 data-nosnippet>2520</a>                length: <span class="number">0</span>,
+<a href=#2521 id=2521 data-nosnippet>2521</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#2522 id=2522 data-nosnippet>2522</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{table_location}/old_file.parquet"</span>),
+<a href=#2523 id=2523 data-nosnippet>2523</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#2524 id=2524 data-nosnippet>2524</a>                schema: new_schema.clone(),
+<a href=#2525 id=2525 data-nosnippet>2525</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>], <span class="comment">// Request both columns 'a' and 'b'
+<a href=#2526 id=2526 data-nosnippet>2526</a>                </span>predicate: <span class="prelude-val">None</span>,
+<a href=#2527 id=2527 data-nosnippet>2527</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#2528 id=2528 data-nosnippet>2528</a>                partition: <span class="prelude-val">None</span>,
+<a href=#2529 id=2529 data-nosnippet>2529</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#2530 id=2530 data-nosnippet>2530</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#2531 id=2531 data-nosnippet>2531</a>            })]
+<a href=#2532 id=2532 data-nosnippet>2532</a>            .into_iter(),
+<a href=#2533 id=2533 data-nosnippet>2533</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#2534 id=2534 data-nosnippet>2534</a>
+<a href=#2535 id=2535 data-nosnippet>2535</a>        <span class="kw">let </span>result = reader
+<a href=#2536 id=2536 data-nosnippet>2536</a>            .read(tasks)
+<a href=#2537 id=2537 data-nosnippet>2537</a>            .unwrap()
+<a href=#2538 id=2538 data-nosnippet>2538</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#2539 id=2539 data-nosnippet>2539</a>            .<span class="kw">await
+<a href=#2540 id=2540 data-nosnippet>2540</a>            </span>.unwrap();
+<a href=#2541 id=2541 data-nosnippet>2541</a>
+<a href=#2542 id=2542 data-nosnippet>2542</a>        <span class="comment">// Verify we got the correct data
+<a href=#2543 id=2543 data-nosnippet>2543</a>        </span><span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
+<a href=#2544 id=2544 data-nosnippet>2544</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
+<a href=#2545 id=2545 data-nosnippet>2545</a>
+<a href=#2546 id=2546 data-nosnippet>2546</a>        <span class="comment">// Should have 2 columns now
+<a href=#2547 id=2547 data-nosnippet>2547</a>        </span><span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
+<a href=#2548 id=2548 data-nosnippet>2548</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">3</span>);
+<a href=#2549 id=2549 data-nosnippet>2549</a>
+<a href=#2550 id=2550 data-nosnippet>2550</a>        <span class="comment">// Column 'a' should have the original data
+<a href=#2551 id=2551 data-nosnippet>2551</a>        </span><span class="kw">let </span>col_a = batch
+<a href=#2552 id=2552 data-nosnippet>2552</a>            .column(<span class="number">0</span>)
+<a href=#2553 id=2553 data-nosnippet>2553</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#2554 id=2554 data-nosnippet>2554</a>        <span class="macro">assert_eq!</span>(col_a.values(), <span class="kw-2">&amp;</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>]);
+<a href=#2555 id=2555 data-nosnippet>2555</a>
+<a href=#2556 id=2556 data-nosnippet>2556</a>        <span class="comment">// Column 'b' should be all NULLs (it didn't exist in the old file)
+<a href=#2557 id=2557 data-nosnippet>2557</a>        </span><span class="kw">let </span>col_b = batch
+<a href=#2558 id=2558 data-nosnippet>2558</a>            .column(<span class="number">1</span>)
+<a href=#2559 id=2559 data-nosnippet>2559</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#2560 id=2560 data-nosnippet>2560</a>        <span class="macro">assert_eq!</span>(col_b.null_count(), <span class="number">3</span>);
+<a href=#2561 id=2561 data-nosnippet>2561</a>        <span class="macro">assert!</span>(col_b.is_null(<span class="number">0</span>));
+<a href=#2562 id=2562 data-nosnippet>2562</a>        <span class="macro">assert!</span>(col_b.is_null(<span class="number">1</span>));
+<a href=#2563 id=2563 data-nosnippet>2563</a>        <span class="macro">assert!</span>(col_b.is_null(<span class="number">2</span>));
+<a href=#2564 id=2564 data-nosnippet>2564</a>    }
+<a href=#2565 id=2565 data-nosnippet>2565</a>
+<a href=#2566 id=2566 data-nosnippet>2566</a>    <span class="doccomment">/// Test for bug where position deletes in later row groups are not applied correctly.
+<a href=#2567 id=2567 data-nosnippet>2567</a>    ///
+<a href=#2568 id=2568 data-nosnippet>2568</a>    /// When a file has multiple row groups and a position delete targets a row in a later
+<a href=#2569 id=2569 data-nosnippet>2569</a>    /// row group, the `build_deletes_row_selection` function had a bug where it would
+<a href=#2570 id=2570 data-nosnippet>2570</a>    /// fail to increment `current_row_group_base_idx` when skipping row groups.
+<a href=#2571 id=2571 data-nosnippet>2571</a>    ///
+<a href=#2572 id=2572 data-nosnippet>2572</a>    /// This test creates:
+<a href=#2573 id=2573 data-nosnippet>2573</a>    /// - A data file with 200 rows split into 2 row groups (0-99, 100-199)
+<a href=#2574 id=2574 data-nosnippet>2574</a>    /// - A position delete file that deletes row 199 (last row in second row group)
+<a href=#2575 id=2575 data-nosnippet>2575</a>    ///
+<a href=#2576 id=2576 data-nosnippet>2576</a>    /// Expected behavior: Should return 199 rows (with id=200 deleted)
+<a href=#2577 id=2577 data-nosnippet>2577</a>    /// Bug behavior: Returns 200 rows (delete is not applied)
+<a href=#2578 id=2578 data-nosnippet>2578</a>    ///
+<a href=#2579 id=2579 data-nosnippet>2579</a>    /// This bug was discovered while running Apache Spark + Apache Iceberg integration tests
+<a href=#2580 id=2580 data-nosnippet>2580</a>    /// through DataFusion Comet. The following Iceberg Java tests failed due to this bug:
+<a href=#2581 id=2581 data-nosnippet>2581</a>    /// - `org.apache.iceberg.spark.extensions.TestMergeOnReadDelete::testDeleteWithMultipleRowGroupsParquet`
+<a href=#2582 id=2582 data-nosnippet>2582</a>    /// - `org.apache.iceberg.spark.extensions.TestMergeOnReadUpdate::testUpdateWithMultipleRowGroupsParquet`
+<a href=#2583 id=2583 data-nosnippet>2583</a>    </span><span class="attr">#[tokio::test]
+<a href=#2584 id=2584 data-nosnippet>2584</a>    </span><span class="kw">async fn </span>test_position_delete_across_multiple_row_groups() {
+<a href=#2585 id=2585 data-nosnippet>2585</a>        <span class="kw">use </span>arrow_array::{Int32Array, Int64Array};
+<a href=#2586 id=2586 data-nosnippet>2586</a>        <span class="kw">use </span>parquet::file::reader::{FileReader, SerializedFileReader};
+<a href=#2587 id=2587 data-nosnippet>2587</a>
+<a href=#2588 id=2588 data-nosnippet>2588</a>        <span class="comment">// Field IDs for positional delete schema
+<a href=#2589 id=2589 data-nosnippet>2589</a>        </span><span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_FILE_PATH: u64 = <span class="number">2147483546</span>;
+<a href=#2590 id=2590 data-nosnippet>2590</a>        <span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_POS: u64 = <span class="number">2147483545</span>;
+<a href=#2591 id=2591 data-nosnippet>2591</a>
+<a href=#2592 id=2592 data-nosnippet>2592</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#2593 id=2593 data-nosnippet>2593</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#2594 id=2594 data-nosnippet>2594</a>
+<a href=#2595 id=2595 data-nosnippet>2595</a>        <span class="comment">// Create table schema with a single 'id' column
+<a href=#2596 id=2596 data-nosnippet>2596</a>        </span><span class="kw">let </span>table_schema = Arc::new(
+<a href=#2597 id=2597 data-nosnippet>2597</a>            Schema::builder()
+<a href=#2598 id=2598 data-nosnippet>2598</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#2599 id=2599 data-nosnippet>2599</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#2600 id=2600 data-nosnippet>2600</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#2601 id=2601 data-nosnippet>2601</a>                ])
+<a href=#2602 id=2602 data-nosnippet>2602</a>                .build()
+<a href=#2603 id=2603 data-nosnippet>2603</a>                .unwrap(),
+<a href=#2604 id=2604 data-nosnippet>2604</a>        );
+<a href=#2605 id=2605 data-nosnippet>2605</a>
+<a href=#2606 id=2606 data-nosnippet>2606</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#2607 id=2607 data-nosnippet>2607</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#2608 id=2608 data-nosnippet>2608</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2609 id=2609 data-nosnippet>2609</a>                <span class="string">"1"</span>.to_string(),
+<a href=#2610 id=2610 data-nosnippet>2610</a>            )])),
+<a href=#2611 id=2611 data-nosnippet>2611</a>        ]));
+<a href=#2612 id=2612 data-nosnippet>2612</a>
+<a href=#2613 id=2613 data-nosnippet>2613</a>        <span class="comment">// Step 1: Create data file with 200 rows in 2 row groups
+<a href=#2614 id=2614 data-nosnippet>2614</a>        // Row group 0: rows 0-99 (ids 1-100)
+<a href=#2615 id=2615 data-nosnippet>2615</a>        // Row group 1: rows 100-199 (ids 101-200)
+<a href=#2616 id=2616 data-nosnippet>2616</a>        </span><span class="kw">let </span>data_file_path = <span class="macro">format!</span>(<span class="string">"{}/data.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#2617 id=2617 data-nosnippet>2617</a>
+<a href=#2618 id=2618 data-nosnippet>2618</a>        <span class="kw">let </span>batch1 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
+<a href=#2619 id=2619 data-nosnippet>2619</a>            Int32Array::from_iter_values(<span class="number">1</span>..=<span class="number">100</span>),
+<a href=#2620 id=2620 data-nosnippet>2620</a>        )])
+<a href=#2621 id=2621 data-nosnippet>2621</a>        .unwrap();
+<a href=#2622 id=2622 data-nosnippet>2622</a>
+<a href=#2623 id=2623 data-nosnippet>2623</a>        <span class="kw">let </span>batch2 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
+<a href=#2624 id=2624 data-nosnippet>2624</a>            Int32Array::from_iter_values(<span class="number">101</span>..=<span class="number">200</span>),
+<a href=#2625 id=2625 data-nosnippet>2625</a>        )])
+<a href=#2626 id=2626 data-nosnippet>2626</a>        .unwrap();
+<a href=#2627 id=2627 data-nosnippet>2627</a>
+<a href=#2628 id=2628 data-nosnippet>2628</a>        <span class="comment">// Force each batch into its own row group
+<a href=#2629 id=2629 data-nosnippet>2629</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
+<a href=#2630 id=2630 data-nosnippet>2630</a>            .set_compression(Compression::SNAPPY)
+<a href=#2631 id=2631 data-nosnippet>2631</a>            .set_max_row_group_size(<span class="number">100</span>)
+<a href=#2632 id=2632 data-nosnippet>2632</a>            .build();
+<a href=#2633 id=2633 data-nosnippet>2633</a>
+<a href=#2634 id=2634 data-nosnippet>2634</a>        <span class="kw">let </span>file = File::create(<span class="kw-2">&amp;</span>data_file_path).unwrap();
+<a href=#2635 id=2635 data-nosnippet>2635</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#2636 id=2636 data-nosnippet>2636</a>        writer.write(<span class="kw-2">&amp;</span>batch1).expect(<span class="string">"Writing batch 1"</span>);
+<a href=#2637 id=2637 data-nosnippet>2637</a>        writer.write(<span class="kw-2">&amp;</span>batch2).expect(<span class="string">"Writing batch 2"</span>);
+<a href=#2638 id=2638 data-nosnippet>2638</a>        writer.close().unwrap();
+<a href=#2639 id=2639 data-nosnippet>2639</a>
+<a href=#2640 id=2640 data-nosnippet>2640</a>        <span class="comment">// Verify we created 2 row groups
+<a href=#2641 id=2641 data-nosnippet>2641</a>        </span><span class="kw">let </span>verify_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
+<a href=#2642 id=2642 data-nosnippet>2642</a>        <span class="kw">let </span>verify_reader = SerializedFileReader::new(verify_file).unwrap();
+<a href=#2643 id=2643 data-nosnippet>2643</a>        <span class="macro">assert_eq!</span>(
+<a href=#2644 id=2644 data-nosnippet>2644</a>            verify_reader.metadata().num_row_groups(),
+<a href=#2645 id=2645 data-nosnippet>2645</a>            <span class="number">2</span>,
+<a href=#2646 id=2646 data-nosnippet>2646</a>            <span class="string">"Should have 2 row groups"
+<a href=#2647 id=2647 data-nosnippet>2647</a>        </span>);
+<a href=#2648 id=2648 data-nosnippet>2648</a>
+<a href=#2649 id=2649 data-nosnippet>2649</a>        <span class="comment">// Step 2: Create position delete file that deletes row 199 (id=200, last row in row group 1)
+<a href=#2650 id=2650 data-nosnippet>2650</a>        </span><span class="kw">let </span>delete_file_path = <span class="macro">format!</span>(<span class="string">"{}/deletes.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#2651 id=2651 data-nosnippet>2651</a>
+<a href=#2652 id=2652 data-nosnippet>2652</a>        <span class="kw">let </span>delete_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#2653 id=2653 data-nosnippet>2653</a>            Field::new(<span class="string">"file_path"</span>, DataType::Utf8, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#2654 id=2654 data-nosnippet>2654</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2655 id=2655 data-nosnippet>2655</a>                FIELD_ID_POSITIONAL_DELETE_FILE_PATH.to_string(),
+<a href=#2656 id=2656 data-nosnippet>2656</a>            )])),
+<a href=#2657 id=2657 data-nosnippet>2657</a>            Field::new(<span class="string">"pos"</span>, DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#2658 id=2658 data-nosnippet>2658</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2659 id=2659 data-nosnippet>2659</a>                FIELD_ID_POSITIONAL_DELETE_POS.to_string(),
+<a href=#2660 id=2660 data-nosnippet>2660</a>            )])),
+<a href=#2661 id=2661 data-nosnippet>2661</a>        ]));
+<a href=#2662 id=2662 data-nosnippet>2662</a>
+<a href=#2663 id=2663 data-nosnippet>2663</a>        <span class="comment">// Delete row at position 199 (0-indexed, so it's the last row: id=200)
+<a href=#2664 id=2664 data-nosnippet>2664</a>        </span><span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[
+<a href=#2665 id=2665 data-nosnippet>2665</a>            Arc::new(StringArray::from_iter_values(<span class="macro">vec!</span>[data_file_path.clone()])),
+<a href=#2666 id=2666 data-nosnippet>2666</a>            Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[<span class="number">199i64</span>])),
+<a href=#2667 id=2667 data-nosnippet>2667</a>        ])
+<a href=#2668 id=2668 data-nosnippet>2668</a>        .unwrap();
+<a href=#2669 id=2669 data-nosnippet>2669</a>
+<a href=#2670 id=2670 data-nosnippet>2670</a>        <span class="kw">let </span>delete_props = WriterProperties::builder()
+<a href=#2671 id=2671 data-nosnippet>2671</a>            .set_compression(Compression::SNAPPY)
+<a href=#2672 id=2672 data-nosnippet>2672</a>            .build();
+<a href=#2673 id=2673 data-nosnippet>2673</a>
+<a href=#2674 id=2674 data-nosnippet>2674</a>        <span class="kw">let </span>delete_file = File::create(<span class="kw-2">&amp;</span>delete_file_path).unwrap();
+<a href=#2675 id=2675 data-nosnippet>2675</a>        <span class="kw">let </span><span class="kw-2">mut </span>delete_writer =
+<a href=#2676 id=2676 data-nosnippet>2676</a>            ArrowWriter::try_new(delete_file, delete_schema, <span class="prelude-val">Some</span>(delete_props)).unwrap();
+<a href=#2677 id=2677 data-nosnippet>2677</a>        delete_writer.write(<span class="kw-2">&amp;</span>delete_batch).unwrap();
+<a href=#2678 id=2678 data-nosnippet>2678</a>        delete_writer.close().unwrap();
+<a href=#2679 id=2679 data-nosnippet>2679</a>
+<a href=#2680 id=2680 data-nosnippet>2680</a>        <span class="comment">// Step 3: Read the data file with the delete applied
+<a href=#2681 id=2681 data-nosnippet>2681</a>        </span><span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#2682 id=2682 data-nosnippet>2682</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
 <a href=#2683 id=2683 data-nosnippet>2683</a>
-<a href=#2684 id=2684 data-nosnippet>2684</a>        <span class="kw">let </span>batch1 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
-<a href=#2685 id=2685 data-nosnippet>2685</a>            Int32Array::from_iter_values(<span class="number">1</span>..=<span class="number">100</span>),
-<a href=#2686 id=2686 data-nosnippet>2686</a>        )])
-<a href=#2687 id=2687 data-nosnippet>2687</a>        .unwrap();
-<a href=#2688 id=2688 data-nosnippet>2688</a>
-<a href=#2689 id=2689 data-nosnippet>2689</a>        <span class="kw">let </span>batch2 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
-<a href=#2690 id=2690 data-nosnippet>2690</a>            Int32Array::from_iter_values(<span class="number">101</span>..=<span class="number">200</span>),
-<a href=#2691 id=2691 data-nosnippet>2691</a>        )])
-<a href=#2692 id=2692 data-nosnippet>2692</a>        .unwrap();
-<a href=#2693 id=2693 data-nosnippet>2693</a>
-<a href=#2694 id=2694 data-nosnippet>2694</a>        <span class="comment">// Force each batch into its own row group
-<a href=#2695 id=2695 data-nosnippet>2695</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
-<a href=#2696 id=2696 data-nosnippet>2696</a>            .set_compression(Compression::SNAPPY)
-<a href=#2697 id=2697 data-nosnippet>2697</a>            .set_max_row_group_size(<span class="number">100</span>)
-<a href=#2698 id=2698 data-nosnippet>2698</a>            .build();
-<a href=#2699 id=2699 data-nosnippet>2699</a>
-<a href=#2700 id=2700 data-nosnippet>2700</a>        <span class="kw">let </span>file = File::create(<span class="kw-2">&amp;</span>data_file_path).unwrap();
-<a href=#2701 id=2701 data-nosnippet>2701</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#2702 id=2702 data-nosnippet>2702</a>        writer.write(<span class="kw-2">&amp;</span>batch1).expect(<span class="string">"Writing batch 1"</span>);
-<a href=#2703 id=2703 data-nosnippet>2703</a>        writer.write(<span class="kw-2">&amp;</span>batch2).expect(<span class="string">"Writing batch 2"</span>);
-<a href=#2704 id=2704 data-nosnippet>2704</a>        writer.close().unwrap();
-<a href=#2705 id=2705 data-nosnippet>2705</a>
-<a href=#2706 id=2706 data-nosnippet>2706</a>        <span class="comment">// Verify we created 2 row groups
-<a href=#2707 id=2707 data-nosnippet>2707</a>        </span><span class="kw">let </span>verify_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
-<a href=#2708 id=2708 data-nosnippet>2708</a>        <span class="kw">let </span>verify_reader = SerializedFileReader::new(verify_file).unwrap();
-<a href=#2709 id=2709 data-nosnippet>2709</a>        <span class="macro">assert_eq!</span>(
-<a href=#2710 id=2710 data-nosnippet>2710</a>            verify_reader.metadata().num_row_groups(),
-<a href=#2711 id=2711 data-nosnippet>2711</a>            <span class="number">2</span>,
-<a href=#2712 id=2712 data-nosnippet>2712</a>            <span class="string">"Should have 2 row groups"
-<a href=#2713 id=2713 data-nosnippet>2713</a>        </span>);
+<a href=#2684 id=2684 data-nosnippet>2684</a>        <span class="kw">let </span>task = FileScanTask {
+<a href=#2685 id=2685 data-nosnippet>2685</a>            start: <span class="number">0</span>,
+<a href=#2686 id=2686 data-nosnippet>2686</a>            length: <span class="number">0</span>,
+<a href=#2687 id=2687 data-nosnippet>2687</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">200</span>),
+<a href=#2688 id=2688 data-nosnippet>2688</a>            data_file_path: data_file_path.clone(),
+<a href=#2689 id=2689 data-nosnippet>2689</a>            data_file_format: DataFileFormat::Parquet,
+<a href=#2690 id=2690 data-nosnippet>2690</a>            schema: table_schema.clone(),
+<a href=#2691 id=2691 data-nosnippet>2691</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
+<a href=#2692 id=2692 data-nosnippet>2692</a>            predicate: <span class="prelude-val">None</span>,
+<a href=#2693 id=2693 data-nosnippet>2693</a>            deletes: <span class="macro">vec!</span>[FileScanTaskDeleteFile {
+<a href=#2694 id=2694 data-nosnippet>2694</a>                file_path: delete_file_path,
+<a href=#2695 id=2695 data-nosnippet>2695</a>                file_type: DataContentType::PositionDeletes,
+<a href=#2696 id=2696 data-nosnippet>2696</a>                partition_spec_id: <span class="number">0</span>,
+<a href=#2697 id=2697 data-nosnippet>2697</a>                equality_ids: <span class="prelude-val">None</span>,
+<a href=#2698 id=2698 data-nosnippet>2698</a>            }],
+<a href=#2699 id=2699 data-nosnippet>2699</a>            partition: <span class="prelude-val">None</span>,
+<a href=#2700 id=2700 data-nosnippet>2700</a>            partition_spec: <span class="prelude-val">None</span>,
+<a href=#2701 id=2701 data-nosnippet>2701</a>            name_mapping: <span class="prelude-val">None</span>,
+<a href=#2702 id=2702 data-nosnippet>2702</a>        };
+<a href=#2703 id=2703 data-nosnippet>2703</a>
+<a href=#2704 id=2704 data-nosnippet>2704</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task)])) <span class="kw">as </span>FileScanTaskStream;
+<a href=#2705 id=2705 data-nosnippet>2705</a>        <span class="kw">let </span>result = reader
+<a href=#2706 id=2706 data-nosnippet>2706</a>            .read(tasks)
+<a href=#2707 id=2707 data-nosnippet>2707</a>            .unwrap()
+<a href=#2708 id=2708 data-nosnippet>2708</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#2709 id=2709 data-nosnippet>2709</a>            .<span class="kw">await
+<a href=#2710 id=2710 data-nosnippet>2710</a>            </span>.unwrap();
+<a href=#2711 id=2711 data-nosnippet>2711</a>
+<a href=#2712 id=2712 data-nosnippet>2712</a>        <span class="comment">// Step 4: Verify we got 199 rows (not 200)
+<a href=#2713 id=2713 data-nosnippet>2713</a>        </span><span class="kw">let </span>total_rows: usize = result.iter().map(|b| b.num_rows()).sum();
 <a href=#2714 id=2714 data-nosnippet>2714</a>
-<a href=#2715 id=2715 data-nosnippet>2715</a>        <span class="comment">// Step 2: Create position delete file that deletes row 199 (id=200, last row in row group 1)
-<a href=#2716 id=2716 data-nosnippet>2716</a>        </span><span class="kw">let </span>delete_file_path = <span class="macro">format!</span>(<span class="string">"{}/deletes.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#2715 id=2715 data-nosnippet>2715</a>        <span class="macro">println!</span>(<span class="string">"Total rows read: {}"</span>, total_rows);
+<a href=#2716 id=2716 data-nosnippet>2716</a>        <span class="macro">println!</span>(<span class="string">"Expected: 199 rows (deleted row 199 which had id=200)"</span>);
 <a href=#2717 id=2717 data-nosnippet>2717</a>
-<a href=#2718 id=2718 data-nosnippet>2718</a>        <span class="kw">let </span>delete_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#2719 id=2719 data-nosnippet>2719</a>            Field::new(<span class="string">"file_path"</span>, DataType::Utf8, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2720 id=2720 data-nosnippet>2720</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2721 id=2721 data-nosnippet>2721</a>                FIELD_ID_POSITIONAL_DELETE_FILE_PATH.to_string(),
-<a href=#2722 id=2722 data-nosnippet>2722</a>            )])),
-<a href=#2723 id=2723 data-nosnippet>2723</a>            Field::new(<span class="string">"pos"</span>, DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2724 id=2724 data-nosnippet>2724</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2725 id=2725 data-nosnippet>2725</a>                FIELD_ID_POSITIONAL_DELETE_POS.to_string(),
-<a href=#2726 id=2726 data-nosnippet>2726</a>            )])),
-<a href=#2727 id=2727 data-nosnippet>2727</a>        ]));
-<a href=#2728 id=2728 data-nosnippet>2728</a>
-<a href=#2729 id=2729 data-nosnippet>2729</a>        <span class="comment">// Delete row at position 199 (0-indexed, so it's the last row: id=200)
-<a href=#2730 id=2730 data-nosnippet>2730</a>        </span><span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[
-<a href=#2731 id=2731 data-nosnippet>2731</a>            Arc::new(StringArray::from_iter_values(<span class="macro">vec!</span>[data_file_path.clone()])),
-<a href=#2732 id=2732 data-nosnippet>2732</a>            Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[<span class="number">199i64</span>])),
-<a href=#2733 id=2733 data-nosnippet>2733</a>        ])
-<a href=#2734 id=2734 data-nosnippet>2734</a>        .unwrap();
-<a href=#2735 id=2735 data-nosnippet>2735</a>
-<a href=#2736 id=2736 data-nosnippet>2736</a>        <span class="kw">let </span>delete_props = WriterProperties::builder()
-<a href=#2737 id=2737 data-nosnippet>2737</a>            .set_compression(Compression::SNAPPY)
-<a href=#2738 id=2738 data-nosnippet>2738</a>            .build();
-<a href=#2739 id=2739 data-nosnippet>2739</a>
-<a href=#2740 id=2740 data-nosnippet>2740</a>        <span class="kw">let </span>delete_file = File::create(<span class="kw-2">&amp;</span>delete_file_path).unwrap();
-<a href=#2741 id=2741 data-nosnippet>2741</a>        <span class="kw">let </span><span class="kw-2">mut </span>delete_writer =
-<a href=#2742 id=2742 data-nosnippet>2742</a>            ArrowWriter::try_new(delete_file, delete_schema, <span class="prelude-val">Some</span>(delete_props)).unwrap();
-<a href=#2743 id=2743 data-nosnippet>2743</a>        delete_writer.write(<span class="kw-2">&amp;</span>delete_batch).unwrap();
-<a href=#2744 id=2744 data-nosnippet>2744</a>        delete_writer.close().unwrap();
-<a href=#2745 id=2745 data-nosnippet>2745</a>
-<a href=#2746 id=2746 data-nosnippet>2746</a>        <span class="comment">// Step 3: Get byte ranges to read ONLY row group 1 (rows 100-199)
-<a href=#2747 id=2747 data-nosnippet>2747</a>        // This exercises the row group selection code path where row group 0 is skipped
-<a href=#2748 id=2748 data-nosnippet>2748</a>        </span><span class="kw">let </span>metadata_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
-<a href=#2749 id=2749 data-nosnippet>2749</a>        <span class="kw">let </span>metadata_reader = SerializedFileReader::new(metadata_file).unwrap();
-<a href=#2750 id=2750 data-nosnippet>2750</a>        <span class="kw">let </span>metadata = metadata_reader.metadata();
+<a href=#2718 id=2718 data-nosnippet>2718</a>        <span class="comment">// This assertion will FAIL before the fix and PASS after the fix
+<a href=#2719 id=2719 data-nosnippet>2719</a>        </span><span class="macro">assert_eq!</span>(
+<a href=#2720 id=2720 data-nosnippet>2720</a>            total_rows, <span class="number">199</span>,
+<a href=#2721 id=2721 data-nosnippet>2721</a>            <span class="string">"Expected 199 rows after deleting row 199, but got {} rows. \
+<a href=#2722 id=2722 data-nosnippet>2722</a>             The bug causes position deletes in later row groups to be ignored."</span>,
+<a href=#2723 id=2723 data-nosnippet>2723</a>            total_rows
+<a href=#2724 id=2724 data-nosnippet>2724</a>        );
+<a href=#2725 id=2725 data-nosnippet>2725</a>
+<a href=#2726 id=2726 data-nosnippet>2726</a>        <span class="comment">// Verify the deleted row (id=200) is not present
+<a href=#2727 id=2727 data-nosnippet>2727</a>        </span><span class="kw">let </span>all_ids: Vec&lt;i32&gt; = result
+<a href=#2728 id=2728 data-nosnippet>2728</a>            .iter()
+<a href=#2729 id=2729 data-nosnippet>2729</a>            .flat_map(|batch| {
+<a href=#2730 id=2730 data-nosnippet>2730</a>                batch
+<a href=#2731 id=2731 data-nosnippet>2731</a>                    .column(<span class="number">0</span>)
+<a href=#2732 id=2732 data-nosnippet>2732</a>                    .as_primitive::&lt;arrow_array::types::Int32Type&gt;()
+<a href=#2733 id=2733 data-nosnippet>2733</a>                    .values()
+<a href=#2734 id=2734 data-nosnippet>2734</a>                    .iter()
+<a href=#2735 id=2735 data-nosnippet>2735</a>                    .copied()
+<a href=#2736 id=2736 data-nosnippet>2736</a>            })
+<a href=#2737 id=2737 data-nosnippet>2737</a>            .collect();
+<a href=#2738 id=2738 data-nosnippet>2738</a>
+<a href=#2739 id=2739 data-nosnippet>2739</a>        <span class="macro">assert!</span>(
+<a href=#2740 id=2740 data-nosnippet>2740</a>            !all_ids.contains(<span class="kw-2">&amp;</span><span class="number">200</span>),
+<a href=#2741 id=2741 data-nosnippet>2741</a>            <span class="string">"Row with id=200 should be deleted but was found in results"
+<a href=#2742 id=2742 data-nosnippet>2742</a>        </span>);
+<a href=#2743 id=2743 data-nosnippet>2743</a>
+<a href=#2744 id=2744 data-nosnippet>2744</a>        <span class="comment">// Verify we have all other ids (1-199)
+<a href=#2745 id=2745 data-nosnippet>2745</a>        </span><span class="kw">let </span>expected_ids: Vec&lt;i32&gt; = (<span class="number">1</span>..=<span class="number">199</span>).collect();
+<a href=#2746 id=2746 data-nosnippet>2746</a>        <span class="macro">assert_eq!</span>(
+<a href=#2747 id=2747 data-nosnippet>2747</a>            all_ids, expected_ids,
+<a href=#2748 id=2748 data-nosnippet>2748</a>            <span class="string">"Should have ids 1-199 but got different values"
+<a href=#2749 id=2749 data-nosnippet>2749</a>        </span>);
+<a href=#2750 id=2750 data-nosnippet>2750</a>    }
 <a href=#2751 id=2751 data-nosnippet>2751</a>
-<a href=#2752 id=2752 data-nosnippet>2752</a>        <span class="kw">let </span>row_group_0 = metadata.row_group(<span class="number">0</span>);
-<a href=#2753 id=2753 data-nosnippet>2753</a>        <span class="kw">let </span>row_group_1 = metadata.row_group(<span class="number">1</span>);
-<a href=#2754 id=2754 data-nosnippet>2754</a>
-<a href=#2755 id=2755 data-nosnippet>2755</a>        <span class="kw">let </span>rg0_start = <span class="number">4u64</span>; <span class="comment">// Parquet files start with 4-byte magic "PAR1"
-<a href=#2756 id=2756 data-nosnippet>2756</a>        </span><span class="kw">let </span>rg1_start = rg0_start + row_group_0.compressed_size() <span class="kw">as </span>u64;
-<a href=#2757 id=2757 data-nosnippet>2757</a>        <span class="kw">let </span>rg1_length = row_group_1.compressed_size() <span class="kw">as </span>u64;
-<a href=#2758 id=2758 data-nosnippet>2758</a>
-<a href=#2759 id=2759 data-nosnippet>2759</a>        <span class="macro">println!</span>(
-<a href=#2760 id=2760 data-nosnippet>2760</a>            <span class="string">"Row group 0: starts at byte {}, {} bytes compressed"</span>,
-<a href=#2761 id=2761 data-nosnippet>2761</a>            rg0_start,
-<a href=#2762 id=2762 data-nosnippet>2762</a>            row_group_0.compressed_size()
-<a href=#2763 id=2763 data-nosnippet>2763</a>        );
-<a href=#2764 id=2764 data-nosnippet>2764</a>        <span class="macro">println!</span>(
-<a href=#2765 id=2765 data-nosnippet>2765</a>            <span class="string">"Row group 1: starts at byte {}, {} bytes compressed"</span>,
-<a href=#2766 id=2766 data-nosnippet>2766</a>            rg1_start,
-<a href=#2767 id=2767 data-nosnippet>2767</a>            row_group_1.compressed_size()
-<a href=#2768 id=2768 data-nosnippet>2768</a>        );
-<a href=#2769 id=2769 data-nosnippet>2769</a>
-<a href=#2770 id=2770 data-nosnippet>2770</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#2771 id=2771 data-nosnippet>2771</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#2772 id=2772 data-nosnippet>2772</a>
-<a href=#2773 id=2773 data-nosnippet>2773</a>        <span class="comment">// Create FileScanTask that reads ONLY row group 1 via byte range filtering
-<a href=#2774 id=2774 data-nosnippet>2774</a>        </span><span class="kw">let </span>task = FileScanTask {
-<a href=#2775 id=2775 data-nosnippet>2775</a>            start: rg1_start,
-<a href=#2776 id=2776 data-nosnippet>2776</a>            length: rg1_length,
-<a href=#2777 id=2777 data-nosnippet>2777</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">100</span>), <span class="comment">// Row group 1 has 100 rows
-<a href=#2778 id=2778 data-nosnippet>2778</a>            </span>data_file_path: data_file_path.clone(),
-<a href=#2779 id=2779 data-nosnippet>2779</a>            data_file_format: DataFileFormat::Parquet,
-<a href=#2780 id=2780 data-nosnippet>2780</a>            schema: table_schema.clone(),
-<a href=#2781 id=2781 data-nosnippet>2781</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
-<a href=#2782 id=2782 data-nosnippet>2782</a>            predicate: <span class="prelude-val">None</span>,
-<a href=#2783 id=2783 data-nosnippet>2783</a>            deletes: <span class="macro">vec!</span>[FileScanTaskDeleteFile {
-<a href=#2784 id=2784 data-nosnippet>2784</a>                file_path: delete_file_path,
-<a href=#2785 id=2785 data-nosnippet>2785</a>                file_type: DataContentType::PositionDeletes,
-<a href=#2786 id=2786 data-nosnippet>2786</a>                partition_spec_id: <span class="number">0</span>,
-<a href=#2787 id=2787 data-nosnippet>2787</a>                equality_ids: <span class="prelude-val">None</span>,
-<a href=#2788 id=2788 data-nosnippet>2788</a>            }],
-<a href=#2789 id=2789 data-nosnippet>2789</a>        };
-<a href=#2790 id=2790 data-nosnippet>2790</a>
-<a href=#2791 id=2791 data-nosnippet>2791</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task)])) <span class="kw">as </span>FileScanTaskStream;
-<a href=#2792 id=2792 data-nosnippet>2792</a>        <span class="kw">let </span>result = reader
-<a href=#2793 id=2793 data-nosnippet>2793</a>            .read(tasks)
-<a href=#2794 id=2794 data-nosnippet>2794</a>            .unwrap()
-<a href=#2795 id=2795 data-nosnippet>2795</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#2796 id=2796 data-nosnippet>2796</a>            .<span class="kw">await
-<a href=#2797 id=2797 data-nosnippet>2797</a>            </span>.unwrap();
-<a href=#2798 id=2798 data-nosnippet>2798</a>
-<a href=#2799 id=2799 data-nosnippet>2799</a>        <span class="comment">// Step 4: Verify we got 99 rows (not 100)
-<a href=#2800 id=2800 data-nosnippet>2800</a>        // Row group 1 has 100 rows (ids 101-200), minus 1 delete (id=200) = 99 rows
-<a href=#2801 id=2801 data-nosnippet>2801</a>        </span><span class="kw">let </span>total_rows: usize = result.iter().map(|b| b.num_rows()).sum();
-<a href=#2802 id=2802 data-nosnippet>2802</a>
-<a href=#2803 id=2803 data-nosnippet>2803</a>        <span class="macro">println!</span>(<span class="string">"Total rows read from row group 1: {}"</span>, total_rows);
-<a href=#2804 id=2804 data-nosnippet>2804</a>        <span class="macro">println!</span>(<span class="string">"Expected: 99 rows (row group 1 has 100 rows, 1 delete at position 199)"</span>);
-<a href=#2805 id=2805 data-nosnippet>2805</a>
-<a href=#2806 id=2806 data-nosnippet>2806</a>        <span class="comment">// This assertion will FAIL before the fix and PASS after the fix
-<a href=#2807 id=2807 data-nosnippet>2807</a>        </span><span class="macro">assert_eq!</span>(
-<a href=#2808 id=2808 data-nosnippet>2808</a>            total_rows, <span class="number">99</span>,
-<a href=#2809 id=2809 data-nosnippet>2809</a>            <span class="string">"Expected 99 rows from row group 1 after deleting position 199, but got {} rows. \
-<a href=#2810 id=2810 data-nosnippet>2810</a>             The bug causes position deletes to be lost when advance_to() is followed by next() \
-<a href=#2811 id=2811 data-nosnippet>2811</a>             when skipping unselected row groups."</span>,
-<a href=#2812 id=2812 data-nosnippet>2812</a>            total_rows
-<a href=#2813 id=2813 data-nosnippet>2813</a>        );
-<a href=#2814 id=2814 data-nosnippet>2814</a>
-<a href=#2815 id=2815 data-nosnippet>2815</a>        <span class="comment">// Verify the deleted row (id=200) is not present
-<a href=#2816 id=2816 data-nosnippet>2816</a>        </span><span class="kw">let </span>all_ids: Vec&lt;i32&gt; = result
-<a href=#2817 id=2817 data-nosnippet>2817</a>            .iter()
-<a href=#2818 id=2818 data-nosnippet>2818</a>            .flat_map(|batch| {
-<a href=#2819 id=2819 data-nosnippet>2819</a>                batch
-<a href=#2820 id=2820 data-nosnippet>2820</a>                    .column(<span class="number">0</span>)
-<a href=#2821 id=2821 data-nosnippet>2821</a>                    .as_primitive::&lt;arrow_array::types::Int32Type&gt;()
-<a href=#2822 id=2822 data-nosnippet>2822</a>                    .values()
-<a href=#2823 id=2823 data-nosnippet>2823</a>                    .iter()
-<a href=#2824 id=2824 data-nosnippet>2824</a>                    .copied()
-<a href=#2825 id=2825 data-nosnippet>2825</a>            })
-<a href=#2826 id=2826 data-nosnippet>2826</a>            .collect();
+<a href=#2752 id=2752 data-nosnippet>2752</a>    <span class="doccomment">/// Test for bug where position deletes are lost when skipping unselected row groups.
+<a href=#2753 id=2753 data-nosnippet>2753</a>    ///
+<a href=#2754 id=2754 data-nosnippet>2754</a>    /// This is a variant of `test_position_delete_across_multiple_row_groups` that exercises
+<a href=#2755 id=2755 data-nosnippet>2755</a>    /// the row group selection code path (`selected_row_groups: Some([...])`).
+<a href=#2756 id=2756 data-nosnippet>2756</a>    ///
+<a href=#2757 id=2757 data-nosnippet>2757</a>    /// When a file has multiple row groups and only some are selected for reading,
+<a href=#2758 id=2758 data-nosnippet>2758</a>    /// the `build_deletes_row_selection` function must correctly skip over deletes in
+<a href=#2759 id=2759 data-nosnippet>2759</a>    /// unselected row groups WITHOUT consuming deletes that belong to selected row groups.
+<a href=#2760 id=2760 data-nosnippet>2760</a>    ///
+<a href=#2761 id=2761 data-nosnippet>2761</a>    /// This test creates:
+<a href=#2762 id=2762 data-nosnippet>2762</a>    /// - A data file with 200 rows split into 2 row groups (0-99, 100-199)
+<a href=#2763 id=2763 data-nosnippet>2763</a>    /// - A position delete file that deletes row 199 (last row in second row group)
+<a href=#2764 id=2764 data-nosnippet>2764</a>    /// - Row group selection that reads ONLY row group 1 (rows 100-199)
+<a href=#2765 id=2765 data-nosnippet>2765</a>    ///
+<a href=#2766 id=2766 data-nosnippet>2766</a>    /// Expected behavior: Should return 99 rows (with row 199 deleted)
+<a href=#2767 id=2767 data-nosnippet>2767</a>    /// Bug behavior: Returns 100 rows (delete is lost when skipping row group 0)
+<a href=#2768 id=2768 data-nosnippet>2768</a>    ///
+<a href=#2769 id=2769 data-nosnippet>2769</a>    /// The bug occurs when processing row group 0 (unselected):
+<a href=#2770 id=2770 data-nosnippet>2770</a>    /// ```rust
+<a href=#2771 id=2771 data-nosnippet>2771</a>    /// delete_vector_iter.advance_to(next_row_group_base_idx); // Position at first delete &gt;= 100
+<a href=#2772 id=2772 data-nosnippet>2772</a>    /// next_deleted_row_idx_opt = delete_vector_iter.next(); // BUG: Consumes delete at 199!
+<a href=#2773 id=2773 data-nosnippet>2773</a>    /// ```
+<a href=#2774 id=2774 data-nosnippet>2774</a>    ///
+<a href=#2775 id=2775 data-nosnippet>2775</a>    /// The fix is to NOT call `next()` after `advance_to()` when skipping unselected row groups,
+<a href=#2776 id=2776 data-nosnippet>2776</a>    /// because `advance_to()` already positions the iterator correctly without consuming elements.
+<a href=#2777 id=2777 data-nosnippet>2777</a>    </span><span class="attr">#[tokio::test]
+<a href=#2778 id=2778 data-nosnippet>2778</a>    </span><span class="kw">async fn </span>test_position_delete_with_row_group_selection() {
+<a href=#2779 id=2779 data-nosnippet>2779</a>        <span class="kw">use </span>arrow_array::{Int32Array, Int64Array};
+<a href=#2780 id=2780 data-nosnippet>2780</a>        <span class="kw">use </span>parquet::file::reader::{FileReader, SerializedFileReader};
+<a href=#2781 id=2781 data-nosnippet>2781</a>
+<a href=#2782 id=2782 data-nosnippet>2782</a>        <span class="comment">// Field IDs for positional delete schema
+<a href=#2783 id=2783 data-nosnippet>2783</a>        </span><span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_FILE_PATH: u64 = <span class="number">2147483546</span>;
+<a href=#2784 id=2784 data-nosnippet>2784</a>        <span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_POS: u64 = <span class="number">2147483545</span>;
+<a href=#2785 id=2785 data-nosnippet>2785</a>
+<a href=#2786 id=2786 data-nosnippet>2786</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#2787 id=2787 data-nosnippet>2787</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#2788 id=2788 data-nosnippet>2788</a>
+<a href=#2789 id=2789 data-nosnippet>2789</a>        <span class="comment">// Create table schema with a single 'id' column
+<a href=#2790 id=2790 data-nosnippet>2790</a>        </span><span class="kw">let </span>table_schema = Arc::new(
+<a href=#2791 id=2791 data-nosnippet>2791</a>            Schema::builder()
+<a href=#2792 id=2792 data-nosnippet>2792</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#2793 id=2793 data-nosnippet>2793</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#2794 id=2794 data-nosnippet>2794</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#2795 id=2795 data-nosnippet>2795</a>                ])
+<a href=#2796 id=2796 data-nosnippet>2796</a>                .build()
+<a href=#2797 id=2797 data-nosnippet>2797</a>                .unwrap(),
+<a href=#2798 id=2798 data-nosnippet>2798</a>        );
+<a href=#2799 id=2799 data-nosnippet>2799</a>
+<a href=#2800 id=2800 data-nosnippet>2800</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#2801 id=2801 data-nosnippet>2801</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#2802 id=2802 data-nosnippet>2802</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2803 id=2803 data-nosnippet>2803</a>                <span class="string">"1"</span>.to_string(),
+<a href=#2804 id=2804 data-nosnippet>2804</a>            )])),
+<a href=#2805 id=2805 data-nosnippet>2805</a>        ]));
+<a href=#2806 id=2806 data-nosnippet>2806</a>
+<a href=#2807 id=2807 data-nosnippet>2807</a>        <span class="comment">// Step 1: Create data file with 200 rows in 2 row groups
+<a href=#2808 id=2808 data-nosnippet>2808</a>        // Row group 0: rows 0-99 (ids 1-100)
+<a href=#2809 id=2809 data-nosnippet>2809</a>        // Row group 1: rows 100-199 (ids 101-200)
+<a href=#2810 id=2810 data-nosnippet>2810</a>        </span><span class="kw">let </span>data_file_path = <span class="macro">format!</span>(<span class="string">"{}/data.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#2811 id=2811 data-nosnippet>2811</a>
+<a href=#2812 id=2812 data-nosnippet>2812</a>        <span class="kw">let </span>batch1 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
+<a href=#2813 id=2813 data-nosnippet>2813</a>            Int32Array::from_iter_values(<span class="number">1</span>..=<span class="number">100</span>),
+<a href=#2814 id=2814 data-nosnippet>2814</a>        )])
+<a href=#2815 id=2815 data-nosnippet>2815</a>        .unwrap();
+<a href=#2816 id=2816 data-nosnippet>2816</a>
+<a href=#2817 id=2817 data-nosnippet>2817</a>        <span class="kw">let </span>batch2 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
+<a href=#2818 id=2818 data-nosnippet>2818</a>            Int32Array::from_iter_values(<span class="number">101</span>..=<span class="number">200</span>),
+<a href=#2819 id=2819 data-nosnippet>2819</a>        )])
+<a href=#2820 id=2820 data-nosnippet>2820</a>        .unwrap();
+<a href=#2821 id=2821 data-nosnippet>2821</a>
+<a href=#2822 id=2822 data-nosnippet>2822</a>        <span class="comment">// Force each batch into its own row group
+<a href=#2823 id=2823 data-nosnippet>2823</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
+<a href=#2824 id=2824 data-nosnippet>2824</a>            .set_compression(Compression::SNAPPY)
+<a href=#2825 id=2825 data-nosnippet>2825</a>            .set_max_row_group_size(<span class="number">100</span>)
+<a href=#2826 id=2826 data-nosnippet>2826</a>            .build();
 <a href=#2827 id=2827 data-nosnippet>2827</a>
-<a href=#2828 id=2828 data-nosnippet>2828</a>        <span class="macro">assert!</span>(
-<a href=#2829 id=2829 data-nosnippet>2829</a>            !all_ids.contains(<span class="kw-2">&amp;</span><span class="number">200</span>),
-<a href=#2830 id=2830 data-nosnippet>2830</a>            <span class="string">"Row with id=200 should be deleted but was found in results"
-<a href=#2831 id=2831 data-nosnippet>2831</a>        </span>);
-<a href=#2832 id=2832 data-nosnippet>2832</a>
-<a href=#2833 id=2833 data-nosnippet>2833</a>        <span class="comment">// Verify we have ids 101-199 (not 101-200)
-<a href=#2834 id=2834 data-nosnippet>2834</a>        </span><span class="kw">let </span>expected_ids: Vec&lt;i32&gt; = (<span class="number">101</span>..=<span class="number">199</span>).collect();
-<a href=#2835 id=2835 data-nosnippet>2835</a>        <span class="macro">assert_eq!</span>(
-<a href=#2836 id=2836 data-nosnippet>2836</a>            all_ids, expected_ids,
-<a href=#2837 id=2837 data-nosnippet>2837</a>            <span class="string">"Should have ids 101-199 but got different values"
-<a href=#2838 id=2838 data-nosnippet>2838</a>        </span>);
-<a href=#2839 id=2839 data-nosnippet>2839</a>    }
-<a href=#2840 id=2840 data-nosnippet>2840</a>    <span class="doccomment">/// Test for bug where stale cached delete causes infinite loop when skipping row groups.
-<a href=#2841 id=2841 data-nosnippet>2841</a>    ///
-<a href=#2842 id=2842 data-nosnippet>2842</a>    /// This test exposes the inverse scenario of `test_position_delete_with_row_group_selection`:
-<a href=#2843 id=2843 data-nosnippet>2843</a>    /// - Position delete targets a row in the SKIPPED row group (not the selected one)
-<a href=#2844 id=2844 data-nosnippet>2844</a>    /// - After calling advance_to(), the cached delete index is stale
-<a href=#2845 id=2845 data-nosnippet>2845</a>    /// - Without updating the cache, the code enters an infinite loop
-<a href=#2846 id=2846 data-nosnippet>2846</a>    ///
-<a href=#2847 id=2847 data-nosnippet>2847</a>    /// This test creates:
-<a href=#2848 id=2848 data-nosnippet>2848</a>    /// - A data file with 200 rows split into 2 row groups (0-99, 100-199)
-<a href=#2849 id=2849 data-nosnippet>2849</a>    /// - A position delete file that deletes row 0 (first row in SKIPPED row group 0)
-<a href=#2850 id=2850 data-nosnippet>2850</a>    /// - Row group selection that reads ONLY row group 1 (rows 100-199)
-<a href=#2851 id=2851 data-nosnippet>2851</a>    ///
-<a href=#2852 id=2852 data-nosnippet>2852</a>    /// The bug occurs when skipping row group 0:
-<a href=#2853 id=2853 data-nosnippet>2853</a>    /// ```rust
-<a href=#2854 id=2854 data-nosnippet>2854</a>    /// let mut next_deleted_row_idx_opt = delete_vector_iter.next(); // Some(0)
-<a href=#2855 id=2855 data-nosnippet>2855</a>    /// // ... skip to row group 1 ...
-<a href=#2856 id=2856 data-nosnippet>2856</a>    /// delete_vector_iter.advance_to(100); // Iterator advances past delete at 0
-<a href=#2857 id=2857 data-nosnippet>2857</a>    /// // BUG: next_deleted_row_idx_opt is still Some(0) - STALE!
-<a href=#2858 id=2858 data-nosnippet>2858</a>    /// // When processing row group 1:
-<a href=#2859 id=2859 data-nosnippet>2859</a>    /// //   current_idx = 100, next_deleted_row_idx = 0, next_row_group_base_idx = 200
-<a href=#2860 id=2860 data-nosnippet>2860</a>    /// //   Loop condition: 0 &lt; 200 (true)
-<a href=#2861 id=2861 data-nosnippet>2861</a>    /// //   But: current_idx (100) &gt; next_deleted_row_idx (0)
-<a href=#2862 id=2862 data-nosnippet>2862</a>    /// //   And: current_idx (100) != next_deleted_row_idx (0)
-<a href=#2863 id=2863 data-nosnippet>2863</a>    /// //   Neither branch executes -&gt; INFINITE LOOP!
-<a href=#2864 id=2864 data-nosnippet>2864</a>    /// ```
-<a href=#2865 id=2865 data-nosnippet>2865</a>    ///
-<a href=#2866 id=2866 data-nosnippet>2866</a>    /// Expected behavior: Should return 100 rows (delete at 0 doesn't affect row group 1)
-<a href=#2867 id=2867 data-nosnippet>2867</a>    /// Bug behavior: Infinite loop in build_deletes_row_selection
-<a href=#2868 id=2868 data-nosnippet>2868</a>    </span><span class="attr">#[tokio::test]
-<a href=#2869 id=2869 data-nosnippet>2869</a>    </span><span class="kw">async fn </span>test_position_delete_in_skipped_row_group() {
-<a href=#2870 id=2870 data-nosnippet>2870</a>        <span class="kw">use </span>arrow_array::{Int32Array, Int64Array};
-<a href=#2871 id=2871 data-nosnippet>2871</a>        <span class="kw">use </span>parquet::file::reader::{FileReader, SerializedFileReader};
-<a href=#2872 id=2872 data-nosnippet>2872</a>
-<a href=#2873 id=2873 data-nosnippet>2873</a>        <span class="comment">// Field IDs for positional delete schema
-<a href=#2874 id=2874 data-nosnippet>2874</a>        </span><span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_FILE_PATH: u64 = <span class="number">2147483546</span>;
-<a href=#2875 id=2875 data-nosnippet>2875</a>        <span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_POS: u64 = <span class="number">2147483545</span>;
-<a href=#2876 id=2876 data-nosnippet>2876</a>
-<a href=#2877 id=2877 data-nosnippet>2877</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#2878 id=2878 data-nosnippet>2878</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#2828 id=2828 data-nosnippet>2828</a>        <span class="kw">let </span>file = File::create(<span class="kw-2">&amp;</span>data_file_path).unwrap();
+<a href=#2829 id=2829 data-nosnippet>2829</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#2830 id=2830 data-nosnippet>2830</a>        writer.write(<span class="kw-2">&amp;</span>batch1).expect(<span class="string">"Writing batch 1"</span>);
+<a href=#2831 id=2831 data-nosnippet>2831</a>        writer.write(<span class="kw-2">&amp;</span>batch2).expect(<span class="string">"Writing batch 2"</span>);
+<a href=#2832 id=2832 data-nosnippet>2832</a>        writer.close().unwrap();
+<a href=#2833 id=2833 data-nosnippet>2833</a>
+<a href=#2834 id=2834 data-nosnippet>2834</a>        <span class="comment">// Verify we created 2 row groups
+<a href=#2835 id=2835 data-nosnippet>2835</a>        </span><span class="kw">let </span>verify_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
+<a href=#2836 id=2836 data-nosnippet>2836</a>        <span class="kw">let </span>verify_reader = SerializedFileReader::new(verify_file).unwrap();
+<a href=#2837 id=2837 data-nosnippet>2837</a>        <span class="macro">assert_eq!</span>(
+<a href=#2838 id=2838 data-nosnippet>2838</a>            verify_reader.metadata().num_row_groups(),
+<a href=#2839 id=2839 data-nosnippet>2839</a>            <span class="number">2</span>,
+<a href=#2840 id=2840 data-nosnippet>2840</a>            <span class="string">"Should have 2 row groups"
+<a href=#2841 id=2841 data-nosnippet>2841</a>        </span>);
+<a href=#2842 id=2842 data-nosnippet>2842</a>
+<a href=#2843 id=2843 data-nosnippet>2843</a>        <span class="comment">// Step 2: Create position delete file that deletes row 199 (id=200, last row in row group 1)
+<a href=#2844 id=2844 data-nosnippet>2844</a>        </span><span class="kw">let </span>delete_file_path = <span class="macro">format!</span>(<span class="string">"{}/deletes.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#2845 id=2845 data-nosnippet>2845</a>
+<a href=#2846 id=2846 data-nosnippet>2846</a>        <span class="kw">let </span>delete_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#2847 id=2847 data-nosnippet>2847</a>            Field::new(<span class="string">"file_path"</span>, DataType::Utf8, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#2848 id=2848 data-nosnippet>2848</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2849 id=2849 data-nosnippet>2849</a>                FIELD_ID_POSITIONAL_DELETE_FILE_PATH.to_string(),
+<a href=#2850 id=2850 data-nosnippet>2850</a>            )])),
+<a href=#2851 id=2851 data-nosnippet>2851</a>            Field::new(<span class="string">"pos"</span>, DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#2852 id=2852 data-nosnippet>2852</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#2853 id=2853 data-nosnippet>2853</a>                FIELD_ID_POSITIONAL_DELETE_POS.to_string(),
+<a href=#2854 id=2854 data-nosnippet>2854</a>            )])),
+<a href=#2855 id=2855 data-nosnippet>2855</a>        ]));
+<a href=#2856 id=2856 data-nosnippet>2856</a>
+<a href=#2857 id=2857 data-nosnippet>2857</a>        <span class="comment">// Delete row at position 199 (0-indexed, so it's the last row: id=200)
+<a href=#2858 id=2858 data-nosnippet>2858</a>        </span><span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[
+<a href=#2859 id=2859 data-nosnippet>2859</a>            Arc::new(StringArray::from_iter_values(<span class="macro">vec!</span>[data_file_path.clone()])),
+<a href=#2860 id=2860 data-nosnippet>2860</a>            Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[<span class="number">199i64</span>])),
+<a href=#2861 id=2861 data-nosnippet>2861</a>        ])
+<a href=#2862 id=2862 data-nosnippet>2862</a>        .unwrap();
+<a href=#2863 id=2863 data-nosnippet>2863</a>
+<a href=#2864 id=2864 data-nosnippet>2864</a>        <span class="kw">let </span>delete_props = WriterProperties::builder()
+<a href=#2865 id=2865 data-nosnippet>2865</a>            .set_compression(Compression::SNAPPY)
+<a href=#2866 id=2866 data-nosnippet>2866</a>            .build();
+<a href=#2867 id=2867 data-nosnippet>2867</a>
+<a href=#2868 id=2868 data-nosnippet>2868</a>        <span class="kw">let </span>delete_file = File::create(<span class="kw-2">&amp;</span>delete_file_path).unwrap();
+<a href=#2869 id=2869 data-nosnippet>2869</a>        <span class="kw">let </span><span class="kw-2">mut </span>delete_writer =
+<a href=#2870 id=2870 data-nosnippet>2870</a>            ArrowWriter::try_new(delete_file, delete_schema, <span class="prelude-val">Some</span>(delete_props)).unwrap();
+<a href=#2871 id=2871 data-nosnippet>2871</a>        delete_writer.write(<span class="kw-2">&amp;</span>delete_batch).unwrap();
+<a href=#2872 id=2872 data-nosnippet>2872</a>        delete_writer.close().unwrap();
+<a href=#2873 id=2873 data-nosnippet>2873</a>
+<a href=#2874 id=2874 data-nosnippet>2874</a>        <span class="comment">// Step 3: Get byte ranges to read ONLY row group 1 (rows 100-199)
+<a href=#2875 id=2875 data-nosnippet>2875</a>        // This exercises the row group selection code path where row group 0 is skipped
+<a href=#2876 id=2876 data-nosnippet>2876</a>        </span><span class="kw">let </span>metadata_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
+<a href=#2877 id=2877 data-nosnippet>2877</a>        <span class="kw">let </span>metadata_reader = SerializedFileReader::new(metadata_file).unwrap();
+<a href=#2878 id=2878 data-nosnippet>2878</a>        <span class="kw">let </span>metadata = metadata_reader.metadata();
 <a href=#2879 id=2879 data-nosnippet>2879</a>
-<a href=#2880 id=2880 data-nosnippet>2880</a>        <span class="comment">// Create table schema with a single 'id' column
-<a href=#2881 id=2881 data-nosnippet>2881</a>        </span><span class="kw">let </span>table_schema = Arc::new(
-<a href=#2882 id=2882 data-nosnippet>2882</a>            Schema::builder()
-<a href=#2883 id=2883 data-nosnippet>2883</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#2884 id=2884 data-nosnippet>2884</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#2885 id=2885 data-nosnippet>2885</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#2886 id=2886 data-nosnippet>2886</a>                ])
-<a href=#2887 id=2887 data-nosnippet>2887</a>                .build()
-<a href=#2888 id=2888 data-nosnippet>2888</a>                .unwrap(),
-<a href=#2889 id=2889 data-nosnippet>2889</a>        );
-<a href=#2890 id=2890 data-nosnippet>2890</a>
-<a href=#2891 id=2891 data-nosnippet>2891</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#2892 id=2892 data-nosnippet>2892</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2893 id=2893 data-nosnippet>2893</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2894 id=2894 data-nosnippet>2894</a>                <span class="string">"1"</span>.to_string(),
-<a href=#2895 id=2895 data-nosnippet>2895</a>            )])),
-<a href=#2896 id=2896 data-nosnippet>2896</a>        ]));
+<a href=#2880 id=2880 data-nosnippet>2880</a>        <span class="kw">let </span>row_group_0 = metadata.row_group(<span class="number">0</span>);
+<a href=#2881 id=2881 data-nosnippet>2881</a>        <span class="kw">let </span>row_group_1 = metadata.row_group(<span class="number">1</span>);
+<a href=#2882 id=2882 data-nosnippet>2882</a>
+<a href=#2883 id=2883 data-nosnippet>2883</a>        <span class="kw">let </span>rg0_start = <span class="number">4u64</span>; <span class="comment">// Parquet files start with 4-byte magic "PAR1"
+<a href=#2884 id=2884 data-nosnippet>2884</a>        </span><span class="kw">let </span>rg1_start = rg0_start + row_group_0.compressed_size() <span class="kw">as </span>u64;
+<a href=#2885 id=2885 data-nosnippet>2885</a>        <span class="kw">let </span>rg1_length = row_group_1.compressed_size() <span class="kw">as </span>u64;
+<a href=#2886 id=2886 data-nosnippet>2886</a>
+<a href=#2887 id=2887 data-nosnippet>2887</a>        <span class="macro">println!</span>(
+<a href=#2888 id=2888 data-nosnippet>2888</a>            <span class="string">"Row group 0: starts at byte {}, {} bytes compressed"</span>,
+<a href=#2889 id=2889 data-nosnippet>2889</a>            rg0_start,
+<a href=#2890 id=2890 data-nosnippet>2890</a>            row_group_0.compressed_size()
+<a href=#2891 id=2891 data-nosnippet>2891</a>        );
+<a href=#2892 id=2892 data-nosnippet>2892</a>        <span class="macro">println!</span>(
+<a href=#2893 id=2893 data-nosnippet>2893</a>            <span class="string">"Row group 1: starts at byte {}, {} bytes compressed"</span>,
+<a href=#2894 id=2894 data-nosnippet>2894</a>            rg1_start,
+<a href=#2895 id=2895 data-nosnippet>2895</a>            row_group_1.compressed_size()
+<a href=#2896 id=2896 data-nosnippet>2896</a>        );
 <a href=#2897 id=2897 data-nosnippet>2897</a>
-<a href=#2898 id=2898 data-nosnippet>2898</a>        <span class="comment">// Step 1: Create data file with 200 rows in 2 row groups
-<a href=#2899 id=2899 data-nosnippet>2899</a>        // Row group 0: rows 0-99 (ids 1-100)
-<a href=#2900 id=2900 data-nosnippet>2900</a>        // Row group 1: rows 100-199 (ids 101-200)
-<a href=#2901 id=2901 data-nosnippet>2901</a>        </span><span class="kw">let </span>data_file_path = <span class="macro">format!</span>(<span class="string">"{}/data.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
-<a href=#2902 id=2902 data-nosnippet>2902</a>
-<a href=#2903 id=2903 data-nosnippet>2903</a>        <span class="kw">let </span>batch1 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
-<a href=#2904 id=2904 data-nosnippet>2904</a>            Int32Array::from_iter_values(<span class="number">1</span>..=<span class="number">100</span>),
-<a href=#2905 id=2905 data-nosnippet>2905</a>        )])
-<a href=#2906 id=2906 data-nosnippet>2906</a>        .unwrap();
-<a href=#2907 id=2907 data-nosnippet>2907</a>
-<a href=#2908 id=2908 data-nosnippet>2908</a>        <span class="kw">let </span>batch2 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
-<a href=#2909 id=2909 data-nosnippet>2909</a>            Int32Array::from_iter_values(<span class="number">101</span>..=<span class="number">200</span>),
-<a href=#2910 id=2910 data-nosnippet>2910</a>        )])
-<a href=#2911 id=2911 data-nosnippet>2911</a>        .unwrap();
-<a href=#2912 id=2912 data-nosnippet>2912</a>
-<a href=#2913 id=2913 data-nosnippet>2913</a>        <span class="comment">// Force each batch into its own row group
-<a href=#2914 id=2914 data-nosnippet>2914</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
-<a href=#2915 id=2915 data-nosnippet>2915</a>            .set_compression(Compression::SNAPPY)
-<a href=#2916 id=2916 data-nosnippet>2916</a>            .set_max_row_group_size(<span class="number">100</span>)
-<a href=#2917 id=2917 data-nosnippet>2917</a>            .build();
-<a href=#2918 id=2918 data-nosnippet>2918</a>
-<a href=#2919 id=2919 data-nosnippet>2919</a>        <span class="kw">let </span>file = File::create(<span class="kw-2">&amp;</span>data_file_path).unwrap();
-<a href=#2920 id=2920 data-nosnippet>2920</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#2921 id=2921 data-nosnippet>2921</a>        writer.write(<span class="kw-2">&amp;</span>batch1).expect(<span class="string">"Writing batch 1"</span>);
-<a href=#2922 id=2922 data-nosnippet>2922</a>        writer.write(<span class="kw-2">&amp;</span>batch2).expect(<span class="string">"Writing batch 2"</span>);
-<a href=#2923 id=2923 data-nosnippet>2923</a>        writer.close().unwrap();
-<a href=#2924 id=2924 data-nosnippet>2924</a>
-<a href=#2925 id=2925 data-nosnippet>2925</a>        <span class="comment">// Verify we created 2 row groups
-<a href=#2926 id=2926 data-nosnippet>2926</a>        </span><span class="kw">let </span>verify_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
-<a href=#2927 id=2927 data-nosnippet>2927</a>        <span class="kw">let </span>verify_reader = SerializedFileReader::new(verify_file).unwrap();
-<a href=#2928 id=2928 data-nosnippet>2928</a>        <span class="macro">assert_eq!</span>(
-<a href=#2929 id=2929 data-nosnippet>2929</a>            verify_reader.metadata().num_row_groups(),
-<a href=#2930 id=2930 data-nosnippet>2930</a>            <span class="number">2</span>,
-<a href=#2931 id=2931 data-nosnippet>2931</a>            <span class="string">"Should have 2 row groups"
-<a href=#2932 id=2932 data-nosnippet>2932</a>        </span>);
+<a href=#2898 id=2898 data-nosnippet>2898</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#2899 id=2899 data-nosnippet>2899</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#2900 id=2900 data-nosnippet>2900</a>
+<a href=#2901 id=2901 data-nosnippet>2901</a>        <span class="comment">// Create FileScanTask that reads ONLY row group 1 via byte range filtering
+<a href=#2902 id=2902 data-nosnippet>2902</a>        </span><span class="kw">let </span>task = FileScanTask {
+<a href=#2903 id=2903 data-nosnippet>2903</a>            start: rg1_start,
+<a href=#2904 id=2904 data-nosnippet>2904</a>            length: rg1_length,
+<a href=#2905 id=2905 data-nosnippet>2905</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">100</span>), <span class="comment">// Row group 1 has 100 rows
+<a href=#2906 id=2906 data-nosnippet>2906</a>            </span>data_file_path: data_file_path.clone(),
+<a href=#2907 id=2907 data-nosnippet>2907</a>            data_file_format: DataFileFormat::Parquet,
+<a href=#2908 id=2908 data-nosnippet>2908</a>            schema: table_schema.clone(),
+<a href=#2909 id=2909 data-nosnippet>2909</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
+<a href=#2910 id=2910 data-nosnippet>2910</a>            predicate: <span class="prelude-val">None</span>,
+<a href=#2911 id=2911 data-nosnippet>2911</a>            deletes: <span class="macro">vec!</span>[FileScanTaskDeleteFile {
+<a href=#2912 id=2912 data-nosnippet>2912</a>                file_path: delete_file_path,
+<a href=#2913 id=2913 data-nosnippet>2913</a>                file_type: DataContentType::PositionDeletes,
+<a href=#2914 id=2914 data-nosnippet>2914</a>                partition_spec_id: <span class="number">0</span>,
+<a href=#2915 id=2915 data-nosnippet>2915</a>                equality_ids: <span class="prelude-val">None</span>,
+<a href=#2916 id=2916 data-nosnippet>2916</a>            }],
+<a href=#2917 id=2917 data-nosnippet>2917</a>            partition: <span class="prelude-val">None</span>,
+<a href=#2918 id=2918 data-nosnippet>2918</a>            partition_spec: <span class="prelude-val">None</span>,
+<a href=#2919 id=2919 data-nosnippet>2919</a>            name_mapping: <span class="prelude-val">None</span>,
+<a href=#2920 id=2920 data-nosnippet>2920</a>        };
+<a href=#2921 id=2921 data-nosnippet>2921</a>
+<a href=#2922 id=2922 data-nosnippet>2922</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task)])) <span class="kw">as </span>FileScanTaskStream;
+<a href=#2923 id=2923 data-nosnippet>2923</a>        <span class="kw">let </span>result = reader
+<a href=#2924 id=2924 data-nosnippet>2924</a>            .read(tasks)
+<a href=#2925 id=2925 data-nosnippet>2925</a>            .unwrap()
+<a href=#2926 id=2926 data-nosnippet>2926</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#2927 id=2927 data-nosnippet>2927</a>            .<span class="kw">await
+<a href=#2928 id=2928 data-nosnippet>2928</a>            </span>.unwrap();
+<a href=#2929 id=2929 data-nosnippet>2929</a>
+<a href=#2930 id=2930 data-nosnippet>2930</a>        <span class="comment">// Step 4: Verify we got 99 rows (not 100)
+<a href=#2931 id=2931 data-nosnippet>2931</a>        // Row group 1 has 100 rows (ids 101-200), minus 1 delete (id=200) = 99 rows
+<a href=#2932 id=2932 data-nosnippet>2932</a>        </span><span class="kw">let </span>total_rows: usize = result.iter().map(|b| b.num_rows()).sum();
 <a href=#2933 id=2933 data-nosnippet>2933</a>
-<a href=#2934 id=2934 data-nosnippet>2934</a>        <span class="comment">// Step 2: Create position delete file that deletes row 0 (id=1, first row in row group 0)
-<a href=#2935 id=2935 data-nosnippet>2935</a>        </span><span class="kw">let </span>delete_file_path = <span class="macro">format!</span>(<span class="string">"{}/deletes.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#2934 id=2934 data-nosnippet>2934</a>        <span class="macro">println!</span>(<span class="string">"Total rows read from row group 1: {}"</span>, total_rows);
+<a href=#2935 id=2935 data-nosnippet>2935</a>        <span class="macro">println!</span>(<span class="string">"Expected: 99 rows (row group 1 has 100 rows, 1 delete at position 199)"</span>);
 <a href=#2936 id=2936 data-nosnippet>2936</a>
-<a href=#2937 id=2937 data-nosnippet>2937</a>        <span class="kw">let </span>delete_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#2938 id=2938 data-nosnippet>2938</a>            Field::new(<span class="string">"file_path"</span>, DataType::Utf8, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2939 id=2939 data-nosnippet>2939</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2940 id=2940 data-nosnippet>2940</a>                FIELD_ID_POSITIONAL_DELETE_FILE_PATH.to_string(),
-<a href=#2941 id=2941 data-nosnippet>2941</a>            )])),
-<a href=#2942 id=2942 data-nosnippet>2942</a>            Field::new(<span class="string">"pos"</span>, DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#2943 id=2943 data-nosnippet>2943</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#2944 id=2944 data-nosnippet>2944</a>                FIELD_ID_POSITIONAL_DELETE_POS.to_string(),
-<a href=#2945 id=2945 data-nosnippet>2945</a>            )])),
-<a href=#2946 id=2946 data-nosnippet>2946</a>        ]));
-<a href=#2947 id=2947 data-nosnippet>2947</a>
-<a href=#2948 id=2948 data-nosnippet>2948</a>        <span class="comment">// Delete row at position 0 (0-indexed, so it's the first row: id=1)
-<a href=#2949 id=2949 data-nosnippet>2949</a>        </span><span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[
-<a href=#2950 id=2950 data-nosnippet>2950</a>            Arc::new(StringArray::from_iter_values(<span class="macro">vec!</span>[data_file_path.clone()])),
-<a href=#2951 id=2951 data-nosnippet>2951</a>            Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[<span class="number">0i64</span>])),
-<a href=#2952 id=2952 data-nosnippet>2952</a>        ])
-<a href=#2953 id=2953 data-nosnippet>2953</a>        .unwrap();
-<a href=#2954 id=2954 data-nosnippet>2954</a>
-<a href=#2955 id=2955 data-nosnippet>2955</a>        <span class="kw">let </span>delete_props = WriterProperties::builder()
-<a href=#2956 id=2956 data-nosnippet>2956</a>            .set_compression(Compression::SNAPPY)
-<a href=#2957 id=2957 data-nosnippet>2957</a>            .build();
+<a href=#2937 id=2937 data-nosnippet>2937</a>        <span class="comment">// This assertion will FAIL before the fix and PASS after the fix
+<a href=#2938 id=2938 data-nosnippet>2938</a>        </span><span class="macro">assert_eq!</span>(
+<a href=#2939 id=2939 data-nosnippet>2939</a>            total_rows, <span class="number">99</span>,
+<a href=#2940 id=2940 data-nosnippet>2940</a>            <span class="string">"Expected 99 rows from row group 1 after deleting position 199, but got {} rows. \
+<a href=#2941 id=2941 data-nosnippet>2941</a>             The bug causes position deletes to be lost when advance_to() is followed by next() \
+<a href=#2942 id=2942 data-nosnippet>2942</a>             when skipping unselected row groups."</span>,
+<a href=#2943 id=2943 data-nosnippet>2943</a>            total_rows
+<a href=#2944 id=2944 data-nosnippet>2944</a>        );
+<a href=#2945 id=2945 data-nosnippet>2945</a>
+<a href=#2946 id=2946 data-nosnippet>2946</a>        <span class="comment">// Verify the deleted row (id=200) is not present
+<a href=#2947 id=2947 data-nosnippet>2947</a>        </span><span class="kw">let </span>all_ids: Vec&lt;i32&gt; = result
+<a href=#2948 id=2948 data-nosnippet>2948</a>            .iter()
+<a href=#2949 id=2949 data-nosnippet>2949</a>            .flat_map(|batch| {
+<a href=#2950 id=2950 data-nosnippet>2950</a>                batch
+<a href=#2951 id=2951 data-nosnippet>2951</a>                    .column(<span class="number">0</span>)
+<a href=#2952 id=2952 data-nosnippet>2952</a>                    .as_primitive::&lt;arrow_array::types::Int32Type&gt;()
+<a href=#2953 id=2953 data-nosnippet>2953</a>                    .values()
+<a href=#2954 id=2954 data-nosnippet>2954</a>                    .iter()
+<a href=#2955 id=2955 data-nosnippet>2955</a>                    .copied()
+<a href=#2956 id=2956 data-nosnippet>2956</a>            })
+<a href=#2957 id=2957 data-nosnippet>2957</a>            .collect();
 <a href=#2958 id=2958 data-nosnippet>2958</a>
-<a href=#2959 id=2959 data-nosnippet>2959</a>        <span class="kw">let </span>delete_file = File::create(<span class="kw-2">&amp;</span>delete_file_path).unwrap();
-<a href=#2960 id=2960 data-nosnippet>2960</a>        <span class="kw">let </span><span class="kw-2">mut </span>delete_writer =
-<a href=#2961 id=2961 data-nosnippet>2961</a>            ArrowWriter::try_new(delete_file, delete_schema, <span class="prelude-val">Some</span>(delete_props)).unwrap();
-<a href=#2962 id=2962 data-nosnippet>2962</a>        delete_writer.write(<span class="kw-2">&amp;</span>delete_batch).unwrap();
-<a href=#2963 id=2963 data-nosnippet>2963</a>        delete_writer.close().unwrap();
-<a href=#2964 id=2964 data-nosnippet>2964</a>
-<a href=#2965 id=2965 data-nosnippet>2965</a>        <span class="comment">// Step 3: Get byte ranges to read ONLY row group 1 (rows 100-199)
-<a href=#2966 id=2966 data-nosnippet>2966</a>        // This exercises the row group selection code path where row group 0 is skipped
-<a href=#2967 id=2967 data-nosnippet>2967</a>        </span><span class="kw">let </span>metadata_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
-<a href=#2968 id=2968 data-nosnippet>2968</a>        <span class="kw">let </span>metadata_reader = SerializedFileReader::new(metadata_file).unwrap();
-<a href=#2969 id=2969 data-nosnippet>2969</a>        <span class="kw">let </span>metadata = metadata_reader.metadata();
-<a href=#2970 id=2970 data-nosnippet>2970</a>
-<a href=#2971 id=2971 data-nosnippet>2971</a>        <span class="kw">let </span>row_group_0 = metadata.row_group(<span class="number">0</span>);
-<a href=#2972 id=2972 data-nosnippet>2972</a>        <span class="kw">let </span>row_group_1 = metadata.row_group(<span class="number">1</span>);
-<a href=#2973 id=2973 data-nosnippet>2973</a>
-<a href=#2974 id=2974 data-nosnippet>2974</a>        <span class="kw">let </span>rg0_start = <span class="number">4u64</span>; <span class="comment">// Parquet files start with 4-byte magic "PAR1"
-<a href=#2975 id=2975 data-nosnippet>2975</a>        </span><span class="kw">let </span>rg1_start = rg0_start + row_group_0.compressed_size() <span class="kw">as </span>u64;
-<a href=#2976 id=2976 data-nosnippet>2976</a>        <span class="kw">let </span>rg1_length = row_group_1.compressed_size() <span class="kw">as </span>u64;
-<a href=#2977 id=2977 data-nosnippet>2977</a>
-<a href=#2978 id=2978 data-nosnippet>2978</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#2979 id=2979 data-nosnippet>2979</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#2980 id=2980 data-nosnippet>2980</a>
-<a href=#2981 id=2981 data-nosnippet>2981</a>        <span class="comment">// Create FileScanTask that reads ONLY row group 1 via byte range filtering
-<a href=#2982 id=2982 data-nosnippet>2982</a>        </span><span class="kw">let </span>task = FileScanTask {
-<a href=#2983 id=2983 data-nosnippet>2983</a>            start: rg1_start,
-<a href=#2984 id=2984 data-nosnippet>2984</a>            length: rg1_length,
-<a href=#2985 id=2985 data-nosnippet>2985</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">100</span>), <span class="comment">// Row group 1 has 100 rows
-<a href=#2986 id=2986 data-nosnippet>2986</a>            </span>data_file_path: data_file_path.clone(),
-<a href=#2987 id=2987 data-nosnippet>2987</a>            data_file_format: DataFileFormat::Parquet,
-<a href=#2988 id=2988 data-nosnippet>2988</a>            schema: table_schema.clone(),
-<a href=#2989 id=2989 data-nosnippet>2989</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
-<a href=#2990 id=2990 data-nosnippet>2990</a>            predicate: <span class="prelude-val">None</span>,
-<a href=#2991 id=2991 data-nosnippet>2991</a>            deletes: <span class="macro">vec!</span>[FileScanTaskDeleteFile {
-<a href=#2992 id=2992 data-nosnippet>2992</a>                file_path: delete_file_path,
-<a href=#2993 id=2993 data-nosnippet>2993</a>                file_type: DataContentType::PositionDeletes,
-<a href=#2994 id=2994 data-nosnippet>2994</a>                partition_spec_id: <span class="number">0</span>,
-<a href=#2995 id=2995 data-nosnippet>2995</a>                equality_ids: <span class="prelude-val">None</span>,
-<a href=#2996 id=2996 data-nosnippet>2996</a>            }],
-<a href=#2997 id=2997 data-nosnippet>2997</a>        };
-<a href=#2998 id=2998 data-nosnippet>2998</a>
-<a href=#2999 id=2999 data-nosnippet>2999</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task)])) <span class="kw">as </span>FileScanTaskStream;
-<a href=#3000 id=3000 data-nosnippet>3000</a>        <span class="kw">let </span>result = reader
-<a href=#3001 id=3001 data-nosnippet>3001</a>            .read(tasks)
-<a href=#3002 id=3002 data-nosnippet>3002</a>            .unwrap()
-<a href=#3003 id=3003 data-nosnippet>3003</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#3004 id=3004 data-nosnippet>3004</a>            .<span class="kw">await
-<a href=#3005 id=3005 data-nosnippet>3005</a>            </span>.unwrap();
-<a href=#3006 id=3006 data-nosnippet>3006</a>
-<a href=#3007 id=3007 data-nosnippet>3007</a>        <span class="comment">// Step 4: Verify we got 100 rows (all of row group 1)
-<a href=#3008 id=3008 data-nosnippet>3008</a>        // The delete at position 0 is in row group 0, which is skipped, so it doesn't affect us
-<a href=#3009 id=3009 data-nosnippet>3009</a>        </span><span class="kw">let </span>total_rows: usize = result.iter().map(|b| b.num_rows()).sum();
+<a href=#2959 id=2959 data-nosnippet>2959</a>        <span class="macro">assert!</span>(
+<a href=#2960 id=2960 data-nosnippet>2960</a>            !all_ids.contains(<span class="kw-2">&amp;</span><span class="number">200</span>),
+<a href=#2961 id=2961 data-nosnippet>2961</a>            <span class="string">"Row with id=200 should be deleted but was found in results"
+<a href=#2962 id=2962 data-nosnippet>2962</a>        </span>);
+<a href=#2963 id=2963 data-nosnippet>2963</a>
+<a href=#2964 id=2964 data-nosnippet>2964</a>        <span class="comment">// Verify we have ids 101-199 (not 101-200)
+<a href=#2965 id=2965 data-nosnippet>2965</a>        </span><span class="kw">let </span>expected_ids: Vec&lt;i32&gt; = (<span class="number">101</span>..=<span class="number">199</span>).collect();
+<a href=#2966 id=2966 data-nosnippet>2966</a>        <span class="macro">assert_eq!</span>(
+<a href=#2967 id=2967 data-nosnippet>2967</a>            all_ids, expected_ids,
+<a href=#2968 id=2968 data-nosnippet>2968</a>            <span class="string">"Should have ids 101-199 but got different values"
+<a href=#2969 id=2969 data-nosnippet>2969</a>        </span>);
+<a href=#2970 id=2970 data-nosnippet>2970</a>    }
+<a href=#2971 id=2971 data-nosnippet>2971</a>    <span class="doccomment">/// Test for bug where stale cached delete causes infinite loop when skipping row groups.
+<a href=#2972 id=2972 data-nosnippet>2972</a>    ///
+<a href=#2973 id=2973 data-nosnippet>2973</a>    /// This test exposes the inverse scenario of `test_position_delete_with_row_group_selection`:
+<a href=#2974 id=2974 data-nosnippet>2974</a>    /// - Position delete targets a row in the SKIPPED row group (not the selected one)
+<a href=#2975 id=2975 data-nosnippet>2975</a>    /// - After calling advance_to(), the cached delete index is stale
+<a href=#2976 id=2976 data-nosnippet>2976</a>    /// - Without updating the cache, the code enters an infinite loop
+<a href=#2977 id=2977 data-nosnippet>2977</a>    ///
+<a href=#2978 id=2978 data-nosnippet>2978</a>    /// This test creates:
+<a href=#2979 id=2979 data-nosnippet>2979</a>    /// - A data file with 200 rows split into 2 row groups (0-99, 100-199)
+<a href=#2980 id=2980 data-nosnippet>2980</a>    /// - A position delete file that deletes row 0 (first row in SKIPPED row group 0)
+<a href=#2981 id=2981 data-nosnippet>2981</a>    /// - Row group selection that reads ONLY row group 1 (rows 100-199)
+<a href=#2982 id=2982 data-nosnippet>2982</a>    ///
+<a href=#2983 id=2983 data-nosnippet>2983</a>    /// The bug occurs when skipping row group 0:
+<a href=#2984 id=2984 data-nosnippet>2984</a>    /// ```rust
+<a href=#2985 id=2985 data-nosnippet>2985</a>    /// let mut next_deleted_row_idx_opt = delete_vector_iter.next(); // Some(0)
+<a href=#2986 id=2986 data-nosnippet>2986</a>    /// // ... skip to row group 1 ...
+<a href=#2987 id=2987 data-nosnippet>2987</a>    /// delete_vector_iter.advance_to(100); // Iterator advances past delete at 0
+<a href=#2988 id=2988 data-nosnippet>2988</a>    /// // BUG: next_deleted_row_idx_opt is still Some(0) - STALE!
+<a href=#2989 id=2989 data-nosnippet>2989</a>    /// // When processing row group 1:
+<a href=#2990 id=2990 data-nosnippet>2990</a>    /// //   current_idx = 100, next_deleted_row_idx = 0, next_row_group_base_idx = 200
+<a href=#2991 id=2991 data-nosnippet>2991</a>    /// //   Loop condition: 0 &lt; 200 (true)
+<a href=#2992 id=2992 data-nosnippet>2992</a>    /// //   But: current_idx (100) &gt; next_deleted_row_idx (0)
+<a href=#2993 id=2993 data-nosnippet>2993</a>    /// //   And: current_idx (100) != next_deleted_row_idx (0)
+<a href=#2994 id=2994 data-nosnippet>2994</a>    /// //   Neither branch executes -&gt; INFINITE LOOP!
+<a href=#2995 id=2995 data-nosnippet>2995</a>    /// ```
+<a href=#2996 id=2996 data-nosnippet>2996</a>    ///
+<a href=#2997 id=2997 data-nosnippet>2997</a>    /// Expected behavior: Should return 100 rows (delete at 0 doesn't affect row group 1)
+<a href=#2998 id=2998 data-nosnippet>2998</a>    /// Bug behavior: Infinite loop in build_deletes_row_selection
+<a href=#2999 id=2999 data-nosnippet>2999</a>    </span><span class="attr">#[tokio::test]
+<a href=#3000 id=3000 data-nosnippet>3000</a>    </span><span class="kw">async fn </span>test_position_delete_in_skipped_row_group() {
+<a href=#3001 id=3001 data-nosnippet>3001</a>        <span class="kw">use </span>arrow_array::{Int32Array, Int64Array};
+<a href=#3002 id=3002 data-nosnippet>3002</a>        <span class="kw">use </span>parquet::file::reader::{FileReader, SerializedFileReader};
+<a href=#3003 id=3003 data-nosnippet>3003</a>
+<a href=#3004 id=3004 data-nosnippet>3004</a>        <span class="comment">// Field IDs for positional delete schema
+<a href=#3005 id=3005 data-nosnippet>3005</a>        </span><span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_FILE_PATH: u64 = <span class="number">2147483546</span>;
+<a href=#3006 id=3006 data-nosnippet>3006</a>        <span class="kw">const </span>FIELD_ID_POSITIONAL_DELETE_POS: u64 = <span class="number">2147483545</span>;
+<a href=#3007 id=3007 data-nosnippet>3007</a>
+<a href=#3008 id=3008 data-nosnippet>3008</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3009 id=3009 data-nosnippet>3009</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
 <a href=#3010 id=3010 data-nosnippet>3010</a>
-<a href=#3011 id=3011 data-nosnippet>3011</a>        <span class="macro">assert_eq!</span>(
-<a href=#3012 id=3012 data-nosnippet>3012</a>            total_rows, <span class="number">100</span>,
-<a href=#3013 id=3013 data-nosnippet>3013</a>            <span class="string">"Expected 100 rows from row group 1 (delete at position 0 is in skipped row group 0). \
-<a href=#3014 id=3014 data-nosnippet>3014</a>             If this hangs or fails, it indicates the cached delete index was not updated after advance_to()."
-<a href=#3015 id=3015 data-nosnippet>3015</a>        </span>);
-<a href=#3016 id=3016 data-nosnippet>3016</a>
-<a href=#3017 id=3017 data-nosnippet>3017</a>        <span class="comment">// Verify we have all ids from row group 1 (101-200)
-<a href=#3018 id=3018 data-nosnippet>3018</a>        </span><span class="kw">let </span>all_ids: Vec&lt;i32&gt; = result
-<a href=#3019 id=3019 data-nosnippet>3019</a>            .iter()
-<a href=#3020 id=3020 data-nosnippet>3020</a>            .flat_map(|batch| {
-<a href=#3021 id=3021 data-nosnippet>3021</a>                batch
-<a href=#3022 id=3022 data-nosnippet>3022</a>                    .column(<span class="number">0</span>)
-<a href=#3023 id=3023 data-nosnippet>3023</a>                    .as_primitive::&lt;arrow_array::types::Int32Type&gt;()
-<a href=#3024 id=3024 data-nosnippet>3024</a>                    .values()
-<a href=#3025 id=3025 data-nosnippet>3025</a>                    .iter()
-<a href=#3026 id=3026 data-nosnippet>3026</a>                    .copied()
-<a href=#3027 id=3027 data-nosnippet>3027</a>            })
-<a href=#3028 id=3028 data-nosnippet>3028</a>            .collect();
-<a href=#3029 id=3029 data-nosnippet>3029</a>
-<a href=#3030 id=3030 data-nosnippet>3030</a>        <span class="kw">let </span>expected_ids: Vec&lt;i32&gt; = (<span class="number">101</span>..=<span class="number">200</span>).collect();
-<a href=#3031 id=3031 data-nosnippet>3031</a>        <span class="macro">assert_eq!</span>(
-<a href=#3032 id=3032 data-nosnippet>3032</a>            all_ids, expected_ids,
-<a href=#3033 id=3033 data-nosnippet>3033</a>            <span class="string">"Should have ids 101-200 (all of row group 1)"
-<a href=#3034 id=3034 data-nosnippet>3034</a>        </span>);
-<a href=#3035 id=3035 data-nosnippet>3035</a>    }
-<a href=#3036 id=3036 data-nosnippet>3036</a>
-<a href=#3037 id=3037 data-nosnippet>3037</a>    <span class="doccomment">/// Test reading Parquet files without field ID metadata (e.g., migrated tables).
-<a href=#3038 id=3038 data-nosnippet>3038</a>    /// This exercises the position-based fallback path.
-<a href=#3039 id=3039 data-nosnippet>3039</a>    ///
-<a href=#3040 id=3040 data-nosnippet>3040</a>    /// Corresponds to Java's ParquetSchemaUtil.addFallbackIds() + pruneColumnsFallback()
-<a href=#3041 id=3041 data-nosnippet>3041</a>    /// in /parquet/src/main/java/org/apache/iceberg/parquet/ParquetSchemaUtil.java
-<a href=#3042 id=3042 data-nosnippet>3042</a>    </span><span class="attr">#[tokio::test]
-<a href=#3043 id=3043 data-nosnippet>3043</a>    </span><span class="kw">async fn </span>test_read_parquet_file_without_field_ids() {
-<a href=#3044 id=3044 data-nosnippet>3044</a>        <span class="kw">let </span>schema = Arc::new(
-<a href=#3045 id=3045 data-nosnippet>3045</a>            Schema::builder()
-<a href=#3046 id=3046 data-nosnippet>3046</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#3047 id=3047 data-nosnippet>3047</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#3048 id=3048 data-nosnippet>3048</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#3049 id=3049 data-nosnippet>3049</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"age"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3050 id=3050 data-nosnippet>3050</a>                ])
-<a href=#3051 id=3051 data-nosnippet>3051</a>                .build()
-<a href=#3052 id=3052 data-nosnippet>3052</a>                .unwrap(),
-<a href=#3053 id=3053 data-nosnippet>3053</a>        );
-<a href=#3054 id=3054 data-nosnippet>3054</a>
-<a href=#3055 id=3055 data-nosnippet>3055</a>        <span class="comment">// Parquet file from a migrated table - no field ID metadata
-<a href=#3056 id=3056 data-nosnippet>3056</a>        </span><span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#3057 id=3057 data-nosnippet>3057</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
-<a href=#3058 id=3058 data-nosnippet>3058</a>            Field::new(<span class="string">"age"</span>, DataType::Int32, <span class="bool-val">false</span>),
-<a href=#3059 id=3059 data-nosnippet>3059</a>        ]));
-<a href=#3060 id=3060 data-nosnippet>3060</a>
-<a href=#3061 id=3061 data-nosnippet>3061</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#3062 id=3062 data-nosnippet>3062</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#3063 id=3063 data-nosnippet>3063</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3011 id=3011 data-nosnippet>3011</a>        <span class="comment">// Create table schema with a single 'id' column
+<a href=#3012 id=3012 data-nosnippet>3012</a>        </span><span class="kw">let </span>table_schema = Arc::new(
+<a href=#3013 id=3013 data-nosnippet>3013</a>            Schema::builder()
+<a href=#3014 id=3014 data-nosnippet>3014</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#3015 id=3015 data-nosnippet>3015</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3016 id=3016 data-nosnippet>3016</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3017 id=3017 data-nosnippet>3017</a>                ])
+<a href=#3018 id=3018 data-nosnippet>3018</a>                .build()
+<a href=#3019 id=3019 data-nosnippet>3019</a>                .unwrap(),
+<a href=#3020 id=3020 data-nosnippet>3020</a>        );
+<a href=#3021 id=3021 data-nosnippet>3021</a>
+<a href=#3022 id=3022 data-nosnippet>3022</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3023 id=3023 data-nosnippet>3023</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#3024 id=3024 data-nosnippet>3024</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#3025 id=3025 data-nosnippet>3025</a>                <span class="string">"1"</span>.to_string(),
+<a href=#3026 id=3026 data-nosnippet>3026</a>            )])),
+<a href=#3027 id=3027 data-nosnippet>3027</a>        ]));
+<a href=#3028 id=3028 data-nosnippet>3028</a>
+<a href=#3029 id=3029 data-nosnippet>3029</a>        <span class="comment">// Step 1: Create data file with 200 rows in 2 row groups
+<a href=#3030 id=3030 data-nosnippet>3030</a>        // Row group 0: rows 0-99 (ids 1-100)
+<a href=#3031 id=3031 data-nosnippet>3031</a>        // Row group 1: rows 100-199 (ids 101-200)
+<a href=#3032 id=3032 data-nosnippet>3032</a>        </span><span class="kw">let </span>data_file_path = <span class="macro">format!</span>(<span class="string">"{}/data.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
+<a href=#3033 id=3033 data-nosnippet>3033</a>
+<a href=#3034 id=3034 data-nosnippet>3034</a>        <span class="kw">let </span>batch1 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
+<a href=#3035 id=3035 data-nosnippet>3035</a>            Int32Array::from_iter_values(<span class="number">1</span>..=<span class="number">100</span>),
+<a href=#3036 id=3036 data-nosnippet>3036</a>        )])
+<a href=#3037 id=3037 data-nosnippet>3037</a>        .unwrap();
+<a href=#3038 id=3038 data-nosnippet>3038</a>
+<a href=#3039 id=3039 data-nosnippet>3039</a>        <span class="kw">let </span>batch2 = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[Arc::new(
+<a href=#3040 id=3040 data-nosnippet>3040</a>            Int32Array::from_iter_values(<span class="number">101</span>..=<span class="number">200</span>),
+<a href=#3041 id=3041 data-nosnippet>3041</a>        )])
+<a href=#3042 id=3042 data-nosnippet>3042</a>        .unwrap();
+<a href=#3043 id=3043 data-nosnippet>3043</a>
+<a href=#3044 id=3044 data-nosnippet>3044</a>        <span class="comment">// Force each batch into its own row group
+<a href=#3045 id=3045 data-nosnippet>3045</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3046 id=3046 data-nosnippet>3046</a>            .set_compression(Compression::SNAPPY)
+<a href=#3047 id=3047 data-nosnippet>3047</a>            .set_max_row_group_size(<span class="number">100</span>)
+<a href=#3048 id=3048 data-nosnippet>3048</a>            .build();
+<a href=#3049 id=3049 data-nosnippet>3049</a>
+<a href=#3050 id=3050 data-nosnippet>3050</a>        <span class="kw">let </span>file = File::create(<span class="kw-2">&amp;</span>data_file_path).unwrap();
+<a href=#3051 id=3051 data-nosnippet>3051</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#3052 id=3052 data-nosnippet>3052</a>        writer.write(<span class="kw-2">&amp;</span>batch1).expect(<span class="string">"Writing batch 1"</span>);
+<a href=#3053 id=3053 data-nosnippet>3053</a>        writer.write(<span class="kw-2">&amp;</span>batch2).expect(<span class="string">"Writing batch 2"</span>);
+<a href=#3054 id=3054 data-nosnippet>3054</a>        writer.close().unwrap();
+<a href=#3055 id=3055 data-nosnippet>3055</a>
+<a href=#3056 id=3056 data-nosnippet>3056</a>        <span class="comment">// Verify we created 2 row groups
+<a href=#3057 id=3057 data-nosnippet>3057</a>        </span><span class="kw">let </span>verify_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
+<a href=#3058 id=3058 data-nosnippet>3058</a>        <span class="kw">let </span>verify_reader = SerializedFileReader::new(verify_file).unwrap();
+<a href=#3059 id=3059 data-nosnippet>3059</a>        <span class="macro">assert_eq!</span>(
+<a href=#3060 id=3060 data-nosnippet>3060</a>            verify_reader.metadata().num_row_groups(),
+<a href=#3061 id=3061 data-nosnippet>3061</a>            <span class="number">2</span>,
+<a href=#3062 id=3062 data-nosnippet>3062</a>            <span class="string">"Should have 2 row groups"
+<a href=#3063 id=3063 data-nosnippet>3063</a>        </span>);
 <a href=#3064 id=3064 data-nosnippet>3064</a>
-<a href=#3065 id=3065 data-nosnippet>3065</a>        <span class="kw">let </span>name_data = <span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>, <span class="string">"Charlie"</span>];
-<a href=#3066 id=3066 data-nosnippet>3066</a>        <span class="kw">let </span>age_data = <span class="macro">vec!</span>[<span class="number">30</span>, <span class="number">25</span>, <span class="number">35</span>];
+<a href=#3065 id=3065 data-nosnippet>3065</a>        <span class="comment">// Step 2: Create position delete file that deletes row 0 (id=1, first row in row group 0)
+<a href=#3066 id=3066 data-nosnippet>3066</a>        </span><span class="kw">let </span>delete_file_path = <span class="macro">format!</span>(<span class="string">"{}/deletes.parquet"</span>, <span class="kw-2">&amp;</span>table_location);
 <a href=#3067 id=3067 data-nosnippet>3067</a>
-<a href=#3068 id=3068 data-nosnippet>3068</a>        <span class="kw">use </span>arrow_array::Int32Array;
-<a href=#3069 id=3069 data-nosnippet>3069</a>        <span class="kw">let </span>name_col = Arc::new(StringArray::from(name_data.clone())) <span class="kw">as </span>ArrayRef;
-<a href=#3070 id=3070 data-nosnippet>3070</a>        <span class="kw">let </span>age_col = Arc::new(Int32Array::from(age_data.clone())) <span class="kw">as </span>ArrayRef;
-<a href=#3071 id=3071 data-nosnippet>3071</a>
-<a href=#3072 id=3072 data-nosnippet>3072</a>        <span class="kw">let </span>to_write = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[name_col, age_col]).unwrap();
-<a href=#3073 id=3073 data-nosnippet>3073</a>
-<a href=#3074 id=3074 data-nosnippet>3074</a>        <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#3075 id=3075 data-nosnippet>3075</a>            .set_compression(Compression::SNAPPY)
-<a href=#3076 id=3076 data-nosnippet>3076</a>            .build();
-<a href=#3077 id=3077 data-nosnippet>3077</a>
-<a href=#3078 id=3078 data-nosnippet>3078</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#3079 id=3079 data-nosnippet>3079</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#3080 id=3080 data-nosnippet>3080</a>
-<a href=#3081 id=3081 data-nosnippet>3081</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
-<a href=#3082 id=3082 data-nosnippet>3082</a>        writer.close().unwrap();
-<a href=#3083 id=3083 data-nosnippet>3083</a>
-<a href=#3084 id=3084 data-nosnippet>3084</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3068 id=3068 data-nosnippet>3068</a>        <span class="kw">let </span>delete_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3069 id=3069 data-nosnippet>3069</a>            Field::new(<span class="string">"file_path"</span>, DataType::Utf8, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#3070 id=3070 data-nosnippet>3070</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#3071 id=3071 data-nosnippet>3071</a>                FIELD_ID_POSITIONAL_DELETE_FILE_PATH.to_string(),
+<a href=#3072 id=3072 data-nosnippet>3072</a>            )])),
+<a href=#3073 id=3073 data-nosnippet>3073</a>            Field::new(<span class="string">"pos"</span>, DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#3074 id=3074 data-nosnippet>3074</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#3075 id=3075 data-nosnippet>3075</a>                FIELD_ID_POSITIONAL_DELETE_POS.to_string(),
+<a href=#3076 id=3076 data-nosnippet>3076</a>            )])),
+<a href=#3077 id=3077 data-nosnippet>3077</a>        ]));
+<a href=#3078 id=3078 data-nosnippet>3078</a>
+<a href=#3079 id=3079 data-nosnippet>3079</a>        <span class="comment">// Delete row at position 0 (0-indexed, so it's the first row: id=1)
+<a href=#3080 id=3080 data-nosnippet>3080</a>        </span><span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[
+<a href=#3081 id=3081 data-nosnippet>3081</a>            Arc::new(StringArray::from_iter_values(<span class="macro">vec!</span>[data_file_path.clone()])),
+<a href=#3082 id=3082 data-nosnippet>3082</a>            Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[<span class="number">0i64</span>])),
+<a href=#3083 id=3083 data-nosnippet>3083</a>        ])
+<a href=#3084 id=3084 data-nosnippet>3084</a>        .unwrap();
 <a href=#3085 id=3085 data-nosnippet>3085</a>
-<a href=#3086 id=3086 data-nosnippet>3086</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#3087 id=3087 data-nosnippet>3087</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#3088 id=3088 data-nosnippet>3088</a>                start: <span class="number">0</span>,
-<a href=#3089 id=3089 data-nosnippet>3089</a>                length: <span class="number">0</span>,
-<a href=#3090 id=3090 data-nosnippet>3090</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#3091 id=3091 data-nosnippet>3091</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
-<a href=#3092 id=3092 data-nosnippet>3092</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#3093 id=3093 data-nosnippet>3093</a>                schema: schema.clone(),
-<a href=#3094 id=3094 data-nosnippet>3094</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>],
-<a href=#3095 id=3095 data-nosnippet>3095</a>                predicate: <span class="prelude-val">None</span>,
-<a href=#3096 id=3096 data-nosnippet>3096</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#3097 id=3097 data-nosnippet>3097</a>            })]
-<a href=#3098 id=3098 data-nosnippet>3098</a>            .into_iter(),
-<a href=#3099 id=3099 data-nosnippet>3099</a>        )) <span class="kw">as </span>FileScanTaskStream;
-<a href=#3100 id=3100 data-nosnippet>3100</a>
-<a href=#3101 id=3101 data-nosnippet>3101</a>        <span class="kw">let </span>result = reader
-<a href=#3102 id=3102 data-nosnippet>3102</a>            .read(tasks)
-<a href=#3103 id=3103 data-nosnippet>3103</a>            .unwrap()
-<a href=#3104 id=3104 data-nosnippet>3104</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#3105 id=3105 data-nosnippet>3105</a>            .<span class="kw">await
-<a href=#3106 id=3106 data-nosnippet>3106</a>            </span>.unwrap();
-<a href=#3107 id=3107 data-nosnippet>3107</a>
-<a href=#3108 id=3108 data-nosnippet>3108</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
-<a href=#3109 id=3109 data-nosnippet>3109</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
-<a href=#3110 id=3110 data-nosnippet>3110</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">3</span>);
-<a href=#3111 id=3111 data-nosnippet>3111</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
-<a href=#3112 id=3112 data-nosnippet>3112</a>
-<a href=#3113 id=3113 data-nosnippet>3113</a>        <span class="comment">// Verify position-based mapping: field_id 1 → position 0, field_id 2 → position 1
-<a href=#3114 id=3114 data-nosnippet>3114</a>        </span><span class="kw">let </span>name_array = batch.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
-<a href=#3115 id=3115 data-nosnippet>3115</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
-<a href=#3116 id=3116 data-nosnippet>3116</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
-<a href=#3117 id=3117 data-nosnippet>3117</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">2</span>), <span class="string">"Charlie"</span>);
-<a href=#3118 id=3118 data-nosnippet>3118</a>
-<a href=#3119 id=3119 data-nosnippet>3119</a>        <span class="kw">let </span>age_array = batch
-<a href=#3120 id=3120 data-nosnippet>3120</a>            .column(<span class="number">1</span>)
-<a href=#3121 id=3121 data-nosnippet>3121</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#3122 id=3122 data-nosnippet>3122</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">0</span>), <span class="number">30</span>);
-<a href=#3123 id=3123 data-nosnippet>3123</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">1</span>), <span class="number">25</span>);
-<a href=#3124 id=3124 data-nosnippet>3124</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">2</span>), <span class="number">35</span>);
-<a href=#3125 id=3125 data-nosnippet>3125</a>    }
-<a href=#3126 id=3126 data-nosnippet>3126</a>
-<a href=#3127 id=3127 data-nosnippet>3127</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with partial projection.
-<a href=#3128 id=3128 data-nosnippet>3128</a>    /// Only a subset of columns are requested, verifying position-based fallback
-<a href=#3129 id=3129 data-nosnippet>3129</a>    /// handles column selection correctly.
-<a href=#3130 id=3130 data-nosnippet>3130</a>    </span><span class="attr">#[tokio::test]
-<a href=#3131 id=3131 data-nosnippet>3131</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_partial_projection() {
-<a href=#3132 id=3132 data-nosnippet>3132</a>        <span class="kw">use </span>arrow_array::Int32Array;
-<a href=#3133 id=3133 data-nosnippet>3133</a>
-<a href=#3134 id=3134 data-nosnippet>3134</a>        <span class="kw">let </span>schema = Arc::new(
-<a href=#3135 id=3135 data-nosnippet>3135</a>            Schema::builder()
-<a href=#3136 id=3136 data-nosnippet>3136</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#3137 id=3137 data-nosnippet>3137</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#3138 id=3138 data-nosnippet>3138</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"col1"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#3139 id=3139 data-nosnippet>3139</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"col2"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3140 id=3140 data-nosnippet>3140</a>                    NestedField::required(<span class="number">3</span>, <span class="string">"col3"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#3141 id=3141 data-nosnippet>3141</a>                    NestedField::required(<span class="number">4</span>, <span class="string">"col4"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3142 id=3142 data-nosnippet>3142</a>                ])
-<a href=#3143 id=3143 data-nosnippet>3143</a>                .build()
-<a href=#3144 id=3144 data-nosnippet>3144</a>                .unwrap(),
-<a href=#3145 id=3145 data-nosnippet>3145</a>        );
-<a href=#3146 id=3146 data-nosnippet>3146</a>
-<a href=#3147 id=3147 data-nosnippet>3147</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#3148 id=3148 data-nosnippet>3148</a>            Field::new(<span class="string">"col1"</span>, DataType::Utf8, <span class="bool-val">false</span>),
-<a href=#3149 id=3149 data-nosnippet>3149</a>            Field::new(<span class="string">"col2"</span>, DataType::Int32, <span class="bool-val">false</span>),
-<a href=#3150 id=3150 data-nosnippet>3150</a>            Field::new(<span class="string">"col3"</span>, DataType::Utf8, <span class="bool-val">false</span>),
-<a href=#3151 id=3151 data-nosnippet>3151</a>            Field::new(<span class="string">"col4"</span>, DataType::Int32, <span class="bool-val">false</span>),
-<a href=#3152 id=3152 data-nosnippet>3152</a>        ]));
-<a href=#3153 id=3153 data-nosnippet>3153</a>
-<a href=#3154 id=3154 data-nosnippet>3154</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#3155 id=3155 data-nosnippet>3155</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#3156 id=3156 data-nosnippet>3156</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#3157 id=3157 data-nosnippet>3157</a>
-<a href=#3158 id=3158 data-nosnippet>3158</a>        <span class="kw">let </span>col1_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"a"</span>, <span class="string">"b"</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3159 id=3159 data-nosnippet>3159</a>        <span class="kw">let </span>col2_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">10</span>, <span class="number">20</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3160 id=3160 data-nosnippet>3160</a>        <span class="kw">let </span>col3_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"c"</span>, <span class="string">"d"</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3161 id=3161 data-nosnippet>3161</a>        <span class="kw">let </span>col4_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">30</span>, <span class="number">40</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3162 id=3162 data-nosnippet>3162</a>
-<a href=#3163 id=3163 data-nosnippet>3163</a>        <span class="kw">let </span>to_write = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
-<a href=#3164 id=3164 data-nosnippet>3164</a>            col1_data, col2_data, col3_data, col4_data,
-<a href=#3165 id=3165 data-nosnippet>3165</a>        ])
-<a href=#3166 id=3166 data-nosnippet>3166</a>        .unwrap();
-<a href=#3167 id=3167 data-nosnippet>3167</a>
-<a href=#3168 id=3168 data-nosnippet>3168</a>        <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#3169 id=3169 data-nosnippet>3169</a>            .set_compression(Compression::SNAPPY)
-<a href=#3170 id=3170 data-nosnippet>3170</a>            .build();
-<a href=#3171 id=3171 data-nosnippet>3171</a>
-<a href=#3172 id=3172 data-nosnippet>3172</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#3173 id=3173 data-nosnippet>3173</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#3174 id=3174 data-nosnippet>3174</a>
-<a href=#3175 id=3175 data-nosnippet>3175</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
-<a href=#3176 id=3176 data-nosnippet>3176</a>        writer.close().unwrap();
-<a href=#3177 id=3177 data-nosnippet>3177</a>
-<a href=#3178 id=3178 data-nosnippet>3178</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#3179 id=3179 data-nosnippet>3179</a>
-<a href=#3180 id=3180 data-nosnippet>3180</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#3181 id=3181 data-nosnippet>3181</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#3182 id=3182 data-nosnippet>3182</a>                start: <span class="number">0</span>,
-<a href=#3183 id=3183 data-nosnippet>3183</a>                length: <span class="number">0</span>,
-<a href=#3184 id=3184 data-nosnippet>3184</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#3185 id=3185 data-nosnippet>3185</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
-<a href=#3186 id=3186 data-nosnippet>3186</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#3187 id=3187 data-nosnippet>3187</a>                schema: schema.clone(),
-<a href=#3188 id=3188 data-nosnippet>3188</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">3</span>],
-<a href=#3189 id=3189 data-nosnippet>3189</a>                predicate: <span class="prelude-val">None</span>,
-<a href=#3190 id=3190 data-nosnippet>3190</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#3191 id=3191 data-nosnippet>3191</a>            })]
-<a href=#3192 id=3192 data-nosnippet>3192</a>            .into_iter(),
-<a href=#3193 id=3193 data-nosnippet>3193</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3086 id=3086 data-nosnippet>3086</a>        <span class="kw">let </span>delete_props = WriterProperties::builder()
+<a href=#3087 id=3087 data-nosnippet>3087</a>            .set_compression(Compression::SNAPPY)
+<a href=#3088 id=3088 data-nosnippet>3088</a>            .build();
+<a href=#3089 id=3089 data-nosnippet>3089</a>
+<a href=#3090 id=3090 data-nosnippet>3090</a>        <span class="kw">let </span>delete_file = File::create(<span class="kw-2">&amp;</span>delete_file_path).unwrap();
+<a href=#3091 id=3091 data-nosnippet>3091</a>        <span class="kw">let </span><span class="kw-2">mut </span>delete_writer =
+<a href=#3092 id=3092 data-nosnippet>3092</a>            ArrowWriter::try_new(delete_file, delete_schema, <span class="prelude-val">Some</span>(delete_props)).unwrap();
+<a href=#3093 id=3093 data-nosnippet>3093</a>        delete_writer.write(<span class="kw-2">&amp;</span>delete_batch).unwrap();
+<a href=#3094 id=3094 data-nosnippet>3094</a>        delete_writer.close().unwrap();
+<a href=#3095 id=3095 data-nosnippet>3095</a>
+<a href=#3096 id=3096 data-nosnippet>3096</a>        <span class="comment">// Step 3: Get byte ranges to read ONLY row group 1 (rows 100-199)
+<a href=#3097 id=3097 data-nosnippet>3097</a>        // This exercises the row group selection code path where row group 0 is skipped
+<a href=#3098 id=3098 data-nosnippet>3098</a>        </span><span class="kw">let </span>metadata_file = File::open(<span class="kw-2">&amp;</span>data_file_path).unwrap();
+<a href=#3099 id=3099 data-nosnippet>3099</a>        <span class="kw">let </span>metadata_reader = SerializedFileReader::new(metadata_file).unwrap();
+<a href=#3100 id=3100 data-nosnippet>3100</a>        <span class="kw">let </span>metadata = metadata_reader.metadata();
+<a href=#3101 id=3101 data-nosnippet>3101</a>
+<a href=#3102 id=3102 data-nosnippet>3102</a>        <span class="kw">let </span>row_group_0 = metadata.row_group(<span class="number">0</span>);
+<a href=#3103 id=3103 data-nosnippet>3103</a>        <span class="kw">let </span>row_group_1 = metadata.row_group(<span class="number">1</span>);
+<a href=#3104 id=3104 data-nosnippet>3104</a>
+<a href=#3105 id=3105 data-nosnippet>3105</a>        <span class="kw">let </span>rg0_start = <span class="number">4u64</span>; <span class="comment">// Parquet files start with 4-byte magic "PAR1"
+<a href=#3106 id=3106 data-nosnippet>3106</a>        </span><span class="kw">let </span>rg1_start = rg0_start + row_group_0.compressed_size() <span class="kw">as </span>u64;
+<a href=#3107 id=3107 data-nosnippet>3107</a>        <span class="kw">let </span>rg1_length = row_group_1.compressed_size() <span class="kw">as </span>u64;
+<a href=#3108 id=3108 data-nosnippet>3108</a>
+<a href=#3109 id=3109 data-nosnippet>3109</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3110 id=3110 data-nosnippet>3110</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3111 id=3111 data-nosnippet>3111</a>
+<a href=#3112 id=3112 data-nosnippet>3112</a>        <span class="comment">// Create FileScanTask that reads ONLY row group 1 via byte range filtering
+<a href=#3113 id=3113 data-nosnippet>3113</a>        </span><span class="kw">let </span>task = FileScanTask {
+<a href=#3114 id=3114 data-nosnippet>3114</a>            start: rg1_start,
+<a href=#3115 id=3115 data-nosnippet>3115</a>            length: rg1_length,
+<a href=#3116 id=3116 data-nosnippet>3116</a>            record_count: <span class="prelude-val">Some</span>(<span class="number">100</span>), <span class="comment">// Row group 1 has 100 rows
+<a href=#3117 id=3117 data-nosnippet>3117</a>            </span>data_file_path: data_file_path.clone(),
+<a href=#3118 id=3118 data-nosnippet>3118</a>            data_file_format: DataFileFormat::Parquet,
+<a href=#3119 id=3119 data-nosnippet>3119</a>            schema: table_schema.clone(),
+<a href=#3120 id=3120 data-nosnippet>3120</a>            project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>],
+<a href=#3121 id=3121 data-nosnippet>3121</a>            predicate: <span class="prelude-val">None</span>,
+<a href=#3122 id=3122 data-nosnippet>3122</a>            deletes: <span class="macro">vec!</span>[FileScanTaskDeleteFile {
+<a href=#3123 id=3123 data-nosnippet>3123</a>                file_path: delete_file_path,
+<a href=#3124 id=3124 data-nosnippet>3124</a>                file_type: DataContentType::PositionDeletes,
+<a href=#3125 id=3125 data-nosnippet>3125</a>                partition_spec_id: <span class="number">0</span>,
+<a href=#3126 id=3126 data-nosnippet>3126</a>                equality_ids: <span class="prelude-val">None</span>,
+<a href=#3127 id=3127 data-nosnippet>3127</a>            }],
+<a href=#3128 id=3128 data-nosnippet>3128</a>            partition: <span class="prelude-val">None</span>,
+<a href=#3129 id=3129 data-nosnippet>3129</a>            partition_spec: <span class="prelude-val">None</span>,
+<a href=#3130 id=3130 data-nosnippet>3130</a>            name_mapping: <span class="prelude-val">None</span>,
+<a href=#3131 id=3131 data-nosnippet>3131</a>        };
+<a href=#3132 id=3132 data-nosnippet>3132</a>
+<a href=#3133 id=3133 data-nosnippet>3133</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(task)])) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3134 id=3134 data-nosnippet>3134</a>        <span class="kw">let </span>result = reader
+<a href=#3135 id=3135 data-nosnippet>3135</a>            .read(tasks)
+<a href=#3136 id=3136 data-nosnippet>3136</a>            .unwrap()
+<a href=#3137 id=3137 data-nosnippet>3137</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3138 id=3138 data-nosnippet>3138</a>            .<span class="kw">await
+<a href=#3139 id=3139 data-nosnippet>3139</a>            </span>.unwrap();
+<a href=#3140 id=3140 data-nosnippet>3140</a>
+<a href=#3141 id=3141 data-nosnippet>3141</a>        <span class="comment">// Step 4: Verify we got 100 rows (all of row group 1)
+<a href=#3142 id=3142 data-nosnippet>3142</a>        // The delete at position 0 is in row group 0, which is skipped, so it doesn't affect us
+<a href=#3143 id=3143 data-nosnippet>3143</a>        </span><span class="kw">let </span>total_rows: usize = result.iter().map(|b| b.num_rows()).sum();
+<a href=#3144 id=3144 data-nosnippet>3144</a>
+<a href=#3145 id=3145 data-nosnippet>3145</a>        <span class="macro">assert_eq!</span>(
+<a href=#3146 id=3146 data-nosnippet>3146</a>            total_rows, <span class="number">100</span>,
+<a href=#3147 id=3147 data-nosnippet>3147</a>            <span class="string">"Expected 100 rows from row group 1 (delete at position 0 is in skipped row group 0). \
+<a href=#3148 id=3148 data-nosnippet>3148</a>             If this hangs or fails, it indicates the cached delete index was not updated after advance_to()."
+<a href=#3149 id=3149 data-nosnippet>3149</a>        </span>);
+<a href=#3150 id=3150 data-nosnippet>3150</a>
+<a href=#3151 id=3151 data-nosnippet>3151</a>        <span class="comment">// Verify we have all ids from row group 1 (101-200)
+<a href=#3152 id=3152 data-nosnippet>3152</a>        </span><span class="kw">let </span>all_ids: Vec&lt;i32&gt; = result
+<a href=#3153 id=3153 data-nosnippet>3153</a>            .iter()
+<a href=#3154 id=3154 data-nosnippet>3154</a>            .flat_map(|batch| {
+<a href=#3155 id=3155 data-nosnippet>3155</a>                batch
+<a href=#3156 id=3156 data-nosnippet>3156</a>                    .column(<span class="number">0</span>)
+<a href=#3157 id=3157 data-nosnippet>3157</a>                    .as_primitive::&lt;arrow_array::types::Int32Type&gt;()
+<a href=#3158 id=3158 data-nosnippet>3158</a>                    .values()
+<a href=#3159 id=3159 data-nosnippet>3159</a>                    .iter()
+<a href=#3160 id=3160 data-nosnippet>3160</a>                    .copied()
+<a href=#3161 id=3161 data-nosnippet>3161</a>            })
+<a href=#3162 id=3162 data-nosnippet>3162</a>            .collect();
+<a href=#3163 id=3163 data-nosnippet>3163</a>
+<a href=#3164 id=3164 data-nosnippet>3164</a>        <span class="kw">let </span>expected_ids: Vec&lt;i32&gt; = (<span class="number">101</span>..=<span class="number">200</span>).collect();
+<a href=#3165 id=3165 data-nosnippet>3165</a>        <span class="macro">assert_eq!</span>(
+<a href=#3166 id=3166 data-nosnippet>3166</a>            all_ids, expected_ids,
+<a href=#3167 id=3167 data-nosnippet>3167</a>            <span class="string">"Should have ids 101-200 (all of row group 1)"
+<a href=#3168 id=3168 data-nosnippet>3168</a>        </span>);
+<a href=#3169 id=3169 data-nosnippet>3169</a>    }
+<a href=#3170 id=3170 data-nosnippet>3170</a>
+<a href=#3171 id=3171 data-nosnippet>3171</a>    <span class="doccomment">/// Test reading Parquet files without field ID metadata (e.g., migrated tables).
+<a href=#3172 id=3172 data-nosnippet>3172</a>    /// This exercises the position-based fallback path.
+<a href=#3173 id=3173 data-nosnippet>3173</a>    ///
+<a href=#3174 id=3174 data-nosnippet>3174</a>    /// Corresponds to Java's ParquetSchemaUtil.addFallbackIds() + pruneColumnsFallback()
+<a href=#3175 id=3175 data-nosnippet>3175</a>    /// in /parquet/src/main/java/org/apache/iceberg/parquet/ParquetSchemaUtil.java
+<a href=#3176 id=3176 data-nosnippet>3176</a>    </span><span class="attr">#[tokio::test]
+<a href=#3177 id=3177 data-nosnippet>3177</a>    </span><span class="kw">async fn </span>test_read_parquet_file_without_field_ids() {
+<a href=#3178 id=3178 data-nosnippet>3178</a>        <span class="kw">let </span>schema = Arc::new(
+<a href=#3179 id=3179 data-nosnippet>3179</a>            Schema::builder()
+<a href=#3180 id=3180 data-nosnippet>3180</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#3181 id=3181 data-nosnippet>3181</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3182 id=3182 data-nosnippet>3182</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#3183 id=3183 data-nosnippet>3183</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"age"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3184 id=3184 data-nosnippet>3184</a>                ])
+<a href=#3185 id=3185 data-nosnippet>3185</a>                .build()
+<a href=#3186 id=3186 data-nosnippet>3186</a>                .unwrap(),
+<a href=#3187 id=3187 data-nosnippet>3187</a>        );
+<a href=#3188 id=3188 data-nosnippet>3188</a>
+<a href=#3189 id=3189 data-nosnippet>3189</a>        <span class="comment">// Parquet file from a migrated table - no field ID metadata
+<a href=#3190 id=3190 data-nosnippet>3190</a>        </span><span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3191 id=3191 data-nosnippet>3191</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
+<a href=#3192 id=3192 data-nosnippet>3192</a>            Field::new(<span class="string">"age"</span>, DataType::Int32, <span class="bool-val">false</span>),
+<a href=#3193 id=3193 data-nosnippet>3193</a>        ]));
 <a href=#3194 id=3194 data-nosnippet>3194</a>
-<a href=#3195 id=3195 data-nosnippet>3195</a>        <span class="kw">let </span>result = reader
-<a href=#3196 id=3196 data-nosnippet>3196</a>            .read(tasks)
-<a href=#3197 id=3197 data-nosnippet>3197</a>            .unwrap()
-<a href=#3198 id=3198 data-nosnippet>3198</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#3199 id=3199 data-nosnippet>3199</a>            .<span class="kw">await
-<a href=#3200 id=3200 data-nosnippet>3200</a>            </span>.unwrap();
+<a href=#3195 id=3195 data-nosnippet>3195</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3196 id=3196 data-nosnippet>3196</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#3197 id=3197 data-nosnippet>3197</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3198 id=3198 data-nosnippet>3198</a>
+<a href=#3199 id=3199 data-nosnippet>3199</a>        <span class="kw">let </span>name_data = <span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>, <span class="string">"Charlie"</span>];
+<a href=#3200 id=3200 data-nosnippet>3200</a>        <span class="kw">let </span>age_data = <span class="macro">vec!</span>[<span class="number">30</span>, <span class="number">25</span>, <span class="number">35</span>];
 <a href=#3201 id=3201 data-nosnippet>3201</a>
-<a href=#3202 id=3202 data-nosnippet>3202</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
-<a href=#3203 id=3203 data-nosnippet>3203</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
-<a href=#3204 id=3204 data-nosnippet>3204</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">2</span>);
-<a href=#3205 id=3205 data-nosnippet>3205</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
-<a href=#3206 id=3206 data-nosnippet>3206</a>
-<a href=#3207 id=3207 data-nosnippet>3207</a>        <span class="kw">let </span>col1_array = batch.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
-<a href=#3208 id=3208 data-nosnippet>3208</a>        <span class="macro">assert_eq!</span>(col1_array.value(<span class="number">0</span>), <span class="string">"a"</span>);
-<a href=#3209 id=3209 data-nosnippet>3209</a>        <span class="macro">assert_eq!</span>(col1_array.value(<span class="number">1</span>), <span class="string">"b"</span>);
-<a href=#3210 id=3210 data-nosnippet>3210</a>
-<a href=#3211 id=3211 data-nosnippet>3211</a>        <span class="kw">let </span>col3_array = batch.column(<span class="number">1</span>).as_string::&lt;i32&gt;();
-<a href=#3212 id=3212 data-nosnippet>3212</a>        <span class="macro">assert_eq!</span>(col3_array.value(<span class="number">0</span>), <span class="string">"c"</span>);
-<a href=#3213 id=3213 data-nosnippet>3213</a>        <span class="macro">assert_eq!</span>(col3_array.value(<span class="number">1</span>), <span class="string">"d"</span>);
-<a href=#3214 id=3214 data-nosnippet>3214</a>    }
-<a href=#3215 id=3215 data-nosnippet>3215</a>
-<a href=#3216 id=3216 data-nosnippet>3216</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with schema evolution.
-<a href=#3217 id=3217 data-nosnippet>3217</a>    /// The Iceberg schema has more fields than the Parquet file, testing that
-<a href=#3218 id=3218 data-nosnippet>3218</a>    /// missing columns are filled with NULLs.
-<a href=#3219 id=3219 data-nosnippet>3219</a>    </span><span class="attr">#[tokio::test]
-<a href=#3220 id=3220 data-nosnippet>3220</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_schema_evolution() {
-<a href=#3221 id=3221 data-nosnippet>3221</a>        <span class="kw">use </span>arrow_array::{Array, Int32Array};
-<a href=#3222 id=3222 data-nosnippet>3222</a>
-<a href=#3223 id=3223 data-nosnippet>3223</a>        <span class="comment">// Schema with field 3 added after the file was written
-<a href=#3224 id=3224 data-nosnippet>3224</a>        </span><span class="kw">let </span>schema = Arc::new(
-<a href=#3225 id=3225 data-nosnippet>3225</a>            Schema::builder()
-<a href=#3226 id=3226 data-nosnippet>3226</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#3227 id=3227 data-nosnippet>3227</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#3228 id=3228 data-nosnippet>3228</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#3229 id=3229 data-nosnippet>3229</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"age"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3230 id=3230 data-nosnippet>3230</a>                    NestedField::optional(<span class="number">3</span>, <span class="string">"city"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#3231 id=3231 data-nosnippet>3231</a>                ])
-<a href=#3232 id=3232 data-nosnippet>3232</a>                .build()
-<a href=#3233 id=3233 data-nosnippet>3233</a>                .unwrap(),
-<a href=#3234 id=3234 data-nosnippet>3234</a>        );
-<a href=#3235 id=3235 data-nosnippet>3235</a>
-<a href=#3236 id=3236 data-nosnippet>3236</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#3237 id=3237 data-nosnippet>3237</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
-<a href=#3238 id=3238 data-nosnippet>3238</a>            Field::new(<span class="string">"age"</span>, DataType::Int32, <span class="bool-val">false</span>),
-<a href=#3239 id=3239 data-nosnippet>3239</a>        ]));
-<a href=#3240 id=3240 data-nosnippet>3240</a>
-<a href=#3241 id=3241 data-nosnippet>3241</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#3242 id=3242 data-nosnippet>3242</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#3243 id=3243 data-nosnippet>3243</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3202 id=3202 data-nosnippet>3202</a>        <span class="kw">use </span>arrow_array::Int32Array;
+<a href=#3203 id=3203 data-nosnippet>3203</a>        <span class="kw">let </span>name_col = Arc::new(StringArray::from(name_data.clone())) <span class="kw">as </span>ArrayRef;
+<a href=#3204 id=3204 data-nosnippet>3204</a>        <span class="kw">let </span>age_col = Arc::new(Int32Array::from(age_data.clone())) <span class="kw">as </span>ArrayRef;
+<a href=#3205 id=3205 data-nosnippet>3205</a>
+<a href=#3206 id=3206 data-nosnippet>3206</a>        <span class="kw">let </span>to_write = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[name_col, age_col]).unwrap();
+<a href=#3207 id=3207 data-nosnippet>3207</a>
+<a href=#3208 id=3208 data-nosnippet>3208</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3209 id=3209 data-nosnippet>3209</a>            .set_compression(Compression::SNAPPY)
+<a href=#3210 id=3210 data-nosnippet>3210</a>            .build();
+<a href=#3211 id=3211 data-nosnippet>3211</a>
+<a href=#3212 id=3212 data-nosnippet>3212</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#3213 id=3213 data-nosnippet>3213</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#3214 id=3214 data-nosnippet>3214</a>
+<a href=#3215 id=3215 data-nosnippet>3215</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
+<a href=#3216 id=3216 data-nosnippet>3216</a>        writer.close().unwrap();
+<a href=#3217 id=3217 data-nosnippet>3217</a>
+<a href=#3218 id=3218 data-nosnippet>3218</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3219 id=3219 data-nosnippet>3219</a>
+<a href=#3220 id=3220 data-nosnippet>3220</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#3221 id=3221 data-nosnippet>3221</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#3222 id=3222 data-nosnippet>3222</a>                start: <span class="number">0</span>,
+<a href=#3223 id=3223 data-nosnippet>3223</a>                length: <span class="number">0</span>,
+<a href=#3224 id=3224 data-nosnippet>3224</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#3225 id=3225 data-nosnippet>3225</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
+<a href=#3226 id=3226 data-nosnippet>3226</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#3227 id=3227 data-nosnippet>3227</a>                schema: schema.clone(),
+<a href=#3228 id=3228 data-nosnippet>3228</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>],
+<a href=#3229 id=3229 data-nosnippet>3229</a>                predicate: <span class="prelude-val">None</span>,
+<a href=#3230 id=3230 data-nosnippet>3230</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#3231 id=3231 data-nosnippet>3231</a>                partition: <span class="prelude-val">None</span>,
+<a href=#3232 id=3232 data-nosnippet>3232</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#3233 id=3233 data-nosnippet>3233</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#3234 id=3234 data-nosnippet>3234</a>            })]
+<a href=#3235 id=3235 data-nosnippet>3235</a>            .into_iter(),
+<a href=#3236 id=3236 data-nosnippet>3236</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3237 id=3237 data-nosnippet>3237</a>
+<a href=#3238 id=3238 data-nosnippet>3238</a>        <span class="kw">let </span>result = reader
+<a href=#3239 id=3239 data-nosnippet>3239</a>            .read(tasks)
+<a href=#3240 id=3240 data-nosnippet>3240</a>            .unwrap()
+<a href=#3241 id=3241 data-nosnippet>3241</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3242 id=3242 data-nosnippet>3242</a>            .<span class="kw">await
+<a href=#3243 id=3243 data-nosnippet>3243</a>            </span>.unwrap();
 <a href=#3244 id=3244 data-nosnippet>3244</a>
-<a href=#3245 id=3245 data-nosnippet>3245</a>        <span class="kw">let </span>name_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3246 id=3246 data-nosnippet>3246</a>        <span class="kw">let </span>age_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">30</span>, <span class="number">25</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3247 id=3247 data-nosnippet>3247</a>
-<a href=#3248 id=3248 data-nosnippet>3248</a>        <span class="kw">let </span>to_write =
-<a href=#3249 id=3249 data-nosnippet>3249</a>            RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[name_data, age_data]).unwrap();
-<a href=#3250 id=3250 data-nosnippet>3250</a>
-<a href=#3251 id=3251 data-nosnippet>3251</a>        <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#3252 id=3252 data-nosnippet>3252</a>            .set_compression(Compression::SNAPPY)
-<a href=#3253 id=3253 data-nosnippet>3253</a>            .build();
-<a href=#3254 id=3254 data-nosnippet>3254</a>
-<a href=#3255 id=3255 data-nosnippet>3255</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#3256 id=3256 data-nosnippet>3256</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#3257 id=3257 data-nosnippet>3257</a>
-<a href=#3258 id=3258 data-nosnippet>3258</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
-<a href=#3259 id=3259 data-nosnippet>3259</a>        writer.close().unwrap();
-<a href=#3260 id=3260 data-nosnippet>3260</a>
-<a href=#3261 id=3261 data-nosnippet>3261</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#3262 id=3262 data-nosnippet>3262</a>
-<a href=#3263 id=3263 data-nosnippet>3263</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#3264 id=3264 data-nosnippet>3264</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#3265 id=3265 data-nosnippet>3265</a>                start: <span class="number">0</span>,
-<a href=#3266 id=3266 data-nosnippet>3266</a>                length: <span class="number">0</span>,
-<a href=#3267 id=3267 data-nosnippet>3267</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#3268 id=3268 data-nosnippet>3268</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
-<a href=#3269 id=3269 data-nosnippet>3269</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#3270 id=3270 data-nosnippet>3270</a>                schema: schema.clone(),
-<a href=#3271 id=3271 data-nosnippet>3271</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>],
-<a href=#3272 id=3272 data-nosnippet>3272</a>                predicate: <span class="prelude-val">None</span>,
-<a href=#3273 id=3273 data-nosnippet>3273</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#3274 id=3274 data-nosnippet>3274</a>            })]
-<a href=#3275 id=3275 data-nosnippet>3275</a>            .into_iter(),
-<a href=#3276 id=3276 data-nosnippet>3276</a>        )) <span class="kw">as </span>FileScanTaskStream;
-<a href=#3277 id=3277 data-nosnippet>3277</a>
-<a href=#3278 id=3278 data-nosnippet>3278</a>        <span class="kw">let </span>result = reader
-<a href=#3279 id=3279 data-nosnippet>3279</a>            .read(tasks)
-<a href=#3280 id=3280 data-nosnippet>3280</a>            .unwrap()
-<a href=#3281 id=3281 data-nosnippet>3281</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#3282 id=3282 data-nosnippet>3282</a>            .<span class="kw">await
-<a href=#3283 id=3283 data-nosnippet>3283</a>            </span>.unwrap();
-<a href=#3284 id=3284 data-nosnippet>3284</a>
-<a href=#3285 id=3285 data-nosnippet>3285</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
-<a href=#3286 id=3286 data-nosnippet>3286</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
-<a href=#3287 id=3287 data-nosnippet>3287</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">2</span>);
-<a href=#3288 id=3288 data-nosnippet>3288</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">3</span>);
-<a href=#3289 id=3289 data-nosnippet>3289</a>
-<a href=#3290 id=3290 data-nosnippet>3290</a>        <span class="kw">let </span>name_array = batch.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
-<a href=#3291 id=3291 data-nosnippet>3291</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
-<a href=#3292 id=3292 data-nosnippet>3292</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
-<a href=#3293 id=3293 data-nosnippet>3293</a>
-<a href=#3294 id=3294 data-nosnippet>3294</a>        <span class="kw">let </span>age_array = batch
-<a href=#3295 id=3295 data-nosnippet>3295</a>            .column(<span class="number">1</span>)
-<a href=#3296 id=3296 data-nosnippet>3296</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#3297 id=3297 data-nosnippet>3297</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">0</span>), <span class="number">30</span>);
-<a href=#3298 id=3298 data-nosnippet>3298</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">1</span>), <span class="number">25</span>);
+<a href=#3245 id=3245 data-nosnippet>3245</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
+<a href=#3246 id=3246 data-nosnippet>3246</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
+<a href=#3247 id=3247 data-nosnippet>3247</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">3</span>);
+<a href=#3248 id=3248 data-nosnippet>3248</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
+<a href=#3249 id=3249 data-nosnippet>3249</a>
+<a href=#3250 id=3250 data-nosnippet>3250</a>        <span class="comment">// Verify position-based mapping: field_id 1 → position 0, field_id 2 → position 1
+<a href=#3251 id=3251 data-nosnippet>3251</a>        </span><span class="kw">let </span>name_array = batch.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
+<a href=#3252 id=3252 data-nosnippet>3252</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
+<a href=#3253 id=3253 data-nosnippet>3253</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
+<a href=#3254 id=3254 data-nosnippet>3254</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">2</span>), <span class="string">"Charlie"</span>);
+<a href=#3255 id=3255 data-nosnippet>3255</a>
+<a href=#3256 id=3256 data-nosnippet>3256</a>        <span class="kw">let </span>age_array = batch
+<a href=#3257 id=3257 data-nosnippet>3257</a>            .column(<span class="number">1</span>)
+<a href=#3258 id=3258 data-nosnippet>3258</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#3259 id=3259 data-nosnippet>3259</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">0</span>), <span class="number">30</span>);
+<a href=#3260 id=3260 data-nosnippet>3260</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">1</span>), <span class="number">25</span>);
+<a href=#3261 id=3261 data-nosnippet>3261</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">2</span>), <span class="number">35</span>);
+<a href=#3262 id=3262 data-nosnippet>3262</a>    }
+<a href=#3263 id=3263 data-nosnippet>3263</a>
+<a href=#3264 id=3264 data-nosnippet>3264</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with partial projection.
+<a href=#3265 id=3265 data-nosnippet>3265</a>    /// Only a subset of columns are requested, verifying position-based fallback
+<a href=#3266 id=3266 data-nosnippet>3266</a>    /// handles column selection correctly.
+<a href=#3267 id=3267 data-nosnippet>3267</a>    </span><span class="attr">#[tokio::test]
+<a href=#3268 id=3268 data-nosnippet>3268</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_partial_projection() {
+<a href=#3269 id=3269 data-nosnippet>3269</a>        <span class="kw">use </span>arrow_array::Int32Array;
+<a href=#3270 id=3270 data-nosnippet>3270</a>
+<a href=#3271 id=3271 data-nosnippet>3271</a>        <span class="kw">let </span>schema = Arc::new(
+<a href=#3272 id=3272 data-nosnippet>3272</a>            Schema::builder()
+<a href=#3273 id=3273 data-nosnippet>3273</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#3274 id=3274 data-nosnippet>3274</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3275 id=3275 data-nosnippet>3275</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"col1"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#3276 id=3276 data-nosnippet>3276</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"col2"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3277 id=3277 data-nosnippet>3277</a>                    NestedField::required(<span class="number">3</span>, <span class="string">"col3"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#3278 id=3278 data-nosnippet>3278</a>                    NestedField::required(<span class="number">4</span>, <span class="string">"col4"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3279 id=3279 data-nosnippet>3279</a>                ])
+<a href=#3280 id=3280 data-nosnippet>3280</a>                .build()
+<a href=#3281 id=3281 data-nosnippet>3281</a>                .unwrap(),
+<a href=#3282 id=3282 data-nosnippet>3282</a>        );
+<a href=#3283 id=3283 data-nosnippet>3283</a>
+<a href=#3284 id=3284 data-nosnippet>3284</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3285 id=3285 data-nosnippet>3285</a>            Field::new(<span class="string">"col1"</span>, DataType::Utf8, <span class="bool-val">false</span>),
+<a href=#3286 id=3286 data-nosnippet>3286</a>            Field::new(<span class="string">"col2"</span>, DataType::Int32, <span class="bool-val">false</span>),
+<a href=#3287 id=3287 data-nosnippet>3287</a>            Field::new(<span class="string">"col3"</span>, DataType::Utf8, <span class="bool-val">false</span>),
+<a href=#3288 id=3288 data-nosnippet>3288</a>            Field::new(<span class="string">"col4"</span>, DataType::Int32, <span class="bool-val">false</span>),
+<a href=#3289 id=3289 data-nosnippet>3289</a>        ]));
+<a href=#3290 id=3290 data-nosnippet>3290</a>
+<a href=#3291 id=3291 data-nosnippet>3291</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3292 id=3292 data-nosnippet>3292</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#3293 id=3293 data-nosnippet>3293</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3294 id=3294 data-nosnippet>3294</a>
+<a href=#3295 id=3295 data-nosnippet>3295</a>        <span class="kw">let </span>col1_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"a"</span>, <span class="string">"b"</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3296 id=3296 data-nosnippet>3296</a>        <span class="kw">let </span>col2_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">10</span>, <span class="number">20</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3297 id=3297 data-nosnippet>3297</a>        <span class="kw">let </span>col3_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"c"</span>, <span class="string">"d"</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3298 id=3298 data-nosnippet>3298</a>        <span class="kw">let </span>col4_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">30</span>, <span class="number">40</span>])) <span class="kw">as </span>ArrayRef;
 <a href=#3299 id=3299 data-nosnippet>3299</a>
-<a href=#3300 id=3300 data-nosnippet>3300</a>        <span class="comment">// Verify missing column filled with NULLs
-<a href=#3301 id=3301 data-nosnippet>3301</a>        </span><span class="kw">let </span>city_array = batch.column(<span class="number">2</span>).as_string::&lt;i32&gt;();
-<a href=#3302 id=3302 data-nosnippet>3302</a>        <span class="macro">assert_eq!</span>(city_array.null_count(), <span class="number">2</span>);
-<a href=#3303 id=3303 data-nosnippet>3303</a>        <span class="macro">assert!</span>(city_array.is_null(<span class="number">0</span>));
-<a href=#3304 id=3304 data-nosnippet>3304</a>        <span class="macro">assert!</span>(city_array.is_null(<span class="number">1</span>));
-<a href=#3305 id=3305 data-nosnippet>3305</a>    }
-<a href=#3306 id=3306 data-nosnippet>3306</a>
-<a href=#3307 id=3307 data-nosnippet>3307</a>    <span class="doccomment">/// Test reading Parquet files without field IDs that have multiple row groups.
-<a href=#3308 id=3308 data-nosnippet>3308</a>    /// This ensures the position-based fallback works correctly across row group boundaries.
-<a href=#3309 id=3309 data-nosnippet>3309</a>    </span><span class="attr">#[tokio::test]
-<a href=#3310 id=3310 data-nosnippet>3310</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_multiple_row_groups() {
-<a href=#3311 id=3311 data-nosnippet>3311</a>        <span class="kw">use </span>arrow_array::Int32Array;
-<a href=#3312 id=3312 data-nosnippet>3312</a>
-<a href=#3313 id=3313 data-nosnippet>3313</a>        <span class="kw">let </span>schema = Arc::new(
-<a href=#3314 id=3314 data-nosnippet>3314</a>            Schema::builder()
-<a href=#3315 id=3315 data-nosnippet>3315</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#3316 id=3316 data-nosnippet>3316</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#3317 id=3317 data-nosnippet>3317</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#3318 id=3318 data-nosnippet>3318</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"value"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3319 id=3319 data-nosnippet>3319</a>                ])
-<a href=#3320 id=3320 data-nosnippet>3320</a>                .build()
-<a href=#3321 id=3321 data-nosnippet>3321</a>                .unwrap(),
-<a href=#3322 id=3322 data-nosnippet>3322</a>        );
-<a href=#3323 id=3323 data-nosnippet>3323</a>
-<a href=#3324 id=3324 data-nosnippet>3324</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#3325 id=3325 data-nosnippet>3325</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
-<a href=#3326 id=3326 data-nosnippet>3326</a>            Field::new(<span class="string">"value"</span>, DataType::Int32, <span class="bool-val">false</span>),
-<a href=#3327 id=3327 data-nosnippet>3327</a>        ]));
-<a href=#3328 id=3328 data-nosnippet>3328</a>
-<a href=#3329 id=3329 data-nosnippet>3329</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#3330 id=3330 data-nosnippet>3330</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#3331 id=3331 data-nosnippet>3331</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#3332 id=3332 data-nosnippet>3332</a>
-<a href=#3333 id=3333 data-nosnippet>3333</a>        <span class="comment">// Small row group size to create multiple row groups
-<a href=#3334 id=3334 data-nosnippet>3334</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
-<a href=#3335 id=3335 data-nosnippet>3335</a>            .set_compression(Compression::SNAPPY)
-<a href=#3336 id=3336 data-nosnippet>3336</a>            .set_write_batch_size(<span class="number">2</span>)
-<a href=#3337 id=3337 data-nosnippet>3337</a>            .set_max_row_group_size(<span class="number">2</span>)
-<a href=#3338 id=3338 data-nosnippet>3338</a>            .build();
-<a href=#3339 id=3339 data-nosnippet>3339</a>
-<a href=#3340 id=3340 data-nosnippet>3340</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#3341 id=3341 data-nosnippet>3341</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#3342 id=3342 data-nosnippet>3342</a>
-<a href=#3343 id=3343 data-nosnippet>3343</a>        <span class="comment">// Write 6 rows in 3 batches (will create 3 row groups)
-<a href=#3344 id=3344 data-nosnippet>3344</a>        </span><span class="kw">for </span>batch_num <span class="kw">in </span><span class="number">0</span>..<span class="number">3 </span>{
-<a href=#3345 id=3345 data-nosnippet>3345</a>            <span class="kw">let </span>name_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[
-<a href=#3346 id=3346 data-nosnippet>3346</a>                <span class="macro">format!</span>(<span class="string">"name_{}"</span>, batch_num * <span class="number">2</span>),
-<a href=#3347 id=3347 data-nosnippet>3347</a>                <span class="macro">format!</span>(<span class="string">"name_{}"</span>, batch_num * <span class="number">2 </span>+ <span class="number">1</span>),
-<a href=#3348 id=3348 data-nosnippet>3348</a>            ])) <span class="kw">as </span>ArrayRef;
-<a href=#3349 id=3349 data-nosnippet>3349</a>            <span class="kw">let </span>value_data =
-<a href=#3350 id=3350 data-nosnippet>3350</a>                Arc::new(Int32Array::from(<span class="macro">vec!</span>[batch_num * <span class="number">2</span>, batch_num * <span class="number">2 </span>+ <span class="number">1</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3351 id=3351 data-nosnippet>3351</a>
-<a href=#3352 id=3352 data-nosnippet>3352</a>            <span class="kw">let </span>batch =
-<a href=#3353 id=3353 data-nosnippet>3353</a>                RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[name_data, value_data]).unwrap();
-<a href=#3354 id=3354 data-nosnippet>3354</a>            writer.write(<span class="kw-2">&amp;</span>batch).expect(<span class="string">"Writing batch"</span>);
-<a href=#3355 id=3355 data-nosnippet>3355</a>        }
-<a href=#3356 id=3356 data-nosnippet>3356</a>        writer.close().unwrap();
-<a href=#3357 id=3357 data-nosnippet>3357</a>
-<a href=#3358 id=3358 data-nosnippet>3358</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#3359 id=3359 data-nosnippet>3359</a>
-<a href=#3360 id=3360 data-nosnippet>3360</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#3361 id=3361 data-nosnippet>3361</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#3362 id=3362 data-nosnippet>3362</a>                start: <span class="number">0</span>,
-<a href=#3363 id=3363 data-nosnippet>3363</a>                length: <span class="number">0</span>,
-<a href=#3364 id=3364 data-nosnippet>3364</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#3365 id=3365 data-nosnippet>3365</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
-<a href=#3366 id=3366 data-nosnippet>3366</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#3367 id=3367 data-nosnippet>3367</a>                schema: schema.clone(),
-<a href=#3368 id=3368 data-nosnippet>3368</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>],
-<a href=#3369 id=3369 data-nosnippet>3369</a>                predicate: <span class="prelude-val">None</span>,
-<a href=#3370 id=3370 data-nosnippet>3370</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#3371 id=3371 data-nosnippet>3371</a>            })]
-<a href=#3372 id=3372 data-nosnippet>3372</a>            .into_iter(),
-<a href=#3373 id=3373 data-nosnippet>3373</a>        )) <span class="kw">as </span>FileScanTaskStream;
-<a href=#3374 id=3374 data-nosnippet>3374</a>
-<a href=#3375 id=3375 data-nosnippet>3375</a>        <span class="kw">let </span>result = reader
-<a href=#3376 id=3376 data-nosnippet>3376</a>            .read(tasks)
-<a href=#3377 id=3377 data-nosnippet>3377</a>            .unwrap()
-<a href=#3378 id=3378 data-nosnippet>3378</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#3379 id=3379 data-nosnippet>3379</a>            .<span class="kw">await
-<a href=#3380 id=3380 data-nosnippet>3380</a>            </span>.unwrap();
-<a href=#3381 id=3381 data-nosnippet>3381</a>
-<a href=#3382 id=3382 data-nosnippet>3382</a>        <span class="macro">assert!</span>(!result.is_empty());
-<a href=#3383 id=3383 data-nosnippet>3383</a>
-<a href=#3384 id=3384 data-nosnippet>3384</a>        <span class="kw">let </span><span class="kw-2">mut </span>all_names = Vec::new();
-<a href=#3385 id=3385 data-nosnippet>3385</a>        <span class="kw">let </span><span class="kw-2">mut </span>all_values = Vec::new();
-<a href=#3386 id=3386 data-nosnippet>3386</a>
-<a href=#3387 id=3387 data-nosnippet>3387</a>        <span class="kw">for </span>batch <span class="kw">in </span><span class="kw-2">&amp;</span>result {
-<a href=#3388 id=3388 data-nosnippet>3388</a>            <span class="kw">let </span>name_array = batch.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
-<a href=#3389 id=3389 data-nosnippet>3389</a>            <span class="kw">let </span>value_array = batch
-<a href=#3390 id=3390 data-nosnippet>3390</a>                .column(<span class="number">1</span>)
-<a href=#3391 id=3391 data-nosnippet>3391</a>                .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#3392 id=3392 data-nosnippet>3392</a>
-<a href=#3393 id=3393 data-nosnippet>3393</a>            <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
-<a href=#3394 id=3394 data-nosnippet>3394</a>                all_names.push(name_array.value(i).to_string());
-<a href=#3395 id=3395 data-nosnippet>3395</a>                all_values.push(value_array.value(i));
-<a href=#3396 id=3396 data-nosnippet>3396</a>            }
-<a href=#3397 id=3397 data-nosnippet>3397</a>        }
-<a href=#3398 id=3398 data-nosnippet>3398</a>
-<a href=#3399 id=3399 data-nosnippet>3399</a>        <span class="macro">assert_eq!</span>(all_names.len(), <span class="number">6</span>);
-<a href=#3400 id=3400 data-nosnippet>3400</a>        <span class="macro">assert_eq!</span>(all_values.len(), <span class="number">6</span>);
-<a href=#3401 id=3401 data-nosnippet>3401</a>
-<a href=#3402 id=3402 data-nosnippet>3402</a>        <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..<span class="number">6 </span>{
-<a href=#3403 id=3403 data-nosnippet>3403</a>            <span class="macro">assert_eq!</span>(all_names[i], <span class="macro">format!</span>(<span class="string">"name_{}"</span>, i));
-<a href=#3404 id=3404 data-nosnippet>3404</a>            <span class="macro">assert_eq!</span>(all_values[i], i <span class="kw">as </span>i32);
-<a href=#3405 id=3405 data-nosnippet>3405</a>        }
-<a href=#3406 id=3406 data-nosnippet>3406</a>    }
-<a href=#3407 id=3407 data-nosnippet>3407</a>
-<a href=#3408 id=3408 data-nosnippet>3408</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with nested types (struct).
-<a href=#3409 id=3409 data-nosnippet>3409</a>    /// Java's pruneColumnsFallback() projects entire top-level columns including nested content.
-<a href=#3410 id=3410 data-nosnippet>3410</a>    /// This test verifies that a top-level struct field is projected correctly with all its nested fields.
-<a href=#3411 id=3411 data-nosnippet>3411</a>    </span><span class="attr">#[tokio::test]
-<a href=#3412 id=3412 data-nosnippet>3412</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_with_struct() {
-<a href=#3413 id=3413 data-nosnippet>3413</a>        <span class="kw">use </span>arrow_array::{Int32Array, StructArray};
-<a href=#3414 id=3414 data-nosnippet>3414</a>        <span class="kw">use </span>arrow_schema::Fields;
-<a href=#3415 id=3415 data-nosnippet>3415</a>
-<a href=#3416 id=3416 data-nosnippet>3416</a>        <span class="kw">let </span>schema = Arc::new(
-<a href=#3417 id=3417 data-nosnippet>3417</a>            Schema::builder()
-<a href=#3418 id=3418 data-nosnippet>3418</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#3419 id=3419 data-nosnippet>3419</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#3420 id=3420 data-nosnippet>3420</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3421 id=3421 data-nosnippet>3421</a>                    NestedField::required(
-<a href=#3422 id=3422 data-nosnippet>3422</a>                        <span class="number">2</span>,
-<a href=#3423 id=3423 data-nosnippet>3423</a>                        <span class="string">"person"</span>,
-<a href=#3424 id=3424 data-nosnippet>3424</a>                        Type::Struct(<span class="kw">crate</span>::spec::StructType::new(<span class="macro">vec!</span>[
-<a href=#3425 id=3425 data-nosnippet>3425</a>                            NestedField::required(
-<a href=#3426 id=3426 data-nosnippet>3426</a>                                <span class="number">3</span>,
-<a href=#3427 id=3427 data-nosnippet>3427</a>                                <span class="string">"name"</span>,
-<a href=#3428 id=3428 data-nosnippet>3428</a>                                Type::Primitive(PrimitiveType::String),
-<a href=#3429 id=3429 data-nosnippet>3429</a>                            )
-<a href=#3430 id=3430 data-nosnippet>3430</a>                            .into(),
-<a href=#3431 id=3431 data-nosnippet>3431</a>                            NestedField::required(<span class="number">4</span>, <span class="string">"age"</span>, Type::Primitive(PrimitiveType::Int))
-<a href=#3432 id=3432 data-nosnippet>3432</a>                                .into(),
-<a href=#3433 id=3433 data-nosnippet>3433</a>                        ])),
-<a href=#3434 id=3434 data-nosnippet>3434</a>                    )
-<a href=#3435 id=3435 data-nosnippet>3435</a>                    .into(),
-<a href=#3436 id=3436 data-nosnippet>3436</a>                ])
-<a href=#3437 id=3437 data-nosnippet>3437</a>                .build()
-<a href=#3438 id=3438 data-nosnippet>3438</a>                .unwrap(),
-<a href=#3439 id=3439 data-nosnippet>3439</a>        );
-<a href=#3440 id=3440 data-nosnippet>3440</a>
-<a href=#3441 id=3441 data-nosnippet>3441</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#3442 id=3442 data-nosnippet>3442</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>),
-<a href=#3443 id=3443 data-nosnippet>3443</a>            Field::new(
-<a href=#3444 id=3444 data-nosnippet>3444</a>                <span class="string">"person"</span>,
-<a href=#3445 id=3445 data-nosnippet>3445</a>                DataType::Struct(Fields::from(<span class="macro">vec!</span>[
-<a href=#3446 id=3446 data-nosnippet>3446</a>                    Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
-<a href=#3447 id=3447 data-nosnippet>3447</a>                    Field::new(<span class="string">"age"</span>, DataType::Int32, <span class="bool-val">false</span>),
-<a href=#3448 id=3448 data-nosnippet>3448</a>                ])),
-<a href=#3449 id=3449 data-nosnippet>3449</a>                <span class="bool-val">false</span>,
-<a href=#3450 id=3450 data-nosnippet>3450</a>            ),
-<a href=#3451 id=3451 data-nosnippet>3451</a>        ]));
-<a href=#3452 id=3452 data-nosnippet>3452</a>
-<a href=#3453 id=3453 data-nosnippet>3453</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#3454 id=3454 data-nosnippet>3454</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#3455 id=3455 data-nosnippet>3455</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#3456 id=3456 data-nosnippet>3456</a>
-<a href=#3457 id=3457 data-nosnippet>3457</a>        <span class="kw">let </span>id_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3458 id=3458 data-nosnippet>3458</a>        <span class="kw">let </span>name_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3459 id=3459 data-nosnippet>3459</a>        <span class="kw">let </span>age_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">30</span>, <span class="number">25</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3460 id=3460 data-nosnippet>3460</a>        <span class="kw">let </span>person_data = Arc::new(StructArray::from(<span class="macro">vec!</span>[
-<a href=#3461 id=3461 data-nosnippet>3461</a>            (
-<a href=#3462 id=3462 data-nosnippet>3462</a>                Arc::new(Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>)),
-<a href=#3463 id=3463 data-nosnippet>3463</a>                name_data,
-<a href=#3464 id=3464 data-nosnippet>3464</a>            ),
-<a href=#3465 id=3465 data-nosnippet>3465</a>            (
-<a href=#3466 id=3466 data-nosnippet>3466</a>                Arc::new(Field::new(<span class="string">"age"</span>, DataType::Int32, <span class="bool-val">false</span>)),
-<a href=#3467 id=3467 data-nosnippet>3467</a>                age_data,
-<a href=#3468 id=3468 data-nosnippet>3468</a>            ),
-<a href=#3469 id=3469 data-nosnippet>3469</a>        ])) <span class="kw">as </span>ArrayRef;
-<a href=#3470 id=3470 data-nosnippet>3470</a>
-<a href=#3471 id=3471 data-nosnippet>3471</a>        <span class="kw">let </span>to_write =
-<a href=#3472 id=3472 data-nosnippet>3472</a>            RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[id_data, person_data]).unwrap();
-<a href=#3473 id=3473 data-nosnippet>3473</a>
-<a href=#3474 id=3474 data-nosnippet>3474</a>        <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#3475 id=3475 data-nosnippet>3475</a>            .set_compression(Compression::SNAPPY)
-<a href=#3476 id=3476 data-nosnippet>3476</a>            .build();
-<a href=#3477 id=3477 data-nosnippet>3477</a>
-<a href=#3478 id=3478 data-nosnippet>3478</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#3479 id=3479 data-nosnippet>3479</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#3480 id=3480 data-nosnippet>3480</a>
-<a href=#3481 id=3481 data-nosnippet>3481</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
-<a href=#3482 id=3482 data-nosnippet>3482</a>        writer.close().unwrap();
-<a href=#3483 id=3483 data-nosnippet>3483</a>
-<a href=#3484 id=3484 data-nosnippet>3484</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3300 id=3300 data-nosnippet>3300</a>        <span class="kw">let </span>to_write = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
+<a href=#3301 id=3301 data-nosnippet>3301</a>            col1_data, col2_data, col3_data, col4_data,
+<a href=#3302 id=3302 data-nosnippet>3302</a>        ])
+<a href=#3303 id=3303 data-nosnippet>3303</a>        .unwrap();
+<a href=#3304 id=3304 data-nosnippet>3304</a>
+<a href=#3305 id=3305 data-nosnippet>3305</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3306 id=3306 data-nosnippet>3306</a>            .set_compression(Compression::SNAPPY)
+<a href=#3307 id=3307 data-nosnippet>3307</a>            .build();
+<a href=#3308 id=3308 data-nosnippet>3308</a>
+<a href=#3309 id=3309 data-nosnippet>3309</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#3310 id=3310 data-nosnippet>3310</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#3311 id=3311 data-nosnippet>3311</a>
+<a href=#3312 id=3312 data-nosnippet>3312</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
+<a href=#3313 id=3313 data-nosnippet>3313</a>        writer.close().unwrap();
+<a href=#3314 id=3314 data-nosnippet>3314</a>
+<a href=#3315 id=3315 data-nosnippet>3315</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3316 id=3316 data-nosnippet>3316</a>
+<a href=#3317 id=3317 data-nosnippet>3317</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#3318 id=3318 data-nosnippet>3318</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#3319 id=3319 data-nosnippet>3319</a>                start: <span class="number">0</span>,
+<a href=#3320 id=3320 data-nosnippet>3320</a>                length: <span class="number">0</span>,
+<a href=#3321 id=3321 data-nosnippet>3321</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#3322 id=3322 data-nosnippet>3322</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
+<a href=#3323 id=3323 data-nosnippet>3323</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#3324 id=3324 data-nosnippet>3324</a>                schema: schema.clone(),
+<a href=#3325 id=3325 data-nosnippet>3325</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">3</span>],
+<a href=#3326 id=3326 data-nosnippet>3326</a>                predicate: <span class="prelude-val">None</span>,
+<a href=#3327 id=3327 data-nosnippet>3327</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#3328 id=3328 data-nosnippet>3328</a>                partition: <span class="prelude-val">None</span>,
+<a href=#3329 id=3329 data-nosnippet>3329</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#3330 id=3330 data-nosnippet>3330</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#3331 id=3331 data-nosnippet>3331</a>            })]
+<a href=#3332 id=3332 data-nosnippet>3332</a>            .into_iter(),
+<a href=#3333 id=3333 data-nosnippet>3333</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3334 id=3334 data-nosnippet>3334</a>
+<a href=#3335 id=3335 data-nosnippet>3335</a>        <span class="kw">let </span>result = reader
+<a href=#3336 id=3336 data-nosnippet>3336</a>            .read(tasks)
+<a href=#3337 id=3337 data-nosnippet>3337</a>            .unwrap()
+<a href=#3338 id=3338 data-nosnippet>3338</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3339 id=3339 data-nosnippet>3339</a>            .<span class="kw">await
+<a href=#3340 id=3340 data-nosnippet>3340</a>            </span>.unwrap();
+<a href=#3341 id=3341 data-nosnippet>3341</a>
+<a href=#3342 id=3342 data-nosnippet>3342</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
+<a href=#3343 id=3343 data-nosnippet>3343</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
+<a href=#3344 id=3344 data-nosnippet>3344</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">2</span>);
+<a href=#3345 id=3345 data-nosnippet>3345</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
+<a href=#3346 id=3346 data-nosnippet>3346</a>
+<a href=#3347 id=3347 data-nosnippet>3347</a>        <span class="kw">let </span>col1_array = batch.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
+<a href=#3348 id=3348 data-nosnippet>3348</a>        <span class="macro">assert_eq!</span>(col1_array.value(<span class="number">0</span>), <span class="string">"a"</span>);
+<a href=#3349 id=3349 data-nosnippet>3349</a>        <span class="macro">assert_eq!</span>(col1_array.value(<span class="number">1</span>), <span class="string">"b"</span>);
+<a href=#3350 id=3350 data-nosnippet>3350</a>
+<a href=#3351 id=3351 data-nosnippet>3351</a>        <span class="kw">let </span>col3_array = batch.column(<span class="number">1</span>).as_string::&lt;i32&gt;();
+<a href=#3352 id=3352 data-nosnippet>3352</a>        <span class="macro">assert_eq!</span>(col3_array.value(<span class="number">0</span>), <span class="string">"c"</span>);
+<a href=#3353 id=3353 data-nosnippet>3353</a>        <span class="macro">assert_eq!</span>(col3_array.value(<span class="number">1</span>), <span class="string">"d"</span>);
+<a href=#3354 id=3354 data-nosnippet>3354</a>    }
+<a href=#3355 id=3355 data-nosnippet>3355</a>
+<a href=#3356 id=3356 data-nosnippet>3356</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with schema evolution.
+<a href=#3357 id=3357 data-nosnippet>3357</a>    /// The Iceberg schema has more fields than the Parquet file, testing that
+<a href=#3358 id=3358 data-nosnippet>3358</a>    /// missing columns are filled with NULLs.
+<a href=#3359 id=3359 data-nosnippet>3359</a>    </span><span class="attr">#[tokio::test]
+<a href=#3360 id=3360 data-nosnippet>3360</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_schema_evolution() {
+<a href=#3361 id=3361 data-nosnippet>3361</a>        <span class="kw">use </span>arrow_array::{Array, Int32Array};
+<a href=#3362 id=3362 data-nosnippet>3362</a>
+<a href=#3363 id=3363 data-nosnippet>3363</a>        <span class="comment">// Schema with field 3 added after the file was written
+<a href=#3364 id=3364 data-nosnippet>3364</a>        </span><span class="kw">let </span>schema = Arc::new(
+<a href=#3365 id=3365 data-nosnippet>3365</a>            Schema::builder()
+<a href=#3366 id=3366 data-nosnippet>3366</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#3367 id=3367 data-nosnippet>3367</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3368 id=3368 data-nosnippet>3368</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#3369 id=3369 data-nosnippet>3369</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"age"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3370 id=3370 data-nosnippet>3370</a>                    NestedField::optional(<span class="number">3</span>, <span class="string">"city"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#3371 id=3371 data-nosnippet>3371</a>                ])
+<a href=#3372 id=3372 data-nosnippet>3372</a>                .build()
+<a href=#3373 id=3373 data-nosnippet>3373</a>                .unwrap(),
+<a href=#3374 id=3374 data-nosnippet>3374</a>        );
+<a href=#3375 id=3375 data-nosnippet>3375</a>
+<a href=#3376 id=3376 data-nosnippet>3376</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3377 id=3377 data-nosnippet>3377</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
+<a href=#3378 id=3378 data-nosnippet>3378</a>            Field::new(<span class="string">"age"</span>, DataType::Int32, <span class="bool-val">false</span>),
+<a href=#3379 id=3379 data-nosnippet>3379</a>        ]));
+<a href=#3380 id=3380 data-nosnippet>3380</a>
+<a href=#3381 id=3381 data-nosnippet>3381</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3382 id=3382 data-nosnippet>3382</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#3383 id=3383 data-nosnippet>3383</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3384 id=3384 data-nosnippet>3384</a>
+<a href=#3385 id=3385 data-nosnippet>3385</a>        <span class="kw">let </span>name_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3386 id=3386 data-nosnippet>3386</a>        <span class="kw">let </span>age_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">30</span>, <span class="number">25</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3387 id=3387 data-nosnippet>3387</a>
+<a href=#3388 id=3388 data-nosnippet>3388</a>        <span class="kw">let </span>to_write =
+<a href=#3389 id=3389 data-nosnippet>3389</a>            RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[name_data, age_data]).unwrap();
+<a href=#3390 id=3390 data-nosnippet>3390</a>
+<a href=#3391 id=3391 data-nosnippet>3391</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3392 id=3392 data-nosnippet>3392</a>            .set_compression(Compression::SNAPPY)
+<a href=#3393 id=3393 data-nosnippet>3393</a>            .build();
+<a href=#3394 id=3394 data-nosnippet>3394</a>
+<a href=#3395 id=3395 data-nosnippet>3395</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#3396 id=3396 data-nosnippet>3396</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#3397 id=3397 data-nosnippet>3397</a>
+<a href=#3398 id=3398 data-nosnippet>3398</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
+<a href=#3399 id=3399 data-nosnippet>3399</a>        writer.close().unwrap();
+<a href=#3400 id=3400 data-nosnippet>3400</a>
+<a href=#3401 id=3401 data-nosnippet>3401</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3402 id=3402 data-nosnippet>3402</a>
+<a href=#3403 id=3403 data-nosnippet>3403</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#3404 id=3404 data-nosnippet>3404</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#3405 id=3405 data-nosnippet>3405</a>                start: <span class="number">0</span>,
+<a href=#3406 id=3406 data-nosnippet>3406</a>                length: <span class="number">0</span>,
+<a href=#3407 id=3407 data-nosnippet>3407</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#3408 id=3408 data-nosnippet>3408</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
+<a href=#3409 id=3409 data-nosnippet>3409</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#3410 id=3410 data-nosnippet>3410</a>                schema: schema.clone(),
+<a href=#3411 id=3411 data-nosnippet>3411</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>],
+<a href=#3412 id=3412 data-nosnippet>3412</a>                predicate: <span class="prelude-val">None</span>,
+<a href=#3413 id=3413 data-nosnippet>3413</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#3414 id=3414 data-nosnippet>3414</a>                partition: <span class="prelude-val">None</span>,
+<a href=#3415 id=3415 data-nosnippet>3415</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#3416 id=3416 data-nosnippet>3416</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#3417 id=3417 data-nosnippet>3417</a>            })]
+<a href=#3418 id=3418 data-nosnippet>3418</a>            .into_iter(),
+<a href=#3419 id=3419 data-nosnippet>3419</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3420 id=3420 data-nosnippet>3420</a>
+<a href=#3421 id=3421 data-nosnippet>3421</a>        <span class="kw">let </span>result = reader
+<a href=#3422 id=3422 data-nosnippet>3422</a>            .read(tasks)
+<a href=#3423 id=3423 data-nosnippet>3423</a>            .unwrap()
+<a href=#3424 id=3424 data-nosnippet>3424</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3425 id=3425 data-nosnippet>3425</a>            .<span class="kw">await
+<a href=#3426 id=3426 data-nosnippet>3426</a>            </span>.unwrap();
+<a href=#3427 id=3427 data-nosnippet>3427</a>
+<a href=#3428 id=3428 data-nosnippet>3428</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
+<a href=#3429 id=3429 data-nosnippet>3429</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
+<a href=#3430 id=3430 data-nosnippet>3430</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">2</span>);
+<a href=#3431 id=3431 data-nosnippet>3431</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">3</span>);
+<a href=#3432 id=3432 data-nosnippet>3432</a>
+<a href=#3433 id=3433 data-nosnippet>3433</a>        <span class="kw">let </span>name_array = batch.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
+<a href=#3434 id=3434 data-nosnippet>3434</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
+<a href=#3435 id=3435 data-nosnippet>3435</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
+<a href=#3436 id=3436 data-nosnippet>3436</a>
+<a href=#3437 id=3437 data-nosnippet>3437</a>        <span class="kw">let </span>age_array = batch
+<a href=#3438 id=3438 data-nosnippet>3438</a>            .column(<span class="number">1</span>)
+<a href=#3439 id=3439 data-nosnippet>3439</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#3440 id=3440 data-nosnippet>3440</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">0</span>), <span class="number">30</span>);
+<a href=#3441 id=3441 data-nosnippet>3441</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">1</span>), <span class="number">25</span>);
+<a href=#3442 id=3442 data-nosnippet>3442</a>
+<a href=#3443 id=3443 data-nosnippet>3443</a>        <span class="comment">// Verify missing column filled with NULLs
+<a href=#3444 id=3444 data-nosnippet>3444</a>        </span><span class="kw">let </span>city_array = batch.column(<span class="number">2</span>).as_string::&lt;i32&gt;();
+<a href=#3445 id=3445 data-nosnippet>3445</a>        <span class="macro">assert_eq!</span>(city_array.null_count(), <span class="number">2</span>);
+<a href=#3446 id=3446 data-nosnippet>3446</a>        <span class="macro">assert!</span>(city_array.is_null(<span class="number">0</span>));
+<a href=#3447 id=3447 data-nosnippet>3447</a>        <span class="macro">assert!</span>(city_array.is_null(<span class="number">1</span>));
+<a href=#3448 id=3448 data-nosnippet>3448</a>    }
+<a href=#3449 id=3449 data-nosnippet>3449</a>
+<a href=#3450 id=3450 data-nosnippet>3450</a>    <span class="doccomment">/// Test reading Parquet files without field IDs that have multiple row groups.
+<a href=#3451 id=3451 data-nosnippet>3451</a>    /// This ensures the position-based fallback works correctly across row group boundaries.
+<a href=#3452 id=3452 data-nosnippet>3452</a>    </span><span class="attr">#[tokio::test]
+<a href=#3453 id=3453 data-nosnippet>3453</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_multiple_row_groups() {
+<a href=#3454 id=3454 data-nosnippet>3454</a>        <span class="kw">use </span>arrow_array::Int32Array;
+<a href=#3455 id=3455 data-nosnippet>3455</a>
+<a href=#3456 id=3456 data-nosnippet>3456</a>        <span class="kw">let </span>schema = Arc::new(
+<a href=#3457 id=3457 data-nosnippet>3457</a>            Schema::builder()
+<a href=#3458 id=3458 data-nosnippet>3458</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#3459 id=3459 data-nosnippet>3459</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3460 id=3460 data-nosnippet>3460</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#3461 id=3461 data-nosnippet>3461</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"value"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3462 id=3462 data-nosnippet>3462</a>                ])
+<a href=#3463 id=3463 data-nosnippet>3463</a>                .build()
+<a href=#3464 id=3464 data-nosnippet>3464</a>                .unwrap(),
+<a href=#3465 id=3465 data-nosnippet>3465</a>        );
+<a href=#3466 id=3466 data-nosnippet>3466</a>
+<a href=#3467 id=3467 data-nosnippet>3467</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3468 id=3468 data-nosnippet>3468</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
+<a href=#3469 id=3469 data-nosnippet>3469</a>            Field::new(<span class="string">"value"</span>, DataType::Int32, <span class="bool-val">false</span>),
+<a href=#3470 id=3470 data-nosnippet>3470</a>        ]));
+<a href=#3471 id=3471 data-nosnippet>3471</a>
+<a href=#3472 id=3472 data-nosnippet>3472</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3473 id=3473 data-nosnippet>3473</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#3474 id=3474 data-nosnippet>3474</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3475 id=3475 data-nosnippet>3475</a>
+<a href=#3476 id=3476 data-nosnippet>3476</a>        <span class="comment">// Small row group size to create multiple row groups
+<a href=#3477 id=3477 data-nosnippet>3477</a>        </span><span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3478 id=3478 data-nosnippet>3478</a>            .set_compression(Compression::SNAPPY)
+<a href=#3479 id=3479 data-nosnippet>3479</a>            .set_write_batch_size(<span class="number">2</span>)
+<a href=#3480 id=3480 data-nosnippet>3480</a>            .set_max_row_group_size(<span class="number">2</span>)
+<a href=#3481 id=3481 data-nosnippet>3481</a>            .build();
+<a href=#3482 id=3482 data-nosnippet>3482</a>
+<a href=#3483 id=3483 data-nosnippet>3483</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#3484 id=3484 data-nosnippet>3484</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
 <a href=#3485 id=3485 data-nosnippet>3485</a>
-<a href=#3486 id=3486 data-nosnippet>3486</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#3487 id=3487 data-nosnippet>3487</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#3488 id=3488 data-nosnippet>3488</a>                start: <span class="number">0</span>,
-<a href=#3489 id=3489 data-nosnippet>3489</a>                length: <span class="number">0</span>,
-<a href=#3490 id=3490 data-nosnippet>3490</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#3491 id=3491 data-nosnippet>3491</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
-<a href=#3492 id=3492 data-nosnippet>3492</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#3493 id=3493 data-nosnippet>3493</a>                schema: schema.clone(),
-<a href=#3494 id=3494 data-nosnippet>3494</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>],
-<a href=#3495 id=3495 data-nosnippet>3495</a>                predicate: <span class="prelude-val">None</span>,
-<a href=#3496 id=3496 data-nosnippet>3496</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#3497 id=3497 data-nosnippet>3497</a>            })]
-<a href=#3498 id=3498 data-nosnippet>3498</a>            .into_iter(),
-<a href=#3499 id=3499 data-nosnippet>3499</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3486 id=3486 data-nosnippet>3486</a>        <span class="comment">// Write 6 rows in 3 batches (will create 3 row groups)
+<a href=#3487 id=3487 data-nosnippet>3487</a>        </span><span class="kw">for </span>batch_num <span class="kw">in </span><span class="number">0</span>..<span class="number">3 </span>{
+<a href=#3488 id=3488 data-nosnippet>3488</a>            <span class="kw">let </span>name_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[
+<a href=#3489 id=3489 data-nosnippet>3489</a>                <span class="macro">format!</span>(<span class="string">"name_{}"</span>, batch_num * <span class="number">2</span>),
+<a href=#3490 id=3490 data-nosnippet>3490</a>                <span class="macro">format!</span>(<span class="string">"name_{}"</span>, batch_num * <span class="number">2 </span>+ <span class="number">1</span>),
+<a href=#3491 id=3491 data-nosnippet>3491</a>            ])) <span class="kw">as </span>ArrayRef;
+<a href=#3492 id=3492 data-nosnippet>3492</a>            <span class="kw">let </span>value_data =
+<a href=#3493 id=3493 data-nosnippet>3493</a>                Arc::new(Int32Array::from(<span class="macro">vec!</span>[batch_num * <span class="number">2</span>, batch_num * <span class="number">2 </span>+ <span class="number">1</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3494 id=3494 data-nosnippet>3494</a>
+<a href=#3495 id=3495 data-nosnippet>3495</a>            <span class="kw">let </span>batch =
+<a href=#3496 id=3496 data-nosnippet>3496</a>                RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[name_data, value_data]).unwrap();
+<a href=#3497 id=3497 data-nosnippet>3497</a>            writer.write(<span class="kw-2">&amp;</span>batch).expect(<span class="string">"Writing batch"</span>);
+<a href=#3498 id=3498 data-nosnippet>3498</a>        }
+<a href=#3499 id=3499 data-nosnippet>3499</a>        writer.close().unwrap();
 <a href=#3500 id=3500 data-nosnippet>3500</a>
-<a href=#3501 id=3501 data-nosnippet>3501</a>        <span class="kw">let </span>result = reader
-<a href=#3502 id=3502 data-nosnippet>3502</a>            .read(tasks)
-<a href=#3503 id=3503 data-nosnippet>3503</a>            .unwrap()
-<a href=#3504 id=3504 data-nosnippet>3504</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#3505 id=3505 data-nosnippet>3505</a>            .<span class="kw">await
-<a href=#3506 id=3506 data-nosnippet>3506</a>            </span>.unwrap();
-<a href=#3507 id=3507 data-nosnippet>3507</a>
-<a href=#3508 id=3508 data-nosnippet>3508</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
-<a href=#3509 id=3509 data-nosnippet>3509</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
-<a href=#3510 id=3510 data-nosnippet>3510</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">2</span>);
-<a href=#3511 id=3511 data-nosnippet>3511</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
-<a href=#3512 id=3512 data-nosnippet>3512</a>
-<a href=#3513 id=3513 data-nosnippet>3513</a>        <span class="kw">let </span>id_array = batch
-<a href=#3514 id=3514 data-nosnippet>3514</a>            .column(<span class="number">0</span>)
-<a href=#3515 id=3515 data-nosnippet>3515</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#3516 id=3516 data-nosnippet>3516</a>        <span class="macro">assert_eq!</span>(id_array.value(<span class="number">0</span>), <span class="number">1</span>);
-<a href=#3517 id=3517 data-nosnippet>3517</a>        <span class="macro">assert_eq!</span>(id_array.value(<span class="number">1</span>), <span class="number">2</span>);
-<a href=#3518 id=3518 data-nosnippet>3518</a>
-<a href=#3519 id=3519 data-nosnippet>3519</a>        <span class="kw">let </span>person_array = batch.column(<span class="number">1</span>).as_struct();
-<a href=#3520 id=3520 data-nosnippet>3520</a>        <span class="macro">assert_eq!</span>(person_array.num_columns(), <span class="number">2</span>);
-<a href=#3521 id=3521 data-nosnippet>3521</a>
-<a href=#3522 id=3522 data-nosnippet>3522</a>        <span class="kw">let </span>name_array = person_array.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
-<a href=#3523 id=3523 data-nosnippet>3523</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
-<a href=#3524 id=3524 data-nosnippet>3524</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
-<a href=#3525 id=3525 data-nosnippet>3525</a>
-<a href=#3526 id=3526 data-nosnippet>3526</a>        <span class="kw">let </span>age_array = person_array
-<a href=#3527 id=3527 data-nosnippet>3527</a>            .column(<span class="number">1</span>)
-<a href=#3528 id=3528 data-nosnippet>3528</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#3529 id=3529 data-nosnippet>3529</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">0</span>), <span class="number">30</span>);
-<a href=#3530 id=3530 data-nosnippet>3530</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">1</span>), <span class="number">25</span>);
-<a href=#3531 id=3531 data-nosnippet>3531</a>    }
+<a href=#3501 id=3501 data-nosnippet>3501</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3502 id=3502 data-nosnippet>3502</a>
+<a href=#3503 id=3503 data-nosnippet>3503</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#3504 id=3504 data-nosnippet>3504</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#3505 id=3505 data-nosnippet>3505</a>                start: <span class="number">0</span>,
+<a href=#3506 id=3506 data-nosnippet>3506</a>                length: <span class="number">0</span>,
+<a href=#3507 id=3507 data-nosnippet>3507</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#3508 id=3508 data-nosnippet>3508</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
+<a href=#3509 id=3509 data-nosnippet>3509</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#3510 id=3510 data-nosnippet>3510</a>                schema: schema.clone(),
+<a href=#3511 id=3511 data-nosnippet>3511</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>],
+<a href=#3512 id=3512 data-nosnippet>3512</a>                predicate: <span class="prelude-val">None</span>,
+<a href=#3513 id=3513 data-nosnippet>3513</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#3514 id=3514 data-nosnippet>3514</a>                partition: <span class="prelude-val">None</span>,
+<a href=#3515 id=3515 data-nosnippet>3515</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#3516 id=3516 data-nosnippet>3516</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#3517 id=3517 data-nosnippet>3517</a>            })]
+<a href=#3518 id=3518 data-nosnippet>3518</a>            .into_iter(),
+<a href=#3519 id=3519 data-nosnippet>3519</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3520 id=3520 data-nosnippet>3520</a>
+<a href=#3521 id=3521 data-nosnippet>3521</a>        <span class="kw">let </span>result = reader
+<a href=#3522 id=3522 data-nosnippet>3522</a>            .read(tasks)
+<a href=#3523 id=3523 data-nosnippet>3523</a>            .unwrap()
+<a href=#3524 id=3524 data-nosnippet>3524</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3525 id=3525 data-nosnippet>3525</a>            .<span class="kw">await
+<a href=#3526 id=3526 data-nosnippet>3526</a>            </span>.unwrap();
+<a href=#3527 id=3527 data-nosnippet>3527</a>
+<a href=#3528 id=3528 data-nosnippet>3528</a>        <span class="macro">assert!</span>(!result.is_empty());
+<a href=#3529 id=3529 data-nosnippet>3529</a>
+<a href=#3530 id=3530 data-nosnippet>3530</a>        <span class="kw">let </span><span class="kw-2">mut </span>all_names = Vec::new();
+<a href=#3531 id=3531 data-nosnippet>3531</a>        <span class="kw">let </span><span class="kw-2">mut </span>all_values = Vec::new();
 <a href=#3532 id=3532 data-nosnippet>3532</a>
-<a href=#3533 id=3533 data-nosnippet>3533</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with schema evolution - column added in the middle.
-<a href=#3534 id=3534 data-nosnippet>3534</a>    /// When a new column is inserted between existing columns in the schema order,
-<a href=#3535 id=3535 data-nosnippet>3535</a>    /// the fallback projection must correctly map field IDs to output positions.
-<a href=#3536 id=3536 data-nosnippet>3536</a>    </span><span class="attr">#[tokio::test]
-<a href=#3537 id=3537 data-nosnippet>3537</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_schema_evolution_add_column_in_middle() {
-<a href=#3538 id=3538 data-nosnippet>3538</a>        <span class="kw">use </span>arrow_array::{Array, Int32Array};
-<a href=#3539 id=3539 data-nosnippet>3539</a>
-<a href=#3540 id=3540 data-nosnippet>3540</a>        <span class="kw">let </span>arrow_schema_old = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#3541 id=3541 data-nosnippet>3541</a>            Field::new(<span class="string">"col0"</span>, DataType::Int32, <span class="bool-val">true</span>),
-<a href=#3542 id=3542 data-nosnippet>3542</a>            Field::new(<span class="string">"col1"</span>, DataType::Int32, <span class="bool-val">true</span>),
-<a href=#3543 id=3543 data-nosnippet>3543</a>        ]));
+<a href=#3533 id=3533 data-nosnippet>3533</a>        <span class="kw">for </span>batch <span class="kw">in </span><span class="kw-2">&amp;</span>result {
+<a href=#3534 id=3534 data-nosnippet>3534</a>            <span class="kw">let </span>name_array = batch.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
+<a href=#3535 id=3535 data-nosnippet>3535</a>            <span class="kw">let </span>value_array = batch
+<a href=#3536 id=3536 data-nosnippet>3536</a>                .column(<span class="number">1</span>)
+<a href=#3537 id=3537 data-nosnippet>3537</a>                .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#3538 id=3538 data-nosnippet>3538</a>
+<a href=#3539 id=3539 data-nosnippet>3539</a>            <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
+<a href=#3540 id=3540 data-nosnippet>3540</a>                all_names.push(name_array.value(i).to_string());
+<a href=#3541 id=3541 data-nosnippet>3541</a>                all_values.push(value_array.value(i));
+<a href=#3542 id=3542 data-nosnippet>3542</a>            }
+<a href=#3543 id=3543 data-nosnippet>3543</a>        }
 <a href=#3544 id=3544 data-nosnippet>3544</a>
-<a href=#3545 id=3545 data-nosnippet>3545</a>        <span class="comment">// New column added between existing columns: col0 (id=1), newCol (id=5), col1 (id=2)
-<a href=#3546 id=3546 data-nosnippet>3546</a>        </span><span class="kw">let </span>schema = Arc::new(
-<a href=#3547 id=3547 data-nosnippet>3547</a>            Schema::builder()
-<a href=#3548 id=3548 data-nosnippet>3548</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#3549 id=3549 data-nosnippet>3549</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#3550 id=3550 data-nosnippet>3550</a>                    NestedField::optional(<span class="number">1</span>, <span class="string">"col0"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3551 id=3551 data-nosnippet>3551</a>                    NestedField::optional(<span class="number">5</span>, <span class="string">"newCol"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3552 id=3552 data-nosnippet>3552</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"col1"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3553 id=3553 data-nosnippet>3553</a>                ])
-<a href=#3554 id=3554 data-nosnippet>3554</a>                .build()
-<a href=#3555 id=3555 data-nosnippet>3555</a>                .unwrap(),
-<a href=#3556 id=3556 data-nosnippet>3556</a>        );
-<a href=#3557 id=3557 data-nosnippet>3557</a>
-<a href=#3558 id=3558 data-nosnippet>3558</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#3559 id=3559 data-nosnippet>3559</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#3560 id=3560 data-nosnippet>3560</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3545 id=3545 data-nosnippet>3545</a>        <span class="macro">assert_eq!</span>(all_names.len(), <span class="number">6</span>);
+<a href=#3546 id=3546 data-nosnippet>3546</a>        <span class="macro">assert_eq!</span>(all_values.len(), <span class="number">6</span>);
+<a href=#3547 id=3547 data-nosnippet>3547</a>
+<a href=#3548 id=3548 data-nosnippet>3548</a>        <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..<span class="number">6 </span>{
+<a href=#3549 id=3549 data-nosnippet>3549</a>            <span class="macro">assert_eq!</span>(all_names[i], <span class="macro">format!</span>(<span class="string">"name_{}"</span>, i));
+<a href=#3550 id=3550 data-nosnippet>3550</a>            <span class="macro">assert_eq!</span>(all_values[i], i <span class="kw">as </span>i32);
+<a href=#3551 id=3551 data-nosnippet>3551</a>        }
+<a href=#3552 id=3552 data-nosnippet>3552</a>    }
+<a href=#3553 id=3553 data-nosnippet>3553</a>
+<a href=#3554 id=3554 data-nosnippet>3554</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with nested types (struct).
+<a href=#3555 id=3555 data-nosnippet>3555</a>    /// Java's pruneColumnsFallback() projects entire top-level columns including nested content.
+<a href=#3556 id=3556 data-nosnippet>3556</a>    /// This test verifies that a top-level struct field is projected correctly with all its nested fields.
+<a href=#3557 id=3557 data-nosnippet>3557</a>    </span><span class="attr">#[tokio::test]
+<a href=#3558 id=3558 data-nosnippet>3558</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_with_struct() {
+<a href=#3559 id=3559 data-nosnippet>3559</a>        <span class="kw">use </span>arrow_array::{Int32Array, StructArray};
+<a href=#3560 id=3560 data-nosnippet>3560</a>        <span class="kw">use </span>arrow_schema::Fields;
 <a href=#3561 id=3561 data-nosnippet>3561</a>
-<a href=#3562 id=3562 data-nosnippet>3562</a>        <span class="kw">let </span>col0_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3563 id=3563 data-nosnippet>3563</a>        <span class="kw">let </span>col1_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">10</span>, <span class="number">20</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3564 id=3564 data-nosnippet>3564</a>
-<a href=#3565 id=3565 data-nosnippet>3565</a>        <span class="kw">let </span>to_write =
-<a href=#3566 id=3566 data-nosnippet>3566</a>            RecordBatch::try_new(arrow_schema_old.clone(), <span class="macro">vec!</span>[col0_data, col1_data]).unwrap();
-<a href=#3567 id=3567 data-nosnippet>3567</a>
-<a href=#3568 id=3568 data-nosnippet>3568</a>        <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#3569 id=3569 data-nosnippet>3569</a>            .set_compression(Compression::SNAPPY)
-<a href=#3570 id=3570 data-nosnippet>3570</a>            .build();
-<a href=#3571 id=3571 data-nosnippet>3571</a>
-<a href=#3572 id=3572 data-nosnippet>3572</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#3573 id=3573 data-nosnippet>3573</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#3574 id=3574 data-nosnippet>3574</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
-<a href=#3575 id=3575 data-nosnippet>3575</a>        writer.close().unwrap();
-<a href=#3576 id=3576 data-nosnippet>3576</a>
-<a href=#3577 id=3577 data-nosnippet>3577</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
-<a href=#3578 id=3578 data-nosnippet>3578</a>
-<a href=#3579 id=3579 data-nosnippet>3579</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#3580 id=3580 data-nosnippet>3580</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#3581 id=3581 data-nosnippet>3581</a>                start: <span class="number">0</span>,
-<a href=#3582 id=3582 data-nosnippet>3582</a>                length: <span class="number">0</span>,
-<a href=#3583 id=3583 data-nosnippet>3583</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#3584 id=3584 data-nosnippet>3584</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
-<a href=#3585 id=3585 data-nosnippet>3585</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#3586 id=3586 data-nosnippet>3586</a>                schema: schema.clone(),
-<a href=#3587 id=3587 data-nosnippet>3587</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">5</span>, <span class="number">2</span>],
-<a href=#3588 id=3588 data-nosnippet>3588</a>                predicate: <span class="prelude-val">None</span>,
-<a href=#3589 id=3589 data-nosnippet>3589</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#3590 id=3590 data-nosnippet>3590</a>            })]
-<a href=#3591 id=3591 data-nosnippet>3591</a>            .into_iter(),
-<a href=#3592 id=3592 data-nosnippet>3592</a>        )) <span class="kw">as </span>FileScanTaskStream;
-<a href=#3593 id=3593 data-nosnippet>3593</a>
-<a href=#3594 id=3594 data-nosnippet>3594</a>        <span class="kw">let </span>result = reader
-<a href=#3595 id=3595 data-nosnippet>3595</a>            .read(tasks)
-<a href=#3596 id=3596 data-nosnippet>3596</a>            .unwrap()
-<a href=#3597 id=3597 data-nosnippet>3597</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#3598 id=3598 data-nosnippet>3598</a>            .<span class="kw">await
-<a href=#3599 id=3599 data-nosnippet>3599</a>            </span>.unwrap();
-<a href=#3600 id=3600 data-nosnippet>3600</a>
-<a href=#3601 id=3601 data-nosnippet>3601</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
-<a href=#3602 id=3602 data-nosnippet>3602</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
-<a href=#3603 id=3603 data-nosnippet>3603</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">2</span>);
-<a href=#3604 id=3604 data-nosnippet>3604</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">3</span>);
-<a href=#3605 id=3605 data-nosnippet>3605</a>
-<a href=#3606 id=3606 data-nosnippet>3606</a>        <span class="kw">let </span>result_col0 = batch
-<a href=#3607 id=3607 data-nosnippet>3607</a>            .column(<span class="number">0</span>)
-<a href=#3608 id=3608 data-nosnippet>3608</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#3609 id=3609 data-nosnippet>3609</a>        <span class="macro">assert_eq!</span>(result_col0.value(<span class="number">0</span>), <span class="number">1</span>);
-<a href=#3610 id=3610 data-nosnippet>3610</a>        <span class="macro">assert_eq!</span>(result_col0.value(<span class="number">1</span>), <span class="number">2</span>);
-<a href=#3611 id=3611 data-nosnippet>3611</a>
-<a href=#3612 id=3612 data-nosnippet>3612</a>        <span class="comment">// New column should be NULL (doesn't exist in old file)
-<a href=#3613 id=3613 data-nosnippet>3613</a>        </span><span class="kw">let </span>result_newcol = batch
-<a href=#3614 id=3614 data-nosnippet>3614</a>            .column(<span class="number">1</span>)
-<a href=#3615 id=3615 data-nosnippet>3615</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#3616 id=3616 data-nosnippet>3616</a>        <span class="macro">assert_eq!</span>(result_newcol.null_count(), <span class="number">2</span>);
-<a href=#3617 id=3617 data-nosnippet>3617</a>        <span class="macro">assert!</span>(result_newcol.is_null(<span class="number">0</span>));
-<a href=#3618 id=3618 data-nosnippet>3618</a>        <span class="macro">assert!</span>(result_newcol.is_null(<span class="number">1</span>));
+<a href=#3562 id=3562 data-nosnippet>3562</a>        <span class="kw">let </span>schema = Arc::new(
+<a href=#3563 id=3563 data-nosnippet>3563</a>            Schema::builder()
+<a href=#3564 id=3564 data-nosnippet>3564</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#3565 id=3565 data-nosnippet>3565</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3566 id=3566 data-nosnippet>3566</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3567 id=3567 data-nosnippet>3567</a>                    NestedField::required(
+<a href=#3568 id=3568 data-nosnippet>3568</a>                        <span class="number">2</span>,
+<a href=#3569 id=3569 data-nosnippet>3569</a>                        <span class="string">"person"</span>,
+<a href=#3570 id=3570 data-nosnippet>3570</a>                        Type::Struct(<span class="kw">crate</span>::spec::StructType::new(<span class="macro">vec!</span>[
+<a href=#3571 id=3571 data-nosnippet>3571</a>                            NestedField::required(
+<a href=#3572 id=3572 data-nosnippet>3572</a>                                <span class="number">3</span>,
+<a href=#3573 id=3573 data-nosnippet>3573</a>                                <span class="string">"name"</span>,
+<a href=#3574 id=3574 data-nosnippet>3574</a>                                Type::Primitive(PrimitiveType::String),
+<a href=#3575 id=3575 data-nosnippet>3575</a>                            )
+<a href=#3576 id=3576 data-nosnippet>3576</a>                            .into(),
+<a href=#3577 id=3577 data-nosnippet>3577</a>                            NestedField::required(<span class="number">4</span>, <span class="string">"age"</span>, Type::Primitive(PrimitiveType::Int))
+<a href=#3578 id=3578 data-nosnippet>3578</a>                                .into(),
+<a href=#3579 id=3579 data-nosnippet>3579</a>                        ])),
+<a href=#3580 id=3580 data-nosnippet>3580</a>                    )
+<a href=#3581 id=3581 data-nosnippet>3581</a>                    .into(),
+<a href=#3582 id=3582 data-nosnippet>3582</a>                ])
+<a href=#3583 id=3583 data-nosnippet>3583</a>                .build()
+<a href=#3584 id=3584 data-nosnippet>3584</a>                .unwrap(),
+<a href=#3585 id=3585 data-nosnippet>3585</a>        );
+<a href=#3586 id=3586 data-nosnippet>3586</a>
+<a href=#3587 id=3587 data-nosnippet>3587</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3588 id=3588 data-nosnippet>3588</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>),
+<a href=#3589 id=3589 data-nosnippet>3589</a>            Field::new(
+<a href=#3590 id=3590 data-nosnippet>3590</a>                <span class="string">"person"</span>,
+<a href=#3591 id=3591 data-nosnippet>3591</a>                DataType::Struct(Fields::from(<span class="macro">vec!</span>[
+<a href=#3592 id=3592 data-nosnippet>3592</a>                    Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
+<a href=#3593 id=3593 data-nosnippet>3593</a>                    Field::new(<span class="string">"age"</span>, DataType::Int32, <span class="bool-val">false</span>),
+<a href=#3594 id=3594 data-nosnippet>3594</a>                ])),
+<a href=#3595 id=3595 data-nosnippet>3595</a>                <span class="bool-val">false</span>,
+<a href=#3596 id=3596 data-nosnippet>3596</a>            ),
+<a href=#3597 id=3597 data-nosnippet>3597</a>        ]));
+<a href=#3598 id=3598 data-nosnippet>3598</a>
+<a href=#3599 id=3599 data-nosnippet>3599</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3600 id=3600 data-nosnippet>3600</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#3601 id=3601 data-nosnippet>3601</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3602 id=3602 data-nosnippet>3602</a>
+<a href=#3603 id=3603 data-nosnippet>3603</a>        <span class="kw">let </span>id_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3604 id=3604 data-nosnippet>3604</a>        <span class="kw">let </span>name_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3605 id=3605 data-nosnippet>3605</a>        <span class="kw">let </span>age_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">30</span>, <span class="number">25</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3606 id=3606 data-nosnippet>3606</a>        <span class="kw">let </span>person_data = Arc::new(StructArray::from(<span class="macro">vec!</span>[
+<a href=#3607 id=3607 data-nosnippet>3607</a>            (
+<a href=#3608 id=3608 data-nosnippet>3608</a>                Arc::new(Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>)),
+<a href=#3609 id=3609 data-nosnippet>3609</a>                name_data,
+<a href=#3610 id=3610 data-nosnippet>3610</a>            ),
+<a href=#3611 id=3611 data-nosnippet>3611</a>            (
+<a href=#3612 id=3612 data-nosnippet>3612</a>                Arc::new(Field::new(<span class="string">"age"</span>, DataType::Int32, <span class="bool-val">false</span>)),
+<a href=#3613 id=3613 data-nosnippet>3613</a>                age_data,
+<a href=#3614 id=3614 data-nosnippet>3614</a>            ),
+<a href=#3615 id=3615 data-nosnippet>3615</a>        ])) <span class="kw">as </span>ArrayRef;
+<a href=#3616 id=3616 data-nosnippet>3616</a>
+<a href=#3617 id=3617 data-nosnippet>3617</a>        <span class="kw">let </span>to_write =
+<a href=#3618 id=3618 data-nosnippet>3618</a>            RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[id_data, person_data]).unwrap();
 <a href=#3619 id=3619 data-nosnippet>3619</a>
-<a href=#3620 id=3620 data-nosnippet>3620</a>        <span class="kw">let </span>result_col1 = batch
-<a href=#3621 id=3621 data-nosnippet>3621</a>            .column(<span class="number">2</span>)
-<a href=#3622 id=3622 data-nosnippet>3622</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
-<a href=#3623 id=3623 data-nosnippet>3623</a>        <span class="macro">assert_eq!</span>(result_col1.value(<span class="number">0</span>), <span class="number">10</span>);
-<a href=#3624 id=3624 data-nosnippet>3624</a>        <span class="macro">assert_eq!</span>(result_col1.value(<span class="number">1</span>), <span class="number">20</span>);
-<a href=#3625 id=3625 data-nosnippet>3625</a>    }
+<a href=#3620 id=3620 data-nosnippet>3620</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3621 id=3621 data-nosnippet>3621</a>            .set_compression(Compression::SNAPPY)
+<a href=#3622 id=3622 data-nosnippet>3622</a>            .build();
+<a href=#3623 id=3623 data-nosnippet>3623</a>
+<a href=#3624 id=3624 data-nosnippet>3624</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#3625 id=3625 data-nosnippet>3625</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
 <a href=#3626 id=3626 data-nosnippet>3626</a>
-<a href=#3627 id=3627 data-nosnippet>3627</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with a filter that eliminates all row groups.
-<a href=#3628 id=3628 data-nosnippet>3628</a>    /// During development of field ID mapping, we saw a panic when row_selection_enabled=true and
-<a href=#3629 id=3629 data-nosnippet>3629</a>    /// all row groups are filtered out.
-<a href=#3630 id=3630 data-nosnippet>3630</a>    </span><span class="attr">#[tokio::test]
-<a href=#3631 id=3631 data-nosnippet>3631</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_filter_eliminates_all_rows() {
-<a href=#3632 id=3632 data-nosnippet>3632</a>        <span class="kw">use </span>arrow_array::{Float64Array, Int32Array};
-<a href=#3633 id=3633 data-nosnippet>3633</a>
-<a href=#3634 id=3634 data-nosnippet>3634</a>        <span class="comment">// Schema with fields that will use fallback IDs 1, 2, 3
-<a href=#3635 id=3635 data-nosnippet>3635</a>        </span><span class="kw">let </span>schema = Arc::new(
-<a href=#3636 id=3636 data-nosnippet>3636</a>            Schema::builder()
-<a href=#3637 id=3637 data-nosnippet>3637</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#3638 id=3638 data-nosnippet>3638</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#3639 id=3639 data-nosnippet>3639</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#3640 id=3640 data-nosnippet>3640</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#3641 id=3641 data-nosnippet>3641</a>                    NestedField::required(<span class="number">3</span>, <span class="string">"value"</span>, Type::Primitive(PrimitiveType::Double))
-<a href=#3642 id=3642 data-nosnippet>3642</a>                        .into(),
-<a href=#3643 id=3643 data-nosnippet>3643</a>                ])
-<a href=#3644 id=3644 data-nosnippet>3644</a>                .build()
-<a href=#3645 id=3645 data-nosnippet>3645</a>                .unwrap(),
-<a href=#3646 id=3646 data-nosnippet>3646</a>        );
-<a href=#3647 id=3647 data-nosnippet>3647</a>
-<a href=#3648 id=3648 data-nosnippet>3648</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#3649 id=3649 data-nosnippet>3649</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>),
-<a href=#3650 id=3650 data-nosnippet>3650</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
-<a href=#3651 id=3651 data-nosnippet>3651</a>            Field::new(<span class="string">"value"</span>, DataType::Float64, <span class="bool-val">false</span>),
-<a href=#3652 id=3652 data-nosnippet>3652</a>        ]));
-<a href=#3653 id=3653 data-nosnippet>3653</a>
-<a href=#3654 id=3654 data-nosnippet>3654</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#3655 id=3655 data-nosnippet>3655</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
-<a href=#3656 id=3656 data-nosnippet>3656</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
-<a href=#3657 id=3657 data-nosnippet>3657</a>
-<a href=#3658 id=3658 data-nosnippet>3658</a>        <span class="comment">// Write data where all ids are &gt;= 10
-<a href=#3659 id=3659 data-nosnippet>3659</a>        </span><span class="kw">let </span>id_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">10</span>, <span class="number">11</span>, <span class="number">12</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3660 id=3660 data-nosnippet>3660</a>        <span class="kw">let </span>name_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"a"</span>, <span class="string">"b"</span>, <span class="string">"c"</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3661 id=3661 data-nosnippet>3661</a>        <span class="kw">let </span>value_data = Arc::new(Float64Array::from(<span class="macro">vec!</span>[<span class="number">100.0</span>, <span class="number">200.0</span>, <span class="number">300.0</span>])) <span class="kw">as </span>ArrayRef;
-<a href=#3662 id=3662 data-nosnippet>3662</a>
-<a href=#3663 id=3663 data-nosnippet>3663</a>        <span class="kw">let </span>to_write =
-<a href=#3664 id=3664 data-nosnippet>3664</a>            RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[id_data, name_data, value_data])
-<a href=#3665 id=3665 data-nosnippet>3665</a>                .unwrap();
-<a href=#3666 id=3666 data-nosnippet>3666</a>
-<a href=#3667 id=3667 data-nosnippet>3667</a>        <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#3668 id=3668 data-nosnippet>3668</a>            .set_compression(Compression::SNAPPY)
-<a href=#3669 id=3669 data-nosnippet>3669</a>            .build();
+<a href=#3627 id=3627 data-nosnippet>3627</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
+<a href=#3628 id=3628 data-nosnippet>3628</a>        writer.close().unwrap();
+<a href=#3629 id=3629 data-nosnippet>3629</a>
+<a href=#3630 id=3630 data-nosnippet>3630</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3631 id=3631 data-nosnippet>3631</a>
+<a href=#3632 id=3632 data-nosnippet>3632</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#3633 id=3633 data-nosnippet>3633</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#3634 id=3634 data-nosnippet>3634</a>                start: <span class="number">0</span>,
+<a href=#3635 id=3635 data-nosnippet>3635</a>                length: <span class="number">0</span>,
+<a href=#3636 id=3636 data-nosnippet>3636</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#3637 id=3637 data-nosnippet>3637</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
+<a href=#3638 id=3638 data-nosnippet>3638</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#3639 id=3639 data-nosnippet>3639</a>                schema: schema.clone(),
+<a href=#3640 id=3640 data-nosnippet>3640</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>],
+<a href=#3641 id=3641 data-nosnippet>3641</a>                predicate: <span class="prelude-val">None</span>,
+<a href=#3642 id=3642 data-nosnippet>3642</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#3643 id=3643 data-nosnippet>3643</a>                partition: <span class="prelude-val">None</span>,
+<a href=#3644 id=3644 data-nosnippet>3644</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#3645 id=3645 data-nosnippet>3645</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#3646 id=3646 data-nosnippet>3646</a>            })]
+<a href=#3647 id=3647 data-nosnippet>3647</a>            .into_iter(),
+<a href=#3648 id=3648 data-nosnippet>3648</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3649 id=3649 data-nosnippet>3649</a>
+<a href=#3650 id=3650 data-nosnippet>3650</a>        <span class="kw">let </span>result = reader
+<a href=#3651 id=3651 data-nosnippet>3651</a>            .read(tasks)
+<a href=#3652 id=3652 data-nosnippet>3652</a>            .unwrap()
+<a href=#3653 id=3653 data-nosnippet>3653</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3654 id=3654 data-nosnippet>3654</a>            .<span class="kw">await
+<a href=#3655 id=3655 data-nosnippet>3655</a>            </span>.unwrap();
+<a href=#3656 id=3656 data-nosnippet>3656</a>
+<a href=#3657 id=3657 data-nosnippet>3657</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
+<a href=#3658 id=3658 data-nosnippet>3658</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
+<a href=#3659 id=3659 data-nosnippet>3659</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">2</span>);
+<a href=#3660 id=3660 data-nosnippet>3660</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
+<a href=#3661 id=3661 data-nosnippet>3661</a>
+<a href=#3662 id=3662 data-nosnippet>3662</a>        <span class="kw">let </span>id_array = batch
+<a href=#3663 id=3663 data-nosnippet>3663</a>            .column(<span class="number">0</span>)
+<a href=#3664 id=3664 data-nosnippet>3664</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#3665 id=3665 data-nosnippet>3665</a>        <span class="macro">assert_eq!</span>(id_array.value(<span class="number">0</span>), <span class="number">1</span>);
+<a href=#3666 id=3666 data-nosnippet>3666</a>        <span class="macro">assert_eq!</span>(id_array.value(<span class="number">1</span>), <span class="number">2</span>);
+<a href=#3667 id=3667 data-nosnippet>3667</a>
+<a href=#3668 id=3668 data-nosnippet>3668</a>        <span class="kw">let </span>person_array = batch.column(<span class="number">1</span>).as_struct();
+<a href=#3669 id=3669 data-nosnippet>3669</a>        <span class="macro">assert_eq!</span>(person_array.num_columns(), <span class="number">2</span>);
 <a href=#3670 id=3670 data-nosnippet>3670</a>
-<a href=#3671 id=3671 data-nosnippet>3671</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
-<a href=#3672 id=3672 data-nosnippet>3672</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#3673 id=3673 data-nosnippet>3673</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
-<a href=#3674 id=3674 data-nosnippet>3674</a>        writer.close().unwrap();
-<a href=#3675 id=3675 data-nosnippet>3675</a>
-<a href=#3676 id=3676 data-nosnippet>3676</a>        <span class="comment">// Filter that eliminates all row groups: id &lt; 5
-<a href=#3677 id=3677 data-nosnippet>3677</a>        </span><span class="kw">let </span>predicate = Reference::new(<span class="string">"id"</span>).less_than(Datum::int(<span class="number">5</span>));
-<a href=#3678 id=3678 data-nosnippet>3678</a>
-<a href=#3679 id=3679 data-nosnippet>3679</a>        <span class="comment">// Enable both row_group_filtering and row_selection - triggered the panic
-<a href=#3680 id=3680 data-nosnippet>3680</a>        </span><span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io)
-<a href=#3681 id=3681 data-nosnippet>3681</a>            .with_row_group_filtering_enabled(<span class="bool-val">true</span>)
-<a href=#3682 id=3682 data-nosnippet>3682</a>            .with_row_selection_enabled(<span class="bool-val">true</span>)
-<a href=#3683 id=3683 data-nosnippet>3683</a>            .build();
-<a href=#3684 id=3684 data-nosnippet>3684</a>
-<a href=#3685 id=3685 data-nosnippet>3685</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
-<a href=#3686 id=3686 data-nosnippet>3686</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
-<a href=#3687 id=3687 data-nosnippet>3687</a>                start: <span class="number">0</span>,
-<a href=#3688 id=3688 data-nosnippet>3688</a>                length: <span class="number">0</span>,
-<a href=#3689 id=3689 data-nosnippet>3689</a>                record_count: <span class="prelude-val">None</span>,
-<a href=#3690 id=3690 data-nosnippet>3690</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
-<a href=#3691 id=3691 data-nosnippet>3691</a>                data_file_format: DataFileFormat::Parquet,
-<a href=#3692 id=3692 data-nosnippet>3692</a>                schema: schema.clone(),
-<a href=#3693 id=3693 data-nosnippet>3693</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>],
-<a href=#3694 id=3694 data-nosnippet>3694</a>                predicate: <span class="prelude-val">Some</span>(predicate.bind(schema, <span class="bool-val">true</span>).unwrap()),
-<a href=#3695 id=3695 data-nosnippet>3695</a>                deletes: <span class="macro">vec!</span>[],
-<a href=#3696 id=3696 data-nosnippet>3696</a>            })]
-<a href=#3697 id=3697 data-nosnippet>3697</a>            .into_iter(),
-<a href=#3698 id=3698 data-nosnippet>3698</a>        )) <span class="kw">as </span>FileScanTaskStream;
-<a href=#3699 id=3699 data-nosnippet>3699</a>
-<a href=#3700 id=3700 data-nosnippet>3700</a>        <span class="comment">// Should no longer panic
-<a href=#3701 id=3701 data-nosnippet>3701</a>        </span><span class="kw">let </span>result = reader
-<a href=#3702 id=3702 data-nosnippet>3702</a>            .read(tasks)
-<a href=#3703 id=3703 data-nosnippet>3703</a>            .unwrap()
-<a href=#3704 id=3704 data-nosnippet>3704</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
-<a href=#3705 id=3705 data-nosnippet>3705</a>            .<span class="kw">await
-<a href=#3706 id=3706 data-nosnippet>3706</a>            </span>.unwrap();
-<a href=#3707 id=3707 data-nosnippet>3707</a>
-<a href=#3708 id=3708 data-nosnippet>3708</a>        <span class="comment">// Should return empty results
-<a href=#3709 id=3709 data-nosnippet>3709</a>        </span><span class="macro">assert!</span>(result.is_empty() || result.iter().all(|batch| batch.num_rows() == <span class="number">0</span>));
-<a href=#3710 id=3710 data-nosnippet>3710</a>    }
-<a href=#3711 id=3711 data-nosnippet>3711</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file
+<a href=#3671 id=3671 data-nosnippet>3671</a>        <span class="kw">let </span>name_array = person_array.column(<span class="number">0</span>).as_string::&lt;i32&gt;();
+<a href=#3672 id=3672 data-nosnippet>3672</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
+<a href=#3673 id=3673 data-nosnippet>3673</a>        <span class="macro">assert_eq!</span>(name_array.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
+<a href=#3674 id=3674 data-nosnippet>3674</a>
+<a href=#3675 id=3675 data-nosnippet>3675</a>        <span class="kw">let </span>age_array = person_array
+<a href=#3676 id=3676 data-nosnippet>3676</a>            .column(<span class="number">1</span>)
+<a href=#3677 id=3677 data-nosnippet>3677</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#3678 id=3678 data-nosnippet>3678</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">0</span>), <span class="number">30</span>);
+<a href=#3679 id=3679 data-nosnippet>3679</a>        <span class="macro">assert_eq!</span>(age_array.value(<span class="number">1</span>), <span class="number">25</span>);
+<a href=#3680 id=3680 data-nosnippet>3680</a>    }
+<a href=#3681 id=3681 data-nosnippet>3681</a>
+<a href=#3682 id=3682 data-nosnippet>3682</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with schema evolution - column added in the middle.
+<a href=#3683 id=3683 data-nosnippet>3683</a>    /// When a new column is inserted between existing columns in the schema order,
+<a href=#3684 id=3684 data-nosnippet>3684</a>    /// the fallback projection must correctly map field IDs to output positions.
+<a href=#3685 id=3685 data-nosnippet>3685</a>    </span><span class="attr">#[tokio::test]
+<a href=#3686 id=3686 data-nosnippet>3686</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_schema_evolution_add_column_in_middle() {
+<a href=#3687 id=3687 data-nosnippet>3687</a>        <span class="kw">use </span>arrow_array::{Array, Int32Array};
+<a href=#3688 id=3688 data-nosnippet>3688</a>
+<a href=#3689 id=3689 data-nosnippet>3689</a>        <span class="kw">let </span>arrow_schema_old = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3690 id=3690 data-nosnippet>3690</a>            Field::new(<span class="string">"col0"</span>, DataType::Int32, <span class="bool-val">true</span>),
+<a href=#3691 id=3691 data-nosnippet>3691</a>            Field::new(<span class="string">"col1"</span>, DataType::Int32, <span class="bool-val">true</span>),
+<a href=#3692 id=3692 data-nosnippet>3692</a>        ]));
+<a href=#3693 id=3693 data-nosnippet>3693</a>
+<a href=#3694 id=3694 data-nosnippet>3694</a>        <span class="comment">// New column added between existing columns: col0 (id=1), newCol (id=5), col1 (id=2)
+<a href=#3695 id=3695 data-nosnippet>3695</a>        </span><span class="kw">let </span>schema = Arc::new(
+<a href=#3696 id=3696 data-nosnippet>3696</a>            Schema::builder()
+<a href=#3697 id=3697 data-nosnippet>3697</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#3698 id=3698 data-nosnippet>3698</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3699 id=3699 data-nosnippet>3699</a>                    NestedField::optional(<span class="number">1</span>, <span class="string">"col0"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3700 id=3700 data-nosnippet>3700</a>                    NestedField::optional(<span class="number">5</span>, <span class="string">"newCol"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3701 id=3701 data-nosnippet>3701</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"col1"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3702 id=3702 data-nosnippet>3702</a>                ])
+<a href=#3703 id=3703 data-nosnippet>3703</a>                .build()
+<a href=#3704 id=3704 data-nosnippet>3704</a>                .unwrap(),
+<a href=#3705 id=3705 data-nosnippet>3705</a>        );
+<a href=#3706 id=3706 data-nosnippet>3706</a>
+<a href=#3707 id=3707 data-nosnippet>3707</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3708 id=3708 data-nosnippet>3708</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#3709 id=3709 data-nosnippet>3709</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3710 id=3710 data-nosnippet>3710</a>
+<a href=#3711 id=3711 data-nosnippet>3711</a>        <span class="kw">let </span>col0_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3712 id=3712 data-nosnippet>3712</a>        <span class="kw">let </span>col1_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">10</span>, <span class="number">20</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3713 id=3713 data-nosnippet>3713</a>
+<a href=#3714 id=3714 data-nosnippet>3714</a>        <span class="kw">let </span>to_write =
+<a href=#3715 id=3715 data-nosnippet>3715</a>            RecordBatch::try_new(arrow_schema_old.clone(), <span class="macro">vec!</span>[col0_data, col1_data]).unwrap();
+<a href=#3716 id=3716 data-nosnippet>3716</a>
+<a href=#3717 id=3717 data-nosnippet>3717</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3718 id=3718 data-nosnippet>3718</a>            .set_compression(Compression::SNAPPY)
+<a href=#3719 id=3719 data-nosnippet>3719</a>            .build();
+<a href=#3720 id=3720 data-nosnippet>3720</a>
+<a href=#3721 id=3721 data-nosnippet>3721</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#3722 id=3722 data-nosnippet>3722</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#3723 id=3723 data-nosnippet>3723</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
+<a href=#3724 id=3724 data-nosnippet>3724</a>        writer.close().unwrap();
+<a href=#3725 id=3725 data-nosnippet>3725</a>
+<a href=#3726 id=3726 data-nosnippet>3726</a>        <span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3727 id=3727 data-nosnippet>3727</a>
+<a href=#3728 id=3728 data-nosnippet>3728</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#3729 id=3729 data-nosnippet>3729</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#3730 id=3730 data-nosnippet>3730</a>                start: <span class="number">0</span>,
+<a href=#3731 id=3731 data-nosnippet>3731</a>                length: <span class="number">0</span>,
+<a href=#3732 id=3732 data-nosnippet>3732</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#3733 id=3733 data-nosnippet>3733</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
+<a href=#3734 id=3734 data-nosnippet>3734</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#3735 id=3735 data-nosnippet>3735</a>                schema: schema.clone(),
+<a href=#3736 id=3736 data-nosnippet>3736</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">5</span>, <span class="number">2</span>],
+<a href=#3737 id=3737 data-nosnippet>3737</a>                predicate: <span class="prelude-val">None</span>,
+<a href=#3738 id=3738 data-nosnippet>3738</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#3739 id=3739 data-nosnippet>3739</a>                partition: <span class="prelude-val">None</span>,
+<a href=#3740 id=3740 data-nosnippet>3740</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#3741 id=3741 data-nosnippet>3741</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#3742 id=3742 data-nosnippet>3742</a>            })]
+<a href=#3743 id=3743 data-nosnippet>3743</a>            .into_iter(),
+<a href=#3744 id=3744 data-nosnippet>3744</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3745 id=3745 data-nosnippet>3745</a>
+<a href=#3746 id=3746 data-nosnippet>3746</a>        <span class="kw">let </span>result = reader
+<a href=#3747 id=3747 data-nosnippet>3747</a>            .read(tasks)
+<a href=#3748 id=3748 data-nosnippet>3748</a>            .unwrap()
+<a href=#3749 id=3749 data-nosnippet>3749</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3750 id=3750 data-nosnippet>3750</a>            .<span class="kw">await
+<a href=#3751 id=3751 data-nosnippet>3751</a>            </span>.unwrap();
+<a href=#3752 id=3752 data-nosnippet>3752</a>
+<a href=#3753 id=3753 data-nosnippet>3753</a>        <span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
+<a href=#3754 id=3754 data-nosnippet>3754</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
+<a href=#3755 id=3755 data-nosnippet>3755</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">2</span>);
+<a href=#3756 id=3756 data-nosnippet>3756</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">3</span>);
+<a href=#3757 id=3757 data-nosnippet>3757</a>
+<a href=#3758 id=3758 data-nosnippet>3758</a>        <span class="kw">let </span>result_col0 = batch
+<a href=#3759 id=3759 data-nosnippet>3759</a>            .column(<span class="number">0</span>)
+<a href=#3760 id=3760 data-nosnippet>3760</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#3761 id=3761 data-nosnippet>3761</a>        <span class="macro">assert_eq!</span>(result_col0.value(<span class="number">0</span>), <span class="number">1</span>);
+<a href=#3762 id=3762 data-nosnippet>3762</a>        <span class="macro">assert_eq!</span>(result_col0.value(<span class="number">1</span>), <span class="number">2</span>);
+<a href=#3763 id=3763 data-nosnippet>3763</a>
+<a href=#3764 id=3764 data-nosnippet>3764</a>        <span class="comment">// New column should be NULL (doesn't exist in old file)
+<a href=#3765 id=3765 data-nosnippet>3765</a>        </span><span class="kw">let </span>result_newcol = batch
+<a href=#3766 id=3766 data-nosnippet>3766</a>            .column(<span class="number">1</span>)
+<a href=#3767 id=3767 data-nosnippet>3767</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#3768 id=3768 data-nosnippet>3768</a>        <span class="macro">assert_eq!</span>(result_newcol.null_count(), <span class="number">2</span>);
+<a href=#3769 id=3769 data-nosnippet>3769</a>        <span class="macro">assert!</span>(result_newcol.is_null(<span class="number">0</span>));
+<a href=#3770 id=3770 data-nosnippet>3770</a>        <span class="macro">assert!</span>(result_newcol.is_null(<span class="number">1</span>));
+<a href=#3771 id=3771 data-nosnippet>3771</a>
+<a href=#3772 id=3772 data-nosnippet>3772</a>        <span class="kw">let </span>result_col1 = batch
+<a href=#3773 id=3773 data-nosnippet>3773</a>            .column(<span class="number">2</span>)
+<a href=#3774 id=3774 data-nosnippet>3774</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#3775 id=3775 data-nosnippet>3775</a>        <span class="macro">assert_eq!</span>(result_col1.value(<span class="number">0</span>), <span class="number">10</span>);
+<a href=#3776 id=3776 data-nosnippet>3776</a>        <span class="macro">assert_eq!</span>(result_col1.value(<span class="number">1</span>), <span class="number">20</span>);
+<a href=#3777 id=3777 data-nosnippet>3777</a>    }
+<a href=#3778 id=3778 data-nosnippet>3778</a>
+<a href=#3779 id=3779 data-nosnippet>3779</a>    <span class="doccomment">/// Test reading Parquet files without field IDs with a filter that eliminates all row groups.
+<a href=#3780 id=3780 data-nosnippet>3780</a>    /// During development of field ID mapping, we saw a panic when row_selection_enabled=true and
+<a href=#3781 id=3781 data-nosnippet>3781</a>    /// all row groups are filtered out.
+<a href=#3782 id=3782 data-nosnippet>3782</a>    </span><span class="attr">#[tokio::test]
+<a href=#3783 id=3783 data-nosnippet>3783</a>    </span><span class="kw">async fn </span>test_read_parquet_without_field_ids_filter_eliminates_all_rows() {
+<a href=#3784 id=3784 data-nosnippet>3784</a>        <span class="kw">use </span>arrow_array::{Float64Array, Int32Array};
+<a href=#3785 id=3785 data-nosnippet>3785</a>
+<a href=#3786 id=3786 data-nosnippet>3786</a>        <span class="comment">// Schema with fields that will use fallback IDs 1, 2, 3
+<a href=#3787 id=3787 data-nosnippet>3787</a>        </span><span class="kw">let </span>schema = Arc::new(
+<a href=#3788 id=3788 data-nosnippet>3788</a>            Schema::builder()
+<a href=#3789 id=3789 data-nosnippet>3789</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#3790 id=3790 data-nosnippet>3790</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3791 id=3791 data-nosnippet>3791</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3792 id=3792 data-nosnippet>3792</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#3793 id=3793 data-nosnippet>3793</a>                    NestedField::required(<span class="number">3</span>, <span class="string">"value"</span>, Type::Primitive(PrimitiveType::Double))
+<a href=#3794 id=3794 data-nosnippet>3794</a>                        .into(),
+<a href=#3795 id=3795 data-nosnippet>3795</a>                ])
+<a href=#3796 id=3796 data-nosnippet>3796</a>                .build()
+<a href=#3797 id=3797 data-nosnippet>3797</a>                .unwrap(),
+<a href=#3798 id=3798 data-nosnippet>3798</a>        );
+<a href=#3799 id=3799 data-nosnippet>3799</a>
+<a href=#3800 id=3800 data-nosnippet>3800</a>        <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3801 id=3801 data-nosnippet>3801</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>),
+<a href=#3802 id=3802 data-nosnippet>3802</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">false</span>),
+<a href=#3803 id=3803 data-nosnippet>3803</a>            Field::new(<span class="string">"value"</span>, DataType::Float64, <span class="bool-val">false</span>),
+<a href=#3804 id=3804 data-nosnippet>3804</a>        ]));
+<a href=#3805 id=3805 data-nosnippet>3805</a>
+<a href=#3806 id=3806 data-nosnippet>3806</a>        <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3807 id=3807 data-nosnippet>3807</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#3808 id=3808 data-nosnippet>3808</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3809 id=3809 data-nosnippet>3809</a>
+<a href=#3810 id=3810 data-nosnippet>3810</a>        <span class="comment">// Write data where all ids are &gt;= 10
+<a href=#3811 id=3811 data-nosnippet>3811</a>        </span><span class="kw">let </span>id_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">10</span>, <span class="number">11</span>, <span class="number">12</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3812 id=3812 data-nosnippet>3812</a>        <span class="kw">let </span>name_data = Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"a"</span>, <span class="string">"b"</span>, <span class="string">"c"</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3813 id=3813 data-nosnippet>3813</a>        <span class="kw">let </span>value_data = Arc::new(Float64Array::from(<span class="macro">vec!</span>[<span class="number">100.0</span>, <span class="number">200.0</span>, <span class="number">300.0</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3814 id=3814 data-nosnippet>3814</a>
+<a href=#3815 id=3815 data-nosnippet>3815</a>        <span class="kw">let </span>to_write =
+<a href=#3816 id=3816 data-nosnippet>3816</a>            RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[id_data, name_data, value_data])
+<a href=#3817 id=3817 data-nosnippet>3817</a>                .unwrap();
+<a href=#3818 id=3818 data-nosnippet>3818</a>
+<a href=#3819 id=3819 data-nosnippet>3819</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3820 id=3820 data-nosnippet>3820</a>            .set_compression(Compression::SNAPPY)
+<a href=#3821 id=3821 data-nosnippet>3821</a>            .build();
+<a href=#3822 id=3822 data-nosnippet>3822</a>
+<a href=#3823 id=3823 data-nosnippet>3823</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#3824 id=3824 data-nosnippet>3824</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#3825 id=3825 data-nosnippet>3825</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
+<a href=#3826 id=3826 data-nosnippet>3826</a>        writer.close().unwrap();
+<a href=#3827 id=3827 data-nosnippet>3827</a>
+<a href=#3828 id=3828 data-nosnippet>3828</a>        <span class="comment">// Filter that eliminates all row groups: id &lt; 5
+<a href=#3829 id=3829 data-nosnippet>3829</a>        </span><span class="kw">let </span>predicate = Reference::new(<span class="string">"id"</span>).less_than(Datum::int(<span class="number">5</span>));
+<a href=#3830 id=3830 data-nosnippet>3830</a>
+<a href=#3831 id=3831 data-nosnippet>3831</a>        <span class="comment">// Enable both row_group_filtering and row_selection - triggered the panic
+<a href=#3832 id=3832 data-nosnippet>3832</a>        </span><span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io)
+<a href=#3833 id=3833 data-nosnippet>3833</a>            .with_row_group_filtering_enabled(<span class="bool-val">true</span>)
+<a href=#3834 id=3834 data-nosnippet>3834</a>            .with_row_selection_enabled(<span class="bool-val">true</span>)
+<a href=#3835 id=3835 data-nosnippet>3835</a>            .build();
+<a href=#3836 id=3836 data-nosnippet>3836</a>
+<a href=#3837 id=3837 data-nosnippet>3837</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#3838 id=3838 data-nosnippet>3838</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#3839 id=3839 data-nosnippet>3839</a>                start: <span class="number">0</span>,
+<a href=#3840 id=3840 data-nosnippet>3840</a>                length: <span class="number">0</span>,
+<a href=#3841 id=3841 data-nosnippet>3841</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#3842 id=3842 data-nosnippet>3842</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/1.parquet"</span>, table_location),
+<a href=#3843 id=3843 data-nosnippet>3843</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#3844 id=3844 data-nosnippet>3844</a>                schema: schema.clone(),
+<a href=#3845 id=3845 data-nosnippet>3845</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>],
+<a href=#3846 id=3846 data-nosnippet>3846</a>                predicate: <span class="prelude-val">Some</span>(predicate.bind(schema, <span class="bool-val">true</span>).unwrap()),
+<a href=#3847 id=3847 data-nosnippet>3847</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#3848 id=3848 data-nosnippet>3848</a>                partition: <span class="prelude-val">None</span>,
+<a href=#3849 id=3849 data-nosnippet>3849</a>                partition_spec: <span class="prelude-val">None</span>,
+<a href=#3850 id=3850 data-nosnippet>3850</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#3851 id=3851 data-nosnippet>3851</a>            })]
+<a href=#3852 id=3852 data-nosnippet>3852</a>            .into_iter(),
+<a href=#3853 id=3853 data-nosnippet>3853</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3854 id=3854 data-nosnippet>3854</a>
+<a href=#3855 id=3855 data-nosnippet>3855</a>        <span class="comment">// Should no longer panic
+<a href=#3856 id=3856 data-nosnippet>3856</a>        </span><span class="kw">let </span>result = reader
+<a href=#3857 id=3857 data-nosnippet>3857</a>            .read(tasks)
+<a href=#3858 id=3858 data-nosnippet>3858</a>            .unwrap()
+<a href=#3859 id=3859 data-nosnippet>3859</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3860 id=3860 data-nosnippet>3860</a>            .<span class="kw">await
+<a href=#3861 id=3861 data-nosnippet>3861</a>            </span>.unwrap();
+<a href=#3862 id=3862 data-nosnippet>3862</a>
+<a href=#3863 id=3863 data-nosnippet>3863</a>        <span class="comment">// Should return empty results
+<a href=#3864 id=3864 data-nosnippet>3864</a>        </span><span class="macro">assert!</span>(result.is_empty() || result.iter().all(|batch| batch.num_rows() == <span class="number">0</span>));
+<a href=#3865 id=3865 data-nosnippet>3865</a>    }
+<a href=#3866 id=3866 data-nosnippet>3866</a>
+<a href=#3867 id=3867 data-nosnippet>3867</a>    <span class="doccomment">/// Test bucket partitioning reads source column from data file (not partition metadata).
+<a href=#3868 id=3868 data-nosnippet>3868</a>    ///
+<a href=#3869 id=3869 data-nosnippet>3869</a>    /// This is an integration test verifying the complete ArrowReader pipeline with bucket partitioning.
+<a href=#3870 id=3870 data-nosnippet>3870</a>    /// It corresponds to TestRuntimeFiltering tests in Iceberg Java (e.g., testRenamedSourceColumnTable).
+<a href=#3871 id=3871 data-nosnippet>3871</a>    ///
+<a href=#3872 id=3872 data-nosnippet>3872</a>    /// # Iceberg Spec Requirements
+<a href=#3873 id=3873 data-nosnippet>3873</a>    ///
+<a href=#3874 id=3874 data-nosnippet>3874</a>    /// Per the Iceberg spec "Column Projection" section:
+<a href=#3875 id=3875 data-nosnippet>3875</a>    /// &gt; "Return the value from partition metadata if an **Identity Transform** exists for the field"
+<a href=#3876 id=3876 data-nosnippet>3876</a>    ///
+<a href=#3877 id=3877 data-nosnippet>3877</a>    /// This means:
+<a href=#3878 id=3878 data-nosnippet>3878</a>    /// - Identity transforms (e.g., `identity(dept)`) use constants from partition metadata
+<a href=#3879 id=3879 data-nosnippet>3879</a>    /// - Non-identity transforms (e.g., `bucket(4, id)`) must read source columns from data files
+<a href=#3880 id=3880 data-nosnippet>3880</a>    /// - Partition metadata for bucket transforms stores bucket numbers (0-3), NOT source values
+<a href=#3881 id=3881 data-nosnippet>3881</a>    ///
+<a href=#3882 id=3882 data-nosnippet>3882</a>    /// Java's PartitionUtil.constantsMap() implements this via:
+<a href=#3883 id=3883 data-nosnippet>3883</a>    /// ```java
+<a href=#3884 id=3884 data-nosnippet>3884</a>    /// if (field.transform().isIdentity()) {
+<a href=#3885 id=3885 data-nosnippet>3885</a>    ///     idToConstant.put(field.sourceId(), converted);
+<a href=#3886 id=3886 data-nosnippet>3886</a>    /// }
+<a href=#3887 id=3887 data-nosnippet>3887</a>    /// ```
+<a href=#3888 id=3888 data-nosnippet>3888</a>    ///
+<a href=#3889 id=3889 data-nosnippet>3889</a>    /// # What This Test Verifies
+<a href=#3890 id=3890 data-nosnippet>3890</a>    ///
+<a href=#3891 id=3891 data-nosnippet>3891</a>    /// This test ensures the full ArrowReader → RecordBatchTransformer pipeline correctly handles
+<a href=#3892 id=3892 data-nosnippet>3892</a>    /// bucket partitioning when FileScanTask provides partition_spec and partition_data:
+<a href=#3893 id=3893 data-nosnippet>3893</a>    ///
+<a href=#3894 id=3894 data-nosnippet>3894</a>    /// - Parquet file has field_id=1 named "id" with actual data [1, 5, 9, 13]
+<a href=#3895 id=3895 data-nosnippet>3895</a>    /// - FileScanTask specifies partition_spec with bucket(4, id) and partition_data with bucket=1
+<a href=#3896 id=3896 data-nosnippet>3896</a>    /// - RecordBatchTransformer.constants_map() excludes bucket-partitioned field from constants
+<a href=#3897 id=3897 data-nosnippet>3897</a>    /// - ArrowReader correctly reads [1, 5, 9, 13] from the data file
+<a href=#3898 id=3898 data-nosnippet>3898</a>    /// - Values are NOT replaced with constant 1 from partition metadata
+<a href=#3899 id=3899 data-nosnippet>3899</a>    ///
+<a href=#3900 id=3900 data-nosnippet>3900</a>    /// # Why This Matters
+<a href=#3901 id=3901 data-nosnippet>3901</a>    ///
+<a href=#3902 id=3902 data-nosnippet>3902</a>    /// Without correct handling:
+<a href=#3903 id=3903 data-nosnippet>3903</a>    /// - Runtime filtering would break (e.g., `WHERE id = 5` would fail)
+<a href=#3904 id=3904 data-nosnippet>3904</a>    /// - Query results would be incorrect (all rows would have id=1)
+<a href=#3905 id=3905 data-nosnippet>3905</a>    /// - Bucket partitioning would be unusable for query optimization
+<a href=#3906 id=3906 data-nosnippet>3906</a>    ///
+<a href=#3907 id=3907 data-nosnippet>3907</a>    /// # References
+<a href=#3908 id=3908 data-nosnippet>3908</a>    /// - Iceberg spec: format/spec.md "Column Projection" + "Partition Transforms"
+<a href=#3909 id=3909 data-nosnippet>3909</a>    /// - Java test: spark/src/test/java/.../TestRuntimeFiltering.java
+<a href=#3910 id=3910 data-nosnippet>3910</a>    /// - Java impl: core/src/main/java/org/apache/iceberg/util/PartitionUtil.java
+<a href=#3911 id=3911 data-nosnippet>3911</a>    </span><span class="attr">#[tokio::test]
+<a href=#3912 id=3912 data-nosnippet>3912</a>    </span><span class="kw">async fn </span>test_bucket_partitioning_reads_source_column_from_file() {
+<a href=#3913 id=3913 data-nosnippet>3913</a>        <span class="kw">use </span>arrow_array::Int32Array;
+<a href=#3914 id=3914 data-nosnippet>3914</a>
+<a href=#3915 id=3915 data-nosnippet>3915</a>        <span class="kw">use </span><span class="kw">crate</span>::spec::{Literal, PartitionSpec, Struct, Transform};
+<a href=#3916 id=3916 data-nosnippet>3916</a>
+<a href=#3917 id=3917 data-nosnippet>3917</a>        <span class="comment">// Iceberg schema with id and name columns
+<a href=#3918 id=3918 data-nosnippet>3918</a>        </span><span class="kw">let </span>schema = Arc::new(
+<a href=#3919 id=3919 data-nosnippet>3919</a>            Schema::builder()
+<a href=#3920 id=3920 data-nosnippet>3920</a>                .with_schema_id(<span class="number">0</span>)
+<a href=#3921 id=3921 data-nosnippet>3921</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#3922 id=3922 data-nosnippet>3922</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#3923 id=3923 data-nosnippet>3923</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#3924 id=3924 data-nosnippet>3924</a>                ])
+<a href=#3925 id=3925 data-nosnippet>3925</a>                .build()
+<a href=#3926 id=3926 data-nosnippet>3926</a>                .unwrap(),
+<a href=#3927 id=3927 data-nosnippet>3927</a>        );
+<a href=#3928 id=3928 data-nosnippet>3928</a>
+<a href=#3929 id=3929 data-nosnippet>3929</a>        <span class="comment">// Partition spec: bucket(4, id)
+<a href=#3930 id=3930 data-nosnippet>3930</a>        </span><span class="kw">let </span>partition_spec = Arc::new(
+<a href=#3931 id=3931 data-nosnippet>3931</a>            PartitionSpec::builder(schema.clone())
+<a href=#3932 id=3932 data-nosnippet>3932</a>                .with_spec_id(<span class="number">0</span>)
+<a href=#3933 id=3933 data-nosnippet>3933</a>                .add_partition_field(<span class="string">"id"</span>, <span class="string">"id_bucket"</span>, Transform::Bucket(<span class="number">4</span>))
+<a href=#3934 id=3934 data-nosnippet>3934</a>                .unwrap()
+<a href=#3935 id=3935 data-nosnippet>3935</a>                .build()
+<a href=#3936 id=3936 data-nosnippet>3936</a>                .unwrap(),
+<a href=#3937 id=3937 data-nosnippet>3937</a>        );
+<a href=#3938 id=3938 data-nosnippet>3938</a>
+<a href=#3939 id=3939 data-nosnippet>3939</a>        <span class="comment">// Partition data: bucket value is 1
+<a href=#3940 id=3940 data-nosnippet>3940</a>        </span><span class="kw">let </span>partition_data = Struct::from_iter(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(Literal::int(<span class="number">1</span>))]);
+<a href=#3941 id=3941 data-nosnippet>3941</a>
+<a href=#3942 id=3942 data-nosnippet>3942</a>        <span class="comment">// Create Arrow schema with field IDs for Parquet file
+<a href=#3943 id=3943 data-nosnippet>3943</a>        </span><span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#3944 id=3944 data-nosnippet>3944</a>            Field::new(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#3945 id=3945 data-nosnippet>3945</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#3946 id=3946 data-nosnippet>3946</a>                <span class="string">"1"</span>.to_string(),
+<a href=#3947 id=3947 data-nosnippet>3947</a>            )])),
+<a href=#3948 id=3948 data-nosnippet>3948</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">true</span>).with_metadata(HashMap::from([(
+<a href=#3949 id=3949 data-nosnippet>3949</a>                PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#3950 id=3950 data-nosnippet>3950</a>                <span class="string">"2"</span>.to_string(),
+<a href=#3951 id=3951 data-nosnippet>3951</a>            )])),
+<a href=#3952 id=3952 data-nosnippet>3952</a>        ]));
+<a href=#3953 id=3953 data-nosnippet>3953</a>
+<a href=#3954 id=3954 data-nosnippet>3954</a>        <span class="comment">// Write Parquet file with data
+<a href=#3955 id=3955 data-nosnippet>3955</a>        </span><span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#3956 id=3956 data-nosnippet>3956</a>        <span class="kw">let </span>table_location = tmp_dir.path().to_str().unwrap().to_string();
+<a href=#3957 id=3957 data-nosnippet>3957</a>        <span class="kw">let </span>file_io = FileIO::from_path(<span class="kw-2">&amp;</span>table_location).unwrap().build().unwrap();
+<a href=#3958 id=3958 data-nosnippet>3958</a>
+<a href=#3959 id=3959 data-nosnippet>3959</a>        <span class="kw">let </span>id_data = Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">5</span>, <span class="number">9</span>, <span class="number">13</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3960 id=3960 data-nosnippet>3960</a>        <span class="kw">let </span>name_data =
+<a href=#3961 id=3961 data-nosnippet>3961</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>, <span class="string">"Charlie"</span>, <span class="string">"Dave"</span>])) <span class="kw">as </span>ArrayRef;
+<a href=#3962 id=3962 data-nosnippet>3962</a>
+<a href=#3963 id=3963 data-nosnippet>3963</a>        <span class="kw">let </span>to_write =
+<a href=#3964 id=3964 data-nosnippet>3964</a>            RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[id_data, name_data]).unwrap();
+<a href=#3965 id=3965 data-nosnippet>3965</a>
+<a href=#3966 id=3966 data-nosnippet>3966</a>        <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#3967 id=3967 data-nosnippet>3967</a>            .set_compression(Compression::SNAPPY)
+<a href=#3968 id=3968 data-nosnippet>3968</a>            .build();
+<a href=#3969 id=3969 data-nosnippet>3969</a>        <span class="kw">let </span>file = File::create(<span class="macro">format!</span>(<span class="string">"{}/data.parquet"</span>, <span class="kw-2">&amp;</span>table_location)).unwrap();
+<a href=#3970 id=3970 data-nosnippet>3970</a>        <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, to_write.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#3971 id=3971 data-nosnippet>3971</a>        writer.write(<span class="kw-2">&amp;</span>to_write).expect(<span class="string">"Writing batch"</span>);
+<a href=#3972 id=3972 data-nosnippet>3972</a>        writer.close().unwrap();
+<a href=#3973 id=3973 data-nosnippet>3973</a>
+<a href=#3974 id=3974 data-nosnippet>3974</a>        <span class="comment">// Read the Parquet file with partition spec and data
+<a href=#3975 id=3975 data-nosnippet>3975</a>        </span><span class="kw">let </span>reader = ArrowReaderBuilder::new(file_io).build();
+<a href=#3976 id=3976 data-nosnippet>3976</a>        <span class="kw">let </span>tasks = Box::pin(futures::stream::iter(
+<a href=#3977 id=3977 data-nosnippet>3977</a>            <span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(FileScanTask {
+<a href=#3978 id=3978 data-nosnippet>3978</a>                start: <span class="number">0</span>,
+<a href=#3979 id=3979 data-nosnippet>3979</a>                length: <span class="number">0</span>,
+<a href=#3980 id=3980 data-nosnippet>3980</a>                record_count: <span class="prelude-val">None</span>,
+<a href=#3981 id=3981 data-nosnippet>3981</a>                data_file_path: <span class="macro">format!</span>(<span class="string">"{}/data.parquet"</span>, table_location),
+<a href=#3982 id=3982 data-nosnippet>3982</a>                data_file_format: DataFileFormat::Parquet,
+<a href=#3983 id=3983 data-nosnippet>3983</a>                schema: schema.clone(),
+<a href=#3984 id=3984 data-nosnippet>3984</a>                project_field_ids: <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>],
+<a href=#3985 id=3985 data-nosnippet>3985</a>                predicate: <span class="prelude-val">None</span>,
+<a href=#3986 id=3986 data-nosnippet>3986</a>                deletes: <span class="macro">vec!</span>[],
+<a href=#3987 id=3987 data-nosnippet>3987</a>                partition: <span class="prelude-val">Some</span>(partition_data),
+<a href=#3988 id=3988 data-nosnippet>3988</a>                partition_spec: <span class="prelude-val">Some</span>(partition_spec),
+<a href=#3989 id=3989 data-nosnippet>3989</a>                name_mapping: <span class="prelude-val">None</span>,
+<a href=#3990 id=3990 data-nosnippet>3990</a>            })]
+<a href=#3991 id=3991 data-nosnippet>3991</a>            .into_iter(),
+<a href=#3992 id=3992 data-nosnippet>3992</a>        )) <span class="kw">as </span>FileScanTaskStream;
+<a href=#3993 id=3993 data-nosnippet>3993</a>
+<a href=#3994 id=3994 data-nosnippet>3994</a>        <span class="kw">let </span>result = reader
+<a href=#3995 id=3995 data-nosnippet>3995</a>            .read(tasks)
+<a href=#3996 id=3996 data-nosnippet>3996</a>            .unwrap()
+<a href=#3997 id=3997 data-nosnippet>3997</a>            .try_collect::&lt;Vec&lt;RecordBatch&gt;&gt;()
+<a href=#3998 id=3998 data-nosnippet>3998</a>            .<span class="kw">await
+<a href=#3999 id=3999 data-nosnippet>3999</a>            </span>.unwrap();
+<a href=#4000 id=4000 data-nosnippet>4000</a>
+<a href=#4001 id=4001 data-nosnippet>4001</a>        <span class="comment">// Verify we got the correct data
+<a href=#4002 id=4002 data-nosnippet>4002</a>        </span><span class="macro">assert_eq!</span>(result.len(), <span class="number">1</span>);
+<a href=#4003 id=4003 data-nosnippet>4003</a>        <span class="kw">let </span>batch = <span class="kw-2">&amp;</span>result[<span class="number">0</span>];
+<a href=#4004 id=4004 data-nosnippet>4004</a>
+<a href=#4005 id=4005 data-nosnippet>4005</a>        <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">2</span>);
+<a href=#4006 id=4006 data-nosnippet>4006</a>        <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">4</span>);
+<a href=#4007 id=4007 data-nosnippet>4007</a>
+<a href=#4008 id=4008 data-nosnippet>4008</a>        <span class="comment">// The id column MUST contain actual values from the Parquet file [1, 5, 9, 13],
+<a href=#4009 id=4009 data-nosnippet>4009</a>        // NOT the constant partition value 1
+<a href=#4010 id=4010 data-nosnippet>4010</a>        </span><span class="kw">let </span>id_col = batch
+<a href=#4011 id=4011 data-nosnippet>4011</a>            .column(<span class="number">0</span>)
+<a href=#4012 id=4012 data-nosnippet>4012</a>            .as_primitive::&lt;arrow_array::types::Int32Type&gt;();
+<a href=#4013 id=4013 data-nosnippet>4013</a>        <span class="macro">assert_eq!</span>(id_col.value(<span class="number">0</span>), <span class="number">1</span>);
+<a href=#4014 id=4014 data-nosnippet>4014</a>        <span class="macro">assert_eq!</span>(id_col.value(<span class="number">1</span>), <span class="number">5</span>);
+<a href=#4015 id=4015 data-nosnippet>4015</a>        <span class="macro">assert_eq!</span>(id_col.value(<span class="number">2</span>), <span class="number">9</span>);
+<a href=#4016 id=4016 data-nosnippet>4016</a>        <span class="macro">assert_eq!</span>(id_col.value(<span class="number">3</span>), <span class="number">13</span>);
+<a href=#4017 id=4017 data-nosnippet>4017</a>
+<a href=#4018 id=4018 data-nosnippet>4018</a>        <span class="kw">let </span>name_col = batch.column(<span class="number">1</span>).as_string::&lt;i32&gt;();
+<a href=#4019 id=4019 data-nosnippet>4019</a>        <span class="macro">assert_eq!</span>(name_col.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
+<a href=#4020 id=4020 data-nosnippet>4020</a>        <span class="macro">assert_eq!</span>(name_col.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
+<a href=#4021 id=4021 data-nosnippet>4021</a>        <span class="macro">assert_eq!</span>(name_col.value(<span class="number">2</span>), <span class="string">"Charlie"</span>);
+<a href=#4022 id=4022 data-nosnippet>4022</a>        <span class="macro">assert_eq!</span>(name_col.value(<span class="number">3</span>), <span class="string">"Dave"</span>);
+<a href=#4023 id=4023 data-nosnippet>4023</a>    }
+<a href=#4024 id=4024 data-nosnippet>4024</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file
diff --git a/api/src/iceberg/arrow/record_batch_transformer.rs.html b/api/src/iceberg/arrow/record_batch_transformer.rs.html
index b9f99b93..475019b1 100644
--- a/api/src/iceberg/arrow/record_batch_transformer.rs.html
+++ b/api/src/iceberg/arrow/record_batch_transformer.rs.html
@@ -1,4 +1,4 @@
-<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Source of the Rust file `crates/iceberg/src/arrow/record_batch_transformer.rs`."><title>record_batch_transformer.rs - source</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de [...]
+<!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Source of the Rust file `crates/iceberg/src/arrow/record_batch_transformer.rs`."><title>record_batch_transformer.rs - source</title><script>if(window.location.protocol!=="file:")document.head.insertAdjacentHTML("beforeend","SourceSerif4-Regular-6b053e98.ttf.woff2,FiraSans-Italic-81dc35de [...]
 <a href=#2 id=2 data-nosnippet>2</a>// or more contributor license agreements.  See the NOTICE file
 <a href=#3 id=3 data-nosnippet>3</a>// distributed with this work for additional information
 <a href=#4 id=4 data-nosnippet>4</a>// regarding copyright ownership.  The ASF licenses this file
@@ -29,671 +29,1417 @@
 <a href=#29 id=29 data-nosnippet>29</a><span class="kw">use </span>parquet::arrow::PARQUET_FIELD_ID_META_KEY;
 <a href=#30 id=30 data-nosnippet>30</a>
 <a href=#31 id=31 data-nosnippet>31</a><span class="kw">use </span><span class="kw">crate</span>::arrow::schema_to_arrow_schema;
-<a href=#32 id=32 data-nosnippet>32</a><span class="kw">use </span><span class="kw">crate</span>::spec::{Literal, PrimitiveLiteral, Schema <span class="kw">as </span>IcebergSchema};
-<a href=#33 id=33 data-nosnippet>33</a><span class="kw">use crate</span>::{Error, ErrorKind, <span class="prelude-ty">Result</span>};
-<a href=#34 id=34 data-nosnippet>34</a>
-<a href=#35 id=35 data-nosnippet>35</a><span class="doccomment">/// Indicates how a particular column in a processed RecordBatch should
-<a href=#36 id=36 data-nosnippet>36</a>/// be sourced.
-<a href=#37 id=37 data-nosnippet>37</a></span><span class="attr">#[derive(Debug)]
-<a href=#38 id=38 data-nosnippet>38</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">enum </span>ColumnSource {
-<a href=#39 id=39 data-nosnippet>39</a>    <span class="comment">// signifies that a column should be passed through unmodified
-<a href=#40 id=40 data-nosnippet>40</a>    // from the file's RecordBatch
-<a href=#41 id=41 data-nosnippet>41</a>    </span>PassThrough {
-<a href=#42 id=42 data-nosnippet>42</a>        source_index: usize,
-<a href=#43 id=43 data-nosnippet>43</a>    },
-<a href=#44 id=44 data-nosnippet>44</a>
-<a href=#45 id=45 data-nosnippet>45</a>    <span class="comment">// signifies that a column from the file's RecordBatch has undergone
-<a href=#46 id=46 data-nosnippet>46</a>    // type promotion so the source column with the given index needs
-<a href=#47 id=47 data-nosnippet>47</a>    // to be promoted to the specified type
-<a href=#48 id=48 data-nosnippet>48</a>    </span>Promote {
-<a href=#49 id=49 data-nosnippet>49</a>        target_type: DataType,
-<a href=#50 id=50 data-nosnippet>50</a>        source_index: usize,
-<a href=#51 id=51 data-nosnippet>51</a>    },
-<a href=#52 id=52 data-nosnippet>52</a>
-<a href=#53 id=53 data-nosnippet>53</a>    <span class="comment">// Signifies that a new column has been inserted before the column
-<a href=#54 id=54 data-nosnippet>54</a>    // with index `index`. (we choose "before" rather than "after" so
-<a href=#55 id=55 data-nosnippet>55</a>    // that we can use usize; if we insert after, then we need to
-<a href=#56 id=56 data-nosnippet>56</a>    // be able to store -1 here to signify that a new
-<a href=#57 id=57 data-nosnippet>57</a>    // column is to be added at the front of the column list).
-<a href=#58 id=58 data-nosnippet>58</a>    // If multiple columns need to be inserted at a given
-<a href=#59 id=59 data-nosnippet>59</a>    // location, they should all be given the same index, as the index
-<a href=#60 id=60 data-nosnippet>60</a>    // here refers to the original RecordBatch, not the interim state after
-<a href=#61 id=61 data-nosnippet>61</a>    // a preceding operation.
-<a href=#62 id=62 data-nosnippet>62</a>    </span>Add {
-<a href=#63 id=63 data-nosnippet>63</a>        target_type: DataType,
-<a href=#64 id=64 data-nosnippet>64</a>        value: <span class="prelude-ty">Option</span>&lt;PrimitiveLiteral&gt;,
-<a href=#65 id=65 data-nosnippet>65</a>    },
-<a href=#66 id=66 data-nosnippet>66</a>    <span class="comment">// The iceberg spec refers to other permissible schema evolution actions
-<a href=#67 id=67 data-nosnippet>67</a>    // (see https://iceberg.apache.org/spec/#schema-evolution):
-<a href=#68 id=68 data-nosnippet>68</a>    // renaming fields, deleting fields and reordering fields.
-<a href=#69 id=69 data-nosnippet>69</a>    // Renames only affect the schema of the RecordBatch rather than the
-<a href=#70 id=70 data-nosnippet>70</a>    // columns themselves, so a single updated cached schema can
-<a href=#71 id=71 data-nosnippet>71</a>    // be re-used and no per-column actions are required.
-<a href=#72 id=72 data-nosnippet>72</a>    // Deletion and Reorder can be achieved without needing this
-<a href=#73 id=73 data-nosnippet>73</a>    // post-processing step by using the projection mask.
-<a href=#74 id=74 data-nosnippet>74</a></span>}
-<a href=#75 id=75 data-nosnippet>75</a>
-<a href=#76 id=76 data-nosnippet>76</a><span class="attr">#[derive(Debug)]
-<a href=#77 id=77 data-nosnippet>77</a></span><span class="kw">enum </span>BatchTransform {
-<a href=#78 id=78 data-nosnippet>78</a>    <span class="comment">// Indicates that no changes need to be performed to the RecordBatches
-<a href=#79 id=79 data-nosnippet>79</a>    // coming in from the stream and that they can be passed through
-<a href=#80 id=80 data-nosnippet>80</a>    // unmodified
-<a href=#81 id=81 data-nosnippet>81</a>    </span>PassThrough,
-<a href=#82 id=82 data-nosnippet>82</a>
-<a href=#83 id=83 data-nosnippet>83</a>    Modify {
-<a href=#84 id=84 data-nosnippet>84</a>        <span class="comment">// Every transformed RecordBatch will have the same schema. We create the
-<a href=#85 id=85 data-nosnippet>85</a>        // target just once and cache it here. Helpfully, Arc&lt;Schema&gt; is needed in
-<a href=#86 id=86 data-nosnippet>86</a>        // the constructor for RecordBatch, so we don't need an expensive copy
-<a href=#87 id=87 data-nosnippet>87</a>        // each time we build a new RecordBatch
-<a href=#88 id=88 data-nosnippet>88</a>        </span>target_schema: Arc&lt;ArrowSchema&gt;,
-<a href=#89 id=89 data-nosnippet>89</a>
-<a href=#90 id=90 data-nosnippet>90</a>        <span class="comment">// Indicates how each column in the target schema is derived.
-<a href=#91 id=91 data-nosnippet>91</a>        </span>operations: Vec&lt;ColumnSource&gt;,
-<a href=#92 id=92 data-nosnippet>92</a>    },
-<a href=#93 id=93 data-nosnippet>93</a>
-<a href=#94 id=94 data-nosnippet>94</a>    <span class="comment">// Sometimes only the schema will need modifying, for example when
-<a href=#95 id=95 data-nosnippet>95</a>    // the column names have changed vs the file, but not the column types.
-<a href=#96 id=96 data-nosnippet>96</a>    // we can avoid a heap allocation per RecordBach in this case by retaining
-<a href=#97 id=97 data-nosnippet>97</a>    // the existing column Vec.
-<a href=#98 id=98 data-nosnippet>98</a>    </span>ModifySchema {
-<a href=#99 id=99 data-nosnippet>99</a>        target_schema: Arc&lt;ArrowSchema&gt;,
+<a href=#32 id=32 data-nosnippet>32</a><span class="kw">use </span><span class="kw">crate</span>::spec::{
+<a href=#33 id=33 data-nosnippet>33</a>    Literal, PartitionSpec, PrimitiveLiteral, Schema <span class="kw">as </span>IcebergSchema, Struct, Transform,
+<a href=#34 id=34 data-nosnippet>34</a>};
+<a href=#35 id=35 data-nosnippet>35</a><span class="kw">use crate</span>::{Error, ErrorKind, <span class="prelude-ty">Result</span>};
+<a href=#36 id=36 data-nosnippet>36</a>
+<a href=#37 id=37 data-nosnippet>37</a><span class="doccomment">/// Build a map of field ID to constant value for identity-partitioned fields.
+<a href=#38 id=38 data-nosnippet>38</a>///
+<a href=#39 id=39 data-nosnippet>39</a>/// Implements Iceberg spec "Column Projection" rule #1: use partition metadata constants
+<a href=#40 id=40 data-nosnippet>40</a>/// only for identity-transformed fields. Non-identity transforms (bucket, truncate, year, etc.)
+<a href=#41 id=41 data-nosnippet>41</a>/// store derived values in partition metadata, so source columns must be read from data files.
+<a href=#42 id=42 data-nosnippet>42</a>///
+<a href=#43 id=43 data-nosnippet>43</a>/// Example: For `bucket(4, id)`, partition metadata has `id_bucket = 2` (bucket number),
+<a href=#44 id=44 data-nosnippet>44</a>/// but the actual `id` values (100, 200, 300) are only in the data file.
+<a href=#45 id=45 data-nosnippet>45</a>///
+<a href=#46 id=46 data-nosnippet>46</a>/// Matches Java's `PartitionUtil.constantsMap()` which filters `if (field.transform().isIdentity())`.
+<a href=#47 id=47 data-nosnippet>47</a>///
+<a href=#48 id=48 data-nosnippet>48</a>/// # References
+<a href=#49 id=49 data-nosnippet>49</a>/// - Spec: https://iceberg.apache.org/spec/#column-projection
+<a href=#50 id=50 data-nosnippet>50</a>/// - Java: core/src/main/java/org/apache/iceberg/util/PartitionUtil.java:constantsMap()
+<a href=#51 id=51 data-nosnippet>51</a></span><span class="kw">fn </span>constants_map(
+<a href=#52 id=52 data-nosnippet>52</a>    partition_spec: <span class="kw-2">&amp;</span>PartitionSpec,
+<a href=#53 id=53 data-nosnippet>53</a>    partition_data: <span class="kw-2">&amp;</span>Struct,
+<a href=#54 id=54 data-nosnippet>54</a>) -&gt; HashMap&lt;i32, PrimitiveLiteral&gt; {
+<a href=#55 id=55 data-nosnippet>55</a>    <span class="kw">let </span><span class="kw-2">mut </span>constants = HashMap::new();
+<a href=#56 id=56 data-nosnippet>56</a>
+<a href=#57 id=57 data-nosnippet>57</a>    <span class="kw">for </span>(pos, field) <span class="kw">in </span>partition_spec.fields().iter().enumerate() {
+<a href=#58 id=58 data-nosnippet>58</a>        <span class="comment">// Only identity transforms should use constant values from partition metadata
+<a href=#59 id=59 data-nosnippet>59</a>        </span><span class="kw">if </span><span class="macro">matches!</span>(field.transform, Transform::Identity) {
+<a href=#60 id=60 data-nosnippet>60</a>            <span class="comment">// Get the partition value for this field
+<a href=#61 id=61 data-nosnippet>61</a>            </span><span class="kw">if let </span><span class="prelude-val">Some</span>(Literal::Primitive(value)) = <span class="kw-2">&amp;</span>partition_data[pos] {
+<a href=#62 id=62 data-nosnippet>62</a>                constants.insert(field.source_id, value.clone());
+<a href=#63 id=63 data-nosnippet>63</a>            }
+<a href=#64 id=64 data-nosnippet>64</a>        }
+<a href=#65 id=65 data-nosnippet>65</a>    }
+<a href=#66 id=66 data-nosnippet>66</a>
+<a href=#67 id=67 data-nosnippet>67</a>    constants
+<a href=#68 id=68 data-nosnippet>68</a>}
+<a href=#69 id=69 data-nosnippet>69</a>
+<a href=#70 id=70 data-nosnippet>70</a><span class="doccomment">/// Indicates how a particular column in a processed RecordBatch should
+<a href=#71 id=71 data-nosnippet>71</a>/// be sourced.
+<a href=#72 id=72 data-nosnippet>72</a></span><span class="attr">#[derive(Debug)]
+<a href=#73 id=73 data-nosnippet>73</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">enum </span>ColumnSource {
+<a href=#74 id=74 data-nosnippet>74</a>    <span class="comment">// signifies that a column should be passed through unmodified
+<a href=#75 id=75 data-nosnippet>75</a>    // from the file's RecordBatch
+<a href=#76 id=76 data-nosnippet>76</a>    </span>PassThrough {
+<a href=#77 id=77 data-nosnippet>77</a>        source_index: usize,
+<a href=#78 id=78 data-nosnippet>78</a>    },
+<a href=#79 id=79 data-nosnippet>79</a>
+<a href=#80 id=80 data-nosnippet>80</a>    <span class="comment">// signifies that a column from the file's RecordBatch has undergone
+<a href=#81 id=81 data-nosnippet>81</a>    // type promotion so the source column with the given index needs
+<a href=#82 id=82 data-nosnippet>82</a>    // to be promoted to the specified type
+<a href=#83 id=83 data-nosnippet>83</a>    </span>Promote {
+<a href=#84 id=84 data-nosnippet>84</a>        target_type: DataType,
+<a href=#85 id=85 data-nosnippet>85</a>        source_index: usize,
+<a href=#86 id=86 data-nosnippet>86</a>    },
+<a href=#87 id=87 data-nosnippet>87</a>
+<a href=#88 id=88 data-nosnippet>88</a>    <span class="comment">// Signifies that a new column has been inserted before the column
+<a href=#89 id=89 data-nosnippet>89</a>    // with index `index`. (we choose "before" rather than "after" so
+<a href=#90 id=90 data-nosnippet>90</a>    // that we can use usize; if we insert after, then we need to
+<a href=#91 id=91 data-nosnippet>91</a>    // be able to store -1 here to signify that a new
+<a href=#92 id=92 data-nosnippet>92</a>    // column is to be added at the front of the column list).
+<a href=#93 id=93 data-nosnippet>93</a>    // If multiple columns need to be inserted at a given
+<a href=#94 id=94 data-nosnippet>94</a>    // location, they should all be given the same index, as the index
+<a href=#95 id=95 data-nosnippet>95</a>    // here refers to the original RecordBatch, not the interim state after
+<a href=#96 id=96 data-nosnippet>96</a>    // a preceding operation.
+<a href=#97 id=97 data-nosnippet>97</a>    </span>Add {
+<a href=#98 id=98 data-nosnippet>98</a>        target_type: DataType,
+<a href=#99 id=99 data-nosnippet>99</a>        value: <span class="prelude-ty">Option</span>&lt;PrimitiveLiteral&gt;,
 <a href=#100 id=100 data-nosnippet>100</a>    },
-<a href=#101 id=101 data-nosnippet>101</a>}
-<a href=#102 id=102 data-nosnippet>102</a>
-<a href=#103 id=103 data-nosnippet>103</a><span class="attr">#[derive(Debug)]
-<a href=#104 id=104 data-nosnippet>104</a></span><span class="kw">enum </span>SchemaComparison {
-<a href=#105 id=105 data-nosnippet>105</a>    Equivalent,
-<a href=#106 id=106 data-nosnippet>106</a>    NameChangesOnly,
-<a href=#107 id=107 data-nosnippet>107</a>    Different,
-<a href=#108 id=108 data-nosnippet>108</a>}
-<a href=#109 id=109 data-nosnippet>109</a>
-<a href=#110 id=110 data-nosnippet>110</a><span class="attr">#[derive(Debug)]
-<a href=#111 id=111 data-nosnippet>111</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">struct </span>RecordBatchTransformer {
-<a href=#112 id=112 data-nosnippet>112</a>    snapshot_schema: Arc&lt;IcebergSchema&gt;,
-<a href=#113 id=113 data-nosnippet>113</a>    projected_iceberg_field_ids: Vec&lt;i32&gt;,
-<a href=#114 id=114 data-nosnippet>114</a>
-<a href=#115 id=115 data-nosnippet>115</a>    <span class="comment">// BatchTransform gets lazily constructed based on the schema of
-<a href=#116 id=116 data-nosnippet>116</a>    // the first RecordBatch we receive from the file
-<a href=#117 id=117 data-nosnippet>117</a>    </span>batch_transform: <span class="prelude-ty">Option</span>&lt;BatchTransform&gt;,
-<a href=#118 id=118 data-nosnippet>118</a>}
-<a href=#119 id=119 data-nosnippet>119</a>
-<a href=#120 id=120 data-nosnippet>120</a><span class="kw">impl </span>RecordBatchTransformer {
-<a href=#121 id=121 data-nosnippet>121</a>    <span class="doccomment">/// Build a RecordBatchTransformer for a given
-<a href=#122 id=122 data-nosnippet>122</a>    /// Iceberg snapshot schema and list of projected field ids.
-<a href=#123 id=123 data-nosnippet>123</a>    </span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>build(
-<a href=#124 id=124 data-nosnippet>124</a>        snapshot_schema: Arc&lt;IcebergSchema&gt;,
-<a href=#125 id=125 data-nosnippet>125</a>        projected_iceberg_field_ids: <span class="kw-2">&amp;</span>[i32],
-<a href=#126 id=126 data-nosnippet>126</a>    ) -&gt; <span class="self">Self </span>{
-<a href=#127 id=127 data-nosnippet>127</a>        <span class="kw">let </span>projected_iceberg_field_ids = projected_iceberg_field_ids.to_vec();
+<a href=#101 id=101 data-nosnippet>101</a>    <span class="comment">// The iceberg spec refers to other permissible schema evolution actions
+<a href=#102 id=102 data-nosnippet>102</a>    // (see https://iceberg.apache.org/spec/#schema-evolution):
+<a href=#103 id=103 data-nosnippet>103</a>    // renaming fields, deleting fields and reordering fields.
+<a href=#104 id=104 data-nosnippet>104</a>    // Renames only affect the schema of the RecordBatch rather than the
+<a href=#105 id=105 data-nosnippet>105</a>    // columns themselves, so a single updated cached schema can
+<a href=#106 id=106 data-nosnippet>106</a>    // be re-used and no per-column actions are required.
+<a href=#107 id=107 data-nosnippet>107</a>    // Deletion and Reorder can be achieved without needing this
+<a href=#108 id=108 data-nosnippet>108</a>    // post-processing step by using the projection mask.
+<a href=#109 id=109 data-nosnippet>109</a></span>}
+<a href=#110 id=110 data-nosnippet>110</a>
+<a href=#111 id=111 data-nosnippet>111</a><span class="attr">#[derive(Debug)]
+<a href=#112 id=112 data-nosnippet>112</a></span><span class="kw">enum </span>BatchTransform {
+<a href=#113 id=113 data-nosnippet>113</a>    <span class="comment">// Indicates that no changes need to be performed to the RecordBatches
+<a href=#114 id=114 data-nosnippet>114</a>    // coming in from the stream and that they can be passed through
+<a href=#115 id=115 data-nosnippet>115</a>    // unmodified
+<a href=#116 id=116 data-nosnippet>116</a>    </span>PassThrough,
+<a href=#117 id=117 data-nosnippet>117</a>
+<a href=#118 id=118 data-nosnippet>118</a>    Modify {
+<a href=#119 id=119 data-nosnippet>119</a>        <span class="comment">// Every transformed RecordBatch will have the same schema. We create the
+<a href=#120 id=120 data-nosnippet>120</a>        // target just once and cache it here. Helpfully, Arc&lt;Schema&gt; is needed in
+<a href=#121 id=121 data-nosnippet>121</a>        // the constructor for RecordBatch, so we don't need an expensive copy
+<a href=#122 id=122 data-nosnippet>122</a>        // each time we build a new RecordBatch
+<a href=#123 id=123 data-nosnippet>123</a>        </span>target_schema: Arc&lt;ArrowSchema&gt;,
+<a href=#124 id=124 data-nosnippet>124</a>
+<a href=#125 id=125 data-nosnippet>125</a>        <span class="comment">// Indicates how each column in the target schema is derived.
+<a href=#126 id=126 data-nosnippet>126</a>        </span>operations: Vec&lt;ColumnSource&gt;,
+<a href=#127 id=127 data-nosnippet>127</a>    },
 <a href=#128 id=128 data-nosnippet>128</a>
-<a href=#129 id=129 data-nosnippet>129</a>        <span class="self">Self </span>{
-<a href=#130 id=130 data-nosnippet>130</a>            snapshot_schema,
-<a href=#131 id=131 data-nosnippet>131</a>            projected_iceberg_field_ids,
-<a href=#132 id=132 data-nosnippet>132</a>            batch_transform: <span class="prelude-val">None</span>,
-<a href=#133 id=133 data-nosnippet>133</a>        }
-<a href=#134 id=134 data-nosnippet>134</a>    }
-<a href=#135 id=135 data-nosnippet>135</a>
-<a href=#136 id=136 data-nosnippet>136</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>process_record_batch(
-<a href=#137 id=137 data-nosnippet>137</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
-<a href=#138 id=138 data-nosnippet>138</a>        record_batch: RecordBatch,
-<a href=#139 id=139 data-nosnippet>139</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;RecordBatch&gt; {
-<a href=#140 id=140 data-nosnippet>140</a>        <span class="prelude-val">Ok</span>(<span class="kw">match </span><span class="kw-2">&amp;</span><span class="self">self</span>.batch_transform {
-<a href=#141 id=141 data-nosnippet>141</a>            <span class="prelude-val">Some</span>(BatchTransform::PassThrough) =&gt; record_batch,
-<a href=#142 id=142 data-nosnippet>142</a>            <span class="prelude-val">Some</span>(BatchTransform::Modify {
-<a href=#143 id=143 data-nosnippet>143</a>                target_schema,
-<a href=#144 id=144 data-nosnippet>144</a>                operations,
-<a href=#145 id=145 data-nosnippet>145</a>            }) =&gt; {
-<a href=#146 id=146 data-nosnippet>146</a>                <span class="kw">let </span>options = RecordBatchOptions::default()
-<a href=#147 id=147 data-nosnippet>147</a>                    .with_match_field_names(<span class="bool-val">false</span>)
-<a href=#148 id=148 data-nosnippet>148</a>                    .with_row_count(<span class="prelude-val">Some</span>(record_batch.num_rows()));
-<a href=#149 id=149 data-nosnippet>149</a>                RecordBatch::try_new_with_options(
-<a href=#150 id=150 data-nosnippet>150</a>                    target_schema.clone(),
-<a href=#151 id=151 data-nosnippet>151</a>                    <span class="self">self</span>.transform_columns(record_batch.columns(), operations)<span class="question-mark">?</span>,
-<a href=#152 id=152 data-nosnippet>152</a>                    <span class="kw-2">&amp;</span>options,
-<a href=#153 id=153 data-nosnippet>153</a>                )<span class="question-mark">?
-<a href=#154 id=154 data-nosnippet>154</a>            </span>}
-<a href=#155 id=155 data-nosnippet>155</a>            <span class="prelude-val">Some</span>(BatchTransform::ModifySchema { target_schema }) =&gt; {
-<a href=#156 id=156 data-nosnippet>156</a>                <span class="kw">let </span>options = RecordBatchOptions::default()
-<a href=#157 id=157 data-nosnippet>157</a>                    .with_match_field_names(<span class="bool-val">false</span>)
-<a href=#158 id=158 data-nosnippet>158</a>                    .with_row_count(<span class="prelude-val">Some</span>(record_batch.num_rows()));
-<a href=#159 id=159 data-nosnippet>159</a>                RecordBatch::try_new_with_options(
-<a href=#160 id=160 data-nosnippet>160</a>                    target_schema.clone(),
-<a href=#161 id=161 data-nosnippet>161</a>                    record_batch.columns().to_vec(),
-<a href=#162 id=162 data-nosnippet>162</a>                    <span class="kw-2">&amp;</span>options,
-<a href=#163 id=163 data-nosnippet>163</a>                )<span class="question-mark">?
-<a href=#164 id=164 data-nosnippet>164</a>            </span>}
-<a href=#165 id=165 data-nosnippet>165</a>            <span class="prelude-val">None </span>=&gt; {
-<a href=#166 id=166 data-nosnippet>166</a>                <span class="self">self</span>.batch_transform = <span class="prelude-val">Some</span>(<span class="self">Self</span>::generate_batch_transform(
-<a href=#167 id=167 data-nosnippet>167</a>                    record_batch.schema_ref(),
-<a href=#168 id=168 data-nosnippet>168</a>                    <span class="self">self</span>.snapshot_schema.as_ref(),
-<a href=#169 id=169 data-nosnippet>169</a>                    <span class="kw-2">&amp;</span><span class="self">self</span>.projected_iceberg_field_ids,
-<a href=#170 id=170 data-nosnippet>170</a>                )<span class="question-mark">?</span>);
-<a href=#171 id=171 data-nosnippet>171</a>
-<a href=#172 id=172 data-nosnippet>172</a>                <span class="self">self</span>.process_record_batch(record_batch)<span class="question-mark">?
-<a href=#173 id=173 data-nosnippet>173</a>            </span>}
-<a href=#174 id=174 data-nosnippet>174</a>        })
-<a href=#175 id=175 data-nosnippet>175</a>    }
-<a href=#176 id=176 data-nosnippet>176</a>
-<a href=#177 id=177 data-nosnippet>177</a>    <span class="comment">// Compare the schema of the incoming RecordBatches to the schema of
-<a href=#178 id=178 data-nosnippet>178</a>    // the Iceberg snapshot to determine what, if any, transformation
-<a href=#179 id=179 data-nosnippet>179</a>    // needs to be applied. If the schemas match, we return BatchTransform::PassThrough
-<a href=#180 id=180 data-nosnippet>180</a>    // to indicate that no changes need to be made. Otherwise, we return a
-<a href=#181 id=181 data-nosnippet>181</a>    // BatchTransform::Modify containing the target RecordBatch schema and
-<a href=#182 id=182 data-nosnippet>182</a>    // the list of `ColumnSource`s that indicate how to source each column in
-<a href=#183 id=183 data-nosnippet>183</a>    // the resulting RecordBatches.
-<a href=#184 id=184 data-nosnippet>184</a>    </span><span class="kw">fn </span>generate_batch_transform(
-<a href=#185 id=185 data-nosnippet>185</a>        source_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
-<a href=#186 id=186 data-nosnippet>186</a>        snapshot_schema: <span class="kw-2">&amp;</span>IcebergSchema,
-<a href=#187 id=187 data-nosnippet>187</a>        projected_iceberg_field_ids: <span class="kw-2">&amp;</span>[i32],
-<a href=#188 id=188 data-nosnippet>188</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;BatchTransform&gt; {
-<a href=#189 id=189 data-nosnippet>189</a>        <span class="kw">let </span>mapped_unprojected_arrow_schema = Arc::new(schema_to_arrow_schema(snapshot_schema)<span class="question-mark">?</span>);
-<a href=#190 id=190 data-nosnippet>190</a>        <span class="kw">let </span>field_id_to_mapped_schema_map =
-<a href=#191 id=191 data-nosnippet>191</a>            <span class="self">Self</span>::build_field_id_to_arrow_schema_map(<span class="kw-2">&amp;</span>mapped_unprojected_arrow_schema)<span class="question-mark">?</span>;
-<a href=#192 id=192 data-nosnippet>192</a>
-<a href=#193 id=193 data-nosnippet>193</a>        <span class="comment">// Create a new arrow schema by selecting fields from mapped_unprojected,
-<a href=#194 id=194 data-nosnippet>194</a>        // in the order of the field ids in projected_iceberg_field_ids
-<a href=#195 id=195 data-nosnippet>195</a>        </span><span class="kw">let </span>fields: <span class="prelude-ty">Result</span>&lt;Vec&lt;<span class="kw">_</span>&gt;&gt; = projected_iceberg_field_ids
-<a href=#196 id=196 data-nosnippet>196</a>            .iter()
-<a href=#197 id=197 data-nosnippet>197</a>            .map(|field_id| {
-<a href=#198 id=198 data-nosnippet>198</a>                <span class="prelude-val">Ok</span>(field_id_to_mapped_schema_map
-<a href=#199 id=199 data-nosnippet>199</a>                    .get(field_id)
-<a href=#200 id=200 data-nosnippet>200</a>                    .ok_or(Error::new(ErrorKind::Unexpected, <span class="string">"field not found"</span>))<span class="question-mark">?
-<a href=#201 id=201 data-nosnippet>201</a>                    </span>.<span class="number">0
-<a href=#202 id=202 data-nosnippet>202</a>                    </span>.clone())
-<a href=#203 id=203 data-nosnippet>203</a>            })
-<a href=#204 id=204 data-nosnippet>204</a>            .collect();
-<a href=#205 id=205 data-nosnippet>205</a>
-<a href=#206 id=206 data-nosnippet>206</a>        <span class="kw">let </span>target_schema = Arc::new(ArrowSchema::new(fields<span class="question-mark">?</span>));
-<a href=#207 id=207 data-nosnippet>207</a>
-<a href=#208 id=208 data-nosnippet>208</a>        <span class="kw">match </span><span class="self">Self</span>::compare_schemas(source_schema, <span class="kw-2">&amp;</span>target_schema) {
-<a href=#209 id=209 data-nosnippet>209</a>            SchemaComparison::Equivalent =&gt; <span class="prelude-val">Ok</span>(BatchTransform::PassThrough),
-<a href=#210 id=210 data-nosnippet>210</a>            SchemaComparison::NameChangesOnly =&gt; <span class="prelude-val">Ok</span>(BatchTransform::ModifySchema { target_schema }),
-<a href=#211 id=211 data-nosnippet>211</a>            SchemaComparison::Different =&gt; <span class="prelude-val">Ok</span>(BatchTransform::Modify {
-<a href=#212 id=212 data-nosnippet>212</a>                operations: <span class="self">Self</span>::generate_transform_operations(
-<a href=#213 id=213 data-nosnippet>213</a>                    source_schema,
-<a href=#214 id=214 data-nosnippet>214</a>                    snapshot_schema,
-<a href=#215 id=215 data-nosnippet>215</a>                    projected_iceberg_field_ids,
-<a href=#216 id=216 data-nosnippet>216</a>                    field_id_to_mapped_schema_map,
-<a href=#217 id=217 data-nosnippet>217</a>                )<span class="question-mark">?</span>,
-<a href=#218 id=218 data-nosnippet>218</a>                target_schema,
-<a href=#219 id=219 data-nosnippet>219</a>            }),
-<a href=#220 id=220 data-nosnippet>220</a>        }
-<a href=#221 id=221 data-nosnippet>221</a>    }
-<a href=#222 id=222 data-nosnippet>222</a>
-<a href=#223 id=223 data-nosnippet>223</a>    <span class="doccomment">/// Compares the source and target schemas
-<a href=#224 id=224 data-nosnippet>224</a>    /// Determines if they have changed in any meaningful way:
-<a href=#225 id=225 data-nosnippet>225</a>    ///  * If they have different numbers of fields, then we need to modify
-<a href=#226 id=226 data-nosnippet>226</a>    ///    the incoming RecordBatch schema AND columns
-<a href=#227 id=227 data-nosnippet>227</a>    ///  * If they have the same number of fields, but some of them differ in
-<a href=#228 id=228 data-nosnippet>228</a>    ///    either data type or nullability, then we need to modify the
-<a href=#229 id=229 data-nosnippet>229</a>    ///    incoming RecordBatch schema AND columns
-<a href=#230 id=230 data-nosnippet>230</a>    ///  * If the schemas differ only in the column names, then we need
-<a href=#231 id=231 data-nosnippet>231</a>    ///    to modify the RecordBatch schema BUT we can keep the
-<a href=#232 id=232 data-nosnippet>232</a>    ///    original column data unmodified
-<a href=#233 id=233 data-nosnippet>233</a>    ///  * If the schemas are identical (or differ only in inconsequential
-<a href=#234 id=234 data-nosnippet>234</a>    ///    ways) then we can pass through the original RecordBatch unmodified
-<a href=#235 id=235 data-nosnippet>235</a>    </span><span class="kw">fn </span>compare_schemas(
-<a href=#236 id=236 data-nosnippet>236</a>        source_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
-<a href=#237 id=237 data-nosnippet>237</a>        target_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
-<a href=#238 id=238 data-nosnippet>238</a>    ) -&gt; SchemaComparison {
-<a href=#239 id=239 data-nosnippet>239</a>        <span class="kw">if </span>source_schema.fields().len() != target_schema.fields().len() {
-<a href=#240 id=240 data-nosnippet>240</a>            <span class="kw">return </span>SchemaComparison::Different;
-<a href=#241 id=241 data-nosnippet>241</a>        }
-<a href=#242 id=242 data-nosnippet>242</a>
-<a href=#243 id=243 data-nosnippet>243</a>        <span class="kw">let </span><span class="kw-2">mut </span>names_changed = <span class="bool-val">false</span>;
+<a href=#129 id=129 data-nosnippet>129</a>    <span class="comment">// Sometimes only the schema will need modifying, for example when
+<a href=#130 id=130 data-nosnippet>130</a>    // the column names have changed vs the file, but not the column types.
+<a href=#131 id=131 data-nosnippet>131</a>    // we can avoid a heap allocation per RecordBach in this case by retaining
+<a href=#132 id=132 data-nosnippet>132</a>    // the existing column Vec.
+<a href=#133 id=133 data-nosnippet>133</a>    </span>ModifySchema {
+<a href=#134 id=134 data-nosnippet>134</a>        target_schema: Arc&lt;ArrowSchema&gt;,
+<a href=#135 id=135 data-nosnippet>135</a>    },
+<a href=#136 id=136 data-nosnippet>136</a>}
+<a href=#137 id=137 data-nosnippet>137</a>
+<a href=#138 id=138 data-nosnippet>138</a><span class="attr">#[derive(Debug)]
+<a href=#139 id=139 data-nosnippet>139</a></span><span class="kw">enum </span>SchemaComparison {
+<a href=#140 id=140 data-nosnippet>140</a>    Equivalent,
+<a href=#141 id=141 data-nosnippet>141</a>    NameChangesOnly,
+<a href=#142 id=142 data-nosnippet>142</a>    Different,
+<a href=#143 id=143 data-nosnippet>143</a>}
+<a href=#144 id=144 data-nosnippet>144</a>
+<a href=#145 id=145 data-nosnippet>145</a><span class="doccomment">/// Builder for RecordBatchTransformer to improve ergonomics when constructing with optional parameters.
+<a href=#146 id=146 data-nosnippet>146</a>///
+<a href=#147 id=147 data-nosnippet>147</a>/// See [`RecordBatchTransformer`] for details on partition spec and partition data.
+<a href=#148 id=148 data-nosnippet>148</a></span><span class="attr">#[derive(Debug)]
+<a href=#149 id=149 data-nosnippet>149</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">struct </span>RecordBatchTransformerBuilder {
+<a href=#150 id=150 data-nosnippet>150</a>    snapshot_schema: Arc&lt;IcebergSchema&gt;,
+<a href=#151 id=151 data-nosnippet>151</a>    projected_iceberg_field_ids: Vec&lt;i32&gt;,
+<a href=#152 id=152 data-nosnippet>152</a>    partition_spec: <span class="prelude-ty">Option</span>&lt;Arc&lt;PartitionSpec&gt;&gt;,
+<a href=#153 id=153 data-nosnippet>153</a>    partition_data: <span class="prelude-ty">Option</span>&lt;Struct&gt;,
+<a href=#154 id=154 data-nosnippet>154</a>}
+<a href=#155 id=155 data-nosnippet>155</a>
+<a href=#156 id=156 data-nosnippet>156</a><span class="kw">impl </span>RecordBatchTransformerBuilder {
+<a href=#157 id=157 data-nosnippet>157</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>new(
+<a href=#158 id=158 data-nosnippet>158</a>        snapshot_schema: Arc&lt;IcebergSchema&gt;,
+<a href=#159 id=159 data-nosnippet>159</a>        projected_iceberg_field_ids: <span class="kw-2">&amp;</span>[i32],
+<a href=#160 id=160 data-nosnippet>160</a>    ) -&gt; <span class="self">Self </span>{
+<a href=#161 id=161 data-nosnippet>161</a>        <span class="self">Self </span>{
+<a href=#162 id=162 data-nosnippet>162</a>            snapshot_schema,
+<a href=#163 id=163 data-nosnippet>163</a>            projected_iceberg_field_ids: projected_iceberg_field_ids.to_vec(),
+<a href=#164 id=164 data-nosnippet>164</a>            partition_spec: <span class="prelude-val">None</span>,
+<a href=#165 id=165 data-nosnippet>165</a>            partition_data: <span class="prelude-val">None</span>,
+<a href=#166 id=166 data-nosnippet>166</a>        }
+<a href=#167 id=167 data-nosnippet>167</a>    }
+<a href=#168 id=168 data-nosnippet>168</a>
+<a href=#169 id=169 data-nosnippet>169</a>    <span class="doccomment">/// Set partition spec and data together for identifying identity-transformed partition columns.
+<a href=#170 id=170 data-nosnippet>170</a>    ///
+<a href=#171 id=171 data-nosnippet>171</a>    /// Both partition_spec and partition_data must be provided together since the spec defines
+<a href=#172 id=172 data-nosnippet>172</a>    /// which fields are identity-partitioned, and the data provides their constant values.
+<a href=#173 id=173 data-nosnippet>173</a>    /// One without the other cannot produce a valid constants map.
+<a href=#174 id=174 data-nosnippet>174</a>    </span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>with_partition(
+<a href=#175 id=175 data-nosnippet>175</a>        <span class="kw-2">mut </span><span class="self">self</span>,
+<a href=#176 id=176 data-nosnippet>176</a>        partition_spec: Arc&lt;PartitionSpec&gt;,
+<a href=#177 id=177 data-nosnippet>177</a>        partition_data: Struct,
+<a href=#178 id=178 data-nosnippet>178</a>    ) -&gt; <span class="self">Self </span>{
+<a href=#179 id=179 data-nosnippet>179</a>        <span class="self">self</span>.partition_spec = <span class="prelude-val">Some</span>(partition_spec);
+<a href=#180 id=180 data-nosnippet>180</a>        <span class="self">self</span>.partition_data = <span class="prelude-val">Some</span>(partition_data);
+<a href=#181 id=181 data-nosnippet>181</a>        <span class="self">self
+<a href=#182 id=182 data-nosnippet>182</a>    </span>}
+<a href=#183 id=183 data-nosnippet>183</a>
+<a href=#184 id=184 data-nosnippet>184</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>build(<span class="self">self</span>) -&gt; RecordBatchTransformer {
+<a href=#185 id=185 data-nosnippet>185</a>        RecordBatchTransformer {
+<a href=#186 id=186 data-nosnippet>186</a>            snapshot_schema: <span class="self">self</span>.snapshot_schema,
+<a href=#187 id=187 data-nosnippet>187</a>            projected_iceberg_field_ids: <span class="self">self</span>.projected_iceberg_field_ids,
+<a href=#188 id=188 data-nosnippet>188</a>            partition_spec: <span class="self">self</span>.partition_spec,
+<a href=#189 id=189 data-nosnippet>189</a>            partition_data: <span class="self">self</span>.partition_data,
+<a href=#190 id=190 data-nosnippet>190</a>            batch_transform: <span class="prelude-val">None</span>,
+<a href=#191 id=191 data-nosnippet>191</a>        }
+<a href=#192 id=192 data-nosnippet>192</a>    }
+<a href=#193 id=193 data-nosnippet>193</a>}
+<a href=#194 id=194 data-nosnippet>194</a>
+<a href=#195 id=195 data-nosnippet>195</a><span class="doccomment">/// Transforms RecordBatches from Parquet files to match the Iceberg table schema.
+<a href=#196 id=196 data-nosnippet>196</a>///
+<a href=#197 id=197 data-nosnippet>197</a>/// Handles schema evolution, column reordering, type promotion, and implements the Iceberg spec's
+<a href=#198 id=198 data-nosnippet>198</a>/// "Column Projection" rules for resolving field IDs "not present" in data files:
+<a href=#199 id=199 data-nosnippet>199</a>/// 1. Return the value from partition metadata if an Identity Transform exists
+<a href=#200 id=200 data-nosnippet>200</a>/// 2. Use schema.name-mapping.default metadata to map field id to columns without field id (applied in ArrowReader)
+<a href=#201 id=201 data-nosnippet>201</a>/// 3. Return the default value if it has a defined initial-default
+<a href=#202 id=202 data-nosnippet>202</a>/// 4. Return null in all other cases
+<a href=#203 id=203 data-nosnippet>203</a>///
+<a href=#204 id=204 data-nosnippet>204</a>/// # Field ID Resolution
+<a href=#205 id=205 data-nosnippet>205</a>///
+<a href=#206 id=206 data-nosnippet>206</a>/// Field ID resolution happens in ArrowReader before data is read (matching Java's ReadConf):
+<a href=#207 id=207 data-nosnippet>207</a>/// - If file has embedded field IDs: trust them (ParquetSchemaUtil.hasIds() = true)
+<a href=#208 id=208 data-nosnippet>208</a>/// - If file lacks IDs and name_mapping exists: apply name mapping (ParquetSchemaUtil.applyNameMapping())
+<a href=#209 id=209 data-nosnippet>209</a>/// - If file lacks IDs and no name_mapping: use position-based fallback (ParquetSchemaUtil.addFallbackIds())
+<a href=#210 id=210 data-nosnippet>210</a>///
+<a href=#211 id=211 data-nosnippet>211</a>/// By the time RecordBatchTransformer processes data, all field IDs are trustworthy.
+<a href=#212 id=212 data-nosnippet>212</a>/// This transformer only handles remaining projection rules (#1, #3, #4) for fields still "not present".
+<a href=#213 id=213 data-nosnippet>213</a>///
+<a href=#214 id=214 data-nosnippet>214</a>/// # Partition Spec and Data
+<a href=#215 id=215 data-nosnippet>215</a>///
+<a href=#216 id=216 data-nosnippet>216</a>/// **Bucket partitioning**: Distinguish identity transforms (use partition metadata constants)
+<a href=#217 id=217 data-nosnippet>217</a>/// from non-identity transforms like bucket (read from data file) to enable runtime filtering on
+<a href=#218 id=218 data-nosnippet>218</a>/// bucket-partitioned columns. For example, `bucket(4, id)` stores only the bucket number in
+<a href=#219 id=219 data-nosnippet>219</a>/// partition metadata, so actual `id` values must be read from the data file.
+<a href=#220 id=220 data-nosnippet>220</a>///
+<a href=#221 id=221 data-nosnippet>221</a>/// # References
+<a href=#222 id=222 data-nosnippet>222</a>/// - Spec: https://iceberg.apache.org/spec/#column-projection
+<a href=#223 id=223 data-nosnippet>223</a>/// - Java: parquet/src/main/java/org/apache/iceberg/parquet/ReadConf.java (field ID resolution)
+<a href=#224 id=224 data-nosnippet>224</a>/// - Java: core/src/main/java/org/apache/iceberg/util/PartitionUtil.java (partition constants)
+<a href=#225 id=225 data-nosnippet>225</a></span><span class="attr">#[derive(Debug)]
+<a href=#226 id=226 data-nosnippet>226</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">struct </span>RecordBatchTransformer {
+<a href=#227 id=227 data-nosnippet>227</a>    snapshot_schema: Arc&lt;IcebergSchema&gt;,
+<a href=#228 id=228 data-nosnippet>228</a>    projected_iceberg_field_ids: Vec&lt;i32&gt;,
+<a href=#229 id=229 data-nosnippet>229</a>
+<a href=#230 id=230 data-nosnippet>230</a>    <span class="doccomment">/// Partition spec for identifying identity-transformed partition columns (spec rule #1).
+<a href=#231 id=231 data-nosnippet>231</a>    /// Only fields with identity transforms use partition data constants; non-identity transforms
+<a href=#232 id=232 data-nosnippet>232</a>    /// (bucket, truncate, etc.) must read source columns from data files.
+<a href=#233 id=233 data-nosnippet>233</a>    </span>partition_spec: <span class="prelude-ty">Option</span>&lt;Arc&lt;PartitionSpec&gt;&gt;,
+<a href=#234 id=234 data-nosnippet>234</a>
+<a href=#235 id=235 data-nosnippet>235</a>    <span class="doccomment">/// Partition data providing constant values for identity-transformed partition columns (spec rule #1).
+<a href=#236 id=236 data-nosnippet>236</a>    /// For example, in a file at path `dept=engineering/file.parquet`, this would contain
+<a href=#237 id=237 data-nosnippet>237</a>    /// the value "engineering" for the dept field.
+<a href=#238 id=238 data-nosnippet>238</a>    </span>partition_data: <span class="prelude-ty">Option</span>&lt;Struct&gt;,
+<a href=#239 id=239 data-nosnippet>239</a>
+<a href=#240 id=240 data-nosnippet>240</a>    <span class="comment">// BatchTransform gets lazily constructed based on the schema of
+<a href=#241 id=241 data-nosnippet>241</a>    // the first RecordBatch we receive from the file
+<a href=#242 id=242 data-nosnippet>242</a>    </span>batch_transform: <span class="prelude-ty">Option</span>&lt;BatchTransform&gt;,
+<a href=#243 id=243 data-nosnippet>243</a>}
 <a href=#244 id=244 data-nosnippet>244</a>
-<a href=#245 id=245 data-nosnippet>245</a>        <span class="kw">for </span>(source_field, target_field) <span class="kw">in </span>source_schema
-<a href=#246 id=246 data-nosnippet>246</a>            .fields()
-<a href=#247 id=247 data-nosnippet>247</a>            .iter()
-<a href=#248 id=248 data-nosnippet>248</a>            .zip(target_schema.fields().iter())
-<a href=#249 id=249 data-nosnippet>249</a>        {
-<a href=#250 id=250 data-nosnippet>250</a>            <span class="kw">if </span>source_field.data_type() != target_field.data_type()
-<a href=#251 id=251 data-nosnippet>251</a>                || source_field.is_nullable() != target_field.is_nullable()
-<a href=#252 id=252 data-nosnippet>252</a>            {
-<a href=#253 id=253 data-nosnippet>253</a>                <span class="kw">return </span>SchemaComparison::Different;
-<a href=#254 id=254 data-nosnippet>254</a>            }
-<a href=#255 id=255 data-nosnippet>255</a>
-<a href=#256 id=256 data-nosnippet>256</a>            <span class="kw">if </span>source_field.name() != target_field.name() {
-<a href=#257 id=257 data-nosnippet>257</a>                names_changed = <span class="bool-val">true</span>;
-<a href=#258 id=258 data-nosnippet>258</a>            }
-<a href=#259 id=259 data-nosnippet>259</a>        }
-<a href=#260 id=260 data-nosnippet>260</a>
-<a href=#261 id=261 data-nosnippet>261</a>        <span class="kw">if </span>names_changed {
-<a href=#262 id=262 data-nosnippet>262</a>            SchemaComparison::NameChangesOnly
-<a href=#263 id=263 data-nosnippet>263</a>        } <span class="kw">else </span>{
-<a href=#264 id=264 data-nosnippet>264</a>            SchemaComparison::Equivalent
-<a href=#265 id=265 data-nosnippet>265</a>        }
-<a href=#266 id=266 data-nosnippet>266</a>    }
-<a href=#267 id=267 data-nosnippet>267</a>
-<a href=#268 id=268 data-nosnippet>268</a>    <span class="kw">fn </span>generate_transform_operations(
-<a href=#269 id=269 data-nosnippet>269</a>        source_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
-<a href=#270 id=270 data-nosnippet>270</a>        snapshot_schema: <span class="kw-2">&amp;</span>IcebergSchema,
-<a href=#271 id=271 data-nosnippet>271</a>        projected_iceberg_field_ids: <span class="kw-2">&amp;</span>[i32],
-<a href=#272 id=272 data-nosnippet>272</a>        field_id_to_mapped_schema_map: HashMap&lt;i32, (FieldRef, usize)&gt;,
-<a href=#273 id=273 data-nosnippet>273</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;ColumnSource&gt;&gt; {
-<a href=#274 id=274 data-nosnippet>274</a>        <span class="kw">let </span>field_id_to_source_schema_map =
-<a href=#275 id=275 data-nosnippet>275</a>            <span class="self">Self</span>::build_field_id_to_arrow_schema_map(source_schema)<span class="question-mark">?</span>;
-<a href=#276 id=276 data-nosnippet>276</a>
-<a href=#277 id=277 data-nosnippet>277</a>        projected_iceberg_field_ids.iter().map(|field_id|{
-<a href=#278 id=278 data-nosnippet>278</a>            <span class="kw">let </span>(target_field, <span class="kw">_</span>) = field_id_to_mapped_schema_map.get(field_id).ok_or(
-<a href=#279 id=279 data-nosnippet>279</a>                Error::new(ErrorKind::Unexpected, <span class="string">"could not find field in schema"</span>)
-<a href=#280 id=280 data-nosnippet>280</a>            )<span class="question-mark">?</span>;
-<a href=#281 id=281 data-nosnippet>281</a>            <span class="kw">let </span>target_type = target_field.data_type();
-<a href=#282 id=282 data-nosnippet>282</a>
-<a href=#283 id=283 data-nosnippet>283</a>            <span class="prelude-val">Ok</span>(<span class="kw">if let </span><span class="prelude-val">Some</span>((source_field, source_index)) = field_id_to_source_schema_map.get(field_id) {
-<a href=#284 id=284 data-nosnippet>284</a>                <span class="comment">// column present in source
-<a href=#285 id=285 data-nosnippet>285</a>
-<a href=#286 id=286 data-nosnippet>286</a>                </span><span class="kw">if </span>source_field.data_type().equals_datatype(target_type) {
-<a href=#287 id=287 data-nosnippet>287</a>                    <span class="comment">// no promotion required
-<a href=#288 id=288 data-nosnippet>288</a>                    </span>ColumnSource::PassThrough {
-<a href=#289 id=289 data-nosnippet>289</a>                        source_index: <span class="kw-2">*</span>source_index
-<a href=#290 id=290 data-nosnippet>290</a>                    }
-<a href=#291 id=291 data-nosnippet>291</a>                } <span class="kw">else </span>{
-<a href=#292 id=292 data-nosnippet>292</a>                    <span class="comment">// promotion required
-<a href=#293 id=293 data-nosnippet>293</a>                    </span>ColumnSource::Promote {
-<a href=#294 id=294 data-nosnippet>294</a>                        target_type: target_type.clone(),
-<a href=#295 id=295 data-nosnippet>295</a>                        source_index: <span class="kw-2">*</span>source_index,
-<a href=#296 id=296 data-nosnippet>296</a>                    }
-<a href=#297 id=297 data-nosnippet>297</a>                }
-<a href=#298 id=298 data-nosnippet>298</a>            } <span class="kw">else </span>{
-<a href=#299 id=299 data-nosnippet>299</a>                <span class="comment">// column must be added
-<a href=#300 id=300 data-nosnippet>300</a>                </span><span class="kw">let </span>iceberg_field = snapshot_schema.field_by_id(<span class="kw-2">*</span>field_id).ok_or(
-<a href=#301 id=301 data-nosnippet>301</a>                    Error::new(ErrorKind::Unexpected, <span class="string">"Field not found in snapshot schema"</span>)
-<a href=#302 id=302 data-nosnippet>302</a>                )<span class="question-mark">?</span>;
-<a href=#303 id=303 data-nosnippet>303</a>
-<a href=#304 id=304 data-nosnippet>304</a>                <span class="kw">let </span>default_value = <span class="kw">if let </span><span class="prelude-val">Some</span>(iceberg_default_value) =
-<a href=#305 id=305 data-nosnippet>305</a>                    <span class="kw-2">&amp;</span>iceberg_field.initial_default
-<a href=#306 id=306 data-nosnippet>306</a>                {
-<a href=#307 id=307 data-nosnippet>307</a>                    <span class="kw">let </span>Literal::Primitive(primitive_literal) = iceberg_default_value <span class="kw">else </span>{
-<a href=#308 id=308 data-nosnippet>308</a>                        <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#309 id=309 data-nosnippet>309</a>                            ErrorKind::Unexpected,
-<a href=#310 id=310 data-nosnippet>310</a>                            <span class="macro">format!</span>(<span class="string">"Default value for column must be primitive type, but encountered {iceberg_default_value:?}"</span>)
-<a href=#311 id=311 data-nosnippet>311</a>                        ));
-<a href=#312 id=312 data-nosnippet>312</a>                    };
-<a href=#313 id=313 data-nosnippet>313</a>                    <span class="prelude-val">Some</span>(primitive_literal.clone())
-<a href=#314 id=314 data-nosnippet>314</a>                } <span class="kw">else </span>{
-<a href=#315 id=315 data-nosnippet>315</a>                    <span class="prelude-val">None
-<a href=#316 id=316 data-nosnippet>316</a>                </span>};
-<a href=#317 id=317 data-nosnippet>317</a>
-<a href=#318 id=318 data-nosnippet>318</a>                ColumnSource::Add {
-<a href=#319 id=319 data-nosnippet>319</a>                    value: default_value,
-<a href=#320 id=320 data-nosnippet>320</a>                    target_type: target_type.clone(),
-<a href=#321 id=321 data-nosnippet>321</a>                }
-<a href=#322 id=322 data-nosnippet>322</a>            })
-<a href=#323 id=323 data-nosnippet>323</a>        }).collect()
-<a href=#324 id=324 data-nosnippet>324</a>    }
-<a href=#325 id=325 data-nosnippet>325</a>
-<a href=#326 id=326 data-nosnippet>326</a>    <span class="kw">fn </span>build_field_id_to_arrow_schema_map(
-<a href=#327 id=327 data-nosnippet>327</a>        source_schema: <span class="kw-2">&amp;</span>SchemaRef,
-<a href=#328 id=328 data-nosnippet>328</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;HashMap&lt;i32, (FieldRef, usize)&gt;&gt; {
-<a href=#329 id=329 data-nosnippet>329</a>        <span class="kw">let </span><span class="kw-2">mut </span>field_id_to_source_schema = HashMap::new();
-<a href=#330 id=330 data-nosnippet>330</a>        <span class="kw">for </span>(source_field_idx, source_field) <span class="kw">in </span>source_schema.fields.iter().enumerate() {
-<a href=#331 id=331 data-nosnippet>331</a>            <span class="kw">let </span>this_field_id = source_field
-<a href=#332 id=332 data-nosnippet>332</a>                .metadata()
-<a href=#333 id=333 data-nosnippet>333</a>                .get(PARQUET_FIELD_ID_META_KEY)
-<a href=#334 id=334 data-nosnippet>334</a>                .ok_or_else(|| {
-<a href=#335 id=335 data-nosnippet>335</a>                    Error::new(
-<a href=#336 id=336 data-nosnippet>336</a>                        ErrorKind::DataInvalid,
-<a href=#337 id=337 data-nosnippet>337</a>                        <span class="string">"field ID not present in parquet metadata"</span>,
-<a href=#338 id=338 data-nosnippet>338</a>                    )
-<a href=#339 id=339 data-nosnippet>339</a>                })<span class="question-mark">?
-<a href=#340 id=340 data-nosnippet>340</a>                </span>.parse()
-<a href=#341 id=341 data-nosnippet>341</a>                .map_err(|e| {
-<a href=#342 id=342 data-nosnippet>342</a>                    Error::new(
-<a href=#343 id=343 data-nosnippet>343</a>                        ErrorKind::DataInvalid,
-<a href=#344 id=344 data-nosnippet>344</a>                        <span class="macro">format!</span>(<span class="string">"field id not parseable as an i32: {e}"</span>),
-<a href=#345 id=345 data-nosnippet>345</a>                    )
-<a href=#346 id=346 data-nosnippet>346</a>                })<span class="question-mark">?</span>;
-<a href=#347 id=347 data-nosnippet>347</a>
-<a href=#348 id=348 data-nosnippet>348</a>            field_id_to_source_schema
-<a href=#349 id=349 data-nosnippet>349</a>                .insert(this_field_id, (source_field.clone(), source_field_idx));
-<a href=#350 id=350 data-nosnippet>350</a>        }
-<a href=#351 id=351 data-nosnippet>351</a>
-<a href=#352 id=352 data-nosnippet>352</a>        <span class="prelude-val">Ok</span>(field_id_to_source_schema)
-<a href=#353 id=353 data-nosnippet>353</a>    }
-<a href=#354 id=354 data-nosnippet>354</a>
-<a href=#355 id=355 data-nosnippet>355</a>    <span class="kw">fn </span>transform_columns(
-<a href=#356 id=356 data-nosnippet>356</a>        <span class="kw-2">&amp;</span><span class="self">self</span>,
-<a href=#357 id=357 data-nosnippet>357</a>        columns: <span class="kw-2">&amp;</span>[Arc&lt;<span class="kw">dyn </span>ArrowArray&gt;],
-<a href=#358 id=358 data-nosnippet>358</a>        operations: <span class="kw-2">&amp;</span>[ColumnSource],
-<a href=#359 id=359 data-nosnippet>359</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;Arc&lt;<span class="kw">dyn </span>ArrowArray&gt;&gt;&gt; {
-<a href=#360 id=360 data-nosnippet>360</a>        <span class="kw">if </span>columns.is_empty() {
-<a href=#361 id=361 data-nosnippet>361</a>            <span class="kw">return </span><span class="prelude-val">Ok</span>(columns.to_vec());
-<a href=#362 id=362 data-nosnippet>362</a>        }
-<a href=#363 id=363 data-nosnippet>363</a>        <span class="kw">let </span>num_rows = columns[<span class="number">0</span>].len();
+<a href=#245 id=245 data-nosnippet>245</a><span class="kw">impl </span>RecordBatchTransformer {
+<a href=#246 id=246 data-nosnippet>246</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>process_record_batch(
+<a href=#247 id=247 data-nosnippet>247</a>        <span class="kw-2">&amp;mut </span><span class="self">self</span>,
+<a href=#248 id=248 data-nosnippet>248</a>        record_batch: RecordBatch,
+<a href=#249 id=249 data-nosnippet>249</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;RecordBatch&gt; {
+<a href=#250 id=250 data-nosnippet>250</a>        <span class="prelude-val">Ok</span>(<span class="kw">match </span><span class="kw-2">&amp;</span><span class="self">self</span>.batch_transform {
+<a href=#251 id=251 data-nosnippet>251</a>            <span class="prelude-val">Some</span>(BatchTransform::PassThrough) =&gt; record_batch,
+<a href=#252 id=252 data-nosnippet>252</a>            <span class="prelude-val">Some</span>(BatchTransform::Modify {
+<a href=#253 id=253 data-nosnippet>253</a>                target_schema,
+<a href=#254 id=254 data-nosnippet>254</a>                operations,
+<a href=#255 id=255 data-nosnippet>255</a>            }) =&gt; {
+<a href=#256 id=256 data-nosnippet>256</a>                <span class="kw">let </span>options = RecordBatchOptions::default()
+<a href=#257 id=257 data-nosnippet>257</a>                    .with_match_field_names(<span class="bool-val">false</span>)
+<a href=#258 id=258 data-nosnippet>258</a>                    .with_row_count(<span class="prelude-val">Some</span>(record_batch.num_rows()));
+<a href=#259 id=259 data-nosnippet>259</a>                RecordBatch::try_new_with_options(
+<a href=#260 id=260 data-nosnippet>260</a>                    Arc::clone(target_schema),
+<a href=#261 id=261 data-nosnippet>261</a>                    <span class="self">self</span>.transform_columns(record_batch.columns(), operations)<span class="question-mark">?</span>,
+<a href=#262 id=262 data-nosnippet>262</a>                    <span class="kw-2">&amp;</span>options,
+<a href=#263 id=263 data-nosnippet>263</a>                )<span class="question-mark">?
+<a href=#264 id=264 data-nosnippet>264</a>            </span>}
+<a href=#265 id=265 data-nosnippet>265</a>            <span class="prelude-val">Some</span>(BatchTransform::ModifySchema { target_schema }) =&gt; {
+<a href=#266 id=266 data-nosnippet>266</a>                <span class="kw">let </span>options = RecordBatchOptions::default()
+<a href=#267 id=267 data-nosnippet>267</a>                    .with_match_field_names(<span class="bool-val">false</span>)
+<a href=#268 id=268 data-nosnippet>268</a>                    .with_row_count(<span class="prelude-val">Some</span>(record_batch.num_rows()));
+<a href=#269 id=269 data-nosnippet>269</a>                RecordBatch::try_new_with_options(
+<a href=#270 id=270 data-nosnippet>270</a>                    Arc::clone(target_schema),
+<a href=#271 id=271 data-nosnippet>271</a>                    record_batch.columns().to_vec(),
+<a href=#272 id=272 data-nosnippet>272</a>                    <span class="kw-2">&amp;</span>options,
+<a href=#273 id=273 data-nosnippet>273</a>                )<span class="question-mark">?
+<a href=#274 id=274 data-nosnippet>274</a>            </span>}
+<a href=#275 id=275 data-nosnippet>275</a>            <span class="prelude-val">None </span>=&gt; {
+<a href=#276 id=276 data-nosnippet>276</a>                <span class="self">self</span>.batch_transform = <span class="prelude-val">Some</span>(<span class="self">Self</span>::generate_batch_transform(
+<a href=#277 id=277 data-nosnippet>277</a>                    record_batch.schema_ref(),
+<a href=#278 id=278 data-nosnippet>278</a>                    <span class="self">self</span>.snapshot_schema.as_ref(),
+<a href=#279 id=279 data-nosnippet>279</a>                    <span class="kw-2">&amp;</span><span class="self">self</span>.projected_iceberg_field_ids,
+<a href=#280 id=280 data-nosnippet>280</a>                    <span class="self">self</span>.partition_spec.as_ref().map(|s| s.as_ref()),
+<a href=#281 id=281 data-nosnippet>281</a>                    <span class="self">self</span>.partition_data.as_ref(),
+<a href=#282 id=282 data-nosnippet>282</a>                )<span class="question-mark">?</span>);
+<a href=#283 id=283 data-nosnippet>283</a>
+<a href=#284 id=284 data-nosnippet>284</a>                <span class="self">self</span>.process_record_batch(record_batch)<span class="question-mark">?
+<a href=#285 id=285 data-nosnippet>285</a>            </span>}
+<a href=#286 id=286 data-nosnippet>286</a>        })
+<a href=#287 id=287 data-nosnippet>287</a>    }
+<a href=#288 id=288 data-nosnippet>288</a>
+<a href=#289 id=289 data-nosnippet>289</a>    <span class="comment">// Compare the schema of the incoming RecordBatches to the schema of
+<a href=#290 id=290 data-nosnippet>290</a>    // the Iceberg snapshot to determine what, if any, transformation
+<a href=#291 id=291 data-nosnippet>291</a>    // needs to be applied. If the schemas match, we return BatchTransform::PassThrough
+<a href=#292 id=292 data-nosnippet>292</a>    // to indicate that no changes need to be made. Otherwise, we return a
+<a href=#293 id=293 data-nosnippet>293</a>    // BatchTransform::Modify containing the target RecordBatch schema and
+<a href=#294 id=294 data-nosnippet>294</a>    // the list of `ColumnSource`s that indicate how to source each column in
+<a href=#295 id=295 data-nosnippet>295</a>    // the resulting RecordBatches.
+<a href=#296 id=296 data-nosnippet>296</a>    </span><span class="kw">fn </span>generate_batch_transform(
+<a href=#297 id=297 data-nosnippet>297</a>        source_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
+<a href=#298 id=298 data-nosnippet>298</a>        snapshot_schema: <span class="kw-2">&amp;</span>IcebergSchema,
+<a href=#299 id=299 data-nosnippet>299</a>        projected_iceberg_field_ids: <span class="kw-2">&amp;</span>[i32],
+<a href=#300 id=300 data-nosnippet>300</a>        partition_spec: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PartitionSpec&gt;,
+<a href=#301 id=301 data-nosnippet>301</a>        partition_data: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>Struct&gt;,
+<a href=#302 id=302 data-nosnippet>302</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;BatchTransform&gt; {
+<a href=#303 id=303 data-nosnippet>303</a>        <span class="kw">let </span>mapped_unprojected_arrow_schema = Arc::new(schema_to_arrow_schema(snapshot_schema)<span class="question-mark">?</span>);
+<a href=#304 id=304 data-nosnippet>304</a>        <span class="kw">let </span>field_id_to_mapped_schema_map =
+<a href=#305 id=305 data-nosnippet>305</a>            <span class="self">Self</span>::build_field_id_to_arrow_schema_map(<span class="kw-2">&amp;</span>mapped_unprojected_arrow_schema)<span class="question-mark">?</span>;
+<a href=#306 id=306 data-nosnippet>306</a>
+<a href=#307 id=307 data-nosnippet>307</a>        <span class="comment">// Create a new arrow schema by selecting fields from mapped_unprojected,
+<a href=#308 id=308 data-nosnippet>308</a>        // in the order of the field ids in projected_iceberg_field_ids
+<a href=#309 id=309 data-nosnippet>309</a>        </span><span class="kw">let </span>fields: <span class="prelude-ty">Result</span>&lt;Vec&lt;<span class="kw">_</span>&gt;&gt; = projected_iceberg_field_ids
+<a href=#310 id=310 data-nosnippet>310</a>            .iter()
+<a href=#311 id=311 data-nosnippet>311</a>            .map(|field_id| {
+<a href=#312 id=312 data-nosnippet>312</a>                <span class="prelude-val">Ok</span>(field_id_to_mapped_schema_map
+<a href=#313 id=313 data-nosnippet>313</a>                    .get(field_id)
+<a href=#314 id=314 data-nosnippet>314</a>                    .ok_or(Error::new(ErrorKind::Unexpected, <span class="string">"field not found"</span>))<span class="question-mark">?
+<a href=#315 id=315 data-nosnippet>315</a>                    </span>.<span class="number">0
+<a href=#316 id=316 data-nosnippet>316</a>                    </span>.clone())
+<a href=#317 id=317 data-nosnippet>317</a>            })
+<a href=#318 id=318 data-nosnippet>318</a>            .collect();
+<a href=#319 id=319 data-nosnippet>319</a>
+<a href=#320 id=320 data-nosnippet>320</a>        <span class="kw">let </span>target_schema = Arc::new(ArrowSchema::new(fields<span class="question-mark">?</span>));
+<a href=#321 id=321 data-nosnippet>321</a>
+<a href=#322 id=322 data-nosnippet>322</a>        <span class="kw">let </span>constants_map = <span class="kw">if let </span>(<span class="prelude-val">Some</span>(spec), <span class="prelude-val">Some</span>(data)) = (partition_spec, partition_data) {
+<a href=#323 id=323 data-nosnippet>323</a>            constants_map(spec, data)
+<a href=#324 id=324 data-nosnippet>324</a>        } <span class="kw">else </span>{
+<a href=#325 id=325 data-nosnippet>325</a>            HashMap::new()
+<a href=#326 id=326 data-nosnippet>326</a>        };
+<a href=#327 id=327 data-nosnippet>327</a>
+<a href=#328 id=328 data-nosnippet>328</a>        <span class="kw">match </span><span class="self">Self</span>::compare_schemas(source_schema, <span class="kw-2">&amp;</span>target_schema) {
+<a href=#329 id=329 data-nosnippet>329</a>            SchemaComparison::Equivalent =&gt; <span class="prelude-val">Ok</span>(BatchTransform::PassThrough),
+<a href=#330 id=330 data-nosnippet>330</a>            SchemaComparison::NameChangesOnly =&gt; <span class="prelude-val">Ok</span>(BatchTransform::ModifySchema { target_schema }),
+<a href=#331 id=331 data-nosnippet>331</a>            SchemaComparison::Different =&gt; <span class="prelude-val">Ok</span>(BatchTransform::Modify {
+<a href=#332 id=332 data-nosnippet>332</a>                operations: <span class="self">Self</span>::generate_transform_operations(
+<a href=#333 id=333 data-nosnippet>333</a>                    source_schema,
+<a href=#334 id=334 data-nosnippet>334</a>                    snapshot_schema,
+<a href=#335 id=335 data-nosnippet>335</a>                    projected_iceberg_field_ids,
+<a href=#336 id=336 data-nosnippet>336</a>                    field_id_to_mapped_schema_map,
+<a href=#337 id=337 data-nosnippet>337</a>                    constants_map,
+<a href=#338 id=338 data-nosnippet>338</a>                    partition_spec,
+<a href=#339 id=339 data-nosnippet>339</a>                )<span class="question-mark">?</span>,
+<a href=#340 id=340 data-nosnippet>340</a>                target_schema,
+<a href=#341 id=341 data-nosnippet>341</a>            }),
+<a href=#342 id=342 data-nosnippet>342</a>        }
+<a href=#343 id=343 data-nosnippet>343</a>    }
+<a href=#344 id=344 data-nosnippet>344</a>
+<a href=#345 id=345 data-nosnippet>345</a>    <span class="doccomment">/// Compares the source and target schemas
+<a href=#346 id=346 data-nosnippet>346</a>    /// Determines if they have changed in any meaningful way:
+<a href=#347 id=347 data-nosnippet>347</a>    ///  * If they have different numbers of fields, then we need to modify
+<a href=#348 id=348 data-nosnippet>348</a>    ///    the incoming RecordBatch schema AND columns
+<a href=#349 id=349 data-nosnippet>349</a>    ///  * If they have the same number of fields, but some of them differ in
+<a href=#350 id=350 data-nosnippet>350</a>    ///    either data type or nullability, then we need to modify the
+<a href=#351 id=351 data-nosnippet>351</a>    ///    incoming RecordBatch schema AND columns
+<a href=#352 id=352 data-nosnippet>352</a>    ///  * If the schemas differ only in the column names, then we need
+<a href=#353 id=353 data-nosnippet>353</a>    ///    to modify the RecordBatch schema BUT we can keep the
+<a href=#354 id=354 data-nosnippet>354</a>    ///    original column data unmodified
+<a href=#355 id=355 data-nosnippet>355</a>    ///  * If the schemas are identical (or differ only in inconsequential
+<a href=#356 id=356 data-nosnippet>356</a>    ///    ways) then we can pass through the original RecordBatch unmodified
+<a href=#357 id=357 data-nosnippet>357</a>    </span><span class="kw">fn </span>compare_schemas(
+<a href=#358 id=358 data-nosnippet>358</a>        source_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
+<a href=#359 id=359 data-nosnippet>359</a>        target_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
+<a href=#360 id=360 data-nosnippet>360</a>    ) -&gt; SchemaComparison {
+<a href=#361 id=361 data-nosnippet>361</a>        <span class="kw">if </span>source_schema.fields().len() != target_schema.fields().len() {
+<a href=#362 id=362 data-nosnippet>362</a>            <span class="kw">return </span>SchemaComparison::Different;
+<a href=#363 id=363 data-nosnippet>363</a>        }
 <a href=#364 id=364 data-nosnippet>364</a>
-<a href=#365 id=365 data-nosnippet>365</a>        operations
-<a href=#366 id=366 data-nosnippet>366</a>            .iter()
-<a href=#367 id=367 data-nosnippet>367</a>            .map(|op| {
-<a href=#368 id=368 data-nosnippet>368</a>                <span class="prelude-val">Ok</span>(<span class="kw">match </span>op {
-<a href=#369 id=369 data-nosnippet>369</a>                    ColumnSource::PassThrough { source_index } =&gt; columns[<span class="kw-2">*</span>source_index].clone(),
-<a href=#370 id=370 data-nosnippet>370</a>
-<a href=#371 id=371 data-nosnippet>371</a>                    ColumnSource::Promote {
-<a href=#372 id=372 data-nosnippet>372</a>                        target_type,
-<a href=#373 id=373 data-nosnippet>373</a>                        source_index,
-<a href=#374 id=374 data-nosnippet>374</a>                    } =&gt; cast(<span class="kw-2">&amp;*</span>columns[<span class="kw-2">*</span>source_index], target_type)<span class="question-mark">?</span>,
-<a href=#375 id=375 data-nosnippet>375</a>
-<a href=#376 id=376 data-nosnippet>376</a>                    ColumnSource::Add { target_type, value } =&gt; {
-<a href=#377 id=377 data-nosnippet>377</a>                        <span class="self">Self</span>::create_column(target_type, value, num_rows)<span class="question-mark">?
-<a href=#378 id=378 data-nosnippet>378</a>                    </span>}
-<a href=#379 id=379 data-nosnippet>379</a>                })
-<a href=#380 id=380 data-nosnippet>380</a>            })
-<a href=#381 id=381 data-nosnippet>381</a>            .collect()
-<a href=#382 id=382 data-nosnippet>382</a>    }
-<a href=#383 id=383 data-nosnippet>383</a>
-<a href=#384 id=384 data-nosnippet>384</a>    <span class="kw">fn </span>create_column(
-<a href=#385 id=385 data-nosnippet>385</a>        target_type: <span class="kw-2">&amp;</span>DataType,
-<a href=#386 id=386 data-nosnippet>386</a>        prim_lit: <span class="kw-2">&amp;</span><span class="prelude-ty">Option</span>&lt;PrimitiveLiteral&gt;,
-<a href=#387 id=387 data-nosnippet>387</a>        num_rows: usize,
-<a href=#388 id=388 data-nosnippet>388</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ArrayRef&gt; {
-<a href=#389 id=389 data-nosnippet>389</a>        <span class="prelude-val">Ok</span>(<span class="kw">match </span>(target_type, prim_lit) {
-<a href=#390 id=390 data-nosnippet>390</a>            (DataType::Boolean, <span class="prelude-val">Some</span>(PrimitiveLiteral::Boolean(value))) =&gt; {
-<a href=#391 id=391 data-nosnippet>391</a>                Arc::new(BooleanArray::from(<span class="macro">vec!</span>[<span class="kw-2">*</span>value; num_rows]))
-<a href=#392 id=392 data-nosnippet>392</a>            }
-<a href=#393 id=393 data-nosnippet>393</a>            (DataType::Boolean, <span class="prelude-val">None</span>) =&gt; {
-<a href=#394 id=394 data-nosnippet>394</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;bool&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
-<a href=#395 id=395 data-nosnippet>395</a>                Arc::new(BooleanArray::from(vals))
-<a href=#396 id=396 data-nosnippet>396</a>            }
-<a href=#397 id=397 data-nosnippet>397</a>            (DataType::Int32, <span class="prelude-val">Some</span>(PrimitiveLiteral::Int(value))) =&gt; {
-<a href=#398 id=398 data-nosnippet>398</a>                Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="kw-2">*</span>value; num_rows]))
-<a href=#399 id=399 data-nosnippet>399</a>            }
-<a href=#400 id=400 data-nosnippet>400</a>            (DataType::Int32, <span class="prelude-val">None</span>) =&gt; {
-<a href=#401 id=401 data-nosnippet>401</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;i32&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
-<a href=#402 id=402 data-nosnippet>402</a>                Arc::new(Int32Array::from(vals))
-<a href=#403 id=403 data-nosnippet>403</a>            }
-<a href=#404 id=404 data-nosnippet>404</a>            (DataType::Date32, <span class="prelude-val">Some</span>(PrimitiveLiteral::Int(value))) =&gt; {
-<a href=#405 id=405 data-nosnippet>405</a>                Arc::new(Date32Array::from(<span class="macro">vec!</span>[<span class="kw-2">*</span>value; num_rows]))
-<a href=#406 id=406 data-nosnippet>406</a>            }
-<a href=#407 id=407 data-nosnippet>407</a>            (DataType::Date32, <span class="prelude-val">None</span>) =&gt; {
-<a href=#408 id=408 data-nosnippet>408</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;i32&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
-<a href=#409 id=409 data-nosnippet>409</a>                Arc::new(Date32Array::from(vals))
-<a href=#410 id=410 data-nosnippet>410</a>            }
-<a href=#411 id=411 data-nosnippet>411</a>            (DataType::Int64, <span class="prelude-val">Some</span>(PrimitiveLiteral::Long(value))) =&gt; {
-<a href=#412 id=412 data-nosnippet>412</a>                Arc::new(Int64Array::from(<span class="macro">vec!</span>[<span class="kw-2">*</span>value; num_rows]))
-<a href=#413 id=413 data-nosnippet>413</a>            }
-<a href=#414 id=414 data-nosnippet>414</a>            (DataType::Int64, <span class="prelude-val">None</span>) =&gt; {
-<a href=#415 id=415 data-nosnippet>415</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;i64&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
-<a href=#416 id=416 data-nosnippet>416</a>                Arc::new(Int64Array::from(vals))
-<a href=#417 id=417 data-nosnippet>417</a>            }
-<a href=#418 id=418 data-nosnippet>418</a>            (DataType::Float32, <span class="prelude-val">Some</span>(PrimitiveLiteral::Float(value))) =&gt; {
-<a href=#419 id=419 data-nosnippet>419</a>                Arc::new(Float32Array::from(<span class="macro">vec!</span>[value.<span class="number">0</span>; num_rows]))
-<a href=#420 id=420 data-nosnippet>420</a>            }
-<a href=#421 id=421 data-nosnippet>421</a>            (DataType::Float32, <span class="prelude-val">None</span>) =&gt; {
-<a href=#422 id=422 data-nosnippet>422</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;f32&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
-<a href=#423 id=423 data-nosnippet>423</a>                Arc::new(Float32Array::from(vals))
-<a href=#424 id=424 data-nosnippet>424</a>            }
-<a href=#425 id=425 data-nosnippet>425</a>            (DataType::Float64, <span class="prelude-val">Some</span>(PrimitiveLiteral::Double(value))) =&gt; {
-<a href=#426 id=426 data-nosnippet>426</a>                Arc::new(Float64Array::from(<span class="macro">vec!</span>[value.<span class="number">0</span>; num_rows]))
-<a href=#427 id=427 data-nosnippet>427</a>            }
-<a href=#428 id=428 data-nosnippet>428</a>            (DataType::Float64, <span class="prelude-val">None</span>) =&gt; {
-<a href=#429 id=429 data-nosnippet>429</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;f64&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
-<a href=#430 id=430 data-nosnippet>430</a>                Arc::new(Float64Array::from(vals))
-<a href=#431 id=431 data-nosnippet>431</a>            }
-<a href=#432 id=432 data-nosnippet>432</a>            (DataType::Utf8, <span class="prelude-val">Some</span>(PrimitiveLiteral::String(value))) =&gt; {
-<a href=#433 id=433 data-nosnippet>433</a>                Arc::new(StringArray::from(<span class="macro">vec!</span>[value.clone(); num_rows]))
-<a href=#434 id=434 data-nosnippet>434</a>            }
-<a href=#435 id=435 data-nosnippet>435</a>            (DataType::Utf8, <span class="prelude-val">None</span>) =&gt; {
-<a href=#436 id=436 data-nosnippet>436</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;String&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
-<a href=#437 id=437 data-nosnippet>437</a>                Arc::new(StringArray::from(vals))
-<a href=#438 id=438 data-nosnippet>438</a>            }
-<a href=#439 id=439 data-nosnippet>439</a>            (DataType::Binary, <span class="prelude-val">Some</span>(PrimitiveLiteral::Binary(value))) =&gt; {
-<a href=#440 id=440 data-nosnippet>440</a>                Arc::new(BinaryArray::from_vec(<span class="macro">vec!</span>[value; num_rows]))
-<a href=#441 id=441 data-nosnippet>441</a>            }
-<a href=#442 id=442 data-nosnippet>442</a>            (DataType::Binary, <span class="prelude-val">None</span>) =&gt; {
-<a href=#443 id=443 data-nosnippet>443</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>[u8]&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
-<a href=#444 id=444 data-nosnippet>444</a>                Arc::new(BinaryArray::from_opt_vec(vals))
-<a href=#445 id=445 data-nosnippet>445</a>            }
-<a href=#446 id=446 data-nosnippet>446</a>            (DataType::Null, <span class="kw">_</span>) =&gt; Arc::new(NullArray::new(num_rows)),
-<a href=#447 id=447 data-nosnippet>447</a>            (dt, <span class="kw">_</span>) =&gt; {
-<a href=#448 id=448 data-nosnippet>448</a>                <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#449 id=449 data-nosnippet>449</a>                    ErrorKind::Unexpected,
-<a href=#450 id=450 data-nosnippet>450</a>                    <span class="macro">format!</span>(<span class="string">"unexpected target column type {dt}"</span>),
-<a href=#451 id=451 data-nosnippet>451</a>                ));
-<a href=#452 id=452 data-nosnippet>452</a>            }
-<a href=#453 id=453 data-nosnippet>453</a>        })
-<a href=#454 id=454 data-nosnippet>454</a>    }
-<a href=#455 id=455 data-nosnippet>455</a>}
-<a href=#456 id=456 data-nosnippet>456</a>
-<a href=#457 id=457 data-nosnippet>457</a><span class="attr">#[cfg(test)]
-<a href=#458 id=458 data-nosnippet>458</a></span><span class="kw">mod </span>test {
-<a href=#459 id=459 data-nosnippet>459</a>    <span class="kw">use </span>std::collections::HashMap;
-<a href=#460 id=460 data-nosnippet>460</a>    <span class="kw">use </span>std::sync::Arc;
-<a href=#461 id=461 data-nosnippet>461</a>
-<a href=#462 id=462 data-nosnippet>462</a>    <span class="kw">use </span>arrow_array::{
-<a href=#463 id=463 data-nosnippet>463</a>        Array, Date32Array, Float32Array, Float64Array, Int32Array, Int64Array, RecordBatch,
-<a href=#464 id=464 data-nosnippet>464</a>        StringArray,
-<a href=#465 id=465 data-nosnippet>465</a>    };
-<a href=#466 id=466 data-nosnippet>466</a>    <span class="kw">use </span>arrow_schema::{DataType, Field, Schema <span class="kw">as </span>ArrowSchema};
-<a href=#467 id=467 data-nosnippet>467</a>    <span class="kw">use </span>parquet::arrow::PARQUET_FIELD_ID_META_KEY;
-<a href=#468 id=468 data-nosnippet>468</a>
-<a href=#469 id=469 data-nosnippet>469</a>    <span class="kw">use </span><span class="kw">crate</span>::arrow::record_batch_transformer::RecordBatchTransformer;
-<a href=#470 id=470 data-nosnippet>470</a>    <span class="kw">use </span><span class="kw">crate</span>::spec::{Literal, NestedField, PrimitiveType, Schema, Type};
-<a href=#471 id=471 data-nosnippet>471</a>
-<a href=#472 id=472 data-nosnippet>472</a>    <span class="attr">#[test]
-<a href=#473 id=473 data-nosnippet>473</a>    </span><span class="kw">fn </span>build_field_id_to_source_schema_map_works() {
-<a href=#474 id=474 data-nosnippet>474</a>        <span class="kw">let </span>arrow_schema = arrow_schema_already_same_as_target();
-<a href=#475 id=475 data-nosnippet>475</a>
-<a href=#476 id=476 data-nosnippet>476</a>        <span class="kw">let </span>result =
-<a href=#477 id=477 data-nosnippet>477</a>            RecordBatchTransformer::build_field_id_to_arrow_schema_map(<span class="kw-2">&amp;</span>arrow_schema).unwrap();
-<a href=#478 id=478 data-nosnippet>478</a>
-<a href=#479 id=479 data-nosnippet>479</a>        <span class="kw">let </span>expected = HashMap::from_iter([
-<a href=#480 id=480 data-nosnippet>480</a>            (<span class="number">10</span>, (arrow_schema.fields()[<span class="number">0</span>].clone(), <span class="number">0</span>)),
-<a href=#481 id=481 data-nosnippet>481</a>            (<span class="number">11</span>, (arrow_schema.fields()[<span class="number">1</span>].clone(), <span class="number">1</span>)),
-<a href=#482 id=482 data-nosnippet>482</a>            (<span class="number">12</span>, (arrow_schema.fields()[<span class="number">2</span>].clone(), <span class="number">2</span>)),
-<a href=#483 id=483 data-nosnippet>483</a>            (<span class="number">14</span>, (arrow_schema.fields()[<span class="number">3</span>].clone(), <span class="number">3</span>)),
-<a href=#484 id=484 data-nosnippet>484</a>            (<span class="number">15</span>, (arrow_schema.fields()[<span class="number">4</span>].clone(), <span class="number">4</span>)),
-<a href=#485 id=485 data-nosnippet>485</a>        ]);
-<a href=#486 id=486 data-nosnippet>486</a>
-<a href=#487 id=487 data-nosnippet>487</a>        <span class="macro">assert!</span>(result.eq(<span class="kw-2">&amp;</span>expected));
-<a href=#488 id=488 data-nosnippet>488</a>    }
-<a href=#489 id=489 data-nosnippet>489</a>
-<a href=#490 id=490 data-nosnippet>490</a>    <span class="attr">#[test]
-<a href=#491 id=491 data-nosnippet>491</a>    </span><span class="kw">fn </span>processor_returns_properly_shaped_record_batch_when_no_schema_migration_required() {
-<a href=#492 id=492 data-nosnippet>492</a>        <span class="kw">let </span>snapshot_schema = Arc::new(iceberg_table_schema());
-<a href=#493 id=493 data-nosnippet>493</a>        <span class="kw">let </span>projected_iceberg_field_ids = [<span class="number">13</span>, <span class="number">14</span>];
-<a href=#494 id=494 data-nosnippet>494</a>
-<a href=#495 id=495 data-nosnippet>495</a>        <span class="kw">let </span><span class="kw-2">mut </span>inst = RecordBatchTransformer::build(snapshot_schema, <span class="kw-2">&amp;</span>projected_iceberg_field_ids);
+<a href=#365 id=365 data-nosnippet>365</a>        <span class="kw">let </span><span class="kw-2">mut </span>names_changed = <span class="bool-val">false</span>;
+<a href=#366 id=366 data-nosnippet>366</a>
+<a href=#367 id=367 data-nosnippet>367</a>        <span class="kw">for </span>(source_field, target_field) <span class="kw">in </span>source_schema
+<a href=#368 id=368 data-nosnippet>368</a>            .fields()
+<a href=#369 id=369 data-nosnippet>369</a>            .iter()
+<a href=#370 id=370 data-nosnippet>370</a>            .zip(target_schema.fields().iter())
+<a href=#371 id=371 data-nosnippet>371</a>        {
+<a href=#372 id=372 data-nosnippet>372</a>            <span class="kw">if </span>source_field.data_type() != target_field.data_type()
+<a href=#373 id=373 data-nosnippet>373</a>                || source_field.is_nullable() != target_field.is_nullable()
+<a href=#374 id=374 data-nosnippet>374</a>            {
+<a href=#375 id=375 data-nosnippet>375</a>                <span class="kw">return </span>SchemaComparison::Different;
+<a href=#376 id=376 data-nosnippet>376</a>            }
+<a href=#377 id=377 data-nosnippet>377</a>
+<a href=#378 id=378 data-nosnippet>378</a>            <span class="kw">if </span>source_field.name() != target_field.name() {
+<a href=#379 id=379 data-nosnippet>379</a>                names_changed = <span class="bool-val">true</span>;
+<a href=#380 id=380 data-nosnippet>380</a>            }
+<a href=#381 id=381 data-nosnippet>381</a>        }
+<a href=#382 id=382 data-nosnippet>382</a>
+<a href=#383 id=383 data-nosnippet>383</a>        <span class="kw">if </span>names_changed {
+<a href=#384 id=384 data-nosnippet>384</a>            SchemaComparison::NameChangesOnly
+<a href=#385 id=385 data-nosnippet>385</a>        } <span class="kw">else </span>{
+<a href=#386 id=386 data-nosnippet>386</a>            SchemaComparison::Equivalent
+<a href=#387 id=387 data-nosnippet>387</a>        }
+<a href=#388 id=388 data-nosnippet>388</a>    }
+<a href=#389 id=389 data-nosnippet>389</a>
+<a href=#390 id=390 data-nosnippet>390</a>    <span class="kw">fn </span>generate_transform_operations(
+<a href=#391 id=391 data-nosnippet>391</a>        source_schema: <span class="kw-2">&amp;</span>ArrowSchemaRef,
+<a href=#392 id=392 data-nosnippet>392</a>        snapshot_schema: <span class="kw-2">&amp;</span>IcebergSchema,
+<a href=#393 id=393 data-nosnippet>393</a>        projected_iceberg_field_ids: <span class="kw-2">&amp;</span>[i32],
+<a href=#394 id=394 data-nosnippet>394</a>        field_id_to_mapped_schema_map: HashMap&lt;i32, (FieldRef, usize)&gt;,
+<a href=#395 id=395 data-nosnippet>395</a>        constants_map: HashMap&lt;i32, PrimitiveLiteral&gt;,
+<a href=#396 id=396 data-nosnippet>396</a>        _partition_spec: <span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>PartitionSpec&gt;,
+<a href=#397 id=397 data-nosnippet>397</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;ColumnSource&gt;&gt; {
+<a href=#398 id=398 data-nosnippet>398</a>        <span class="kw">let </span>field_id_to_source_schema_map =
+<a href=#399 id=399 data-nosnippet>399</a>            <span class="self">Self</span>::build_field_id_to_arrow_schema_map(source_schema)<span class="question-mark">?</span>;
+<a href=#400 id=400 data-nosnippet>400</a>
+<a href=#401 id=401 data-nosnippet>401</a>        projected_iceberg_field_ids
+<a href=#402 id=402 data-nosnippet>402</a>            .iter()
+<a href=#403 id=403 data-nosnippet>403</a>            .map(|field_id| {
+<a href=#404 id=404 data-nosnippet>404</a>                <span class="kw">let </span>(target_field, <span class="kw">_</span>) =
+<a href=#405 id=405 data-nosnippet>405</a>                    field_id_to_mapped_schema_map
+<a href=#406 id=406 data-nosnippet>406</a>                        .get(field_id)
+<a href=#407 id=407 data-nosnippet>407</a>                        .ok_or(Error::new(
+<a href=#408 id=408 data-nosnippet>408</a>                            ErrorKind::Unexpected,
+<a href=#409 id=409 data-nosnippet>409</a>                            <span class="string">"could not find field in schema"</span>,
+<a href=#410 id=410 data-nosnippet>410</a>                        ))<span class="question-mark">?</span>;
+<a href=#411 id=411 data-nosnippet>411</a>                <span class="kw">let </span>target_type = target_field.data_type();
+<a href=#412 id=412 data-nosnippet>412</a>
+<a href=#413 id=413 data-nosnippet>413</a>                <span class="kw">let </span>iceberg_field = snapshot_schema.field_by_id(<span class="kw-2">*</span>field_id).ok_or(Error::new(
+<a href=#414 id=414 data-nosnippet>414</a>                    ErrorKind::Unexpected,
+<a href=#415 id=415 data-nosnippet>415</a>                    <span class="string">"Field not found in snapshot schema"</span>,
+<a href=#416 id=416 data-nosnippet>416</a>                ))<span class="question-mark">?</span>;
+<a href=#417 id=417 data-nosnippet>417</a>
+<a href=#418 id=418 data-nosnippet>418</a>                <span class="comment">// Iceberg spec's "Column Projection" rules (https://iceberg.apache.org/spec/#column-projection).
+<a href=#419 id=419 data-nosnippet>419</a>                // For fields "not present" in data files:
+<a href=#420 id=420 data-nosnippet>420</a>                // 1. Use partition metadata (identity transforms only)
+<a href=#421 id=421 data-nosnippet>421</a>                // 2. Use name mapping
+<a href=#422 id=422 data-nosnippet>422</a>                // 3. Use initial_default
+<a href=#423 id=423 data-nosnippet>423</a>                // 4. Return null
+<a href=#424 id=424 data-nosnippet>424</a>                //
+<a href=#425 id=425 data-nosnippet>425</a>                // Why check partition constants before Parquet field IDs (Java: BaseParquetReaders.java:299):
+<a href=#426 id=426 data-nosnippet>426</a>                // In add_files scenarios, partition columns may exist in BOTH Parquet AND partition metadata.
+<a href=#427 id=427 data-nosnippet>427</a>                // Partition metadata is authoritative - it defines which partition this file belongs to.
+<a href=#428 id=428 data-nosnippet>428</a>
+<a href=#429 id=429 data-nosnippet>429</a>                // Field ID resolution now happens in ArrowReader via:
+<a href=#430 id=430 data-nosnippet>430</a>                // 1. Embedded field IDs (ParquetSchemaUtil.hasIds() = true) - trust them
+<a href=#431 id=431 data-nosnippet>431</a>                // 2. Name mapping (ParquetSchemaUtil.applyNameMapping()) - applied upfront
+<a href=#432 id=432 data-nosnippet>432</a>                // 3. Position-based fallback (ParquetSchemaUtil.addFallbackIds()) - applied upfront
+<a href=#433 id=433 data-nosnippet>433</a>                //
+<a href=#434 id=434 data-nosnippet>434</a>                // At this point, all field IDs in the source schema are trustworthy.
+<a href=#435 id=435 data-nosnippet>435</a>                // No conflict detection needed - schema resolution happened in reader.rs.
+<a href=#436 id=436 data-nosnippet>436</a>                </span><span class="kw">let </span>field_by_id = field_id_to_source_schema_map.get(field_id).map(
+<a href=#437 id=437 data-nosnippet>437</a>                    |(source_field, source_index)| {
+<a href=#438 id=438 data-nosnippet>438</a>                        <span class="kw">if </span>source_field.data_type().equals_datatype(target_type) {
+<a href=#439 id=439 data-nosnippet>439</a>                            ColumnSource::PassThrough {
+<a href=#440 id=440 data-nosnippet>440</a>                                source_index: <span class="kw-2">*</span>source_index,
+<a href=#441 id=441 data-nosnippet>441</a>                            }
+<a href=#442 id=442 data-nosnippet>442</a>                        } <span class="kw">else </span>{
+<a href=#443 id=443 data-nosnippet>443</a>                            ColumnSource::Promote {
+<a href=#444 id=444 data-nosnippet>444</a>                                target_type: target_type.clone(),
+<a href=#445 id=445 data-nosnippet>445</a>                                source_index: <span class="kw-2">*</span>source_index,
+<a href=#446 id=446 data-nosnippet>446</a>                            }
+<a href=#447 id=447 data-nosnippet>447</a>                        }
+<a href=#448 id=448 data-nosnippet>448</a>                    },
+<a href=#449 id=449 data-nosnippet>449</a>                );
+<a href=#450 id=450 data-nosnippet>450</a>
+<a href=#451 id=451 data-nosnippet>451</a>                <span class="comment">// Apply spec's fallback steps for "not present" fields.
+<a href=#452 id=452 data-nosnippet>452</a>                </span><span class="kw">let </span>column_source = <span class="kw">if let </span><span class="prelude-val">Some</span>(constant_value) = constants_map.get(field_id) {
+<a href=#453 id=453 data-nosnippet>453</a>                    <span class="comment">// Rule #1: Identity partition constant
+<a href=#454 id=454 data-nosnippet>454</a>                    </span>ColumnSource::Add {
+<a href=#455 id=455 data-nosnippet>455</a>                        value: <span class="prelude-val">Some</span>(constant_value.clone()),
+<a href=#456 id=456 data-nosnippet>456</a>                        target_type: target_type.clone(),
+<a href=#457 id=457 data-nosnippet>457</a>                    }
+<a href=#458 id=458 data-nosnippet>458</a>                } <span class="kw">else if let </span><span class="prelude-val">Some</span>(source) = field_by_id {
+<a href=#459 id=459 data-nosnippet>459</a>                    source
+<a href=#460 id=460 data-nosnippet>460</a>                } <span class="kw">else </span>{
+<a href=#461 id=461 data-nosnippet>461</a>                    <span class="comment">// Rules #2, #3 and #4:
+<a href=#462 id=462 data-nosnippet>462</a>                    // Rule #2 (name mapping) was already applied in reader.rs if needed.
+<a href=#463 id=463 data-nosnippet>463</a>                    // If field_id is still not found, the column doesn't exist in the Parquet file.
+<a href=#464 id=464 data-nosnippet>464</a>                    // Fall through to rule #3 (initial_default) or rule #4 (null).
+<a href=#465 id=465 data-nosnippet>465</a>                    </span><span class="kw">let </span>default_value = iceberg_field.initial_default.as_ref().and_then(|lit| {
+<a href=#466 id=466 data-nosnippet>466</a>                        <span class="kw">if let </span>Literal::Primitive(prim) = lit {
+<a href=#467 id=467 data-nosnippet>467</a>                            <span class="prelude-val">Some</span>(prim.clone())
+<a href=#468 id=468 data-nosnippet>468</a>                        } <span class="kw">else </span>{
+<a href=#469 id=469 data-nosnippet>469</a>                            <span class="prelude-val">None
+<a href=#470 id=470 data-nosnippet>470</a>                        </span>}
+<a href=#471 id=471 data-nosnippet>471</a>                    });
+<a href=#472 id=472 data-nosnippet>472</a>                    ColumnSource::Add {
+<a href=#473 id=473 data-nosnippet>473</a>                        value: default_value,
+<a href=#474 id=474 data-nosnippet>474</a>                        target_type: target_type.clone(),
+<a href=#475 id=475 data-nosnippet>475</a>                    }
+<a href=#476 id=476 data-nosnippet>476</a>                };
+<a href=#477 id=477 data-nosnippet>477</a>
+<a href=#478 id=478 data-nosnippet>478</a>                <span class="prelude-val">Ok</span>(column_source)
+<a href=#479 id=479 data-nosnippet>479</a>            })
+<a href=#480 id=480 data-nosnippet>480</a>            .collect()
+<a href=#481 id=481 data-nosnippet>481</a>    }
+<a href=#482 id=482 data-nosnippet>482</a>
+<a href=#483 id=483 data-nosnippet>483</a>    <span class="kw">fn </span>build_field_id_to_arrow_schema_map(
+<a href=#484 id=484 data-nosnippet>484</a>        source_schema: <span class="kw-2">&amp;</span>SchemaRef,
+<a href=#485 id=485 data-nosnippet>485</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;HashMap&lt;i32, (FieldRef, usize)&gt;&gt; {
+<a href=#486 id=486 data-nosnippet>486</a>        <span class="kw">let </span><span class="kw-2">mut </span>field_id_to_source_schema = HashMap::new();
+<a href=#487 id=487 data-nosnippet>487</a>        <span class="kw">for </span>(source_field_idx, source_field) <span class="kw">in </span>source_schema.fields.iter().enumerate() {
+<a href=#488 id=488 data-nosnippet>488</a>            <span class="comment">// Check if field has a field ID in metadata
+<a href=#489 id=489 data-nosnippet>489</a>            </span><span class="kw">if let </span><span class="prelude-val">Some</span>(field_id_str) = source_field.metadata().get(PARQUET_FIELD_ID_META_KEY) {
+<a href=#490 id=490 data-nosnippet>490</a>                <span class="kw">let </span>this_field_id = field_id_str.parse().map_err(|e| {
+<a href=#491 id=491 data-nosnippet>491</a>                    Error::new(
+<a href=#492 id=492 data-nosnippet>492</a>                        ErrorKind::DataInvalid,
+<a href=#493 id=493 data-nosnippet>493</a>                        <span class="macro">format!</span>(<span class="string">"field id not parseable as an i32: {}"</span>, e),
+<a href=#494 id=494 data-nosnippet>494</a>                    )
+<a href=#495 id=495 data-nosnippet>495</a>                })<span class="question-mark">?</span>;
 <a href=#496 id=496 data-nosnippet>496</a>
-<a href=#497 id=497 data-nosnippet>497</a>        <span class="kw">let </span>result = inst
-<a href=#498 id=498 data-nosnippet>498</a>            .process_record_batch(source_record_batch_no_migration_required())
-<a href=#499 id=499 data-nosnippet>499</a>            .unwrap();
-<a href=#500 id=500 data-nosnippet>500</a>
-<a href=#501 id=501 data-nosnippet>501</a>        <span class="kw">let </span>expected = source_record_batch_no_migration_required();
+<a href=#497 id=497 data-nosnippet>497</a>                field_id_to_source_schema
+<a href=#498 id=498 data-nosnippet>498</a>                    .insert(this_field_id, (source_field.clone(), source_field_idx));
+<a href=#499 id=499 data-nosnippet>499</a>            }
+<a href=#500 id=500 data-nosnippet>500</a>            <span class="comment">// If field doesn't have a field ID, skip it - name mapping will handle it
+<a href=#501 id=501 data-nosnippet>501</a>        </span>}
 <a href=#502 id=502 data-nosnippet>502</a>
-<a href=#503 id=503 data-nosnippet>503</a>        <span class="macro">assert_eq!</span>(result, expected);
+<a href=#503 id=503 data-nosnippet>503</a>        <span class="prelude-val">Ok</span>(field_id_to_source_schema)
 <a href=#504 id=504 data-nosnippet>504</a>    }
 <a href=#505 id=505 data-nosnippet>505</a>
-<a href=#506 id=506 data-nosnippet>506</a>    <span class="attr">#[test]
-<a href=#507 id=507 data-nosnippet>507</a>    </span><span class="kw">fn </span>processor_returns_properly_shaped_record_batch_when_schema_migration_required() {
-<a href=#508 id=508 data-nosnippet>508</a>        <span class="kw">let </span>snapshot_schema = Arc::new(iceberg_table_schema());
-<a href=#509 id=509 data-nosnippet>509</a>        <span class="kw">let </span>projected_iceberg_field_ids = [<span class="number">10</span>, <span class="number">11</span>, <span class="number">12</span>, <span class="number">14</span>, <span class="number">15</span>]; <span class="comment">// a, b, c, e, f
-<a href=#510 id=510 data-nosnippet>510</a>
-<a href=#511 id=511 data-nosnippet>511</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>inst = RecordBatchTransformer::build(snapshot_schema, <span class="kw-2">&amp;</span>projected_iceberg_field_ids);
-<a href=#512 id=512 data-nosnippet>512</a>
-<a href=#513 id=513 data-nosnippet>513</a>        <span class="kw">let </span>result = inst.process_record_batch(source_record_batch()).unwrap();
-<a href=#514 id=514 data-nosnippet>514</a>
-<a href=#515 id=515 data-nosnippet>515</a>        <span class="kw">let </span>expected = expected_record_batch_migration_required();
-<a href=#516 id=516 data-nosnippet>516</a>
-<a href=#517 id=517 data-nosnippet>517</a>        <span class="macro">assert_eq!</span>(result, expected);
-<a href=#518 id=518 data-nosnippet>518</a>    }
-<a href=#519 id=519 data-nosnippet>519</a>
-<a href=#520 id=520 data-nosnippet>520</a>    <span class="attr">#[test]
-<a href=#521 id=521 data-nosnippet>521</a>    </span><span class="kw">fn </span>schema_evolution_adds_date_column_with_nulls() {
-<a href=#522 id=522 data-nosnippet>522</a>        <span class="comment">// Reproduces TestSelect.readAndWriteWithBranchAfterSchemaChange from iceberg-spark.
-<a href=#523 id=523 data-nosnippet>523</a>        // When reading old snapshots after adding a DATE column, the transformer must
-<a href=#524 id=524 data-nosnippet>524</a>        // populate the new column with NULL values since old files lack this field.
-<a href=#525 id=525 data-nosnippet>525</a>        </span><span class="kw">let </span>snapshot_schema = Arc::new(
-<a href=#526 id=526 data-nosnippet>526</a>            Schema::builder()
-<a href=#527 id=527 data-nosnippet>527</a>                .with_schema_id(<span class="number">1</span>)
-<a href=#528 id=528 data-nosnippet>528</a>                .with_fields(<span class="macro">vec!</span>[
-<a href=#529 id=529 data-nosnippet>529</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#530 id=530 data-nosnippet>530</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#531 id=531 data-nosnippet>531</a>                    NestedField::optional(<span class="number">3</span>, <span class="string">"date_col"</span>, Type::Primitive(PrimitiveType::Date))
-<a href=#532 id=532 data-nosnippet>532</a>                        .into(),
-<a href=#533 id=533 data-nosnippet>533</a>                ])
-<a href=#534 id=534 data-nosnippet>534</a>                .build()
-<a href=#535 id=535 data-nosnippet>535</a>                .unwrap(),
-<a href=#536 id=536 data-nosnippet>536</a>        );
-<a href=#537 id=537 data-nosnippet>537</a>        <span class="kw">let </span>projected_iceberg_field_ids = [<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>];
-<a href=#538 id=538 data-nosnippet>538</a>
-<a href=#539 id=539 data-nosnippet>539</a>        <span class="kw">let </span><span class="kw-2">mut </span>transformer =
-<a href=#540 id=540 data-nosnippet>540</a>            RecordBatchTransformer::build(snapshot_schema, <span class="kw-2">&amp;</span>projected_iceberg_field_ids);
-<a href=#541 id=541 data-nosnippet>541</a>
-<a href=#542 id=542 data-nosnippet>542</a>        <span class="kw">let </span>file_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#543 id=543 data-nosnippet>543</a>            simple_field(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"1"</span>),
-<a href=#544 id=544 data-nosnippet>544</a>            simple_field(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"2"</span>),
-<a href=#545 id=545 data-nosnippet>545</a>        ]));
-<a href=#546 id=546 data-nosnippet>546</a>
-<a href=#547 id=547 data-nosnippet>547</a>        <span class="kw">let </span>file_batch = RecordBatch::try_new(file_schema, <span class="macro">vec!</span>[
-<a href=#548 id=548 data-nosnippet>548</a>            Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>])),
-<a href=#549 id=549 data-nosnippet>549</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[
-<a href=#550 id=550 data-nosnippet>550</a>                <span class="prelude-val">Some</span>(<span class="string">"Alice"</span>),
-<a href=#551 id=551 data-nosnippet>551</a>                <span class="prelude-val">Some</span>(<span class="string">"Bob"</span>),
-<a href=#552 id=552 data-nosnippet>552</a>                <span class="prelude-val">Some</span>(<span class="string">"Charlie"</span>),
-<a href=#553 id=553 data-nosnippet>553</a>            ])),
-<a href=#554 id=554 data-nosnippet>554</a>        ])
-<a href=#555 id=555 data-nosnippet>555</a>        .unwrap();
-<a href=#556 id=556 data-nosnippet>556</a>
-<a href=#557 id=557 data-nosnippet>557</a>        <span class="kw">let </span>result = transformer.process_record_batch(file_batch).unwrap();
-<a href=#558 id=558 data-nosnippet>558</a>
-<a href=#559 id=559 data-nosnippet>559</a>        <span class="macro">assert_eq!</span>(result.num_columns(), <span class="number">3</span>);
-<a href=#560 id=560 data-nosnippet>560</a>        <span class="macro">assert_eq!</span>(result.num_rows(), <span class="number">3</span>);
-<a href=#561 id=561 data-nosnippet>561</a>
-<a href=#562 id=562 data-nosnippet>562</a>        <span class="kw">let </span>id_column = result
-<a href=#563 id=563 data-nosnippet>563</a>            .column(<span class="number">0</span>)
-<a href=#564 id=564 data-nosnippet>564</a>            .as_any()
-<a href=#565 id=565 data-nosnippet>565</a>            .downcast_ref::&lt;Int32Array&gt;()
-<a href=#566 id=566 data-nosnippet>566</a>            .unwrap();
-<a href=#567 id=567 data-nosnippet>567</a>        <span class="macro">assert_eq!</span>(id_column.values(), <span class="kw-2">&amp;</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>]);
-<a href=#568 id=568 data-nosnippet>568</a>
-<a href=#569 id=569 data-nosnippet>569</a>        <span class="kw">let </span>name_column = result
-<a href=#570 id=570 data-nosnippet>570</a>            .column(<span class="number">1</span>)
-<a href=#571 id=571 data-nosnippet>571</a>            .as_any()
-<a href=#572 id=572 data-nosnippet>572</a>            .downcast_ref::&lt;StringArray&gt;()
-<a href=#573 id=573 data-nosnippet>573</a>            .unwrap();
-<a href=#574 id=574 data-nosnippet>574</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
-<a href=#575 id=575 data-nosnippet>575</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
-<a href=#576 id=576 data-nosnippet>576</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">2</span>), <span class="string">"Charlie"</span>);
-<a href=#577 id=577 data-nosnippet>577</a>
-<a href=#578 id=578 data-nosnippet>578</a>        <span class="kw">let </span>date_column = result
-<a href=#579 id=579 data-nosnippet>579</a>            .column(<span class="number">2</span>)
-<a href=#580 id=580 data-nosnippet>580</a>            .as_any()
-<a href=#581 id=581 data-nosnippet>581</a>            .downcast_ref::&lt;Date32Array&gt;()
-<a href=#582 id=582 data-nosnippet>582</a>            .unwrap();
-<a href=#583 id=583 data-nosnippet>583</a>        <span class="macro">assert!</span>(date_column.is_null(<span class="number">0</span>));
-<a href=#584 id=584 data-nosnippet>584</a>        <span class="macro">assert!</span>(date_column.is_null(<span class="number">1</span>));
-<a href=#585 id=585 data-nosnippet>585</a>        <span class="macro">assert!</span>(date_column.is_null(<span class="number">2</span>));
-<a href=#586 id=586 data-nosnippet>586</a>    }
-<a href=#587 id=587 data-nosnippet>587</a>
-<a href=#588 id=588 data-nosnippet>588</a>    <span class="kw">pub fn </span>source_record_batch() -&gt; RecordBatch {
-<a href=#589 id=589 data-nosnippet>589</a>        RecordBatch::try_new(
-<a href=#590 id=590 data-nosnippet>590</a>            arrow_schema_promotion_addition_and_renaming_required(),
-<a href=#591 id=591 data-nosnippet>591</a>            <span class="macro">vec!</span>[
-<a href=#592 id=592 data-nosnippet>592</a>                Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">1001</span>), <span class="prelude-val">Some</span>(<span class="number">1002</span>), <span class="prelude-val">Some</span>(<span class="number">1003</span>)])), <span class="comment">// b
-<a href=#593 id=593 data-nosnippet>593</a>                </span>Arc::new(Float32Array::from(<span class="macro">vec!</span>[
-<a href=#594 id=594 data-nosnippet>594</a>                    <span class="prelude-val">Some</span>(<span class="number">12.125</span>),
-<a href=#595 id=595 data-nosnippet>595</a>                    <span class="prelude-val">Some</span>(<span class="number">23.375</span>),
-<a href=#596 id=596 data-nosnippet>596</a>                    <span class="prelude-val">Some</span>(<span class="number">34.875</span>),
-<a href=#597 id=597 data-nosnippet>597</a>                ])), <span class="comment">// c
-<a href=#598 id=598 data-nosnippet>598</a>                </span>Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">2001</span>), <span class="prelude-val">Some</span>(<span class="number">2002</span>), <span class="prelude-val">Some</span>(<span class="number">2003</span>)])), <span class="comment">// d
-<a href=#599 id=599 data-nosnippet>599</a>                </span>Arc::new(StringArray::from(<span class="macro">vec!</span>[
-<a href=#600 id=600 data-nosnippet>600</a>                    <span class="prelude-val">Some</span>(<span class="string">"Apache"</span>),
-<a href=#601 id=601 data-nosnippet>601</a>                    <span class="prelude-val">Some</span>(<span class="string">"Iceberg"</span>),
-<a href=#602 id=602 data-nosnippet>602</a>                    <span class="prelude-val">Some</span>(<span class="string">"Rocks"</span>),
-<a href=#603 id=603 data-nosnippet>603</a>                ])), <span class="comment">// e
-<a href=#604 id=604 data-nosnippet>604</a>            </span>],
-<a href=#605 id=605 data-nosnippet>605</a>        )
-<a href=#606 id=606 data-nosnippet>606</a>        .unwrap()
-<a href=#607 id=607 data-nosnippet>607</a>    }
-<a href=#608 id=608 data-nosnippet>608</a>
-<a href=#609 id=609 data-nosnippet>609</a>    <span class="kw">pub fn </span>source_record_batch_no_migration_required() -&gt; RecordBatch {
-<a href=#610 id=610 data-nosnippet>610</a>        RecordBatch::try_new(
-<a href=#611 id=611 data-nosnippet>611</a>            arrow_schema_no_promotion_addition_or_renaming_required(),
-<a href=#612 id=612 data-nosnippet>612</a>            <span class="macro">vec!</span>[
-<a href=#613 id=613 data-nosnippet>613</a>                Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">2001</span>), <span class="prelude-val">Some</span>(<span class="number">2002</span>), <span class="prelude-val">Some</span>(<span class="number">2003</span>)])), <span class="comment">// d
-<a href=#614 id=614 data-nosnippet>614</a>                </span>Arc::new(StringArray::from(<span class="macro">vec!</span>[
-<a href=#615 id=615 data-nosnippet>615</a>                    <span class="prelude-val">Some</span>(<span class="string">"Apache"</span>),
-<a href=#616 id=616 data-nosnippet>616</a>                    <span class="prelude-val">Some</span>(<span class="string">"Iceberg"</span>),
-<a href=#617 id=617 data-nosnippet>617</a>                    <span class="prelude-val">Some</span>(<span class="string">"Rocks"</span>),
-<a href=#618 id=618 data-nosnippet>618</a>                ])), <span class="comment">// e
-<a href=#619 id=619 data-nosnippet>619</a>            </span>],
-<a href=#620 id=620 data-nosnippet>620</a>        )
-<a href=#621 id=621 data-nosnippet>621</a>        .unwrap()
-<a href=#622 id=622 data-nosnippet>622</a>    }
-<a href=#623 id=623 data-nosnippet>623</a>
-<a href=#624 id=624 data-nosnippet>624</a>    <span class="kw">pub fn </span>expected_record_batch_migration_required() -&gt; RecordBatch {
-<a href=#625 id=625 data-nosnippet>625</a>        RecordBatch::try_new(arrow_schema_already_same_as_target(), <span class="macro">vec!</span>[
-<a href=#626 id=626 data-nosnippet>626</a>            Arc::new(StringArray::from(Vec::&lt;<span class="prelude-ty">Option</span>&lt;String&gt;&gt;::from([
-<a href=#627 id=627 data-nosnippet>627</a>                <span class="prelude-val">None</span>, <span class="prelude-val">None</span>, <span class="prelude-val">None</span>,
-<a href=#628 id=628 data-nosnippet>628</a>            ]))), <span class="comment">// a
-<a href=#629 id=629 data-nosnippet>629</a>            </span>Arc::new(Int64Array::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">1001</span>), <span class="prelude-val">Some</span>(<span class="number">1002</span>), <span class="prelude-val">Some</span>(<span class="number">1003</span>)])), <span class="comment">// b
-<a href=#630 id=630 data-nosnippet>630</a>            </span>Arc::new(Float64Array::from(<span class="macro">vec!</span>[
-<a href=#631 id=631 data-nosnippet>631</a>                <span class="prelude-val">Some</span>(<span class="number">12.125</span>),
-<a href=#632 id=632 data-nosnippet>632</a>                <span class="prelude-val">Some</span>(<span class="number">23.375</span>),
-<a href=#633 id=633 data-nosnippet>633</a>                <span class="prelude-val">Some</span>(<span class="number">34.875</span>),
-<a href=#634 id=634 data-nosnippet>634</a>            ])), <span class="comment">// c
-<a href=#635 id=635 data-nosnippet>635</a>            </span>Arc::new(StringArray::from(<span class="macro">vec!</span>[
-<a href=#636 id=636 data-nosnippet>636</a>                <span class="prelude-val">Some</span>(<span class="string">"Apache"</span>),
-<a href=#637 id=637 data-nosnippet>637</a>                <span class="prelude-val">Some</span>(<span class="string">"Iceberg"</span>),
-<a href=#638 id=638 data-nosnippet>638</a>                <span class="prelude-val">Some</span>(<span class="string">"Rocks"</span>),
-<a href=#639 id=639 data-nosnippet>639</a>            ])), <span class="comment">// e (d skipped by projection)
-<a href=#640 id=640 data-nosnippet>640</a>            </span>Arc::new(StringArray::from(<span class="macro">vec!</span>[
-<a href=#641 id=641 data-nosnippet>641</a>                <span class="prelude-val">Some</span>(<span class="string">"(╯°□°)╯"</span>),
-<a href=#642 id=642 data-nosnippet>642</a>                <span class="prelude-val">Some</span>(<span class="string">"(╯°□°)╯"</span>),
-<a href=#643 id=643 data-nosnippet>643</a>                <span class="prelude-val">Some</span>(<span class="string">"(╯°□°)╯"</span>),
-<a href=#644 id=644 data-nosnippet>644</a>            ])), <span class="comment">// f
-<a href=#645 id=645 data-nosnippet>645</a>        </span>])
-<a href=#646 id=646 data-nosnippet>646</a>        .unwrap()
-<a href=#647 id=647 data-nosnippet>647</a>    }
-<a href=#648 id=648 data-nosnippet>648</a>
-<a href=#649 id=649 data-nosnippet>649</a>    <span class="kw">pub fn </span>iceberg_table_schema() -&gt; Schema {
-<a href=#650 id=650 data-nosnippet>650</a>        Schema::builder()
-<a href=#651 id=651 data-nosnippet>651</a>            .with_schema_id(<span class="number">2</span>)
-<a href=#652 id=652 data-nosnippet>652</a>            .with_fields(<span class="macro">vec!</span>[
-<a href=#653 id=653 data-nosnippet>653</a>                NestedField::optional(<span class="number">10</span>, <span class="string">"a"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#654 id=654 data-nosnippet>654</a>                NestedField::required(<span class="number">11</span>, <span class="string">"b"</span>, Type::Primitive(PrimitiveType::Long)).into(),
-<a href=#655 id=655 data-nosnippet>655</a>                NestedField::required(<span class="number">12</span>, <span class="string">"c"</span>, Type::Primitive(PrimitiveType::Double)).into(),
-<a href=#656 id=656 data-nosnippet>656</a>                NestedField::required(<span class="number">13</span>, <span class="string">"d"</span>, Type::Primitive(PrimitiveType::Int)).into(),
-<a href=#657 id=657 data-nosnippet>657</a>                NestedField::optional(<span class="number">14</span>, <span class="string">"e"</span>, Type::Primitive(PrimitiveType::String)).into(),
-<a href=#658 id=658 data-nosnippet>658</a>                NestedField::required(<span class="number">15</span>, <span class="string">"f"</span>, Type::Primitive(PrimitiveType::String))
-<a href=#659 id=659 data-nosnippet>659</a>                    .with_initial_default(Literal::string(<span class="string">"(╯°□°)╯"</span>))
-<a href=#660 id=660 data-nosnippet>660</a>                    .into(),
-<a href=#661 id=661 data-nosnippet>661</a>            ])
-<a href=#662 id=662 data-nosnippet>662</a>            .build()
-<a href=#663 id=663 data-nosnippet>663</a>            .unwrap()
-<a href=#664 id=664 data-nosnippet>664</a>    }
+<a href=#506 id=506 data-nosnippet>506</a>    <span class="kw">fn </span>transform_columns(
+<a href=#507 id=507 data-nosnippet>507</a>        <span class="kw-2">&amp;</span><span class="self">self</span>,
+<a href=#508 id=508 data-nosnippet>508</a>        columns: <span class="kw-2">&amp;</span>[Arc&lt;<span class="kw">dyn </span>ArrowArray&gt;],
+<a href=#509 id=509 data-nosnippet>509</a>        operations: <span class="kw-2">&amp;</span>[ColumnSource],
+<a href=#510 id=510 data-nosnippet>510</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Vec&lt;Arc&lt;<span class="kw">dyn </span>ArrowArray&gt;&gt;&gt; {
+<a href=#511 id=511 data-nosnippet>511</a>        <span class="kw">if </span>columns.is_empty() {
+<a href=#512 id=512 data-nosnippet>512</a>            <span class="kw">return </span><span class="prelude-val">Ok</span>(columns.to_vec());
+<a href=#513 id=513 data-nosnippet>513</a>        }
+<a href=#514 id=514 data-nosnippet>514</a>        <span class="kw">let </span>num_rows = columns[<span class="number">0</span>].len();
+<a href=#515 id=515 data-nosnippet>515</a>
+<a href=#516 id=516 data-nosnippet>516</a>        operations
+<a href=#517 id=517 data-nosnippet>517</a>            .iter()
+<a href=#518 id=518 data-nosnippet>518</a>            .map(|op| {
+<a href=#519 id=519 data-nosnippet>519</a>                <span class="prelude-val">Ok</span>(<span class="kw">match </span>op {
+<a href=#520 id=520 data-nosnippet>520</a>                    ColumnSource::PassThrough { source_index } =&gt; columns[<span class="kw-2">*</span>source_index].clone(),
+<a href=#521 id=521 data-nosnippet>521</a>
+<a href=#522 id=522 data-nosnippet>522</a>                    ColumnSource::Promote {
+<a href=#523 id=523 data-nosnippet>523</a>                        target_type,
+<a href=#524 id=524 data-nosnippet>524</a>                        source_index,
+<a href=#525 id=525 data-nosnippet>525</a>                    } =&gt; cast(<span class="kw-2">&amp;*</span>columns[<span class="kw-2">*</span>source_index], target_type)<span class="question-mark">?</span>,
+<a href=#526 id=526 data-nosnippet>526</a>
+<a href=#527 id=527 data-nosnippet>527</a>                    ColumnSource::Add { target_type, value } =&gt; {
+<a href=#528 id=528 data-nosnippet>528</a>                        <span class="self">Self</span>::create_column(target_type, value, num_rows)<span class="question-mark">?
+<a href=#529 id=529 data-nosnippet>529</a>                    </span>}
+<a href=#530 id=530 data-nosnippet>530</a>                })
+<a href=#531 id=531 data-nosnippet>531</a>            })
+<a href=#532 id=532 data-nosnippet>532</a>            .collect()
+<a href=#533 id=533 data-nosnippet>533</a>    }
+<a href=#534 id=534 data-nosnippet>534</a>
+<a href=#535 id=535 data-nosnippet>535</a>    <span class="kw">fn </span>create_column(
+<a href=#536 id=536 data-nosnippet>536</a>        target_type: <span class="kw-2">&amp;</span>DataType,
+<a href=#537 id=537 data-nosnippet>537</a>        prim_lit: <span class="kw-2">&amp;</span><span class="prelude-ty">Option</span>&lt;PrimitiveLiteral&gt;,
+<a href=#538 id=538 data-nosnippet>538</a>        num_rows: usize,
+<a href=#539 id=539 data-nosnippet>539</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;ArrayRef&gt; {
+<a href=#540 id=540 data-nosnippet>540</a>        <span class="prelude-val">Ok</span>(<span class="kw">match </span>(target_type, prim_lit) {
+<a href=#541 id=541 data-nosnippet>541</a>            (DataType::Boolean, <span class="prelude-val">Some</span>(PrimitiveLiteral::Boolean(value))) =&gt; {
+<a href=#542 id=542 data-nosnippet>542</a>                Arc::new(BooleanArray::from(<span class="macro">vec!</span>[<span class="kw-2">*</span>value; num_rows]))
+<a href=#543 id=543 data-nosnippet>543</a>            }
+<a href=#544 id=544 data-nosnippet>544</a>            (DataType::Boolean, <span class="prelude-val">None</span>) =&gt; {
+<a href=#545 id=545 data-nosnippet>545</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;bool&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
+<a href=#546 id=546 data-nosnippet>546</a>                Arc::new(BooleanArray::from(vals))
+<a href=#547 id=547 data-nosnippet>547</a>            }
+<a href=#548 id=548 data-nosnippet>548</a>            (DataType::Int32, <span class="prelude-val">Some</span>(PrimitiveLiteral::Int(value))) =&gt; {
+<a href=#549 id=549 data-nosnippet>549</a>                Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="kw-2">*</span>value; num_rows]))
+<a href=#550 id=550 data-nosnippet>550</a>            }
+<a href=#551 id=551 data-nosnippet>551</a>            (DataType::Int32, <span class="prelude-val">None</span>) =&gt; {
+<a href=#552 id=552 data-nosnippet>552</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;i32&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
+<a href=#553 id=553 data-nosnippet>553</a>                Arc::new(Int32Array::from(vals))
+<a href=#554 id=554 data-nosnippet>554</a>            }
+<a href=#555 id=555 data-nosnippet>555</a>            (DataType::Date32, <span class="prelude-val">Some</span>(PrimitiveLiteral::Int(value))) =&gt; {
+<a href=#556 id=556 data-nosnippet>556</a>                Arc::new(Date32Array::from(<span class="macro">vec!</span>[<span class="kw-2">*</span>value; num_rows]))
+<a href=#557 id=557 data-nosnippet>557</a>            }
+<a href=#558 id=558 data-nosnippet>558</a>            (DataType::Date32, <span class="prelude-val">None</span>) =&gt; {
+<a href=#559 id=559 data-nosnippet>559</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;i32&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
+<a href=#560 id=560 data-nosnippet>560</a>                Arc::new(Date32Array::from(vals))
+<a href=#561 id=561 data-nosnippet>561</a>            }
+<a href=#562 id=562 data-nosnippet>562</a>            (DataType::Int64, <span class="prelude-val">Some</span>(PrimitiveLiteral::Long(value))) =&gt; {
+<a href=#563 id=563 data-nosnippet>563</a>                Arc::new(Int64Array::from(<span class="macro">vec!</span>[<span class="kw-2">*</span>value; num_rows]))
+<a href=#564 id=564 data-nosnippet>564</a>            }
+<a href=#565 id=565 data-nosnippet>565</a>            (DataType::Int64, <span class="prelude-val">None</span>) =&gt; {
+<a href=#566 id=566 data-nosnippet>566</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;i64&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
+<a href=#567 id=567 data-nosnippet>567</a>                Arc::new(Int64Array::from(vals))
+<a href=#568 id=568 data-nosnippet>568</a>            }
+<a href=#569 id=569 data-nosnippet>569</a>            (DataType::Float32, <span class="prelude-val">Some</span>(PrimitiveLiteral::Float(value))) =&gt; {
+<a href=#570 id=570 data-nosnippet>570</a>                Arc::new(Float32Array::from(<span class="macro">vec!</span>[value.<span class="number">0</span>; num_rows]))
+<a href=#571 id=571 data-nosnippet>571</a>            }
+<a href=#572 id=572 data-nosnippet>572</a>            (DataType::Float32, <span class="prelude-val">None</span>) =&gt; {
+<a href=#573 id=573 data-nosnippet>573</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;f32&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
+<a href=#574 id=574 data-nosnippet>574</a>                Arc::new(Float32Array::from(vals))
+<a href=#575 id=575 data-nosnippet>575</a>            }
+<a href=#576 id=576 data-nosnippet>576</a>            (DataType::Float64, <span class="prelude-val">Some</span>(PrimitiveLiteral::Double(value))) =&gt; {
+<a href=#577 id=577 data-nosnippet>577</a>                Arc::new(Float64Array::from(<span class="macro">vec!</span>[value.<span class="number">0</span>; num_rows]))
+<a href=#578 id=578 data-nosnippet>578</a>            }
+<a href=#579 id=579 data-nosnippet>579</a>            (DataType::Float64, <span class="prelude-val">None</span>) =&gt; {
+<a href=#580 id=580 data-nosnippet>580</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;f64&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
+<a href=#581 id=581 data-nosnippet>581</a>                Arc::new(Float64Array::from(vals))
+<a href=#582 id=582 data-nosnippet>582</a>            }
+<a href=#583 id=583 data-nosnippet>583</a>            (DataType::Utf8, <span class="prelude-val">Some</span>(PrimitiveLiteral::String(value))) =&gt; {
+<a href=#584 id=584 data-nosnippet>584</a>                Arc::new(StringArray::from(<span class="macro">vec!</span>[value.clone(); num_rows]))
+<a href=#585 id=585 data-nosnippet>585</a>            }
+<a href=#586 id=586 data-nosnippet>586</a>            (DataType::Utf8, <span class="prelude-val">None</span>) =&gt; {
+<a href=#587 id=587 data-nosnippet>587</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;String&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
+<a href=#588 id=588 data-nosnippet>588</a>                Arc::new(StringArray::from(vals))
+<a href=#589 id=589 data-nosnippet>589</a>            }
+<a href=#590 id=590 data-nosnippet>590</a>            (DataType::Binary, <span class="prelude-val">Some</span>(PrimitiveLiteral::Binary(value))) =&gt; {
+<a href=#591 id=591 data-nosnippet>591</a>                Arc::new(BinaryArray::from_vec(<span class="macro">vec!</span>[value; num_rows]))
+<a href=#592 id=592 data-nosnippet>592</a>            }
+<a href=#593 id=593 data-nosnippet>593</a>            (DataType::Binary, <span class="prelude-val">None</span>) =&gt; {
+<a href=#594 id=594 data-nosnippet>594</a>                <span class="kw">let </span>vals: Vec&lt;<span class="prelude-ty">Option</span>&lt;<span class="kw-2">&amp;</span>[u8]&gt;&gt; = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; num_rows];
+<a href=#595 id=595 data-nosnippet>595</a>                Arc::new(BinaryArray::from_opt_vec(vals))
+<a href=#596 id=596 data-nosnippet>596</a>            }
+<a href=#597 id=597 data-nosnippet>597</a>            (DataType::Null, <span class="kw">_</span>) =&gt; Arc::new(NullArray::new(num_rows)),
+<a href=#598 id=598 data-nosnippet>598</a>            (dt, <span class="kw">_</span>) =&gt; {
+<a href=#599 id=599 data-nosnippet>599</a>                <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#600 id=600 data-nosnippet>600</a>                    ErrorKind::Unexpected,
+<a href=#601 id=601 data-nosnippet>601</a>                    <span class="macro">format!</span>(<span class="string">"unexpected target column type {}"</span>, dt),
+<a href=#602 id=602 data-nosnippet>602</a>                ));
+<a href=#603 id=603 data-nosnippet>603</a>            }
+<a href=#604 id=604 data-nosnippet>604</a>        })
+<a href=#605 id=605 data-nosnippet>605</a>    }
+<a href=#606 id=606 data-nosnippet>606</a>}
+<a href=#607 id=607 data-nosnippet>607</a>
+<a href=#608 id=608 data-nosnippet>608</a><span class="attr">#[cfg(test)]
+<a href=#609 id=609 data-nosnippet>609</a></span><span class="kw">mod </span>test {
+<a href=#610 id=610 data-nosnippet>610</a>    <span class="kw">use </span>std::collections::HashMap;
+<a href=#611 id=611 data-nosnippet>611</a>    <span class="kw">use </span>std::sync::Arc;
+<a href=#612 id=612 data-nosnippet>612</a>
+<a href=#613 id=613 data-nosnippet>613</a>    <span class="kw">use </span>arrow_array::{
+<a href=#614 id=614 data-nosnippet>614</a>        Array, Date32Array, Float32Array, Float64Array, Int32Array, Int64Array, RecordBatch,
+<a href=#615 id=615 data-nosnippet>615</a>        StringArray,
+<a href=#616 id=616 data-nosnippet>616</a>    };
+<a href=#617 id=617 data-nosnippet>617</a>    <span class="kw">use </span>arrow_schema::{DataType, Field, Schema <span class="kw">as </span>ArrowSchema};
+<a href=#618 id=618 data-nosnippet>618</a>    <span class="kw">use </span>parquet::arrow::PARQUET_FIELD_ID_META_KEY;
+<a href=#619 id=619 data-nosnippet>619</a>
+<a href=#620 id=620 data-nosnippet>620</a>    <span class="kw">use </span><span class="kw">crate</span>::arrow::record_batch_transformer::{
+<a href=#621 id=621 data-nosnippet>621</a>        RecordBatchTransformer, RecordBatchTransformerBuilder,
+<a href=#622 id=622 data-nosnippet>622</a>    };
+<a href=#623 id=623 data-nosnippet>623</a>    <span class="kw">use </span><span class="kw">crate</span>::spec::{Literal, NestedField, PrimitiveType, Schema, Struct, Type};
+<a href=#624 id=624 data-nosnippet>624</a>
+<a href=#625 id=625 data-nosnippet>625</a>    <span class="attr">#[test]
+<a href=#626 id=626 data-nosnippet>626</a>    </span><span class="kw">fn </span>build_field_id_to_source_schema_map_works() {
+<a href=#627 id=627 data-nosnippet>627</a>        <span class="kw">let </span>arrow_schema = arrow_schema_already_same_as_target();
+<a href=#628 id=628 data-nosnippet>628</a>
+<a href=#629 id=629 data-nosnippet>629</a>        <span class="kw">let </span>result =
+<a href=#630 id=630 data-nosnippet>630</a>            RecordBatchTransformer::build_field_id_to_arrow_schema_map(<span class="kw-2">&amp;</span>arrow_schema).unwrap();
+<a href=#631 id=631 data-nosnippet>631</a>
+<a href=#632 id=632 data-nosnippet>632</a>        <span class="kw">let </span>expected = HashMap::from_iter([
+<a href=#633 id=633 data-nosnippet>633</a>            (<span class="number">10</span>, (arrow_schema.fields()[<span class="number">0</span>].clone(), <span class="number">0</span>)),
+<a href=#634 id=634 data-nosnippet>634</a>            (<span class="number">11</span>, (arrow_schema.fields()[<span class="number">1</span>].clone(), <span class="number">1</span>)),
+<a href=#635 id=635 data-nosnippet>635</a>            (<span class="number">12</span>, (arrow_schema.fields()[<span class="number">2</span>].clone(), <span class="number">2</span>)),
+<a href=#636 id=636 data-nosnippet>636</a>            (<span class="number">14</span>, (arrow_schema.fields()[<span class="number">3</span>].clone(), <span class="number">3</span>)),
+<a href=#637 id=637 data-nosnippet>637</a>            (<span class="number">15</span>, (arrow_schema.fields()[<span class="number">4</span>].clone(), <span class="number">4</span>)),
+<a href=#638 id=638 data-nosnippet>638</a>        ]);
+<a href=#639 id=639 data-nosnippet>639</a>
+<a href=#640 id=640 data-nosnippet>640</a>        <span class="macro">assert!</span>(result.eq(<span class="kw-2">&amp;</span>expected));
+<a href=#641 id=641 data-nosnippet>641</a>    }
+<a href=#642 id=642 data-nosnippet>642</a>
+<a href=#643 id=643 data-nosnippet>643</a>    <span class="attr">#[test]
+<a href=#644 id=644 data-nosnippet>644</a>    </span><span class="kw">fn </span>processor_returns_properly_shaped_record_batch_when_no_schema_migration_required() {
+<a href=#645 id=645 data-nosnippet>645</a>        <span class="kw">let </span>snapshot_schema = Arc::new(iceberg_table_schema());
+<a href=#646 id=646 data-nosnippet>646</a>        <span class="kw">let </span>projected_iceberg_field_ids = [<span class="number">13</span>, <span class="number">14</span>];
+<a href=#647 id=647 data-nosnippet>647</a>
+<a href=#648 id=648 data-nosnippet>648</a>        <span class="kw">let </span><span class="kw-2">mut </span>inst =
+<a href=#649 id=649 data-nosnippet>649</a>            RecordBatchTransformerBuilder::new(snapshot_schema, <span class="kw-2">&amp;</span>projected_iceberg_field_ids)
+<a href=#650 id=650 data-nosnippet>650</a>                .build();
+<a href=#651 id=651 data-nosnippet>651</a>
+<a href=#652 id=652 data-nosnippet>652</a>        <span class="kw">let </span>result = inst
+<a href=#653 id=653 data-nosnippet>653</a>            .process_record_batch(source_record_batch_no_migration_required())
+<a href=#654 id=654 data-nosnippet>654</a>            .unwrap();
+<a href=#655 id=655 data-nosnippet>655</a>
+<a href=#656 id=656 data-nosnippet>656</a>        <span class="kw">let </span>expected = source_record_batch_no_migration_required();
+<a href=#657 id=657 data-nosnippet>657</a>
+<a href=#658 id=658 data-nosnippet>658</a>        <span class="macro">assert_eq!</span>(result, expected);
+<a href=#659 id=659 data-nosnippet>659</a>    }
+<a href=#660 id=660 data-nosnippet>660</a>
+<a href=#661 id=661 data-nosnippet>661</a>    <span class="attr">#[test]
+<a href=#662 id=662 data-nosnippet>662</a>    </span><span class="kw">fn </span>processor_returns_properly_shaped_record_batch_when_schema_migration_required() {
+<a href=#663 id=663 data-nosnippet>663</a>        <span class="kw">let </span>snapshot_schema = Arc::new(iceberg_table_schema());
+<a href=#664 id=664 data-nosnippet>664</a>        <span class="kw">let </span>projected_iceberg_field_ids = [<span class="number">10</span>, <span class="number">11</span>, <span class="number">12</span>, <span class="number">14</span>, <span class="number">15</span>]; <span class="comment">// a, b, c, e, f
 <a href=#665 id=665 data-nosnippet>665</a>
-<a href=#666 id=666 data-nosnippet>666</a>    <span class="kw">fn </span>arrow_schema_already_same_as_target() -&gt; Arc&lt;ArrowSchema&gt; {
-<a href=#667 id=667 data-nosnippet>667</a>        Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#668 id=668 data-nosnippet>668</a>            simple_field(<span class="string">"a"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"10"</span>),
-<a href=#669 id=669 data-nosnippet>669</a>            simple_field(<span class="string">"b"</span>, DataType::Int64, <span class="bool-val">false</span>, <span class="string">"11"</span>),
-<a href=#670 id=670 data-nosnippet>670</a>            simple_field(<span class="string">"c"</span>, DataType::Float64, <span class="bool-val">false</span>, <span class="string">"12"</span>),
-<a href=#671 id=671 data-nosnippet>671</a>            simple_field(<span class="string">"e"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"14"</span>),
-<a href=#672 id=672 data-nosnippet>672</a>            simple_field(<span class="string">"f"</span>, DataType::Utf8, <span class="bool-val">false</span>, <span class="string">"15"</span>),
-<a href=#673 id=673 data-nosnippet>673</a>        ]))
-<a href=#674 id=674 data-nosnippet>674</a>    }
-<a href=#675 id=675 data-nosnippet>675</a>
-<a href=#676 id=676 data-nosnippet>676</a>    <span class="kw">fn </span>arrow_schema_promotion_addition_and_renaming_required() -&gt; Arc&lt;ArrowSchema&gt; {
-<a href=#677 id=677 data-nosnippet>677</a>        Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#678 id=678 data-nosnippet>678</a>            simple_field(<span class="string">"b"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"11"</span>),
-<a href=#679 id=679 data-nosnippet>679</a>            simple_field(<span class="string">"c"</span>, DataType::Float32, <span class="bool-val">false</span>, <span class="string">"12"</span>),
-<a href=#680 id=680 data-nosnippet>680</a>            simple_field(<span class="string">"d"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"13"</span>),
-<a href=#681 id=681 data-nosnippet>681</a>            simple_field(<span class="string">"e_old"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"14"</span>),
-<a href=#682 id=682 data-nosnippet>682</a>        ]))
-<a href=#683 id=683 data-nosnippet>683</a>    }
-<a href=#684 id=684 data-nosnippet>684</a>
-<a href=#685 id=685 data-nosnippet>685</a>    <span class="kw">fn </span>arrow_schema_no_promotion_addition_or_renaming_required() -&gt; Arc&lt;ArrowSchema&gt; {
-<a href=#686 id=686 data-nosnippet>686</a>        Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#687 id=687 data-nosnippet>687</a>            simple_field(<span class="string">"d"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"13"</span>),
-<a href=#688 id=688 data-nosnippet>688</a>            simple_field(<span class="string">"e"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"14"</span>),
-<a href=#689 id=689 data-nosnippet>689</a>        ]))
-<a href=#690 id=690 data-nosnippet>690</a>    }
-<a href=#691 id=691 data-nosnippet>691</a>
-<a href=#692 id=692 data-nosnippet>692</a>    <span class="doccomment">/// Create a simple arrow field with metadata.
-<a href=#693 id=693 data-nosnippet>693</a>    </span><span class="kw">fn </span>simple_field(name: <span class="kw-2">&amp;</span>str, ty: DataType, nullable: bool, value: <span class="kw-2">&amp;</span>str) -&gt; Field {
-<a href=#694 id=694 data-nosnippet>694</a>        Field::new(name, ty, nullable).with_metadata(HashMap::from([(
-<a href=#695 id=695 data-nosnippet>695</a>            PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#696 id=696 data-nosnippet>696</a>            value.to_string(),
-<a href=#697 id=697 data-nosnippet>697</a>        )]))
-<a href=#698 id=698 data-nosnippet>698</a>    }
-<a href=#699 id=699 data-nosnippet>699</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file
+<a href=#666 id=666 data-nosnippet>666</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>inst =
+<a href=#667 id=667 data-nosnippet>667</a>            RecordBatchTransformerBuilder::new(snapshot_schema, <span class="kw-2">&amp;</span>projected_iceberg_field_ids)
+<a href=#668 id=668 data-nosnippet>668</a>                .build();
+<a href=#669 id=669 data-nosnippet>669</a>
+<a href=#670 id=670 data-nosnippet>670</a>        <span class="kw">let </span>result = inst.process_record_batch(source_record_batch()).unwrap();
+<a href=#671 id=671 data-nosnippet>671</a>
+<a href=#672 id=672 data-nosnippet>672</a>        <span class="kw">let </span>expected = expected_record_batch_migration_required();
+<a href=#673 id=673 data-nosnippet>673</a>
+<a href=#674 id=674 data-nosnippet>674</a>        <span class="macro">assert_eq!</span>(result, expected);
+<a href=#675 id=675 data-nosnippet>675</a>    }
+<a href=#676 id=676 data-nosnippet>676</a>
+<a href=#677 id=677 data-nosnippet>677</a>    <span class="attr">#[test]
+<a href=#678 id=678 data-nosnippet>678</a>    </span><span class="kw">fn </span>schema_evolution_adds_date_column_with_nulls() {
+<a href=#679 id=679 data-nosnippet>679</a>        <span class="comment">// Reproduces TestSelect.readAndWriteWithBranchAfterSchemaChange from iceberg-spark.
+<a href=#680 id=680 data-nosnippet>680</a>        // When reading old snapshots after adding a DATE column, the transformer must
+<a href=#681 id=681 data-nosnippet>681</a>        // populate the new column with NULL values since old files lack this field.
+<a href=#682 id=682 data-nosnippet>682</a>        </span><span class="kw">let </span>snapshot_schema = Arc::new(
+<a href=#683 id=683 data-nosnippet>683</a>            Schema::builder()
+<a href=#684 id=684 data-nosnippet>684</a>                .with_schema_id(<span class="number">1</span>)
+<a href=#685 id=685 data-nosnippet>685</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#686 id=686 data-nosnippet>686</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#687 id=687 data-nosnippet>687</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#688 id=688 data-nosnippet>688</a>                    NestedField::optional(<span class="number">3</span>, <span class="string">"date_col"</span>, Type::Primitive(PrimitiveType::Date))
+<a href=#689 id=689 data-nosnippet>689</a>                        .into(),
+<a href=#690 id=690 data-nosnippet>690</a>                ])
+<a href=#691 id=691 data-nosnippet>691</a>                .build()
+<a href=#692 id=692 data-nosnippet>692</a>                .unwrap(),
+<a href=#693 id=693 data-nosnippet>693</a>        );
+<a href=#694 id=694 data-nosnippet>694</a>        <span class="kw">let </span>projected_iceberg_field_ids = [<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>];
+<a href=#695 id=695 data-nosnippet>695</a>
+<a href=#696 id=696 data-nosnippet>696</a>        <span class="kw">let </span><span class="kw-2">mut </span>transformer =
+<a href=#697 id=697 data-nosnippet>697</a>            RecordBatchTransformerBuilder::new(snapshot_schema, <span class="kw-2">&amp;</span>projected_iceberg_field_ids)
+<a href=#698 id=698 data-nosnippet>698</a>                .build();
+<a href=#699 id=699 data-nosnippet>699</a>
+<a href=#700 id=700 data-nosnippet>700</a>        <span class="kw">let </span>file_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#701 id=701 data-nosnippet>701</a>            simple_field(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"1"</span>),
+<a href=#702 id=702 data-nosnippet>702</a>            simple_field(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"2"</span>),
+<a href=#703 id=703 data-nosnippet>703</a>        ]));
+<a href=#704 id=704 data-nosnippet>704</a>
+<a href=#705 id=705 data-nosnippet>705</a>        <span class="kw">let </span>file_batch = RecordBatch::try_new(file_schema, <span class="macro">vec!</span>[
+<a href=#706 id=706 data-nosnippet>706</a>            Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>])),
+<a href=#707 id=707 data-nosnippet>707</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[
+<a href=#708 id=708 data-nosnippet>708</a>                <span class="prelude-val">Some</span>(<span class="string">"Alice"</span>),
+<a href=#709 id=709 data-nosnippet>709</a>                <span class="prelude-val">Some</span>(<span class="string">"Bob"</span>),
+<a href=#710 id=710 data-nosnippet>710</a>                <span class="prelude-val">Some</span>(<span class="string">"Charlie"</span>),
+<a href=#711 id=711 data-nosnippet>711</a>            ])),
+<a href=#712 id=712 data-nosnippet>712</a>        ])
+<a href=#713 id=713 data-nosnippet>713</a>        .unwrap();
+<a href=#714 id=714 data-nosnippet>714</a>
+<a href=#715 id=715 data-nosnippet>715</a>        <span class="kw">let </span>result = transformer.process_record_batch(file_batch).unwrap();
+<a href=#716 id=716 data-nosnippet>716</a>
+<a href=#717 id=717 data-nosnippet>717</a>        <span class="macro">assert_eq!</span>(result.num_columns(), <span class="number">3</span>);
+<a href=#718 id=718 data-nosnippet>718</a>        <span class="macro">assert_eq!</span>(result.num_rows(), <span class="number">3</span>);
+<a href=#719 id=719 data-nosnippet>719</a>
+<a href=#720 id=720 data-nosnippet>720</a>        <span class="kw">let </span>id_column = result
+<a href=#721 id=721 data-nosnippet>721</a>            .column(<span class="number">0</span>)
+<a href=#722 id=722 data-nosnippet>722</a>            .as_any()
+<a href=#723 id=723 data-nosnippet>723</a>            .downcast_ref::&lt;Int32Array&gt;()
+<a href=#724 id=724 data-nosnippet>724</a>            .unwrap();
+<a href=#725 id=725 data-nosnippet>725</a>        <span class="macro">assert_eq!</span>(id_column.values(), <span class="kw-2">&amp;</span>[<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>]);
+<a href=#726 id=726 data-nosnippet>726</a>
+<a href=#727 id=727 data-nosnippet>727</a>        <span class="kw">let </span>name_column = result
+<a href=#728 id=728 data-nosnippet>728</a>            .column(<span class="number">1</span>)
+<a href=#729 id=729 data-nosnippet>729</a>            .as_any()
+<a href=#730 id=730 data-nosnippet>730</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#731 id=731 data-nosnippet>731</a>            .unwrap();
+<a href=#732 id=732 data-nosnippet>732</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
+<a href=#733 id=733 data-nosnippet>733</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
+<a href=#734 id=734 data-nosnippet>734</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">2</span>), <span class="string">"Charlie"</span>);
+<a href=#735 id=735 data-nosnippet>735</a>
+<a href=#736 id=736 data-nosnippet>736</a>        <span class="kw">let </span>date_column = result
+<a href=#737 id=737 data-nosnippet>737</a>            .column(<span class="number">2</span>)
+<a href=#738 id=738 data-nosnippet>738</a>            .as_any()
+<a href=#739 id=739 data-nosnippet>739</a>            .downcast_ref::&lt;Date32Array&gt;()
+<a href=#740 id=740 data-nosnippet>740</a>            .unwrap();
+<a href=#741 id=741 data-nosnippet>741</a>        <span class="macro">assert!</span>(date_column.is_null(<span class="number">0</span>));
+<a href=#742 id=742 data-nosnippet>742</a>        <span class="macro">assert!</span>(date_column.is_null(<span class="number">1</span>));
+<a href=#743 id=743 data-nosnippet>743</a>        <span class="macro">assert!</span>(date_column.is_null(<span class="number">2</span>));
+<a href=#744 id=744 data-nosnippet>744</a>    }
+<a href=#745 id=745 data-nosnippet>745</a>
+<a href=#746 id=746 data-nosnippet>746</a>    <span class="kw">pub fn </span>source_record_batch() -&gt; RecordBatch {
+<a href=#747 id=747 data-nosnippet>747</a>        RecordBatch::try_new(
+<a href=#748 id=748 data-nosnippet>748</a>            arrow_schema_promotion_addition_and_renaming_required(),
+<a href=#749 id=749 data-nosnippet>749</a>            <span class="macro">vec!</span>[
+<a href=#750 id=750 data-nosnippet>750</a>                Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">1001</span>), <span class="prelude-val">Some</span>(<span class="number">1002</span>), <span class="prelude-val">Some</span>(<span class="number">1003</span>)])), <span class="comment">// b
+<a href=#751 id=751 data-nosnippet>751</a>                </span>Arc::new(Float32Array::from(<span class="macro">vec!</span>[
+<a href=#752 id=752 data-nosnippet>752</a>                    <span class="prelude-val">Some</span>(<span class="number">12.125</span>),
+<a href=#753 id=753 data-nosnippet>753</a>                    <span class="prelude-val">Some</span>(<span class="number">23.375</span>),
+<a href=#754 id=754 data-nosnippet>754</a>                    <span class="prelude-val">Some</span>(<span class="number">34.875</span>),
+<a href=#755 id=755 data-nosnippet>755</a>                ])), <span class="comment">// c
+<a href=#756 id=756 data-nosnippet>756</a>                </span>Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">2001</span>), <span class="prelude-val">Some</span>(<span class="number">2002</span>), <span class="prelude-val">Some</span>(<span class="number">2003</span>)])), <span class="comment">// d
+<a href=#757 id=757 data-nosnippet>757</a>                </span>Arc::new(StringArray::from(<span class="macro">vec!</span>[
+<a href=#758 id=758 data-nosnippet>758</a>                    <span class="prelude-val">Some</span>(<span class="string">"Apache"</span>),
+<a href=#759 id=759 data-nosnippet>759</a>                    <span class="prelude-val">Some</span>(<span class="string">"Iceberg"</span>),
+<a href=#760 id=760 data-nosnippet>760</a>                    <span class="prelude-val">Some</span>(<span class="string">"Rocks"</span>),
+<a href=#761 id=761 data-nosnippet>761</a>                ])), <span class="comment">// e
+<a href=#762 id=762 data-nosnippet>762</a>            </span>],
+<a href=#763 id=763 data-nosnippet>763</a>        )
+<a href=#764 id=764 data-nosnippet>764</a>        .unwrap()
+<a href=#765 id=765 data-nosnippet>765</a>    }
+<a href=#766 id=766 data-nosnippet>766</a>
+<a href=#767 id=767 data-nosnippet>767</a>    <span class="kw">pub fn </span>source_record_batch_no_migration_required() -&gt; RecordBatch {
+<a href=#768 id=768 data-nosnippet>768</a>        RecordBatch::try_new(
+<a href=#769 id=769 data-nosnippet>769</a>            arrow_schema_no_promotion_addition_or_renaming_required(),
+<a href=#770 id=770 data-nosnippet>770</a>            <span class="macro">vec!</span>[
+<a href=#771 id=771 data-nosnippet>771</a>                Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">2001</span>), <span class="prelude-val">Some</span>(<span class="number">2002</span>), <span class="prelude-val">Some</span>(<span class="number">2003</span>)])), <span class="comment">// d
+<a href=#772 id=772 data-nosnippet>772</a>                </span>Arc::new(StringArray::from(<span class="macro">vec!</span>[
+<a href=#773 id=773 data-nosnippet>773</a>                    <span class="prelude-val">Some</span>(<span class="string">"Apache"</span>),
+<a href=#774 id=774 data-nosnippet>774</a>                    <span class="prelude-val">Some</span>(<span class="string">"Iceberg"</span>),
+<a href=#775 id=775 data-nosnippet>775</a>                    <span class="prelude-val">Some</span>(<span class="string">"Rocks"</span>),
+<a href=#776 id=776 data-nosnippet>776</a>                ])), <span class="comment">// e
+<a href=#777 id=777 data-nosnippet>777</a>            </span>],
+<a href=#778 id=778 data-nosnippet>778</a>        )
+<a href=#779 id=779 data-nosnippet>779</a>        .unwrap()
+<a href=#780 id=780 data-nosnippet>780</a>    }
+<a href=#781 id=781 data-nosnippet>781</a>
+<a href=#782 id=782 data-nosnippet>782</a>    <span class="kw">pub fn </span>expected_record_batch_migration_required() -&gt; RecordBatch {
+<a href=#783 id=783 data-nosnippet>783</a>        RecordBatch::try_new(arrow_schema_already_same_as_target(), <span class="macro">vec!</span>[
+<a href=#784 id=784 data-nosnippet>784</a>            Arc::new(StringArray::from(Vec::&lt;<span class="prelude-ty">Option</span>&lt;String&gt;&gt;::from([
+<a href=#785 id=785 data-nosnippet>785</a>                <span class="prelude-val">None</span>, <span class="prelude-val">None</span>, <span class="prelude-val">None</span>,
+<a href=#786 id=786 data-nosnippet>786</a>            ]))), <span class="comment">// a
+<a href=#787 id=787 data-nosnippet>787</a>            </span>Arc::new(Int64Array::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">1001</span>), <span class="prelude-val">Some</span>(<span class="number">1002</span>), <span class="prelude-val">Some</span>(<span class="number">1003</span>)])), <span class="comment">// b
+<a href=#788 id=788 data-nosnippet>788</a>            </span>Arc::new(Float64Array::from(<span class="macro">vec!</span>[
+<a href=#789 id=789 data-nosnippet>789</a>                <span class="prelude-val">Some</span>(<span class="number">12.125</span>),
+<a href=#790 id=790 data-nosnippet>790</a>                <span class="prelude-val">Some</span>(<span class="number">23.375</span>),
+<a href=#791 id=791 data-nosnippet>791</a>                <span class="prelude-val">Some</span>(<span class="number">34.875</span>),
+<a href=#792 id=792 data-nosnippet>792</a>            ])), <span class="comment">// c
+<a href=#793 id=793 data-nosnippet>793</a>            </span>Arc::new(StringArray::from(<span class="macro">vec!</span>[
+<a href=#794 id=794 data-nosnippet>794</a>                <span class="prelude-val">Some</span>(<span class="string">"Apache"</span>),
+<a href=#795 id=795 data-nosnippet>795</a>                <span class="prelude-val">Some</span>(<span class="string">"Iceberg"</span>),
+<a href=#796 id=796 data-nosnippet>796</a>                <span class="prelude-val">Some</span>(<span class="string">"Rocks"</span>),
+<a href=#797 id=797 data-nosnippet>797</a>            ])), <span class="comment">// e (d skipped by projection)
+<a href=#798 id=798 data-nosnippet>798</a>            </span>Arc::new(StringArray::from(<span class="macro">vec!</span>[
+<a href=#799 id=799 data-nosnippet>799</a>                <span class="prelude-val">Some</span>(<span class="string">"(╯°□°)╯"</span>),
+<a href=#800 id=800 data-nosnippet>800</a>                <span class="prelude-val">Some</span>(<span class="string">"(╯°□°)╯"</span>),
+<a href=#801 id=801 data-nosnippet>801</a>                <span class="prelude-val">Some</span>(<span class="string">"(╯°□°)╯"</span>),
+<a href=#802 id=802 data-nosnippet>802</a>            ])), <span class="comment">// f
+<a href=#803 id=803 data-nosnippet>803</a>        </span>])
+<a href=#804 id=804 data-nosnippet>804</a>        .unwrap()
+<a href=#805 id=805 data-nosnippet>805</a>    }
+<a href=#806 id=806 data-nosnippet>806</a>
+<a href=#807 id=807 data-nosnippet>807</a>    <span class="kw">pub fn </span>iceberg_table_schema() -&gt; Schema {
+<a href=#808 id=808 data-nosnippet>808</a>        Schema::builder()
+<a href=#809 id=809 data-nosnippet>809</a>            .with_schema_id(<span class="number">2</span>)
+<a href=#810 id=810 data-nosnippet>810</a>            .with_fields(<span class="macro">vec!</span>[
+<a href=#811 id=811 data-nosnippet>811</a>                NestedField::optional(<span class="number">10</span>, <span class="string">"a"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#812 id=812 data-nosnippet>812</a>                NestedField::required(<span class="number">11</span>, <span class="string">"b"</span>, Type::Primitive(PrimitiveType::Long)).into(),
+<a href=#813 id=813 data-nosnippet>813</a>                NestedField::required(<span class="number">12</span>, <span class="string">"c"</span>, Type::Primitive(PrimitiveType::Double)).into(),
+<a href=#814 id=814 data-nosnippet>814</a>                NestedField::required(<span class="number">13</span>, <span class="string">"d"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#815 id=815 data-nosnippet>815</a>                NestedField::optional(<span class="number">14</span>, <span class="string">"e"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#816 id=816 data-nosnippet>816</a>                NestedField::required(<span class="number">15</span>, <span class="string">"f"</span>, Type::Primitive(PrimitiveType::String))
+<a href=#817 id=817 data-nosnippet>817</a>                    .with_initial_default(Literal::string(<span class="string">"(╯°□°)╯"</span>))
+<a href=#818 id=818 data-nosnippet>818</a>                    .into(),
+<a href=#819 id=819 data-nosnippet>819</a>            ])
+<a href=#820 id=820 data-nosnippet>820</a>            .build()
+<a href=#821 id=821 data-nosnippet>821</a>            .unwrap()
+<a href=#822 id=822 data-nosnippet>822</a>    }
+<a href=#823 id=823 data-nosnippet>823</a>
+<a href=#824 id=824 data-nosnippet>824</a>    <span class="kw">fn </span>arrow_schema_already_same_as_target() -&gt; Arc&lt;ArrowSchema&gt; {
+<a href=#825 id=825 data-nosnippet>825</a>        Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#826 id=826 data-nosnippet>826</a>            simple_field(<span class="string">"a"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"10"</span>),
+<a href=#827 id=827 data-nosnippet>827</a>            simple_field(<span class="string">"b"</span>, DataType::Int64, <span class="bool-val">false</span>, <span class="string">"11"</span>),
+<a href=#828 id=828 data-nosnippet>828</a>            simple_field(<span class="string">"c"</span>, DataType::Float64, <span class="bool-val">false</span>, <span class="string">"12"</span>),
+<a href=#829 id=829 data-nosnippet>829</a>            simple_field(<span class="string">"e"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"14"</span>),
+<a href=#830 id=830 data-nosnippet>830</a>            simple_field(<span class="string">"f"</span>, DataType::Utf8, <span class="bool-val">false</span>, <span class="string">"15"</span>),
+<a href=#831 id=831 data-nosnippet>831</a>        ]))
+<a href=#832 id=832 data-nosnippet>832</a>    }
+<a href=#833 id=833 data-nosnippet>833</a>
+<a href=#834 id=834 data-nosnippet>834</a>    <span class="kw">fn </span>arrow_schema_promotion_addition_and_renaming_required() -&gt; Arc&lt;ArrowSchema&gt; {
+<a href=#835 id=835 data-nosnippet>835</a>        Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#836 id=836 data-nosnippet>836</a>            simple_field(<span class="string">"b"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"11"</span>),
+<a href=#837 id=837 data-nosnippet>837</a>            simple_field(<span class="string">"c"</span>, DataType::Float32, <span class="bool-val">false</span>, <span class="string">"12"</span>),
+<a href=#838 id=838 data-nosnippet>838</a>            simple_field(<span class="string">"d"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"13"</span>),
+<a href=#839 id=839 data-nosnippet>839</a>            simple_field(<span class="string">"e_old"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"14"</span>),
+<a href=#840 id=840 data-nosnippet>840</a>        ]))
+<a href=#841 id=841 data-nosnippet>841</a>    }
+<a href=#842 id=842 data-nosnippet>842</a>
+<a href=#843 id=843 data-nosnippet>843</a>    <span class="kw">fn </span>arrow_schema_no_promotion_addition_or_renaming_required() -&gt; Arc&lt;ArrowSchema&gt; {
+<a href=#844 id=844 data-nosnippet>844</a>        Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#845 id=845 data-nosnippet>845</a>            simple_field(<span class="string">"d"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"13"</span>),
+<a href=#846 id=846 data-nosnippet>846</a>            simple_field(<span class="string">"e"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"14"</span>),
+<a href=#847 id=847 data-nosnippet>847</a>        ]))
+<a href=#848 id=848 data-nosnippet>848</a>    }
+<a href=#849 id=849 data-nosnippet>849</a>
+<a href=#850 id=850 data-nosnippet>850</a>    <span class="doccomment">/// Create a simple arrow field with metadata.
+<a href=#851 id=851 data-nosnippet>851</a>    </span><span class="kw">fn </span>simple_field(name: <span class="kw-2">&amp;</span>str, ty: DataType, nullable: bool, value: <span class="kw-2">&amp;</span>str) -&gt; Field {
+<a href=#852 id=852 data-nosnippet>852</a>        Field::new(name, ty, nullable).with_metadata(HashMap::from([(
+<a href=#853 id=853 data-nosnippet>853</a>            PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#854 id=854 data-nosnippet>854</a>            value.to_string(),
+<a href=#855 id=855 data-nosnippet>855</a>        )]))
+<a href=#856 id=856 data-nosnippet>856</a>    }
+<a href=#857 id=857 data-nosnippet>857</a>
+<a href=#858 id=858 data-nosnippet>858</a>    <span class="doccomment">/// Test for add_files with Parquet files that have NO field IDs (Hive tables).
+<a href=#859 id=859 data-nosnippet>859</a>    ///
+<a href=#860 id=860 data-nosnippet>860</a>    /// This reproduces the scenario from Iceberg spec where:
+<a href=#861 id=861 data-nosnippet>861</a>    /// - Hive-style partitioned Parquet files are imported via add_files procedure
+<a href=#862 id=862 data-nosnippet>862</a>    /// - Parquet files originally DO NOT have field IDs (typical for Hive tables)
+<a href=#863 id=863 data-nosnippet>863</a>    /// - ArrowReader applies name mapping to assign correct Iceberg field IDs
+<a href=#864 id=864 data-nosnippet>864</a>    /// - Iceberg schema assigns field IDs: id (1), name (2), dept (3), subdept (4)
+<a href=#865 id=865 data-nosnippet>865</a>    /// - Partition columns (id, dept) have initial_default values
+<a href=#866 id=866 data-nosnippet>866</a>    ///
+<a href=#867 id=867 data-nosnippet>867</a>    /// Per the Iceberg spec (https://iceberg.apache.org/spec/#column-projection),
+<a href=#868 id=868 data-nosnippet>868</a>    /// this scenario requires `schema.name-mapping.default` from table metadata
+<a href=#869 id=869 data-nosnippet>869</a>    /// to correctly map Parquet columns by name to Iceberg field IDs.
+<a href=#870 id=870 data-nosnippet>870</a>    /// This mapping is now applied in ArrowReader before data is processed.
+<a href=#871 id=871 data-nosnippet>871</a>    ///
+<a href=#872 id=872 data-nosnippet>872</a>    /// Expected behavior:
+<a href=#873 id=873 data-nosnippet>873</a>    /// 1. id=1 (from initial_default) - spec rule #3
+<a href=#874 id=874 data-nosnippet>874</a>    /// 2. name="John Doe" (from Parquet with field_id=2 assigned by reader) - found by field ID
+<a href=#875 id=875 data-nosnippet>875</a>    /// 3. dept="hr" (from initial_default) - spec rule #3
+<a href=#876 id=876 data-nosnippet>876</a>    /// 4. subdept="communications" (from Parquet with field_id=4 assigned by reader) - found by field ID
+<a href=#877 id=877 data-nosnippet>877</a>    </span><span class="attr">#[test]
+<a href=#878 id=878 data-nosnippet>878</a>    </span><span class="kw">fn </span>add_files_with_name_mapping_applied_in_reader() {
+<a href=#879 id=879 data-nosnippet>879</a>        <span class="comment">// Iceberg schema after add_files: id (partition), name, dept (partition), subdept
+<a href=#880 id=880 data-nosnippet>880</a>        </span><span class="kw">let </span>snapshot_schema = Arc::new(
+<a href=#881 id=881 data-nosnippet>881</a>            Schema::builder()
+<a href=#882 id=882 data-nosnippet>882</a>                .with_schema_id(<span class="number">0</span>)
+<a href=#883 id=883 data-nosnippet>883</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#884 id=884 data-nosnippet>884</a>                    NestedField::optional(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int))
+<a href=#885 id=885 data-nosnippet>885</a>                        .with_initial_default(Literal::int(<span class="number">1</span>))
+<a href=#886 id=886 data-nosnippet>886</a>                        .into(),
+<a href=#887 id=887 data-nosnippet>887</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#888 id=888 data-nosnippet>888</a>                    NestedField::optional(<span class="number">3</span>, <span class="string">"dept"</span>, Type::Primitive(PrimitiveType::String))
+<a href=#889 id=889 data-nosnippet>889</a>                        .with_initial_default(Literal::string(<span class="string">"hr"</span>))
+<a href=#890 id=890 data-nosnippet>890</a>                        .into(),
+<a href=#891 id=891 data-nosnippet>891</a>                    NestedField::optional(<span class="number">4</span>, <span class="string">"subdept"</span>, Type::Primitive(PrimitiveType::String))
+<a href=#892 id=892 data-nosnippet>892</a>                        .into(),
+<a href=#893 id=893 data-nosnippet>893</a>                ])
+<a href=#894 id=894 data-nosnippet>894</a>                .build()
+<a href=#895 id=895 data-nosnippet>895</a>                .unwrap(),
+<a href=#896 id=896 data-nosnippet>896</a>        );
+<a href=#897 id=897 data-nosnippet>897</a>
+<a href=#898 id=898 data-nosnippet>898</a>        <span class="comment">// Simulate ArrowReader having applied name mapping:
+<a href=#899 id=899 data-nosnippet>899</a>        // Original Parquet: name, subdept (NO field IDs)
+<a href=#900 id=900 data-nosnippet>900</a>        // After reader.rs applies name mapping: name (field_id=2), subdept (field_id=4)
+<a href=#901 id=901 data-nosnippet>901</a>        //
+<a href=#902 id=902 data-nosnippet>902</a>        // Note: Partition columns (id, dept) are NOT in the Parquet file - they're in directory paths
+<a href=#903 id=903 data-nosnippet>903</a>        </span><span class="kw">use </span>std::collections::HashMap;
+<a href=#904 id=904 data-nosnippet>904</a>        <span class="kw">let </span>parquet_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#905 id=905 data-nosnippet>905</a>            Field::new(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">true</span>).with_metadata(HashMap::from([(
+<a href=#906 id=906 data-nosnippet>906</a>                <span class="string">"PARQUET:field_id"</span>.to_string(),
+<a href=#907 id=907 data-nosnippet>907</a>                <span class="string">"2"</span>.to_string(),
+<a href=#908 id=908 data-nosnippet>908</a>            )])),
+<a href=#909 id=909 data-nosnippet>909</a>            Field::new(<span class="string">"subdept"</span>, DataType::Utf8, <span class="bool-val">true</span>).with_metadata(HashMap::from([(
+<a href=#910 id=910 data-nosnippet>910</a>                <span class="string">"PARQUET:field_id"</span>.to_string(),
+<a href=#911 id=911 data-nosnippet>911</a>                <span class="string">"4"</span>.to_string(),
+<a href=#912 id=912 data-nosnippet>912</a>            )])),
+<a href=#913 id=913 data-nosnippet>913</a>        ]));
+<a href=#914 id=914 data-nosnippet>914</a>
+<a href=#915 id=915 data-nosnippet>915</a>        <span class="kw">let </span>projected_field_ids = [<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>, <span class="number">4</span>]; <span class="comment">// id, name, dept, subdept
+<a href=#916 id=916 data-nosnippet>916</a>
+<a href=#917 id=917 data-nosnippet>917</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>transformer =
+<a href=#918 id=918 data-nosnippet>918</a>            RecordBatchTransformerBuilder::new(snapshot_schema, <span class="kw-2">&amp;</span>projected_field_ids).build();
+<a href=#919 id=919 data-nosnippet>919</a>
+<a href=#920 id=920 data-nosnippet>920</a>        <span class="comment">// Create a Parquet RecordBatch with data for: name="John Doe", subdept="communications"
+<a href=#921 id=921 data-nosnippet>921</a>        </span><span class="kw">let </span>parquet_batch = RecordBatch::try_new(parquet_schema, <span class="macro">vec!</span>[
+<a href=#922 id=922 data-nosnippet>922</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"John Doe"</span>])),
+<a href=#923 id=923 data-nosnippet>923</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"communications"</span>])),
+<a href=#924 id=924 data-nosnippet>924</a>        ])
+<a href=#925 id=925 data-nosnippet>925</a>        .unwrap();
+<a href=#926 id=926 data-nosnippet>926</a>
+<a href=#927 id=927 data-nosnippet>927</a>        <span class="kw">let </span>result = transformer.process_record_batch(parquet_batch).unwrap();
+<a href=#928 id=928 data-nosnippet>928</a>
+<a href=#929 id=929 data-nosnippet>929</a>        <span class="comment">// Verify the transformed RecordBatch has:
+<a href=#930 id=930 data-nosnippet>930</a>        // - id=1 (from initial_default, not from Parquet)
+<a href=#931 id=931 data-nosnippet>931</a>        // - name="John Doe" (from Parquet with correct field_id=2)
+<a href=#932 id=932 data-nosnippet>932</a>        // - dept="hr" (from initial_default, not from Parquet)
+<a href=#933 id=933 data-nosnippet>933</a>        // - subdept="communications" (from Parquet with correct field_id=4)
+<a href=#934 id=934 data-nosnippet>934</a>        </span><span class="macro">assert_eq!</span>(result.num_columns(), <span class="number">4</span>);
+<a href=#935 id=935 data-nosnippet>935</a>        <span class="macro">assert_eq!</span>(result.num_rows(), <span class="number">1</span>);
+<a href=#936 id=936 data-nosnippet>936</a>
+<a href=#937 id=937 data-nosnippet>937</a>        <span class="kw">let </span>id_column = result
+<a href=#938 id=938 data-nosnippet>938</a>            .column(<span class="number">0</span>)
+<a href=#939 id=939 data-nosnippet>939</a>            .as_any()
+<a href=#940 id=940 data-nosnippet>940</a>            .downcast_ref::&lt;Int32Array&gt;()
+<a href=#941 id=941 data-nosnippet>941</a>            .unwrap();
+<a href=#942 id=942 data-nosnippet>942</a>        <span class="macro">assert_eq!</span>(id_column.value(<span class="number">0</span>), <span class="number">1</span>);
+<a href=#943 id=943 data-nosnippet>943</a>
+<a href=#944 id=944 data-nosnippet>944</a>        <span class="kw">let </span>name_column = result
+<a href=#945 id=945 data-nosnippet>945</a>            .column(<span class="number">1</span>)
+<a href=#946 id=946 data-nosnippet>946</a>            .as_any()
+<a href=#947 id=947 data-nosnippet>947</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#948 id=948 data-nosnippet>948</a>            .unwrap();
+<a href=#949 id=949 data-nosnippet>949</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">0</span>), <span class="string">"John Doe"</span>);
+<a href=#950 id=950 data-nosnippet>950</a>
+<a href=#951 id=951 data-nosnippet>951</a>        <span class="kw">let </span>dept_column = result
+<a href=#952 id=952 data-nosnippet>952</a>            .column(<span class="number">2</span>)
+<a href=#953 id=953 data-nosnippet>953</a>            .as_any()
+<a href=#954 id=954 data-nosnippet>954</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#955 id=955 data-nosnippet>955</a>            .unwrap();
+<a href=#956 id=956 data-nosnippet>956</a>        <span class="macro">assert_eq!</span>(dept_column.value(<span class="number">0</span>), <span class="string">"hr"</span>);
+<a href=#957 id=957 data-nosnippet>957</a>
+<a href=#958 id=958 data-nosnippet>958</a>        <span class="kw">let </span>subdept_column = result
+<a href=#959 id=959 data-nosnippet>959</a>            .column(<span class="number">3</span>)
+<a href=#960 id=960 data-nosnippet>960</a>            .as_any()
+<a href=#961 id=961 data-nosnippet>961</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#962 id=962 data-nosnippet>962</a>            .unwrap();
+<a href=#963 id=963 data-nosnippet>963</a>        <span class="macro">assert_eq!</span>(subdept_column.value(<span class="number">0</span>), <span class="string">"communications"</span>);
+<a href=#964 id=964 data-nosnippet>964</a>    }
+<a href=#965 id=965 data-nosnippet>965</a>
+<a href=#966 id=966 data-nosnippet>966</a>    <span class="doccomment">/// Test for bucket partitioning where source columns must be read from data files.
+<a href=#967 id=967 data-nosnippet>967</a>    ///
+<a href=#968 id=968 data-nosnippet>968</a>    /// This test verifies correct implementation of the Iceberg spec's "Column Projection" rules:
+<a href=#969 id=969 data-nosnippet>969</a>    /// &gt; "Return the value from partition metadata if an **Identity Transform** exists for the field"
+<a href=#970 id=970 data-nosnippet>970</a>    ///
+<a href=#971 id=971 data-nosnippet>971</a>    /// # Why this test is critical
+<a href=#972 id=972 data-nosnippet>972</a>    ///
+<a href=#973 id=973 data-nosnippet>973</a>    /// The key insight is that partition metadata stores TRANSFORMED values, not source values:
+<a href=#974 id=974 data-nosnippet>974</a>    /// - For `bucket(4, id)`, partition metadata has `id_bucket = 2` (the bucket number)
+<a href=#975 id=975 data-nosnippet>975</a>    /// - The actual `id` column values (100, 200, 300) are ONLY in the data file
+<a href=#976 id=976 data-nosnippet>976</a>    ///
+<a href=#977 id=977 data-nosnippet>977</a>    /// If iceberg-rust incorrectly treated bucket-partitioned fields as constants, it would:
+<a href=#978 id=978 data-nosnippet>978</a>    /// 1. Replace all `id` values with the constant `2` from partition metadata
+<a href=#979 id=979 data-nosnippet>979</a>    /// 2. Break runtime filtering (e.g., `WHERE id = 100` would match no rows)
+<a href=#980 id=980 data-nosnippet>980</a>    /// 3. Return incorrect query results
+<a href=#981 id=981 data-nosnippet>981</a>    ///
+<a href=#982 id=982 data-nosnippet>982</a>    /// # What this test verifies
+<a href=#983 id=983 data-nosnippet>983</a>    ///
+<a href=#984 id=984 data-nosnippet>984</a>    /// - Bucket-partitioned fields (e.g., `bucket(4, id)`) are read from the data file
+<a href=#985 id=985 data-nosnippet>985</a>    /// - The source column `id` contains actual values (100, 200, 300), not constants
+<a href=#986 id=986 data-nosnippet>986</a>    /// - Java's `PartitionUtil.constantsMap()` behavior is correctly replicated:
+<a href=#987 id=987 data-nosnippet>987</a>    ///   ```java
+<a href=#988 id=988 data-nosnippet>988</a>    ///   if (field.transform().isIdentity()) {  // FALSE for bucket transforms
+<a href=#989 id=989 data-nosnippet>989</a>    ///       idToConstant.put(field.sourceId(), converted);
+<a href=#990 id=990 data-nosnippet>990</a>    ///   }
+<a href=#991 id=991 data-nosnippet>991</a>    ///   ```
+<a href=#992 id=992 data-nosnippet>992</a>    ///
+<a href=#993 id=993 data-nosnippet>993</a>    /// # Real-world impact
+<a href=#994 id=994 data-nosnippet>994</a>    ///
+<a href=#995 id=995 data-nosnippet>995</a>    /// This reproduces the failure scenario from Iceberg Java's TestRuntimeFiltering:
+<a href=#996 id=996 data-nosnippet>996</a>    /// - Tables partitioned by `bucket(N, col)` are common for load balancing
+<a href=#997 id=997 data-nosnippet>997</a>    /// - Queries filter on the source column: `SELECT * FROM tbl WHERE col = value`
+<a href=#998 id=998 data-nosnippet>998</a>    /// - Runtime filtering pushes predicates down to Iceberg file scans
+<a href=#999 id=999 data-nosnippet>999</a>    /// - Without this fix, the filter would match against constant partition values instead of data
+<a href=#1000 id=1000 data-nosnippet>1000</a>    ///
+<a href=#1001 id=1001 data-nosnippet>1001</a>    /// # References
+<a href=#1002 id=1002 data-nosnippet>1002</a>    /// - Iceberg spec: format/spec.md "Column Projection" + "Partition Transforms"
+<a href=#1003 id=1003 data-nosnippet>1003</a>    /// - Java impl: core/src/main/java/org/apache/iceberg/util/PartitionUtil.java
+<a href=#1004 id=1004 data-nosnippet>1004</a>    /// - Java test: spark/src/test/java/.../TestRuntimeFiltering.java
+<a href=#1005 id=1005 data-nosnippet>1005</a>    </span><span class="attr">#[test]
+<a href=#1006 id=1006 data-nosnippet>1006</a>    </span><span class="kw">fn </span>bucket_partitioning_reads_source_column_from_file() {
+<a href=#1007 id=1007 data-nosnippet>1007</a>        <span class="kw">use </span><span class="kw">crate</span>::spec::{Struct, Transform};
+<a href=#1008 id=1008 data-nosnippet>1008</a>
+<a href=#1009 id=1009 data-nosnippet>1009</a>        <span class="comment">// Table schema: id (data column), name (data column), id_bucket (partition column)
+<a href=#1010 id=1010 data-nosnippet>1010</a>        </span><span class="kw">let </span>snapshot_schema = Arc::new(
+<a href=#1011 id=1011 data-nosnippet>1011</a>            Schema::builder()
+<a href=#1012 id=1012 data-nosnippet>1012</a>                .with_schema_id(<span class="number">0</span>)
+<a href=#1013 id=1013 data-nosnippet>1013</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#1014 id=1014 data-nosnippet>1014</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#1015 id=1015 data-nosnippet>1015</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#1016 id=1016 data-nosnippet>1016</a>                ])
+<a href=#1017 id=1017 data-nosnippet>1017</a>                .build()
+<a href=#1018 id=1018 data-nosnippet>1018</a>                .unwrap(),
+<a href=#1019 id=1019 data-nosnippet>1019</a>        );
+<a href=#1020 id=1020 data-nosnippet>1020</a>
+<a href=#1021 id=1021 data-nosnippet>1021</a>        <span class="comment">// Partition spec: bucket(4, id) - the id field is bucketed
+<a href=#1022 id=1022 data-nosnippet>1022</a>        </span><span class="kw">let </span>partition_spec = Arc::new(
+<a href=#1023 id=1023 data-nosnippet>1023</a>            <span class="kw">crate</span>::spec::PartitionSpec::builder(snapshot_schema.clone())
+<a href=#1024 id=1024 data-nosnippet>1024</a>                .with_spec_id(<span class="number">0</span>)
+<a href=#1025 id=1025 data-nosnippet>1025</a>                .add_partition_field(<span class="string">"id"</span>, <span class="string">"id_bucket"</span>, Transform::Bucket(<span class="number">4</span>))
+<a href=#1026 id=1026 data-nosnippet>1026</a>                .unwrap()
+<a href=#1027 id=1027 data-nosnippet>1027</a>                .build()
+<a href=#1028 id=1028 data-nosnippet>1028</a>                .unwrap(),
+<a href=#1029 id=1029 data-nosnippet>1029</a>        );
+<a href=#1030 id=1030 data-nosnippet>1030</a>
+<a href=#1031 id=1031 data-nosnippet>1031</a>        <span class="comment">// Partition data: bucket value is 2
+<a href=#1032 id=1032 data-nosnippet>1032</a>        // In Iceberg, partition data is a Struct where each field corresponds to a partition field
+<a href=#1033 id=1033 data-nosnippet>1033</a>        </span><span class="kw">let </span>partition_data = Struct::from_iter(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(Literal::int(<span class="number">2</span>))]);
+<a href=#1034 id=1034 data-nosnippet>1034</a>
+<a href=#1035 id=1035 data-nosnippet>1035</a>        <span class="comment">// Parquet file contains both id and name columns
+<a href=#1036 id=1036 data-nosnippet>1036</a>        </span><span class="kw">let </span>parquet_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#1037 id=1037 data-nosnippet>1037</a>            simple_field(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"1"</span>),
+<a href=#1038 id=1038 data-nosnippet>1038</a>            simple_field(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"2"</span>),
+<a href=#1039 id=1039 data-nosnippet>1039</a>        ]));
+<a href=#1040 id=1040 data-nosnippet>1040</a>
+<a href=#1041 id=1041 data-nosnippet>1041</a>        <span class="kw">let </span>projected_field_ids = [<span class="number">1</span>, <span class="number">2</span>]; <span class="comment">// id, name
+<a href=#1042 id=1042 data-nosnippet>1042</a>
+<a href=#1043 id=1043 data-nosnippet>1043</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>transformer =
+<a href=#1044 id=1044 data-nosnippet>1044</a>            RecordBatchTransformerBuilder::new(snapshot_schema, <span class="kw-2">&amp;</span>projected_field_ids)
+<a href=#1045 id=1045 data-nosnippet>1045</a>                .with_partition(partition_spec, partition_data)
+<a href=#1046 id=1046 data-nosnippet>1046</a>                .build();
+<a href=#1047 id=1047 data-nosnippet>1047</a>
+<a href=#1048 id=1048 data-nosnippet>1048</a>        <span class="comment">// Create a Parquet RecordBatch with actual data
+<a href=#1049 id=1049 data-nosnippet>1049</a>        // The id column MUST be read from here, not treated as a constant
+<a href=#1050 id=1050 data-nosnippet>1050</a>        </span><span class="kw">let </span>parquet_batch = RecordBatch::try_new(parquet_schema, <span class="macro">vec!</span>[
+<a href=#1051 id=1051 data-nosnippet>1051</a>            Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>])),
+<a href=#1052 id=1052 data-nosnippet>1052</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>, <span class="string">"Charlie"</span>])),
+<a href=#1053 id=1053 data-nosnippet>1053</a>        ])
+<a href=#1054 id=1054 data-nosnippet>1054</a>        .unwrap();
+<a href=#1055 id=1055 data-nosnippet>1055</a>
+<a href=#1056 id=1056 data-nosnippet>1056</a>        <span class="kw">let </span>result = transformer.process_record_batch(parquet_batch).unwrap();
+<a href=#1057 id=1057 data-nosnippet>1057</a>
+<a href=#1058 id=1058 data-nosnippet>1058</a>        <span class="comment">// Verify the transformed RecordBatch correctly reads id from the file
+<a href=#1059 id=1059 data-nosnippet>1059</a>        // (NOT as a constant from partition metadata)
+<a href=#1060 id=1060 data-nosnippet>1060</a>        </span><span class="macro">assert_eq!</span>(result.num_columns(), <span class="number">2</span>);
+<a href=#1061 id=1061 data-nosnippet>1061</a>        <span class="macro">assert_eq!</span>(result.num_rows(), <span class="number">3</span>);
+<a href=#1062 id=1062 data-nosnippet>1062</a>
+<a href=#1063 id=1063 data-nosnippet>1063</a>        <span class="kw">let </span>id_column = result
+<a href=#1064 id=1064 data-nosnippet>1064</a>            .column(<span class="number">0</span>)
+<a href=#1065 id=1065 data-nosnippet>1065</a>            .as_any()
+<a href=#1066 id=1066 data-nosnippet>1066</a>            .downcast_ref::&lt;Int32Array&gt;()
+<a href=#1067 id=1067 data-nosnippet>1067</a>            .unwrap();
+<a href=#1068 id=1068 data-nosnippet>1068</a>        <span class="comment">// These values MUST come from the Parquet file, not be replaced by constants
+<a href=#1069 id=1069 data-nosnippet>1069</a>        </span><span class="macro">assert_eq!</span>(id_column.value(<span class="number">0</span>), <span class="number">100</span>);
+<a href=#1070 id=1070 data-nosnippet>1070</a>        <span class="macro">assert_eq!</span>(id_column.value(<span class="number">1</span>), <span class="number">200</span>);
+<a href=#1071 id=1071 data-nosnippet>1071</a>        <span class="macro">assert_eq!</span>(id_column.value(<span class="number">2</span>), <span class="number">300</span>);
+<a href=#1072 id=1072 data-nosnippet>1072</a>
+<a href=#1073 id=1073 data-nosnippet>1073</a>        <span class="kw">let </span>name_column = result
+<a href=#1074 id=1074 data-nosnippet>1074</a>            .column(<span class="number">1</span>)
+<a href=#1075 id=1075 data-nosnippet>1075</a>            .as_any()
+<a href=#1076 id=1076 data-nosnippet>1076</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#1077 id=1077 data-nosnippet>1077</a>            .unwrap();
+<a href=#1078 id=1078 data-nosnippet>1078</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
+<a href=#1079 id=1079 data-nosnippet>1079</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
+<a href=#1080 id=1080 data-nosnippet>1080</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">2</span>), <span class="string">"Charlie"</span>);
+<a href=#1081 id=1081 data-nosnippet>1081</a>    }
+<a href=#1082 id=1082 data-nosnippet>1082</a>
+<a href=#1083 id=1083 data-nosnippet>1083</a>    <span class="doccomment">/// Test that identity-transformed partition fields ARE treated as constants.
+<a href=#1084 id=1084 data-nosnippet>1084</a>    ///
+<a href=#1085 id=1085 data-nosnippet>1085</a>    /// This is the complement to `bucket_partitioning_reads_source_column_from_file`,
+<a href=#1086 id=1086 data-nosnippet>1086</a>    /// verifying that constants_map() correctly identifies identity-transformed
+<a href=#1087 id=1087 data-nosnippet>1087</a>    /// partition fields per the Iceberg spec.
+<a href=#1088 id=1088 data-nosnippet>1088</a>    ///
+<a href=#1089 id=1089 data-nosnippet>1089</a>    /// # Spec requirement (format/spec.md "Column Projection")
+<a href=#1090 id=1090 data-nosnippet>1090</a>    ///
+<a href=#1091 id=1091 data-nosnippet>1091</a>    /// &gt; "Return the value from partition metadata if an Identity Transform exists for the field
+<a href=#1092 id=1092 data-nosnippet>1092</a>    /// &gt;  and the partition value is present in the `partition` struct on `data_file` object
+<a href=#1093 id=1093 data-nosnippet>1093</a>    /// &gt;  in the manifest. This allows for metadata only migrations of Hive tables."
+<a href=#1094 id=1094 data-nosnippet>1094</a>    ///
+<a href=#1095 id=1095 data-nosnippet>1095</a>    /// # Why identity transforms use constants
+<a href=#1096 id=1096 data-nosnippet>1096</a>    ///
+<a href=#1097 id=1097 data-nosnippet>1097</a>    /// Unlike bucket/truncate/year/etc., identity transforms don't modify the value:
+<a href=#1098 id=1098 data-nosnippet>1098</a>    /// - `identity(dept)` stores the actual `dept` value in partition metadata
+<a href=#1099 id=1099 data-nosnippet>1099</a>    /// - Partition metadata has `dept = "engineering"` (the real value, not a hash/bucket)
+<a href=#1100 id=1100 data-nosnippet>1100</a>    /// - This value can be used directly without reading the data file
+<a href=#1101 id=1101 data-nosnippet>1101</a>    ///
+<a href=#1102 id=1102 data-nosnippet>1102</a>    /// # Performance benefit
+<a href=#1103 id=1103 data-nosnippet>1103</a>    ///
+<a href=#1104 id=1104 data-nosnippet>1104</a>    /// For Hive migrations where partition columns aren't in data files:
+<a href=#1105 id=1105 data-nosnippet>1105</a>    /// - Partition metadata provides the column values
+<a href=#1106 id=1106 data-nosnippet>1106</a>    /// - No need to read from data files (metadata-only query optimization)
+<a href=#1107 id=1107 data-nosnippet>1107</a>    /// - Common pattern: `dept=engineering/subdept=backend/file.parquet`
+<a href=#1108 id=1108 data-nosnippet>1108</a>    ///   - `dept` and `subdept` are in directory structure, not in `file.parquet`
+<a href=#1109 id=1109 data-nosnippet>1109</a>    ///   - Iceberg populates these from partition metadata as constants
+<a href=#1110 id=1110 data-nosnippet>1110</a>    ///
+<a href=#1111 id=1111 data-nosnippet>1111</a>    /// # What this test verifies
+<a href=#1112 id=1112 data-nosnippet>1112</a>    ///
+<a href=#1113 id=1113 data-nosnippet>1113</a>    /// - Identity-partitioned fields use constants from partition metadata
+<a href=#1114 id=1114 data-nosnippet>1114</a>    /// - The `dept` column is populated with `"engineering"` (not read from file)
+<a href=#1115 id=1115 data-nosnippet>1115</a>    /// - Java's `PartitionUtil.constantsMap()` behavior is matched:
+<a href=#1116 id=1116 data-nosnippet>1116</a>    ///   ```java
+<a href=#1117 id=1117 data-nosnippet>1117</a>    ///   if (field.transform().isIdentity()) {  // TRUE for identity
+<a href=#1118 id=1118 data-nosnippet>1118</a>    ///       idToConstant.put(field.sourceId(), converted);
+<a href=#1119 id=1119 data-nosnippet>1119</a>    ///   }
+<a href=#1120 id=1120 data-nosnippet>1120</a>    ///   ```
+<a href=#1121 id=1121 data-nosnippet>1121</a>    ///
+<a href=#1122 id=1122 data-nosnippet>1122</a>    /// # References
+<a href=#1123 id=1123 data-nosnippet>1123</a>    /// - Iceberg spec: format/spec.md "Column Projection"
+<a href=#1124 id=1124 data-nosnippet>1124</a>    /// - Java impl: core/src/main/java/org/apache/iceberg/util/PartitionUtil.java
+<a href=#1125 id=1125 data-nosnippet>1125</a>    </span><span class="attr">#[test]
+<a href=#1126 id=1126 data-nosnippet>1126</a>    </span><span class="kw">fn </span>identity_partition_uses_constant_from_metadata() {
+<a href=#1127 id=1127 data-nosnippet>1127</a>        <span class="kw">use </span><span class="kw">crate</span>::spec::{Struct, Transform};
+<a href=#1128 id=1128 data-nosnippet>1128</a>
+<a href=#1129 id=1129 data-nosnippet>1129</a>        <span class="comment">// Table schema: id (data column), dept (partition column), name (data column)
+<a href=#1130 id=1130 data-nosnippet>1130</a>        </span><span class="kw">let </span>snapshot_schema = Arc::new(
+<a href=#1131 id=1131 data-nosnippet>1131</a>            Schema::builder()
+<a href=#1132 id=1132 data-nosnippet>1132</a>                .with_schema_id(<span class="number">0</span>)
+<a href=#1133 id=1133 data-nosnippet>1133</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#1134 id=1134 data-nosnippet>1134</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#1135 id=1135 data-nosnippet>1135</a>                    NestedField::required(<span class="number">2</span>, <span class="string">"dept"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#1136 id=1136 data-nosnippet>1136</a>                    NestedField::optional(<span class="number">3</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#1137 id=1137 data-nosnippet>1137</a>                ])
+<a href=#1138 id=1138 data-nosnippet>1138</a>                .build()
+<a href=#1139 id=1139 data-nosnippet>1139</a>                .unwrap(),
+<a href=#1140 id=1140 data-nosnippet>1140</a>        );
+<a href=#1141 id=1141 data-nosnippet>1141</a>
+<a href=#1142 id=1142 data-nosnippet>1142</a>        <span class="comment">// Partition spec: identity(dept) - the dept field uses identity transform
+<a href=#1143 id=1143 data-nosnippet>1143</a>        </span><span class="kw">let </span>partition_spec = Arc::new(
+<a href=#1144 id=1144 data-nosnippet>1144</a>            <span class="kw">crate</span>::spec::PartitionSpec::builder(snapshot_schema.clone())
+<a href=#1145 id=1145 data-nosnippet>1145</a>                .with_spec_id(<span class="number">0</span>)
+<a href=#1146 id=1146 data-nosnippet>1146</a>                .add_partition_field(<span class="string">"dept"</span>, <span class="string">"dept"</span>, Transform::Identity)
+<a href=#1147 id=1147 data-nosnippet>1147</a>                .unwrap()
+<a href=#1148 id=1148 data-nosnippet>1148</a>                .build()
+<a href=#1149 id=1149 data-nosnippet>1149</a>                .unwrap(),
+<a href=#1150 id=1150 data-nosnippet>1150</a>        );
+<a href=#1151 id=1151 data-nosnippet>1151</a>
+<a href=#1152 id=1152 data-nosnippet>1152</a>        <span class="comment">// Partition data: dept="engineering"
+<a href=#1153 id=1153 data-nosnippet>1153</a>        </span><span class="kw">let </span>partition_data = Struct::from_iter(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(Literal::string(<span class="string">"engineering"</span>))]);
+<a href=#1154 id=1154 data-nosnippet>1154</a>
+<a href=#1155 id=1155 data-nosnippet>1155</a>        <span class="comment">// Parquet file contains only id and name (dept is in partition path)
+<a href=#1156 id=1156 data-nosnippet>1156</a>        </span><span class="kw">let </span>parquet_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#1157 id=1157 data-nosnippet>1157</a>            simple_field(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"1"</span>),
+<a href=#1158 id=1158 data-nosnippet>1158</a>            simple_field(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"3"</span>),
+<a href=#1159 id=1159 data-nosnippet>1159</a>        ]));
+<a href=#1160 id=1160 data-nosnippet>1160</a>
+<a href=#1161 id=1161 data-nosnippet>1161</a>        <span class="kw">let </span>projected_field_ids = [<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>]; <span class="comment">// id, dept, name
+<a href=#1162 id=1162 data-nosnippet>1162</a>
+<a href=#1163 id=1163 data-nosnippet>1163</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>transformer =
+<a href=#1164 id=1164 data-nosnippet>1164</a>            RecordBatchTransformerBuilder::new(snapshot_schema, <span class="kw-2">&amp;</span>projected_field_ids)
+<a href=#1165 id=1165 data-nosnippet>1165</a>                .with_partition(partition_spec, partition_data)
+<a href=#1166 id=1166 data-nosnippet>1166</a>                .build();
+<a href=#1167 id=1167 data-nosnippet>1167</a>
+<a href=#1168 id=1168 data-nosnippet>1168</a>        <span class="kw">let </span>parquet_batch = RecordBatch::try_new(parquet_schema, <span class="macro">vec!</span>[
+<a href=#1169 id=1169 data-nosnippet>1169</a>            Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">100</span>, <span class="number">200</span>])),
+<a href=#1170 id=1170 data-nosnippet>1170</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>])),
+<a href=#1171 id=1171 data-nosnippet>1171</a>        ])
+<a href=#1172 id=1172 data-nosnippet>1172</a>        .unwrap();
+<a href=#1173 id=1173 data-nosnippet>1173</a>
+<a href=#1174 id=1174 data-nosnippet>1174</a>        <span class="kw">let </span>result = transformer.process_record_batch(parquet_batch).unwrap();
+<a href=#1175 id=1175 data-nosnippet>1175</a>
+<a href=#1176 id=1176 data-nosnippet>1176</a>        <span class="comment">// Verify the dept column is populated with the constant from partition metadata
+<a href=#1177 id=1177 data-nosnippet>1177</a>        </span><span class="macro">assert_eq!</span>(result.num_columns(), <span class="number">3</span>);
+<a href=#1178 id=1178 data-nosnippet>1178</a>        <span class="macro">assert_eq!</span>(result.num_rows(), <span class="number">2</span>);
+<a href=#1179 id=1179 data-nosnippet>1179</a>
+<a href=#1180 id=1180 data-nosnippet>1180</a>        <span class="kw">let </span>id_column = result
+<a href=#1181 id=1181 data-nosnippet>1181</a>            .column(<span class="number">0</span>)
+<a href=#1182 id=1182 data-nosnippet>1182</a>            .as_any()
+<a href=#1183 id=1183 data-nosnippet>1183</a>            .downcast_ref::&lt;Int32Array&gt;()
+<a href=#1184 id=1184 data-nosnippet>1184</a>            .unwrap();
+<a href=#1185 id=1185 data-nosnippet>1185</a>        <span class="macro">assert_eq!</span>(id_column.value(<span class="number">0</span>), <span class="number">100</span>);
+<a href=#1186 id=1186 data-nosnippet>1186</a>        <span class="macro">assert_eq!</span>(id_column.value(<span class="number">1</span>), <span class="number">200</span>);
+<a href=#1187 id=1187 data-nosnippet>1187</a>
+<a href=#1188 id=1188 data-nosnippet>1188</a>        <span class="kw">let </span>dept_column = result
+<a href=#1189 id=1189 data-nosnippet>1189</a>            .column(<span class="number">1</span>)
+<a href=#1190 id=1190 data-nosnippet>1190</a>            .as_any()
+<a href=#1191 id=1191 data-nosnippet>1191</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#1192 id=1192 data-nosnippet>1192</a>            .unwrap();
+<a href=#1193 id=1193 data-nosnippet>1193</a>        <span class="comment">// This value MUST come from partition metadata (constant)
+<a href=#1194 id=1194 data-nosnippet>1194</a>        </span><span class="macro">assert_eq!</span>(dept_column.value(<span class="number">0</span>), <span class="string">"engineering"</span>);
+<a href=#1195 id=1195 data-nosnippet>1195</a>        <span class="macro">assert_eq!</span>(dept_column.value(<span class="number">1</span>), <span class="string">"engineering"</span>);
+<a href=#1196 id=1196 data-nosnippet>1196</a>
+<a href=#1197 id=1197 data-nosnippet>1197</a>        <span class="kw">let </span>name_column = result
+<a href=#1198 id=1198 data-nosnippet>1198</a>            .column(<span class="number">2</span>)
+<a href=#1199 id=1199 data-nosnippet>1199</a>            .as_any()
+<a href=#1200 id=1200 data-nosnippet>1200</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#1201 id=1201 data-nosnippet>1201</a>            .unwrap();
+<a href=#1202 id=1202 data-nosnippet>1202</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
+<a href=#1203 id=1203 data-nosnippet>1203</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
+<a href=#1204 id=1204 data-nosnippet>1204</a>    }
+<a href=#1205 id=1205 data-nosnippet>1205</a>
+<a href=#1206 id=1206 data-nosnippet>1206</a>    <span class="doccomment">/// Test bucket partitioning with renamed source column.
+<a href=#1207 id=1207 data-nosnippet>1207</a>    ///
+<a href=#1208 id=1208 data-nosnippet>1208</a>    /// This verifies correct behavior for TestRuntimeFiltering.testRenamedSourceColumnTable() in Iceberg Java.
+<a href=#1209 id=1209 data-nosnippet>1209</a>    /// When a source column is renamed after partitioning is established, field-ID-based mapping
+<a href=#1210 id=1210 data-nosnippet>1210</a>    /// must still correctly identify the column in Parquet files.
+<a href=#1211 id=1211 data-nosnippet>1211</a>    ///
+<a href=#1212 id=1212 data-nosnippet>1212</a>    /// # Scenario
+<a href=#1213 id=1213 data-nosnippet>1213</a>    ///
+<a href=#1214 id=1214 data-nosnippet>1214</a>    /// 1. Table created with `bucket(4, id)` partitioning
+<a href=#1215 id=1215 data-nosnippet>1215</a>    /// 2. Data written to Parquet files (field_id=1, name="id")
+<a href=#1216 id=1216 data-nosnippet>1216</a>    /// 3. Column renamed: `ALTER TABLE ... RENAME COLUMN id TO row_id`
+<a href=#1217 id=1217 data-nosnippet>1217</a>    /// 4. Iceberg schema now has: field_id=1, name="row_id"
+<a href=#1218 id=1218 data-nosnippet>1218</a>    /// 5. Parquet files still have: field_id=1, name="id"
+<a href=#1219 id=1219 data-nosnippet>1219</a>    ///
+<a href=#1220 id=1220 data-nosnippet>1220</a>    /// # Expected Behavior Per Iceberg Spec
+<a href=#1221 id=1221 data-nosnippet>1221</a>    ///
+<a href=#1222 id=1222 data-nosnippet>1222</a>    /// Per the Iceberg spec "Column Projection" section and Java's PartitionUtil.constantsMap():
+<a href=#1223 id=1223 data-nosnippet>1223</a>    /// - Bucket transforms are NON-identity, so partition metadata stores bucket numbers (0-3), not source values
+<a href=#1224 id=1224 data-nosnippet>1224</a>    /// - Source columns for non-identity transforms MUST be read from data files
+<a href=#1225 id=1225 data-nosnippet>1225</a>    /// - Field-ID-based mapping should find the column by field_id=1 (ignoring name mismatch)
+<a href=#1226 id=1226 data-nosnippet>1226</a>    /// - Runtime filtering on `row_id` should work correctly
+<a href=#1227 id=1227 data-nosnippet>1227</a>    ///
+<a href=#1228 id=1228 data-nosnippet>1228</a>    /// # What This Tests
+<a href=#1229 id=1229 data-nosnippet>1229</a>    ///
+<a href=#1230 id=1230 data-nosnippet>1230</a>    /// This test ensures that when FileScanTask provides partition_spec and partition_data:
+<a href=#1231 id=1231 data-nosnippet>1231</a>    /// - constants_map() correctly identifies that bucket(4, row_id) is NOT an identity transform
+<a href=#1232 id=1232 data-nosnippet>1232</a>    /// - The source column (field_id=1) is NOT added to constants_map
+<a href=#1233 id=1233 data-nosnippet>1233</a>    /// - Field-ID-based mapping reads actual values from the Parquet file
+<a href=#1234 id=1234 data-nosnippet>1234</a>    /// - Values [100, 200, 300] are read, not replaced with bucket constant 2
+<a href=#1235 id=1235 data-nosnippet>1235</a>    ///
+<a href=#1236 id=1236 data-nosnippet>1236</a>    /// # References
+<a href=#1237 id=1237 data-nosnippet>1237</a>    /// - Java test: spark/src/test/java/.../TestRuntimeFiltering.java::testRenamedSourceColumnTable
+<a href=#1238 id=1238 data-nosnippet>1238</a>    /// - Java impl: core/src/main/java/org/apache/iceberg/util/PartitionUtil.java::constantsMap()
+<a href=#1239 id=1239 data-nosnippet>1239</a>    /// - Iceberg spec: format/spec.md "Column Projection" section
+<a href=#1240 id=1240 data-nosnippet>1240</a>    </span><span class="attr">#[test]
+<a href=#1241 id=1241 data-nosnippet>1241</a>    </span><span class="kw">fn </span>test_bucket_partitioning_with_renamed_source_column() {
+<a href=#1242 id=1242 data-nosnippet>1242</a>        <span class="kw">use </span><span class="kw">crate</span>::spec::{Struct, Transform};
+<a href=#1243 id=1243 data-nosnippet>1243</a>
+<a href=#1244 id=1244 data-nosnippet>1244</a>        <span class="comment">// Iceberg schema after rename: row_id (was id), name
+<a href=#1245 id=1245 data-nosnippet>1245</a>        </span><span class="kw">let </span>snapshot_schema = Arc::new(
+<a href=#1246 id=1246 data-nosnippet>1246</a>            Schema::builder()
+<a href=#1247 id=1247 data-nosnippet>1247</a>                .with_schema_id(<span class="number">0</span>)
+<a href=#1248 id=1248 data-nosnippet>1248</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#1249 id=1249 data-nosnippet>1249</a>                    NestedField::required(<span class="number">1</span>, <span class="string">"row_id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#1250 id=1250 data-nosnippet>1250</a>                    NestedField::optional(<span class="number">2</span>, <span class="string">"name"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#1251 id=1251 data-nosnippet>1251</a>                ])
+<a href=#1252 id=1252 data-nosnippet>1252</a>                .build()
+<a href=#1253 id=1253 data-nosnippet>1253</a>                .unwrap(),
+<a href=#1254 id=1254 data-nosnippet>1254</a>        );
+<a href=#1255 id=1255 data-nosnippet>1255</a>
+<a href=#1256 id=1256 data-nosnippet>1256</a>        <span class="comment">// Partition spec: bucket(4, row_id) - but source_id still points to field_id=1
+<a href=#1257 id=1257 data-nosnippet>1257</a>        </span><span class="kw">let </span>partition_spec = Arc::new(
+<a href=#1258 id=1258 data-nosnippet>1258</a>            <span class="kw">crate</span>::spec::PartitionSpec::builder(snapshot_schema.clone())
+<a href=#1259 id=1259 data-nosnippet>1259</a>                .with_spec_id(<span class="number">0</span>)
+<a href=#1260 id=1260 data-nosnippet>1260</a>                .add_partition_field(<span class="string">"row_id"</span>, <span class="string">"row_id_bucket"</span>, Transform::Bucket(<span class="number">4</span>))
+<a href=#1261 id=1261 data-nosnippet>1261</a>                .unwrap()
+<a href=#1262 id=1262 data-nosnippet>1262</a>                .build()
+<a href=#1263 id=1263 data-nosnippet>1263</a>                .unwrap(),
+<a href=#1264 id=1264 data-nosnippet>1264</a>        );
+<a href=#1265 id=1265 data-nosnippet>1265</a>
+<a href=#1266 id=1266 data-nosnippet>1266</a>        <span class="comment">// Partition data: bucket value is 2
+<a href=#1267 id=1267 data-nosnippet>1267</a>        </span><span class="kw">let </span>partition_data = Struct::from_iter(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(Literal::int(<span class="number">2</span>))]);
+<a href=#1268 id=1268 data-nosnippet>1268</a>
+<a href=#1269 id=1269 data-nosnippet>1269</a>        <span class="comment">// Parquet file has OLD column name "id" but SAME field_id=1
+<a href=#1270 id=1270 data-nosnippet>1270</a>        // Field-ID-based mapping should find this despite name mismatch
+<a href=#1271 id=1271 data-nosnippet>1271</a>        </span><span class="kw">let </span>parquet_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#1272 id=1272 data-nosnippet>1272</a>            simple_field(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"1"</span>),
+<a href=#1273 id=1273 data-nosnippet>1273</a>            simple_field(<span class="string">"name"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"2"</span>),
+<a href=#1274 id=1274 data-nosnippet>1274</a>        ]));
+<a href=#1275 id=1275 data-nosnippet>1275</a>
+<a href=#1276 id=1276 data-nosnippet>1276</a>        <span class="kw">let </span>projected_field_ids = [<span class="number">1</span>, <span class="number">2</span>]; <span class="comment">// row_id (field_id=1), name (field_id=2)
+<a href=#1277 id=1277 data-nosnippet>1277</a>
+<a href=#1278 id=1278 data-nosnippet>1278</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>transformer =
+<a href=#1279 id=1279 data-nosnippet>1279</a>            RecordBatchTransformerBuilder::new(snapshot_schema, <span class="kw-2">&amp;</span>projected_field_ids)
+<a href=#1280 id=1280 data-nosnippet>1280</a>                .with_partition(partition_spec, partition_data)
+<a href=#1281 id=1281 data-nosnippet>1281</a>                .build();
+<a href=#1282 id=1282 data-nosnippet>1282</a>
+<a href=#1283 id=1283 data-nosnippet>1283</a>        <span class="comment">// Create a Parquet RecordBatch with actual data
+<a href=#1284 id=1284 data-nosnippet>1284</a>        // Despite column rename, data should be read via field_id=1
+<a href=#1285 id=1285 data-nosnippet>1285</a>        </span><span class="kw">let </span>parquet_batch = RecordBatch::try_new(parquet_schema, <span class="macro">vec!</span>[
+<a href=#1286 id=1286 data-nosnippet>1286</a>            Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>])),
+<a href=#1287 id=1287 data-nosnippet>1287</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"Alice"</span>, <span class="string">"Bob"</span>, <span class="string">"Charlie"</span>])),
+<a href=#1288 id=1288 data-nosnippet>1288</a>        ])
+<a href=#1289 id=1289 data-nosnippet>1289</a>        .unwrap();
+<a href=#1290 id=1290 data-nosnippet>1290</a>
+<a href=#1291 id=1291 data-nosnippet>1291</a>        <span class="kw">let </span>result = transformer.process_record_batch(parquet_batch).unwrap();
+<a href=#1292 id=1292 data-nosnippet>1292</a>
+<a href=#1293 id=1293 data-nosnippet>1293</a>        <span class="comment">// Verify the transformed RecordBatch correctly reads data despite name mismatch
+<a href=#1294 id=1294 data-nosnippet>1294</a>        </span><span class="macro">assert_eq!</span>(result.num_columns(), <span class="number">2</span>);
+<a href=#1295 id=1295 data-nosnippet>1295</a>        <span class="macro">assert_eq!</span>(result.num_rows(), <span class="number">3</span>);
+<a href=#1296 id=1296 data-nosnippet>1296</a>
+<a href=#1297 id=1297 data-nosnippet>1297</a>        <span class="kw">let </span>row_id_column = result
+<a href=#1298 id=1298 data-nosnippet>1298</a>            .column(<span class="number">0</span>)
+<a href=#1299 id=1299 data-nosnippet>1299</a>            .as_any()
+<a href=#1300 id=1300 data-nosnippet>1300</a>            .downcast_ref::&lt;Int32Array&gt;()
+<a href=#1301 id=1301 data-nosnippet>1301</a>            .unwrap();
+<a href=#1302 id=1302 data-nosnippet>1302</a>        <span class="comment">// These values MUST come from the Parquet file via field_id=1,
+<a href=#1303 id=1303 data-nosnippet>1303</a>        // not be replaced by the bucket constant (2)
+<a href=#1304 id=1304 data-nosnippet>1304</a>        </span><span class="macro">assert_eq!</span>(row_id_column.value(<span class="number">0</span>), <span class="number">100</span>);
+<a href=#1305 id=1305 data-nosnippet>1305</a>        <span class="macro">assert_eq!</span>(row_id_column.value(<span class="number">1</span>), <span class="number">200</span>);
+<a href=#1306 id=1306 data-nosnippet>1306</a>        <span class="macro">assert_eq!</span>(row_id_column.value(<span class="number">2</span>), <span class="number">300</span>);
+<a href=#1307 id=1307 data-nosnippet>1307</a>
+<a href=#1308 id=1308 data-nosnippet>1308</a>        <span class="kw">let </span>name_column = result
+<a href=#1309 id=1309 data-nosnippet>1309</a>            .column(<span class="number">1</span>)
+<a href=#1310 id=1310 data-nosnippet>1310</a>            .as_any()
+<a href=#1311 id=1311 data-nosnippet>1311</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#1312 id=1312 data-nosnippet>1312</a>            .unwrap();
+<a href=#1313 id=1313 data-nosnippet>1313</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">0</span>), <span class="string">"Alice"</span>);
+<a href=#1314 id=1314 data-nosnippet>1314</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">1</span>), <span class="string">"Bob"</span>);
+<a href=#1315 id=1315 data-nosnippet>1315</a>        <span class="macro">assert_eq!</span>(name_column.value(<span class="number">2</span>), <span class="string">"Charlie"</span>);
+<a href=#1316 id=1316 data-nosnippet>1316</a>    }
+<a href=#1317 id=1317 data-nosnippet>1317</a>
+<a href=#1318 id=1318 data-nosnippet>1318</a>    <span class="doccomment">/// Comprehensive integration test that verifies all 4 Iceberg spec rules work correctly.
+<a href=#1319 id=1319 data-nosnippet>1319</a>    ///
+<a href=#1320 id=1320 data-nosnippet>1320</a>    /// Per the Iceberg spec (https://iceberg.apache.org/spec/#column-projection),
+<a href=#1321 id=1321 data-nosnippet>1321</a>    /// "Values for field ids which are not present in a data file must be resolved
+<a href=#1322 id=1322 data-nosnippet>1322</a>    /// according the following rules:"
+<a href=#1323 id=1323 data-nosnippet>1323</a>    ///
+<a href=#1324 id=1324 data-nosnippet>1324</a>    /// This test creates a scenario where each rule is exercised:
+<a href=#1325 id=1325 data-nosnippet>1325</a>    /// - Rule #1: dept (identity-partitioned) -&gt; constant from partition metadata
+<a href=#1326 id=1326 data-nosnippet>1326</a>    /// - Rule #2: data (via name mapping) -&gt; read from Parquet file by name
+<a href=#1327 id=1327 data-nosnippet>1327</a>    /// - Rule #3: category (initial_default) -&gt; use default value
+<a href=#1328 id=1328 data-nosnippet>1328</a>    /// - Rule #4: notes (no default) -&gt; return null
+<a href=#1329 id=1329 data-nosnippet>1329</a>    ///
+<a href=#1330 id=1330 data-nosnippet>1330</a>    /// # References
+<a href=#1331 id=1331 data-nosnippet>1331</a>    /// - Iceberg spec: format/spec.md "Column Projection" section
+<a href=#1332 id=1332 data-nosnippet>1332</a>    </span><span class="attr">#[test]
+<a href=#1333 id=1333 data-nosnippet>1333</a>    </span><span class="kw">fn </span>test_all_four_spec_rules() {
+<a href=#1334 id=1334 data-nosnippet>1334</a>        <span class="kw">use </span><span class="kw">crate</span>::spec::Transform;
+<a href=#1335 id=1335 data-nosnippet>1335</a>
+<a href=#1336 id=1336 data-nosnippet>1336</a>        <span class="comment">// Iceberg schema with columns designed to exercise each spec rule
+<a href=#1337 id=1337 data-nosnippet>1337</a>        </span><span class="kw">let </span>snapshot_schema = Arc::new(
+<a href=#1338 id=1338 data-nosnippet>1338</a>            Schema::builder()
+<a href=#1339 id=1339 data-nosnippet>1339</a>                .with_schema_id(<span class="number">0</span>)
+<a href=#1340 id=1340 data-nosnippet>1340</a>                .with_fields(<span class="macro">vec!</span>[
+<a href=#1341 id=1341 data-nosnippet>1341</a>                    <span class="comment">// Field in Parquet by field ID (normal case)
+<a href=#1342 id=1342 data-nosnippet>1342</a>                    </span>NestedField::required(<span class="number">1</span>, <span class="string">"id"</span>, Type::Primitive(PrimitiveType::Int)).into(),
+<a href=#1343 id=1343 data-nosnippet>1343</a>                    <span class="comment">// Rule #1: Identity-partitioned field - should use partition metadata
+<a href=#1344 id=1344 data-nosnippet>1344</a>                    </span>NestedField::required(<span class="number">2</span>, <span class="string">"dept"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#1345 id=1345 data-nosnippet>1345</a>                    <span class="comment">// Rule #2: Field resolved by name mapping (ArrowReader already applied)
+<a href=#1346 id=1346 data-nosnippet>1346</a>                    </span>NestedField::required(<span class="number">3</span>, <span class="string">"data"</span>, Type::Primitive(PrimitiveType::String)).into(),
+<a href=#1347 id=1347 data-nosnippet>1347</a>                    <span class="comment">// Rule #3: Field with initial_default
+<a href=#1348 id=1348 data-nosnippet>1348</a>                    </span>NestedField::optional(<span class="number">4</span>, <span class="string">"category"</span>, Type::Primitive(PrimitiveType::String))
+<a href=#1349 id=1349 data-nosnippet>1349</a>                        .with_initial_default(Literal::string(<span class="string">"default_category"</span>))
+<a href=#1350 id=1350 data-nosnippet>1350</a>                        .into(),
+<a href=#1351 id=1351 data-nosnippet>1351</a>                    <span class="comment">// Rule #4: Field with no default - should be null
+<a href=#1352 id=1352 data-nosnippet>1352</a>                    </span>NestedField::optional(<span class="number">5</span>, <span class="string">"notes"</span>, Type::Primitive(PrimitiveType::String))
+<a href=#1353 id=1353 data-nosnippet>1353</a>                        .into(),
+<a href=#1354 id=1354 data-nosnippet>1354</a>                ])
+<a href=#1355 id=1355 data-nosnippet>1355</a>                .build()
+<a href=#1356 id=1356 data-nosnippet>1356</a>                .unwrap(),
+<a href=#1357 id=1357 data-nosnippet>1357</a>        );
+<a href=#1358 id=1358 data-nosnippet>1358</a>
+<a href=#1359 id=1359 data-nosnippet>1359</a>        <span class="comment">// Partition spec: identity transform on dept
+<a href=#1360 id=1360 data-nosnippet>1360</a>        </span><span class="kw">let </span>partition_spec = Arc::new(
+<a href=#1361 id=1361 data-nosnippet>1361</a>            <span class="kw">crate</span>::spec::PartitionSpec::builder(snapshot_schema.clone())
+<a href=#1362 id=1362 data-nosnippet>1362</a>                .with_spec_id(<span class="number">0</span>)
+<a href=#1363 id=1363 data-nosnippet>1363</a>                .add_partition_field(<span class="string">"dept"</span>, <span class="string">"dept"</span>, Transform::Identity)
+<a href=#1364 id=1364 data-nosnippet>1364</a>                .unwrap()
+<a href=#1365 id=1365 data-nosnippet>1365</a>                .build()
+<a href=#1366 id=1366 data-nosnippet>1366</a>                .unwrap(),
+<a href=#1367 id=1367 data-nosnippet>1367</a>        );
+<a href=#1368 id=1368 data-nosnippet>1368</a>
+<a href=#1369 id=1369 data-nosnippet>1369</a>        <span class="comment">// Partition data: dept="engineering"
+<a href=#1370 id=1370 data-nosnippet>1370</a>        </span><span class="kw">let </span>partition_data = Struct::from_iter(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(Literal::string(<span class="string">"engineering"</span>))]);
+<a href=#1371 id=1371 data-nosnippet>1371</a>
+<a href=#1372 id=1372 data-nosnippet>1372</a>        <span class="comment">// Parquet schema: simulates post-ArrowReader state where name mapping already applied
+<a href=#1373 id=1373 data-nosnippet>1373</a>        // Has id (field_id=1) and data (field_id=3, assigned by ArrowReader via name mapping)
+<a href=#1374 id=1374 data-nosnippet>1374</a>        // Missing: dept (in partition), category (has default), notes (no default)
+<a href=#1375 id=1375 data-nosnippet>1375</a>        </span><span class="kw">let </span>parquet_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#1376 id=1376 data-nosnippet>1376</a>            simple_field(<span class="string">"id"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"1"</span>),
+<a href=#1377 id=1377 data-nosnippet>1377</a>            simple_field(<span class="string">"data"</span>, DataType::Utf8, <span class="bool-val">false</span>, <span class="string">"3"</span>),
+<a href=#1378 id=1378 data-nosnippet>1378</a>        ]));
+<a href=#1379 id=1379 data-nosnippet>1379</a>
+<a href=#1380 id=1380 data-nosnippet>1380</a>        <span class="kw">let </span>projected_field_ids = [<span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>, <span class="number">4</span>, <span class="number">5</span>]; <span class="comment">// id, dept, data, category, notes
+<a href=#1381 id=1381 data-nosnippet>1381</a>
+<a href=#1382 id=1382 data-nosnippet>1382</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>transformer =
+<a href=#1383 id=1383 data-nosnippet>1383</a>            RecordBatchTransformerBuilder::new(snapshot_schema, <span class="kw-2">&amp;</span>projected_field_ids)
+<a href=#1384 id=1384 data-nosnippet>1384</a>                .with_partition(partition_spec, partition_data)
+<a href=#1385 id=1385 data-nosnippet>1385</a>                .build();
+<a href=#1386 id=1386 data-nosnippet>1386</a>
+<a href=#1387 id=1387 data-nosnippet>1387</a>        <span class="kw">let </span>parquet_batch = RecordBatch::try_new(parquet_schema, <span class="macro">vec!</span>[
+<a href=#1388 id=1388 data-nosnippet>1388</a>            Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">100</span>, <span class="number">200</span>])),
+<a href=#1389 id=1389 data-nosnippet>1389</a>            Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="string">"value1"</span>, <span class="string">"value2"</span>])),
+<a href=#1390 id=1390 data-nosnippet>1390</a>        ])
+<a href=#1391 id=1391 data-nosnippet>1391</a>        .unwrap();
+<a href=#1392 id=1392 data-nosnippet>1392</a>
+<a href=#1393 id=1393 data-nosnippet>1393</a>        <span class="kw">let </span>result = transformer.process_record_batch(parquet_batch).unwrap();
+<a href=#1394 id=1394 data-nosnippet>1394</a>
+<a href=#1395 id=1395 data-nosnippet>1395</a>        <span class="macro">assert_eq!</span>(result.num_columns(), <span class="number">5</span>);
+<a href=#1396 id=1396 data-nosnippet>1396</a>        <span class="macro">assert_eq!</span>(result.num_rows(), <span class="number">2</span>);
+<a href=#1397 id=1397 data-nosnippet>1397</a>
+<a href=#1398 id=1398 data-nosnippet>1398</a>        <span class="comment">// Verify each column demonstrates the correct spec rule:
+<a href=#1399 id=1399 data-nosnippet>1399</a>
+<a href=#1400 id=1400 data-nosnippet>1400</a>        // Normal case: id from Parquet by field ID
+<a href=#1401 id=1401 data-nosnippet>1401</a>        </span><span class="kw">let </span>id_column = result
+<a href=#1402 id=1402 data-nosnippet>1402</a>            .column(<span class="number">0</span>)
+<a href=#1403 id=1403 data-nosnippet>1403</a>            .as_any()
+<a href=#1404 id=1404 data-nosnippet>1404</a>            .downcast_ref::&lt;Int32Array&gt;()
+<a href=#1405 id=1405 data-nosnippet>1405</a>            .unwrap();
+<a href=#1406 id=1406 data-nosnippet>1406</a>        <span class="macro">assert_eq!</span>(id_column.value(<span class="number">0</span>), <span class="number">100</span>);
+<a href=#1407 id=1407 data-nosnippet>1407</a>        <span class="macro">assert_eq!</span>(id_column.value(<span class="number">1</span>), <span class="number">200</span>);
+<a href=#1408 id=1408 data-nosnippet>1408</a>
+<a href=#1409 id=1409 data-nosnippet>1409</a>        <span class="comment">// Rule #1: dept from partition metadata (identity transform)
+<a href=#1410 id=1410 data-nosnippet>1410</a>        </span><span class="kw">let </span>dept_column = result
+<a href=#1411 id=1411 data-nosnippet>1411</a>            .column(<span class="number">1</span>)
+<a href=#1412 id=1412 data-nosnippet>1412</a>            .as_any()
+<a href=#1413 id=1413 data-nosnippet>1413</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#1414 id=1414 data-nosnippet>1414</a>            .unwrap();
+<a href=#1415 id=1415 data-nosnippet>1415</a>        <span class="macro">assert_eq!</span>(dept_column.value(<span class="number">0</span>), <span class="string">"engineering"</span>);
+<a href=#1416 id=1416 data-nosnippet>1416</a>        <span class="macro">assert_eq!</span>(dept_column.value(<span class="number">1</span>), <span class="string">"engineering"</span>);
+<a href=#1417 id=1417 data-nosnippet>1417</a>
+<a href=#1418 id=1418 data-nosnippet>1418</a>        <span class="comment">// Rule #2: data from Parquet via name mapping
+<a href=#1419 id=1419 data-nosnippet>1419</a>        </span><span class="kw">let </span>data_column = result
+<a href=#1420 id=1420 data-nosnippet>1420</a>            .column(<span class="number">2</span>)
+<a href=#1421 id=1421 data-nosnippet>1421</a>            .as_any()
+<a href=#1422 id=1422 data-nosnippet>1422</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#1423 id=1423 data-nosnippet>1423</a>            .unwrap();
+<a href=#1424 id=1424 data-nosnippet>1424</a>        <span class="macro">assert_eq!</span>(data_column.value(<span class="number">0</span>), <span class="string">"value1"</span>);
+<a href=#1425 id=1425 data-nosnippet>1425</a>        <span class="macro">assert_eq!</span>(data_column.value(<span class="number">1</span>), <span class="string">"value2"</span>);
+<a href=#1426 id=1426 data-nosnippet>1426</a>
+<a href=#1427 id=1427 data-nosnippet>1427</a>        <span class="comment">// Rule #3: category from initial_default
+<a href=#1428 id=1428 data-nosnippet>1428</a>        </span><span class="kw">let </span>category_column = result
+<a href=#1429 id=1429 data-nosnippet>1429</a>            .column(<span class="number">3</span>)
+<a href=#1430 id=1430 data-nosnippet>1430</a>            .as_any()
+<a href=#1431 id=1431 data-nosnippet>1431</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#1432 id=1432 data-nosnippet>1432</a>            .unwrap();
+<a href=#1433 id=1433 data-nosnippet>1433</a>        <span class="macro">assert_eq!</span>(category_column.value(<span class="number">0</span>), <span class="string">"default_category"</span>);
+<a href=#1434 id=1434 data-nosnippet>1434</a>        <span class="macro">assert_eq!</span>(category_column.value(<span class="number">1</span>), <span class="string">"default_category"</span>);
+<a href=#1435 id=1435 data-nosnippet>1435</a>
+<a href=#1436 id=1436 data-nosnippet>1436</a>        <span class="comment">// Rule #4: notes is null (no default, not in Parquet, not in partition)
+<a href=#1437 id=1437 data-nosnippet>1437</a>        </span><span class="kw">let </span>notes_column = result
+<a href=#1438 id=1438 data-nosnippet>1438</a>            .column(<span class="number">4</span>)
+<a href=#1439 id=1439 data-nosnippet>1439</a>            .as_any()
+<a href=#1440 id=1440 data-nosnippet>1440</a>            .downcast_ref::&lt;StringArray&gt;()
+<a href=#1441 id=1441 data-nosnippet>1441</a>            .unwrap();
+<a href=#1442 id=1442 data-nosnippet>1442</a>        <span class="macro">assert!</span>(notes_column.is_null(<span class="number">0</span>));
+<a href=#1443 id=1443 data-nosnippet>1443</a>        <span class="macro">assert!</span>(notes_column.is_null(<span class="number">1</span>));
+<a href=#1444 id=1444 data-nosnippet>1444</a>    }
+<a href=#1445 id=1445 data-nosnippet>1445</a>}</code></pre></div></section></main></body></html>
\ No newline at end of file
diff --git a/api/src/iceberg/scan/context.rs.html b/api/src/iceberg/scan/context.rs.html
index 0995c062..154a6c11 100644
--- a/api/src/iceberg/scan/context.rs.html
+++ b/api/src/iceberg/scan/context.rs.html
@@ -128,138 +128,145 @@
 <a href=#128 id=128 data-nosnippet>128</a>                .map(|x| x.as_ref().snapshot_bound_predicate.clone()),
 <a href=#129 id=129 data-nosnippet>129</a>
 <a href=#130 id=130 data-nosnippet>130</a>            deletes,
-<a href=#131 id=131 data-nosnippet>131</a>        })
-<a href=#132 id=132 data-nosnippet>132</a>    }
-<a href=#133 id=133 data-nosnippet>133</a>}
-<a href=#134 id=134 data-nosnippet>134</a>
-<a href=#135 id=135 data-nosnippet>135</a><span class="doccomment">/// PlanContext wraps a [`SnapshotRef`] alongside all the other
-<a href=#136 id=136 data-nosnippet>136</a>/// objects that are required to perform a scan file plan.
-<a href=#137 id=137 data-nosnippet>137</a></span><span class="attr">#[derive(Debug)]
-<a href=#138 id=138 data-nosnippet>138</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">struct </span>PlanContext {
-<a href=#139 id=139 data-nosnippet>139</a>    <span class="kw">pub </span>snapshot: SnapshotRef,
-<a href=#140 id=140 data-nosnippet>140</a>
-<a href=#141 id=141 data-nosnippet>141</a>    <span class="kw">pub </span>table_metadata: TableMetadataRef,
-<a href=#142 id=142 data-nosnippet>142</a>    <span class="kw">pub </span>snapshot_schema: SchemaRef,
-<a href=#143 id=143 data-nosnippet>143</a>    <span class="kw">pub </span>case_sensitive: bool,
-<a href=#144 id=144 data-nosnippet>144</a>    <span class="kw">pub </span>predicate: <span class="prelude-ty">Option</span>&lt;Arc&lt;Predicate&gt;&gt;,
-<a href=#145 id=145 data-nosnippet>145</a>    <span class="kw">pub </span>snapshot_bound_predicate: <span class="prelude-ty">Option</span>&lt;Arc&lt;BoundPredicate&gt;&gt;,
-<a href=#146 id=146 data-nosnippet>146</a>    <span class="kw">pub </span>object_cache: Arc&lt;ObjectCache&gt;,
-<a href=#147 id=147 data-nosnippet>147</a>    <span class="kw">pub </span>field_ids: Arc&lt;Vec&lt;i32&gt;&gt;,
-<a href=#148 id=148 data-nosnippet>148</a>
-<a href=#149 id=149 data-nosnippet>149</a>    <span class="kw">pub </span>partition_filter_cache: Arc&lt;PartitionFilterCache&gt;,
-<a href=#150 id=150 data-nosnippet>150</a>    <span class="kw">pub </span>manifest_evaluator_cache: Arc&lt;ManifestEvaluatorCache&gt;,
-<a href=#151 id=151 data-nosnippet>151</a>    <span class="kw">pub </span>expression_evaluator_cache: Arc&lt;ExpressionEvaluatorCache&gt;,
-<a href=#152 id=152 data-nosnippet>152</a>}
-<a href=#153 id=153 data-nosnippet>153</a>
-<a href=#154 id=154 data-nosnippet>154</a><span class="kw">impl </span>PlanContext {
-<a href=#155 id=155 data-nosnippet>155</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">async fn </span>get_manifest_list(<span class="kw-2">&amp;</span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Arc&lt;ManifestList&gt;&gt; {
-<a href=#156 id=156 data-nosnippet>156</a>        <span class="self">self</span>.object_cache
-<a href=#157 id=157 data-nosnippet>157</a>            .as_ref()
-<a href=#158 id=158 data-nosnippet>158</a>            .get_manifest_list(<span class="kw-2">&amp;</span><span class="self">self</span>.snapshot, <span class="kw-2">&amp;</span><span class="self">self</span>.table_metadata)
-<a href=#159 id=159 data-nosnippet>159</a>            .<span class="kw">await
-<a href=#160 id=160 data-nosnippet>160</a>    </span>}
-<a href=#161 id=161 data-nosnippet>161</a>
-<a href=#162 id=162 data-nosnippet>162</a>    <span class="kw">fn </span>get_partition_filter(<span class="kw-2">&amp;</span><span class="self">self</span>, manifest_file: <span class="kw-2">&amp;</span>ManifestFile) -&gt; <span class="prelude-ty">Result</span>&lt;Arc&lt;BoundPredicate&gt;&gt; {
-<a href=#163 id=163 data-nosnippet>163</a>        <span class="kw">let </span>partition_spec_id = manifest_file.partition_spec_id;
-<a href=#164 id=164 data-nosnippet>164</a>
-<a href=#165 id=165 data-nosnippet>165</a>        <span class="kw">let </span>partition_filter = <span class="self">self</span>.partition_filter_cache.get(
-<a href=#166 id=166 data-nosnippet>166</a>            partition_spec_id,
-<a href=#167 id=167 data-nosnippet>167</a>            <span class="kw-2">&amp;</span><span class="self">self</span>.table_metadata,
-<a href=#168 id=168 data-nosnippet>168</a>            <span class="kw-2">&amp;</span><span class="self">self</span>.snapshot_schema,
-<a href=#169 id=169 data-nosnippet>169</a>            <span class="self">self</span>.case_sensitive,
-<a href=#170 id=170 data-nosnippet>170</a>            <span class="self">self</span>.predicate
-<a href=#171 id=171 data-nosnippet>171</a>                .as_ref()
-<a href=#172 id=172 data-nosnippet>172</a>                .ok_or(Error::new(
-<a href=#173 id=173 data-nosnippet>173</a>                    ErrorKind::Unexpected,
-<a href=#174 id=174 data-nosnippet>174</a>                    <span class="string">"Expected a predicate but none present"</span>,
-<a href=#175 id=175 data-nosnippet>175</a>                ))<span class="question-mark">?
-<a href=#176 id=176 data-nosnippet>176</a>                </span>.as_ref()
-<a href=#177 id=177 data-nosnippet>177</a>                .bind(<span class="self">self</span>.snapshot_schema.clone(), <span class="self">self</span>.case_sensitive)<span class="question-mark">?</span>,
-<a href=#178 id=178 data-nosnippet>178</a>        )<span class="question-mark">?</span>;
-<a href=#179 id=179 data-nosnippet>179</a>
-<a href=#180 id=180 data-nosnippet>180</a>        <span class="prelude-val">Ok</span>(partition_filter)
-<a href=#181 id=181 data-nosnippet>181</a>    }
-<a href=#182 id=182 data-nosnippet>182</a>
-<a href=#183 id=183 data-nosnippet>183</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>build_manifest_file_contexts(
-<a href=#184 id=184 data-nosnippet>184</a>        <span class="kw-2">&amp;</span><span class="self">self</span>,
-<a href=#185 id=185 data-nosnippet>185</a>        manifest_list: Arc&lt;ManifestList&gt;,
-<a href=#186 id=186 data-nosnippet>186</a>        tx_data: Sender&lt;ManifestEntryContext&gt;,
-<a href=#187 id=187 data-nosnippet>187</a>        delete_file_idx: DeleteFileIndex,
-<a href=#188 id=188 data-nosnippet>188</a>        delete_file_tx: Sender&lt;ManifestEntryContext&gt;,
-<a href=#189 id=189 data-nosnippet>189</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;<span class="kw">impl </span>Iterator&lt;Item = <span class="prelude-ty">Result</span>&lt;ManifestFileContext&gt;&gt; + <span class="lifetime">'static</span>&gt;&gt; {
-<a href=#190 id=190 data-nosnippet>190</a>        <span class="kw">let </span>manifest_files = manifest_list.entries().iter();
-<a href=#191 id=191 data-nosnippet>191</a>
-<a href=#192 id=192 data-nosnippet>192</a>        <span class="comment">// TODO: Ideally we could ditch this intermediate Vec as we return an iterator.
-<a href=#193 id=193 data-nosnippet>193</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>filtered_mfcs = <span class="macro">vec!</span>[];
-<a href=#194 id=194 data-nosnippet>194</a>        <span class="kw">for </span>manifest_file <span class="kw">in </span>manifest_files {
-<a href=#195 id=195 data-nosnippet>195</a>            <span class="kw">let </span>tx = <span class="kw">if </span>manifest_file.content == ManifestContentType::Deletes {
-<a href=#196 id=196 data-nosnippet>196</a>                delete_file_tx.clone()
-<a href=#197 id=197 data-nosnippet>197</a>            } <span class="kw">else </span>{
-<a href=#198 id=198 data-nosnippet>198</a>                tx_data.clone()
-<a href=#199 id=199 data-nosnippet>199</a>            };
-<a href=#200 id=200 data-nosnippet>200</a>
-<a href=#201 id=201 data-nosnippet>201</a>            <span class="kw">let </span>partition_bound_predicate = <span class="kw">if </span><span class="self">self</span>.predicate.is_some() {
-<a href=#202 id=202 data-nosnippet>202</a>                <span class="kw">let </span>partition_bound_predicate = <span class="self">self</span>.get_partition_filter(manifest_file)<span class="question-mark">?</span>;
-<a href=#203 id=203 data-nosnippet>203</a>
-<a href=#204 id=204 data-nosnippet>204</a>                <span class="comment">// evaluate the ManifestFile against the partition filter. Skip
-<a href=#205 id=205 data-nosnippet>205</a>                // if it cannot contain any matching rows
-<a href=#206 id=206 data-nosnippet>206</a>                </span><span class="kw">if </span>!<span class="self">self
-<a href=#207 id=207 data-nosnippet>207</a>                    </span>.manifest_evaluator_cache
-<a href=#208 id=208 data-nosnippet>208</a>                    .get(
-<a href=#209 id=209 data-nosnippet>209</a>                        manifest_file.partition_spec_id,
-<a href=#210 id=210 data-nosnippet>210</a>                        partition_bound_predicate.clone(),
-<a href=#211 id=211 data-nosnippet>211</a>                    )
-<a href=#212 id=212 data-nosnippet>212</a>                    .eval(manifest_file)<span class="question-mark">?
-<a href=#213 id=213 data-nosnippet>213</a>                </span>{
-<a href=#214 id=214 data-nosnippet>214</a>                    <span class="kw">continue</span>;
-<a href=#215 id=215 data-nosnippet>215</a>                }
-<a href=#216 id=216 data-nosnippet>216</a>
-<a href=#217 id=217 data-nosnippet>217</a>                <span class="prelude-val">Some</span>(partition_bound_predicate)
-<a href=#218 id=218 data-nosnippet>218</a>            } <span class="kw">else </span>{
-<a href=#219 id=219 data-nosnippet>219</a>                <span class="prelude-val">None
-<a href=#220 id=220 data-nosnippet>220</a>            </span>};
-<a href=#221 id=221 data-nosnippet>221</a>
-<a href=#222 id=222 data-nosnippet>222</a>            <span class="kw">let </span>mfc = <span class="self">self</span>.create_manifest_file_context(
-<a href=#223 id=223 data-nosnippet>223</a>                manifest_file,
-<a href=#224 id=224 data-nosnippet>224</a>                partition_bound_predicate,
-<a href=#225 id=225 data-nosnippet>225</a>                tx,
-<a href=#226 id=226 data-nosnippet>226</a>                delete_file_idx.clone(),
-<a href=#227 id=227 data-nosnippet>227</a>            );
+<a href=#131 id=131 data-nosnippet>131</a>
+<a href=#132 id=132 data-nosnippet>132</a>            <span class="comment">// Include partition data and spec from manifest entry
+<a href=#133 id=133 data-nosnippet>133</a>            </span>partition: <span class="prelude-val">Some</span>(<span class="self">self</span>.manifest_entry.data_file.partition.clone()),
+<a href=#134 id=134 data-nosnippet>134</a>            <span class="comment">// TODO: Pass actual PartitionSpec through context chain for native flow
+<a href=#135 id=135 data-nosnippet>135</a>            </span>partition_spec: <span class="prelude-val">None</span>,
+<a href=#136 id=136 data-nosnippet>136</a>            <span class="comment">// TODO: Extract name_mapping from table metadata property "schema.name-mapping.default"
+<a href=#137 id=137 data-nosnippet>137</a>            </span>name_mapping: <span class="prelude-val">None</span>,
+<a href=#138 id=138 data-nosnippet>138</a>        })
+<a href=#139 id=139 data-nosnippet>139</a>    }
+<a href=#140 id=140 data-nosnippet>140</a>}
+<a href=#141 id=141 data-nosnippet>141</a>
+<a href=#142 id=142 data-nosnippet>142</a><span class="doccomment">/// PlanContext wraps a [`SnapshotRef`] alongside all the other
+<a href=#143 id=143 data-nosnippet>143</a>/// objects that are required to perform a scan file plan.
+<a href=#144 id=144 data-nosnippet>144</a></span><span class="attr">#[derive(Debug)]
+<a href=#145 id=145 data-nosnippet>145</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">struct </span>PlanContext {
+<a href=#146 id=146 data-nosnippet>146</a>    <span class="kw">pub </span>snapshot: SnapshotRef,
+<a href=#147 id=147 data-nosnippet>147</a>
+<a href=#148 id=148 data-nosnippet>148</a>    <span class="kw">pub </span>table_metadata: TableMetadataRef,
+<a href=#149 id=149 data-nosnippet>149</a>    <span class="kw">pub </span>snapshot_schema: SchemaRef,
+<a href=#150 id=150 data-nosnippet>150</a>    <span class="kw">pub </span>case_sensitive: bool,
+<a href=#151 id=151 data-nosnippet>151</a>    <span class="kw">pub </span>predicate: <span class="prelude-ty">Option</span>&lt;Arc&lt;Predicate&gt;&gt;,
+<a href=#152 id=152 data-nosnippet>152</a>    <span class="kw">pub </span>snapshot_bound_predicate: <span class="prelude-ty">Option</span>&lt;Arc&lt;BoundPredicate&gt;&gt;,
+<a href=#153 id=153 data-nosnippet>153</a>    <span class="kw">pub </span>object_cache: Arc&lt;ObjectCache&gt;,
+<a href=#154 id=154 data-nosnippet>154</a>    <span class="kw">pub </span>field_ids: Arc&lt;Vec&lt;i32&gt;&gt;,
+<a href=#155 id=155 data-nosnippet>155</a>
+<a href=#156 id=156 data-nosnippet>156</a>    <span class="kw">pub </span>partition_filter_cache: Arc&lt;PartitionFilterCache&gt;,
+<a href=#157 id=157 data-nosnippet>157</a>    <span class="kw">pub </span>manifest_evaluator_cache: Arc&lt;ManifestEvaluatorCache&gt;,
+<a href=#158 id=158 data-nosnippet>158</a>    <span class="kw">pub </span>expression_evaluator_cache: Arc&lt;ExpressionEvaluatorCache&gt;,
+<a href=#159 id=159 data-nosnippet>159</a>}
+<a href=#160 id=160 data-nosnippet>160</a>
+<a href=#161 id=161 data-nosnippet>161</a><span class="kw">impl </span>PlanContext {
+<a href=#162 id=162 data-nosnippet>162</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">async fn </span>get_manifest_list(<span class="kw-2">&amp;</span><span class="self">self</span>) -&gt; <span class="prelude-ty">Result</span>&lt;Arc&lt;ManifestList&gt;&gt; {
+<a href=#163 id=163 data-nosnippet>163</a>        <span class="self">self</span>.object_cache
+<a href=#164 id=164 data-nosnippet>164</a>            .as_ref()
+<a href=#165 id=165 data-nosnippet>165</a>            .get_manifest_list(<span class="kw-2">&amp;</span><span class="self">self</span>.snapshot, <span class="kw-2">&amp;</span><span class="self">self</span>.table_metadata)
+<a href=#166 id=166 data-nosnippet>166</a>            .<span class="kw">await
+<a href=#167 id=167 data-nosnippet>167</a>    </span>}
+<a href=#168 id=168 data-nosnippet>168</a>
+<a href=#169 id=169 data-nosnippet>169</a>    <span class="kw">fn </span>get_partition_filter(<span class="kw-2">&amp;</span><span class="self">self</span>, manifest_file: <span class="kw-2">&amp;</span>ManifestFile) -&gt; <span class="prelude-ty">Result</span>&lt;Arc&lt;BoundPredicate&gt;&gt; {
+<a href=#170 id=170 data-nosnippet>170</a>        <span class="kw">let </span>partition_spec_id = manifest_file.partition_spec_id;
+<a href=#171 id=171 data-nosnippet>171</a>
+<a href=#172 id=172 data-nosnippet>172</a>        <span class="kw">let </span>partition_filter = <span class="self">self</span>.partition_filter_cache.get(
+<a href=#173 id=173 data-nosnippet>173</a>            partition_spec_id,
+<a href=#174 id=174 data-nosnippet>174</a>            <span class="kw-2">&amp;</span><span class="self">self</span>.table_metadata,
+<a href=#175 id=175 data-nosnippet>175</a>            <span class="kw-2">&amp;</span><span class="self">self</span>.snapshot_schema,
+<a href=#176 id=176 data-nosnippet>176</a>            <span class="self">self</span>.case_sensitive,
+<a href=#177 id=177 data-nosnippet>177</a>            <span class="self">self</span>.predicate
+<a href=#178 id=178 data-nosnippet>178</a>                .as_ref()
+<a href=#179 id=179 data-nosnippet>179</a>                .ok_or(Error::new(
+<a href=#180 id=180 data-nosnippet>180</a>                    ErrorKind::Unexpected,
+<a href=#181 id=181 data-nosnippet>181</a>                    <span class="string">"Expected a predicate but none present"</span>,
+<a href=#182 id=182 data-nosnippet>182</a>                ))<span class="question-mark">?
+<a href=#183 id=183 data-nosnippet>183</a>                </span>.as_ref()
+<a href=#184 id=184 data-nosnippet>184</a>                .bind(<span class="self">self</span>.snapshot_schema.clone(), <span class="self">self</span>.case_sensitive)<span class="question-mark">?</span>,
+<a href=#185 id=185 data-nosnippet>185</a>        )<span class="question-mark">?</span>;
+<a href=#186 id=186 data-nosnippet>186</a>
+<a href=#187 id=187 data-nosnippet>187</a>        <span class="prelude-val">Ok</span>(partition_filter)
+<a href=#188 id=188 data-nosnippet>188</a>    }
+<a href=#189 id=189 data-nosnippet>189</a>
+<a href=#190 id=190 data-nosnippet>190</a>    <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>build_manifest_file_contexts(
+<a href=#191 id=191 data-nosnippet>191</a>        <span class="kw-2">&amp;</span><span class="self">self</span>,
+<a href=#192 id=192 data-nosnippet>192</a>        manifest_list: Arc&lt;ManifestList&gt;,
+<a href=#193 id=193 data-nosnippet>193</a>        tx_data: Sender&lt;ManifestEntryContext&gt;,
+<a href=#194 id=194 data-nosnippet>194</a>        delete_file_idx: DeleteFileIndex,
+<a href=#195 id=195 data-nosnippet>195</a>        delete_file_tx: Sender&lt;ManifestEntryContext&gt;,
+<a href=#196 id=196 data-nosnippet>196</a>    ) -&gt; <span class="prelude-ty">Result</span>&lt;Box&lt;<span class="kw">impl </span>Iterator&lt;Item = <span class="prelude-ty">Result</span>&lt;ManifestFileContext&gt;&gt; + <span class="lifetime">'static</span>&gt;&gt; {
+<a href=#197 id=197 data-nosnippet>197</a>        <span class="kw">let </span>manifest_files = manifest_list.entries().iter();
+<a href=#198 id=198 data-nosnippet>198</a>
+<a href=#199 id=199 data-nosnippet>199</a>        <span class="comment">// TODO: Ideally we could ditch this intermediate Vec as we return an iterator.
+<a href=#200 id=200 data-nosnippet>200</a>        </span><span class="kw">let </span><span class="kw-2">mut </span>filtered_mfcs = <span class="macro">vec!</span>[];
+<a href=#201 id=201 data-nosnippet>201</a>        <span class="kw">for </span>manifest_file <span class="kw">in </span>manifest_files {
+<a href=#202 id=202 data-nosnippet>202</a>            <span class="kw">let </span>tx = <span class="kw">if </span>manifest_file.content == ManifestContentType::Deletes {
+<a href=#203 id=203 data-nosnippet>203</a>                delete_file_tx.clone()
+<a href=#204 id=204 data-nosnippet>204</a>            } <span class="kw">else </span>{
+<a href=#205 id=205 data-nosnippet>205</a>                tx_data.clone()
+<a href=#206 id=206 data-nosnippet>206</a>            };
+<a href=#207 id=207 data-nosnippet>207</a>
+<a href=#208 id=208 data-nosnippet>208</a>            <span class="kw">let </span>partition_bound_predicate = <span class="kw">if </span><span class="self">self</span>.predicate.is_some() {
+<a href=#209 id=209 data-nosnippet>209</a>                <span class="kw">let </span>partition_bound_predicate = <span class="self">self</span>.get_partition_filter(manifest_file)<span class="question-mark">?</span>;
+<a href=#210 id=210 data-nosnippet>210</a>
+<a href=#211 id=211 data-nosnippet>211</a>                <span class="comment">// evaluate the ManifestFile against the partition filter. Skip
+<a href=#212 id=212 data-nosnippet>212</a>                // if it cannot contain any matching rows
+<a href=#213 id=213 data-nosnippet>213</a>                </span><span class="kw">if </span>!<span class="self">self
+<a href=#214 id=214 data-nosnippet>214</a>                    </span>.manifest_evaluator_cache
+<a href=#215 id=215 data-nosnippet>215</a>                    .get(
+<a href=#216 id=216 data-nosnippet>216</a>                        manifest_file.partition_spec_id,
+<a href=#217 id=217 data-nosnippet>217</a>                        partition_bound_predicate.clone(),
+<a href=#218 id=218 data-nosnippet>218</a>                    )
+<a href=#219 id=219 data-nosnippet>219</a>                    .eval(manifest_file)<span class="question-mark">?
+<a href=#220 id=220 data-nosnippet>220</a>                </span>{
+<a href=#221 id=221 data-nosnippet>221</a>                    <span class="kw">continue</span>;
+<a href=#222 id=222 data-nosnippet>222</a>                }
+<a href=#223 id=223 data-nosnippet>223</a>
+<a href=#224 id=224 data-nosnippet>224</a>                <span class="prelude-val">Some</span>(partition_bound_predicate)
+<a href=#225 id=225 data-nosnippet>225</a>            } <span class="kw">else </span>{
+<a href=#226 id=226 data-nosnippet>226</a>                <span class="prelude-val">None
+<a href=#227 id=227 data-nosnippet>227</a>            </span>};
 <a href=#228 id=228 data-nosnippet>228</a>
-<a href=#229 id=229 data-nosnippet>229</a>            filtered_mfcs.push(<span class="prelude-val">Ok</span>(mfc));
-<a href=#230 id=230 data-nosnippet>230</a>        }
-<a href=#231 id=231 data-nosnippet>231</a>
-<a href=#232 id=232 data-nosnippet>232</a>        <span class="prelude-val">Ok</span>(Box::new(filtered_mfcs.into_iter()))
-<a href=#233 id=233 data-nosnippet>233</a>    }
-<a href=#234 id=234 data-nosnippet>234</a>
-<a href=#235 id=235 data-nosnippet>235</a>    <span class="kw">fn </span>create_manifest_file_context(
-<a href=#236 id=236 data-nosnippet>236</a>        <span class="kw-2">&amp;</span><span class="self">self</span>,
-<a href=#237 id=237 data-nosnippet>237</a>        manifest_file: <span class="kw-2">&amp;</span>ManifestFile,
-<a href=#238 id=238 data-nosnippet>238</a>        partition_filter: <span class="prelude-ty">Option</span>&lt;Arc&lt;BoundPredicate&gt;&gt;,
-<a href=#239 id=239 data-nosnippet>239</a>        sender: Sender&lt;ManifestEntryContext&gt;,
-<a href=#240 id=240 data-nosnippet>240</a>        delete_file_index: DeleteFileIndex,
-<a href=#241 id=241 data-nosnippet>241</a>    ) -&gt; ManifestFileContext {
-<a href=#242 id=242 data-nosnippet>242</a>        <span class="kw">let </span>bound_predicates =
-<a href=#243 id=243 data-nosnippet>243</a>            <span class="kw">if let </span>(<span class="prelude-val">Some</span>(<span class="kw-2">ref </span>partition_bound_predicate), <span class="prelude-val">Some</span>(snapshot_bound_predicate)) =
-<a href=#244 id=244 data-nosnippet>244</a>                (partition_filter, <span class="kw-2">&amp;</span><span class="self">self</span>.snapshot_bound_predicate)
-<a href=#245 id=245 data-nosnippet>245</a>            {
-<a href=#246 id=246 data-nosnippet>246</a>                <span class="prelude-val">Some</span>(Arc::new(BoundPredicates {
-<a href=#247 id=247 data-nosnippet>247</a>                    partition_bound_predicate: partition_bound_predicate.as_ref().clone(),
-<a href=#248 id=248 data-nosnippet>248</a>                    snapshot_bound_predicate: snapshot_bound_predicate.as_ref().clone(),
-<a href=#249 id=249 data-nosnippet>249</a>                }))
-<a href=#250 id=250 data-nosnippet>250</a>            } <span class="kw">else </span>{
-<a href=#251 id=251 data-nosnippet>251</a>                <span class="prelude-val">None
-<a href=#252 id=252 data-nosnippet>252</a>            </span>};
-<a href=#253 id=253 data-nosnippet>253</a>
-<a href=#254 id=254 data-nosnippet>254</a>        ManifestFileContext {
-<a href=#255 id=255 data-nosnippet>255</a>            manifest_file: manifest_file.clone(),
-<a href=#256 id=256 data-nosnippet>256</a>            bound_predicates,
-<a href=#257 id=257 data-nosnippet>257</a>            sender,
-<a href=#258 id=258 data-nosnippet>258</a>            object_cache: <span class="self">self</span>.object_cache.clone(),
-<a href=#259 id=259 data-nosnippet>259</a>            snapshot_schema: <span class="self">self</span>.snapshot_schema.clone(),
-<a href=#260 id=260 data-nosnippet>260</a>            field_ids: <span class="self">self</span>.field_ids.clone(),
-<a href=#261 id=261 data-nosnippet>261</a>            expression_evaluator_cache: <span class="self">self</span>.expression_evaluator_cache.clone(),
-<a href=#262 id=262 data-nosnippet>262</a>            delete_file_index,
-<a href=#263 id=263 data-nosnippet>263</a>        }
... 357 lines suppressed ...