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 2026/07/08 07:18:21 UTC
(iceberg-rust) branch gh-pages updated: deploy: 22c256e985a68b3a71dd3890d3ebd6bedc3e7b0d
This is an automated email from the ASF dual-hosted git repository.
github-actions[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 3253512be deploy: 22c256e985a68b3a71dd3890d3ebd6bedc3e7b0d
3253512be is described below
commit 3253512be65fa1f7e9b82fd646cbed1cae75e8f5
Author: blackmwk <bl...@users.noreply.github.com>
AuthorDate: Wed Jul 8 07:18:15 2026 +0000
deploy: 22c256e985a68b3a71dd3890d3ebd6bedc3e7b0d
---
api/iceberg_catalog_hms/struct.HmsCatalog.html | 96 +-
.../struct.HmsCatalogBuilder.html | 4 +-
api/search.index/path/0e7338738470.js | 1 -
api/search.index/path/ed9380482964.js | 1 +
api/search.index/root.js | 2 +-
.../arrow/caching_delete_file_loader.rs.html | 2233 +++++++++++---------
api/trait.impl/iceberg/catalog/trait.Catalog.js | 4 +-
.../iceberg/catalog/trait.CatalogBuilder.js | 4 +-
8 files changed, 1233 insertions(+), 1112 deletions(-)
diff --git a/api/iceberg_catalog_hms/struct.HmsCatalog.html b/api/iceberg_catalog_hms/struct.HmsCatalog.html
index 8e75213c2..c75e3d33f 100644
--- a/api/iceberg_catalog_hms/struct.HmsCatalog.html
+++ b/api/iceberg_catalog_hms/struct.HmsCatalog.html
@@ -1,19 +1,19 @@
<!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="Hive metastore Catalog."><title>HmsCatalog in iceberg_catalog_hms - 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-Medi [...]
-</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-HmsCatalog" class="impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#195-247">Source</a><a href="#impl-HmsCatalog" class="anchor">§</a><h3 class="code-header">impl <a class="struct" href="struct.HmsCatalog.html" title="struc [...]
-</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-Catalog-for-HmsCatalog" class="impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#250-715">Source</a><a href="#impl-Catalog-for-HmsCatalog" class="anchor">§</a><h3 class="code-hea [...]
+</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-HmsCatalog" class="impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#195-247">Source</a><a href="#impl-HmsCatalog" class="anchor">§</a><h3 class="code-header">impl <a class="struct" href="struct.HmsCatalog.html" title="struc [...]
+</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-Catalog-for-HmsCatalog" class="impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#250-715">Source</a><a href="#impl-Catalog-for-HmsCatalog" class="anchor">§</a><h3 class="code-hea [...]
&'life0 self,
- parent: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a><&'life1 NamespaceIdent>,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="struct" href="https://doc.rust-lang.org/nig [...]
+ parent: <a class="enum" href="https://doc.rust-lang.org/nightly/core/option/enum.Option.html" title="enum core::option::Option">Option</a><&'life1 <a class="struct" href="../iceberg/catalog/struct.NamespaceIdent.html" title="struct iceberg::catalog::NamespaceIdent">NamespaceIdent</a>>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>HMS doesn’t support nested namespaces.</p>
<p>We will return empty list if parent is some.</p>
<p>Align with java implementation: <a href="https://github.com/apache/iceberg/blob/9bd62f79f8cd973c39d14e89163cb1c707470ed2/hive-metastore/src/main/java/org/apache/iceberg/hive/HiveCatalog.java#L305C26-L330">https://github.com/apache/iceberg/blob/9bd62f79f8cd973c39d14e89163cb1c707470ed2/hive-metastore/src/main/java/org/apache/iceberg/hive/HiveCatalog.java#L305C26-L330</a></p>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.create_namespace" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#292-312">Source</a><a href="#method.create_namespace" class="anchor">§</a><h4 class="code-header">fn <a class="fn">create_namespace</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.create_namespace" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#292-312">Source</a><a href="#method.create_namespace" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.create_namespace" class="fn">create_namespace</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- namespace: &'life1 NamespaceIdent,
+ namespace: &'life1 <a class="struct" href="../iceberg/catalog/struct.NamespaceIdent.html" title="struct iceberg::catalog::NamespaceIdent">NamespaceIdent</a>,
properties: <a class="struct" href="https://doc.rust-lang.org/nightly/std/collections/hash/map/struct.HashMap.html" title="struct std::collections::hash::map::HashMap">HashMap</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>, <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>>,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<Namespace>> + <a class="trait" href="https://do [...]
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Creates a new namespace with the given identifier and properties.</p>
@@ -30,10 +30,10 @@ successfully converted into a database configuration.</li>
<li>Errors from the underlying database creation process, converted using
<code>from_thrift_error</code>.</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.get_namespace" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#324-354">Source</a><a href="#method.get_namespace" class="anchor">§</a><h4 class="code-header">fn <a class="fn">get_namespace</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.get_namespace" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#324-354">Source</a><a href="#method.get_namespace" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.get_namespace" class="fn">get_namespace</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- namespace: &'life1 NamespaceIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<Namespace>> + <a class="trait" href="https://do [...]
+ namespace: &'life1 <a class="struct" href="../iceberg/catalog/struct.NamespaceIdent.html" title="struct iceberg::catalog::NamespaceIdent">NamespaceIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Retrieves a namespace by its identifier.</p>
@@ -46,10 +46,10 @@ Constructs a <code>Namespace</code> object with the retrieved data and returns i
<li>If there is an error querying the database, returned by
<code>from_thrift_error</code>.</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.namespace_exists" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#368-385">Source</a><a href="#method.namespace_exists" class="anchor">§</a><h4 class="code-header">fn <a class="fn">namespace_exists</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.namespace_exists" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#368-385">Source</a><a href="#method.namespace_exists" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.namespace_exists" class="fn">namespace_exists</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- namespace: &'life1 NamespaceIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="primitive" href="https://doc.rust-lang.org/ [...]
+ namespace: &'life1 <a class="struct" href="../iceberg/catalog/struct.NamespaceIdent.html" title="struct iceberg::catalog::NamespaceIdent">NamespaceIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Checks if a namespace exists within the Hive Metastore.</p>
@@ -64,11 +64,11 @@ to determine if the specified namespace (database) exists.</p>
<li><code>Err(...)</code> if an error occurs during validation or the Hive Metastore
query, with the error encapsulating the issue.</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.update_namespace" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#397-427">Source</a><a href="#method.update_namespace" class="anchor">§</a><h4 class="code-header">fn <a class="fn">update_namespace</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.update_namespace" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#397-427">Source</a><a href="#method.update_namespace" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.update_namespace" class="fn">update_namespace</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- namespace: &'life1 NamespaceIdent,
+ namespace: &'life1 <a class="struct" href="../iceberg/catalog/struct.NamespaceIdent.html" title="struct iceberg::catalog::NamespaceIdent">NamespaceIdent</a>,
properties: <a class="struct" href="https://doc.rust-lang.org/nightly/std/collections/hash/map/struct.HashMap.html" title="struct std::collections::hash::map::HashMap">HashMap</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>, <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>>,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="primitive" href="https://doc.rust-lang.org/ [...]
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Asynchronously updates properties of an existing namespace.</p>
@@ -79,10 +79,10 @@ in the Hive Metastore.</p>
<p>Returns <code>Ok(())</code> if the namespace update is successful. If the
namespace cannot be updated due to missing information or an error
during the update process, an <code>Err(...)</code> is returned.</p>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.drop_namespace" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#436-453">Source</a><a href="#method.drop_namespace" class="anchor">§</a><h4 class="code-header">fn <a class="fn">drop_namespace</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.drop_namespace" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#436-453">Source</a><a href="#method.drop_namespace" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.drop_namespace" class="fn">drop_namespace</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- namespace: &'life1 NamespaceIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="primitive" href="https://doc.rust-lang.org/ [...]
+ namespace: &'life1 <a class="struct" href="../iceberg/catalog/struct.NamespaceIdent.html" title="struct iceberg::catalog::NamespaceIdent">NamespaceIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Asynchronously drops a namespace from the Hive Metastore.</p>
@@ -93,10 +93,10 @@ during the update process, an <code>Err(...)</code> is returned.</p>
<li><code>Err(...)</code> signifies failure to drop the namespace due to validation
errors, connectivity issues, or Hive Metastore constraints.</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.list_tables" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#464-487">Source</a><a href="#method.list_tables" class="anchor">§</a><h4 class="code-header">fn <a class="fn">list_tables</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.list_tables" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#464-487">Source</a><a href="#method.list_tables" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.list_tables" class="fn">list_tables</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- namespace: &'life1 NamespaceIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="struct" href="https://doc.rust-lang.org/nig [...]
+ namespace: &'life1 <a class="struct" href="../iceberg/catalog/struct.NamespaceIdent.html" title="struct iceberg::catalog::NamespaceIdent">NamespaceIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Asynchronously lists all tables within a specified namespace.</p>
@@ -108,11 +108,11 @@ representing a table within the specified namespace.</li>
<li><code>Err(...)</code> if an error occurs during namespace validation or while
querying the database.</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.create_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#501-549">Source</a><a href="#method.create_table" class="anchor">§</a><h4 class="code-header">fn <a class="fn">create_table</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.create_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#501-549">Source</a><a href="#method.create_table" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.create_table" class="fn">create_table</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- namespace: &'life1 NamespaceIdent,
- creation: TableCreation,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<Table>> + <a class="trait" href="https://doc.ru [...]
+ namespace: &'life1 <a class="struct" href="../iceberg/catalog/struct.NamespaceIdent.html" title="struct iceberg::catalog::NamespaceIdent">NamespaceIdent</a>,
+ creation: <a class="struct" href="../iceberg/catalog/struct.TableCreation.html" title="struct iceberg::catalog::TableCreation">TableCreation</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Creates a new table within a specified namespace using the provided
@@ -125,10 +125,10 @@ table.</p>
namespace identifiers, failure to determine a default storage location,
issues generating or writing table metadata, and errors communicating
with the Hive Metastore.</p>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.load_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#563-588">Source</a><a href="#method.load_table" class="anchor">§</a><h4 class="code-header">fn <a class="fn">load_table</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.load_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#563-588">Source</a><a href="#method.load_table" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.load_table" class="fn">load_table</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- table: &'life1 TableIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<Table>> + <a class="trait" href="https://doc.ru [...]
+ table: &'life1 <a class="struct" href="../iceberg/catalog/struct.TableIdent.html" title="struct iceberg::catalog::TableIdent">TableIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Loads a table from the Hive Metastore and constructs a <code>Table</code> object
@@ -143,10 +143,10 @@ based on its metadata.</p>
<li>Absence of metadata location information in the table’s properties.</li>
<li>Issues reading or deserializing the table’s metadata file.</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.drop_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#600-622">Source</a><a href="#method.drop_table" class="anchor">§</a><h4 class="code-header">fn <a class="fn">drop_table</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.drop_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#600-622">Source</a><a href="#method.drop_table" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.drop_table" class="fn">drop_table</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- table: &'life1 TableIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="primitive" href="https://doc.rust-lang.org/ [...]
+ table: &'life1 <a class="struct" href="../iceberg/catalog/struct.TableIdent.html" title="struct iceberg::catalog::TableIdent">TableIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Asynchronously drops a table from the database.</p>
@@ -160,10 +160,10 @@ attempting to drop the table. This includes scenarios where
the table does not exist.</li>
<li>Any network or communication error occurs with the database backend.</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.table_exists" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#637-657">Source</a><a href="#method.table_exists" class="anchor">§</a><h4 class="code-header">fn <a class="fn">table_exists</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.table_exists" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#637-657">Source</a><a href="#method.table_exists" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.table_exists" class="fn">table_exists</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- table: &'life1 TableIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="primitive" href="https://doc.rust-lang.org/ [...]
+ table: &'life1 <a class="struct" href="../iceberg/catalog/struct.TableIdent.html" title="struct iceberg::catalog::TableIdent">TableIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,</div></h4></section></summary><div class="docblock"><p>Asynchronously checks the existence of a specified table
@@ -174,11 +174,11 @@ in the database.</p>
<li><code>Ok(false)</code> if the table does not exist in the database.</li>
<li><code>Err(...)</code> if an error occurs during the process</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.rename_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#665-696">Source</a><a href="#method.rename_table" class="anchor">§</a><h4 class="code-header">fn <a class="fn">rename_table</a><'life0, 'life1, 'life2, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.rename_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#665-696">Source</a><a href="#method.rename_table" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.rename_table" class="fn">rename_table</a><'life0, 'life1, 'life2, 'async_trait>(
&'life0 self,
- src: &'life1 TableIdent,
- dest: &'life2 TableIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="primitive" href="https://doc.rust-lang.org/ [...]
+ src: &'life1 <a class="struct" href="../iceberg/catalog/struct.TableIdent.html" title="struct iceberg::catalog::TableIdent">TableIdent</a>,
+ dest: &'life2 <a class="struct" href="../iceberg/catalog/struct.TableIdent.html" title="struct iceberg::catalog::TableIdent">TableIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
'life1: 'async_trait,
@@ -189,23 +189,23 @@ or moves it between namespaces (databases).</p>
<li><code>Ok(())</code> on successful rename or move of the table.</li>
<li><code>Err(...)</code> if an error occurs during the process.</li>
</ul>
-</div></details><details class="toggle method-toggle" open><summary><section id="method.purge_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#624-628">Source</a><a href="#method.purge_table" class="anchor">§</a><h4 class="code-header">fn <a class="fn">purge_table</a><'life0, 'life1, 'async_trait>(
+</div></details><details class="toggle method-toggle" open><summary><section id="method.purge_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#624-628">Source</a><a href="#method.purge_table" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.purge_table" class="fn">purge_table</a><'life0, 'life1, 'async_trait>(
&'life0 self,
- table: &'life1 TableIdent,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<<a class="primitive" href="https://doc.rust-lang.org/ [...]
+ table: &'life1 <a class="struct" href="../iceberg/catalog/struct.TableIdent.html" title="struct iceberg::catalog::TableIdent">TableIdent</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
- 'life1: 'async_trait,</div></h4></section></summary><div class='docblock'>Drop a table from the catalog and delete the underlying table data. <a>Read more</a></div></details><details class="toggle method-toggle" open><summary><section id="method.register_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#698-707">Source</a><a href="#method.register_table" class="anchor">§</a><h4 class="code-header">fn <a class="fn">register_tabl [...]
+ 'life1: 'async_trait,</div></h4></section></summary><div class='docblock'>Drop a table from the catalog and delete the underlying table data. <a href="../iceberg/catalog/trait.Catalog.html#tymethod.purge_table">Read more</a></div></details><details class="toggle method-toggle" open><summary><section id="method.register_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#698-707">Source</a><a href="#method.register_table" class="a [...]
&'life0 self,
- _table_ident: &'life1 TableIdent,
+ _table_ident: &'life1 <a class="struct" href="../iceberg/catalog/struct.TableIdent.html" title="struct iceberg::catalog::TableIdent">TableIdent</a>,
_metadata_location: <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<Table>> + <a class="trait" href="https://doc.ru [...]
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,
- 'life1: 'async_trait,</div></h4></section></summary><div class='docblock'>Register an existing table to the catalog.</div></details><details class="toggle method-toggle" open><summary><section id="method.update_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#709-714">Source</a><a href="#method.update_table" class="anchor">§</a><h4 class="code-header">fn <a class="fn">update_table</a><'life0, 'async_trait>(
+ 'life1: 'async_trait,</div></h4></section></summary><div class='docblock'>Register an existing table to the catalog.</div></details><details class="toggle method-toggle" open><summary><section id="method.update_table" class="method trait-impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#709-714">Source</a><a href="#method.update_table" class="anchor">§</a><h4 class="code-header">fn <a href="../iceberg/catalog/trait.Catalog.html#tymethod.update_table" cla [...]
&'life0 self,
- _commit: TableCommit,
-) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<Table>> + <a class="trait" href="https://doc.ru [...]
+ _commit: <a class="struct" href="../iceberg/catalog/struct.TableCommit.html" title="struct iceberg::catalog::TableCommit">TableCommit</a>,
+) -> <a class="struct" href="https://doc.rust-lang.org/nightly/core/pin/struct.Pin.html" title="struct core::pin::Pin">Pin</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/boxed/struct.Box.html" title="struct alloc::boxed::Box">Box</a><dyn <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title= [...]
Self: 'async_trait,
'life0: 'async_trait,</div></h4></section></summary><div class='docblock'>Update a table to the catalog.</div></details></div></details><details class="toggle implementors-toggle" open><summary><section id="impl-Debug-for-HmsCatalog" class="impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#187-193">Source</a><a href="#impl-Debug-for-HmsCatalog" class="anchor">§</a><h3 class="code-header">impl <a class="trait" href="https://doc.rust-lang.org/nightly/core/ [...]
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#142">Source</a><a href="#method.type_id" class="anchor">§</a><h4 class="code-header">fn <a href [...]
diff --git a/api/iceberg_catalog_hms/struct.HmsCatalogBuilder.html b/api/iceberg_catalog_hms/struct.HmsCatalogBuilder.html
index 94b1f17f3..efa9e11fc 100644
--- a/api/iceberg_catalog_hms/struct.HmsCatalogBuilder.html
+++ b/api/iceberg_catalog_hms/struct.HmsCatalogBuilder.html
@@ -1,9 +1,9 @@
<!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="Builder for `HmsCatalog`."><title>HmsCatalogBuilder in iceberg_catalog_hms - 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,Fira [...]
-</div></details><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-CatalogBuilder-for-HmsCatalogBuilder" class="impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#78-154">Source</a><a href="#impl-CatalogBuilder-for-HmsCatalogBuilder" class="anchor">§</a><h3 class="cod [...]
+</div></details><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-CatalogBuilder-for-HmsCatalogBuilder" class="impl"><a class="src rightside" href="../src/iceberg_catalog_hms/catalog.rs.html#78-154">Source</a><a href="#impl-CatalogBuilder-for-HmsCatalogBuilder" class="anchor">§</a><h3 class="cod [...]
self,
name: impl <a class="trait" href="https://doc.rust-lang.org/nightly/core/convert/trait.Into.html" title="trait core::convert::Into">Into</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>>,
props: <a class="struct" href="https://doc.rust-lang.org/nightly/std/collections/hash/map/struct.HashMap.html" title="struct std::collections::hash::map::HashMap">HashMap</a><<a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>, <a class="struct" href="https://doc.rust-lang.org/nightly/alloc/string/struct.String.html" title="struct alloc::string::String">String</a>>,
-) -> impl <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = Result<Self::C>> + <a class="trait" href="https://doc.rust-lang.org/nightly/core/marker/trait.Send.html" title="trait core::marker::Send">Send</a></h4></section></summary><div class='docblock'>Create a new catalog instance.</div></details></div></details><details class="toggle implementors-toggle" open><summ [...]
+) -> impl <a class="trait" href="https://doc.rust-lang.org/nightly/core/future/future/trait.Future.html" title="trait core::future::future::Future">Future</a><Output = <a class="type" href="../iceberg/error/type.Result.html" title="type iceberg::error::Result">Result</a><Self::<a class="associatedtype" href="../iceberg/catalog/trait.CatalogBuilder.html#associatedtype.C" title="type iceberg::catalog::CatalogBuilder::C">C</a>>> + <a class="trait" href="https://doc.rust-lang. [...]
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#142">Source</a><a href="#method.type_id" class="anchor">§</a><h4 class="code-header">fn <a href [...]
T: CastFrom<U>,</div></h4></section></summary><div class='docblock'>Casts <code>self</code> to type <code>T</code>. The <a href="https://doc.rust-lang.org/reference/expressions/operator-expr.html#semantics">semantics of numeric casting</a> with the <code>as</code> operator are followed, so <code><T as As>::as_::<U></code> can be used in the same way as <code>T as U</code> for numeric conversions. <a>Read more</a></div></details></div></details><details class="toggle [...]
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>(&self) -> &(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 [...]
diff --git a/api/search.index/path/0e7338738470.js b/api/search.index/path/0e7338738470.js
deleted file mode 100644
index 1675a2431..000000000
--- a/api/search.index/path/0e7338738470.js
+++ /dev/null
@@ -1 +0,0 @@
-rd_("g[17,\"\"]000f[1,\"\"]0Aa[10,\"bnum::cast\"]Bn[15,\"iceberg_storage_opendal::OpenDalStorage\"]Eg[5,\"tower_http::follow_redirect::policy::or\",\"tower_http::follow_redirect::policy::or\"]1Ce[15,\"iceberg_storage_opendal::OpenDalStorageFactory\"]4Bb[2,\"iceberg::io::storage::config\"]m[2,\"iceberg\"]Ai[2,\"iceberg::encryption\"]2Am[2,\"iceberg_storage_opendal\"]8Ei[5,\"tower_http::follow_redirect::policy::and\",\"tower_http::follow_redirect::policy::and\"]A`[10,\"core::any\"]Ao[5,\"a [...]
\ No newline at end of file
diff --git a/api/search.index/path/ed9380482964.js b/api/search.index/path/ed9380482964.js
new file mode 100644
index 000000000..8d8b96e5a
--- /dev/null
+++ b/api/search.index/path/ed9380482964.js
@@ -0,0 +1 @@
+rd_("g[17,\"\"]000f[1,\"\"]0Aa[10,\"bnum::cast\"]Bn[15,\"iceberg_storage_opendal::OpenDalStorage\"]Eg[5,\"tower_http::follow_redirect::policy::or\",\"tower_http::follow_redirect::policy::or\"]1Ce[15,\"iceberg_storage_opendal::OpenDalStorageFactory\"]4Bb[2,\"iceberg::io::storage::config\"]m[2,\"iceberg\"]Ai[2,\"iceberg::encryption\"]2Am[2,\"iceberg_storage_opendal\"]8Ei[5,\"tower_http::follow_redirect::policy::and\",\"tower_http::follow_redirect::policy::and\"]A`[10,\"core::any\"]Ao[5,\"a [...]
\ No newline at end of file
diff --git a/api/search.index/root.js b/api/search.index/root.js
index cf0522364..49d9634ac 100644
--- a/api/search.index/root.js
+++ b/api/search.index/root.js
@@ -1 +1 @@
-rr_('{"normalizedName":{"I":"RQBAAAAEixJlcQGrBqMLAA4kDhUAQAAAIFQDMQSQBDSOBAIOFQBAAAAAAg42RhWQHzFAAACDAVQDMQSQBAUBwAAAAlQaBTUcYWUxQAAAqwajCwAOJA5jAASgEAAADgKwcB5SAH41OGRt9iAAAABTAeAB3QBfAHIJVQJAAAAxqwajCwAOJA4AjgQBAQ4yNjkeAB8AcgGxDzodQR8FAcAAAAMrHQBCGmFzAQEBoGAAABm4ZDswAAABAADaAAMAmgQAALAEAACuCNgABQHDAACXDh+SZh5kczUAQgAABZgYY4sJjAmNCb0JBABCdGQAAAZlZmF1bHQVGmSzBFYLAMHIoAAAAAGDXRzDOJDloFAAAB2PsCAPqQqaAAAAAQhOHAAA0roYAACxBQBSAAAGnR9khx8FAFQAAAaeH2SLH/ODAnB0NQBEAACWkh9k7B7tHmsfbB9bA6BgAAAa1WF [...]
\ No newline at end of file
+rr_('{"normalizedName":{"I":"RQBAAAAEixJlcQGrBqMLAA4kDhUAQAAAIFQDMQSQBDSOBAIOFQBAAAAAAg42RhWQHzFAAACDAVQDMQSQBAUBwAAAAlQaBTUcYWUxQAAAqwajCwAOJA5jAASgEAAADgKwcB5SAH41OGRt9iAAAABTAeAB3QBfAHIJVQJAAAAxqwajCwAOJA4AjgQBAQ4yNjkeAB8AcgGxDzodQR8FAcAAAAMrHQBCGmFzAQEBoGAAABm4ZDswAAABAADaAAMAmgQAALAEAACuCNgABQHDAACXDh+SZh5kczUAQgAABZgYY4sJjAmNCb0JBABCdGQAAAZlZmF1bHQVGmSzBFYLAMHIoAAAAAGDXRzDOJDloFAAAB2PsCAPqQqaAAAAAQhOHAAA0roYAACxBQBSAAAGnR9khx8FAFQAAAaeH2SLH/ODAnB0NQBEAACWkh9k7B7tHmsfbB9bA6BgAAAa1WF [...]
\ 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 54b1112f0..49c5903b8 100644
--- a/api/src/iceberg/arrow/caching_delete_file_loader.rs.html
+++ b/api/src/iceberg/arrow/caching_delete_file_loader.rs.html
@@ -16,1077 +16,1198 @@
<a href=#16 id=16 data-nosnippet>16</a>// under the License.
<a href=#17 id=17 data-nosnippet>17</a>
<a href=#18 id=18 data-nosnippet>18</a></span><span class="kw">use </span>std::collections::{HashMap, HashSet};
-<a href=#19 id=19 data-nosnippet>19</a><span class="kw">use </span>std::ops::Not;
-<a href=#20 id=20 data-nosnippet>20</a><span class="kw">use </span>std::sync::Arc;
-<a href=#21 id=21 data-nosnippet>21</a>
-<a href=#22 id=22 data-nosnippet>22</a><span class="kw">use </span>arrow_array::{Array, ArrayRef, Int64Array, StringArray, StructArray};
-<a href=#23 id=23 data-nosnippet>23</a><span class="kw">use </span>futures::{StreamExt, TryStreamExt};
-<a href=#24 id=24 data-nosnippet>24</a><span class="kw">use </span>tokio::sync::oneshot::{Receiver, channel};
-<a href=#25 id=25 data-nosnippet>25</a>
-<a href=#26 id=26 data-nosnippet>26</a><span class="kw">use </span><span class="kw">super</span>::delete_filter::{DeleteFilter, PosDelLoadAction};
-<a href=#27 id=27 data-nosnippet>27</a><span class="kw">use </span><span class="kw">crate</span>::arrow::delete_file_loader::BasicDeleteFileLoader;
-<a href=#28 id=28 data-nosnippet>28</a><span class="kw">use </span><span class="kw">crate</span>::arrow::scan_metrics::ScanMetrics;
-<a href=#29 id=29 data-nosnippet>29</a><span class="kw">use </span><span class="kw">crate</span>::arrow::{arrow_primitive_to_literal, arrow_schema_to_schema};
-<a href=#30 id=30 data-nosnippet>30</a><span class="kw">use </span><span class="kw">crate</span>::delete_vector::DeleteVector;
-<a href=#31 id=31 data-nosnippet>31</a><span class="kw">use </span><span class="kw">crate</span>::expr::Predicate::AlwaysTrue;
-<a href=#32 id=32 data-nosnippet>32</a><span class="kw">use </span><span class="kw">crate</span>::expr::{Predicate, Reference};
-<a href=#33 id=33 data-nosnippet>33</a><span class="kw">use </span><span class="kw">crate</span>::io::FileIO;
-<a href=#34 id=34 data-nosnippet>34</a><span class="kw">use </span><span class="kw">crate</span>::runtime::Runtime;
-<a href=#35 id=35 data-nosnippet>35</a><span class="kw">use </span><span class="kw">crate</span>::scan::{ArrowRecordBatchStream, FileScanTaskDeleteFile};
-<a href=#36 id=36 data-nosnippet>36</a><span class="kw">use </span><span class="kw">crate</span>::spec::{
-<a href=#37 id=37 data-nosnippet>37</a> DataContentType, Datum, ListType, MapType, NestedField, NestedFieldRef, PartnerAccessor,
-<a href=#38 id=38 data-nosnippet>38</a> PrimitiveType, Schema, SchemaRef, SchemaWithPartnerVisitor, StructType, Type,
-<a href=#39 id=39 data-nosnippet>39</a> visit_schema_with_partner,
-<a href=#40 id=40 data-nosnippet>40</a>};
-<a href=#41 id=41 data-nosnippet>41</a><span class="kw">use </span>crate::{Error, ErrorKind, <span class="prelude-ty">Result</span>};
-<a href=#42 id=42 data-nosnippet>42</a>
-<a href=#43 id=43 data-nosnippet>43</a><span class="attr">#[derive(Clone, Debug)]
-<a href=#44 id=44 data-nosnippet>44</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">struct </span>CachingDeleteFileLoader {
-<a href=#45 id=45 data-nosnippet>45</a> basic_delete_file_loader: BasicDeleteFileLoader,
-<a href=#46 id=46 data-nosnippet>46</a> concurrency_limit_data_files: usize,
-<a href=#47 id=47 data-nosnippet>47</a> <span class="doccomment">/// Shared filter state to allow caching loaded deletes across multiple
-<a href=#48 id=48 data-nosnippet>48</a> /// calls to `load_deletes` (e.g., across multiple file scan tasks).
-<a href=#49 id=49 data-nosnippet>49</a> </span>delete_filter: DeleteFilter,
-<a href=#50 id=50 data-nosnippet>50</a> runtime: Runtime,
-<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">// Intermediate context during processing of a delete file task.
-<a href=#54 id=54 data-nosnippet>54</a></span><span class="kw">enum </span>DeleteFileContext {
-<a href=#55 id=55 data-nosnippet>55</a> <span class="comment">// TODO: Delete Vector loader from Puffin files
-<a href=#56 id=56 data-nosnippet>56</a> </span>ExistingEqDel,
-<a href=#57 id=57 data-nosnippet>57</a> ExistingPosDel,
-<a href=#58 id=58 data-nosnippet>58</a> PosDels {
-<a href=#59 id=59 data-nosnippet>59</a> file_path: String,
-<a href=#60 id=60 data-nosnippet>60</a> stream: ArrowRecordBatchStream,
-<a href=#61 id=61 data-nosnippet>61</a> },
-<a href=#62 id=62 data-nosnippet>62</a> FreshEqDel {
-<a href=#63 id=63 data-nosnippet>63</a> batch_stream: ArrowRecordBatchStream,
-<a href=#64 id=64 data-nosnippet>64</a> equality_ids: HashSet<i32>,
-<a href=#65 id=65 data-nosnippet>65</a> sender: tokio::sync::oneshot::Sender<Predicate>,
-<a href=#66 id=66 data-nosnippet>66</a> },
-<a href=#67 id=67 data-nosnippet>67</a>}
-<a href=#68 id=68 data-nosnippet>68</a>
-<a href=#69 id=69 data-nosnippet>69</a><span class="comment">// Final result of the processing of a delete file task before
-<a href=#70 id=70 data-nosnippet>70</a>// results are fully merged into the DeleteFileManager's state
-<a href=#71 id=71 data-nosnippet>71</a></span><span class="kw">enum </span>ParsedDeleteFileContext {
-<a href=#72 id=72 data-nosnippet>72</a> DelVecs {
-<a href=#73 id=73 data-nosnippet>73</a> file_path: String,
-<a href=#74 id=74 data-nosnippet>74</a> results: HashMap<String, DeleteVector>,
-<a href=#75 id=75 data-nosnippet>75</a> },
-<a href=#76 id=76 data-nosnippet>76</a> EqDel,
-<a href=#77 id=77 data-nosnippet>77</a> ExistingPosDel,
-<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="attr">#[allow(unused_variables)]
-<a href=#81 id=81 data-nosnippet>81</a></span><span class="kw">impl </span>CachingDeleteFileLoader {
-<a href=#82 id=82 data-nosnippet>82</a> <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>new(
-<a href=#83 id=83 data-nosnippet>83</a> file_io: FileIO,
-<a href=#84 id=84 data-nosnippet>84</a> concurrency_limit_data_files: usize,
-<a href=#85 id=85 data-nosnippet>85</a> runtime: Runtime,
-<a href=#86 id=86 data-nosnippet>86</a> ) -> <span class="self">Self </span>{
-<a href=#87 id=87 data-nosnippet>87</a> <span class="kw">let </span>scan_metrics = ScanMetrics::new();
-<a href=#88 id=88 data-nosnippet>88</a> CachingDeleteFileLoader {
-<a href=#89 id=89 data-nosnippet>89</a> basic_delete_file_loader: BasicDeleteFileLoader::new(file_io, scan_metrics),
-<a href=#90 id=90 data-nosnippet>90</a> concurrency_limit_data_files,
-<a href=#91 id=91 data-nosnippet>91</a> delete_filter: DeleteFilter::new(runtime.clone()),
-<a href=#92 id=92 data-nosnippet>92</a> runtime,
-<a href=#93 id=93 data-nosnippet>93</a> }
-<a href=#94 id=94 data-nosnippet>94</a> }
-<a href=#95 id=95 data-nosnippet>95</a>
-<a href=#96 id=96 data-nosnippet>96</a> <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>with_scan_metrics(<span class="kw-2">mut </span><span class="self">self</span>, scan_metrics: ScanMetrics) -> <span class="self">Self </span>{
-<a href=#97 id=97 data-nosnippet>97</a> <span class="self">self</span>.basic_delete_file_loader = BasicDeleteFileLoader::new(
-<a href=#98 id=98 data-nosnippet>98</a> <span class="self">self</span>.basic_delete_file_loader.file_io().clone(),
-<a href=#99 id=99 data-nosnippet>99</a> scan_metrics,
-<a href=#100 id=100 data-nosnippet>100</a> );
-<a href=#101 id=101 data-nosnippet>101</a> <span class="self">self
-<a href=#102 id=102 data-nosnippet>102</a> </span>}
-<a href=#103 id=103 data-nosnippet>103</a>
-<a href=#104 id=104 data-nosnippet>104</a> <span class="doccomment">/// Initiates loading of all deletes for all the specified tasks
-<a href=#105 id=105 data-nosnippet>105</a> ///
-<a href=#106 id=106 data-nosnippet>106</a> /// Returned future completes once all positional deletes and delete vectors
-<a href=#107 id=107 data-nosnippet>107</a> /// have loaded. EQ deletes are not waited for in this method but the returned
-<a href=#108 id=108 data-nosnippet>108</a> /// DeleteFilter will await their loading when queried for them.
-<a href=#109 id=109 data-nosnippet>109</a> ///
-<a href=#110 id=110 data-nosnippet>110</a> /// * Create a single stream of all delete file tasks irrespective of type,
-<a href=#111 id=111 data-nosnippet>111</a> /// so that we can respect the combined concurrency limit
-<a href=#112 id=112 data-nosnippet>112</a> /// * We then process each in two phases: load and parse.
-<a href=#113 id=113 data-nosnippet>113</a> /// * for positional deletes the load phase instantiates an ArrowRecordBatchStream to
-<a href=#114 id=114 data-nosnippet>114</a> /// stream the file contents out
-<a href=#115 id=115 data-nosnippet>115</a> /// * for eq deletes, we first check if the EQ delete is already loaded or being loaded by
-<a href=#116 id=116 data-nosnippet>116</a> /// another concurrently processing data file scan task. If it is, we skip it.
-<a href=#117 id=117 data-nosnippet>117</a> /// If not, the DeleteFilter is updated to contain a notifier to prevent other data file
-<a href=#118 id=118 data-nosnippet>118</a> /// tasks from starting to load the same equality delete file. We spawn a task to load
-<a href=#119 id=119 data-nosnippet>119</a> /// the EQ delete's record batch stream, convert it to a predicate, update the delete filter,
-<a href=#120 id=120 data-nosnippet>120</a> /// and notify any task that was waiting for it.
-<a href=#121 id=121 data-nosnippet>121</a> /// * When this gets updated to add support for delete vectors, the load phase will return
-<a href=#122 id=122 data-nosnippet>122</a> /// a PuffinReader for them.
-<a href=#123 id=123 data-nosnippet>123</a> /// * The parse phase parses each record batch stream according to its associated data type.
-<a href=#124 id=124 data-nosnippet>124</a> /// The result of this is a map of data file paths to delete vectors for the positional
-<a href=#125 id=125 data-nosnippet>125</a> /// delete tasks (and in future for the delete vector tasks). For equality delete
-<a href=#126 id=126 data-nosnippet>126</a> /// file tasks, this results in an unbound Predicate.
-<a href=#127 id=127 data-nosnippet>127</a> /// * The unbound Predicates resulting from equality deletes are sent to their associated oneshot
-<a href=#128 id=128 data-nosnippet>128</a> /// channel to store them in the right place in the delete file managers state.
-<a href=#129 id=129 data-nosnippet>129</a> /// * The results of all of these futures are awaited on in parallel with the specified
-<a href=#130 id=130 data-nosnippet>130</a> /// level of concurrency and collected into a vec. We then combine all the delete
-<a href=#131 id=131 data-nosnippet>131</a> /// vector maps that resulted from any positional delete or delete vector files into a
-<a href=#132 id=132 data-nosnippet>132</a> /// single map and persist it in the state.
+<a href=#19 id=19 data-nosnippet>19</a><span class="kw">use </span>std::sync::Arc;
+<a href=#20 id=20 data-nosnippet>20</a>
+<a href=#21 id=21 data-nosnippet>21</a><span class="kw">use </span>arrow_array::{Array, ArrayRef, Int64Array, StringArray, StructArray};
+<a href=#22 id=22 data-nosnippet>22</a><span class="kw">use </span>futures::{StreamExt, TryStreamExt};
+<a href=#23 id=23 data-nosnippet>23</a><span class="kw">use </span>tokio::sync::oneshot::{Receiver, channel};
+<a href=#24 id=24 data-nosnippet>24</a>
+<a href=#25 id=25 data-nosnippet>25</a><span class="kw">use </span><span class="kw">super</span>::delete_filter::{DeleteFilter, PosDelLoadAction};
+<a href=#26 id=26 data-nosnippet>26</a><span class="kw">use </span><span class="kw">crate</span>::arrow::delete_file_loader::BasicDeleteFileLoader;
+<a href=#27 id=27 data-nosnippet>27</a><span class="kw">use </span><span class="kw">crate</span>::arrow::scan_metrics::ScanMetrics;
+<a href=#28 id=28 data-nosnippet>28</a><span class="kw">use </span><span class="kw">crate</span>::arrow::{arrow_primitive_to_literal, arrow_schema_to_schema};
+<a href=#29 id=29 data-nosnippet>29</a><span class="kw">use </span><span class="kw">crate</span>::delete_vector::DeleteVector;
+<a href=#30 id=30 data-nosnippet>30</a><span class="kw">use </span><span class="kw">crate</span>::expr::Predicate::AlwaysTrue;
+<a href=#31 id=31 data-nosnippet>31</a><span class="kw">use </span><span class="kw">crate</span>::expr::{Predicate, Reference};
+<a href=#32 id=32 data-nosnippet>32</a><span class="kw">use </span><span class="kw">crate</span>::io::FileIO;
+<a href=#33 id=33 data-nosnippet>33</a><span class="kw">use </span><span class="kw">crate</span>::runtime::Runtime;
+<a href=#34 id=34 data-nosnippet>34</a><span class="kw">use </span><span class="kw">crate</span>::scan::{ArrowRecordBatchStream, FileScanTaskDeleteFile};
+<a href=#35 id=35 data-nosnippet>35</a><span class="kw">use </span><span class="kw">crate</span>::spec::{
+<a href=#36 id=36 data-nosnippet>36</a> DataContentType, Datum, ListType, MapType, NestedField, NestedFieldRef, PartnerAccessor,
+<a href=#37 id=37 data-nosnippet>37</a> PrimitiveType, Schema, SchemaRef, SchemaWithPartnerVisitor, StructType, Type,
+<a href=#38 id=38 data-nosnippet>38</a> visit_schema_with_partner,
+<a href=#39 id=39 data-nosnippet>39</a>};
+<a href=#40 id=40 data-nosnippet>40</a><span class="kw">use </span>crate::{Error, ErrorKind, <span class="prelude-ty">Result</span>};
+<a href=#41 id=41 data-nosnippet>41</a>
+<a href=#42 id=42 data-nosnippet>42</a><span class="attr">#[derive(Clone, Debug)]
+<a href=#43 id=43 data-nosnippet>43</a></span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">struct </span>CachingDeleteFileLoader {
+<a href=#44 id=44 data-nosnippet>44</a> basic_delete_file_loader: BasicDeleteFileLoader,
+<a href=#45 id=45 data-nosnippet>45</a> concurrency_limit_data_files: usize,
+<a href=#46 id=46 data-nosnippet>46</a> <span class="doccomment">/// Shared filter state to allow caching loaded deletes across multiple
+<a href=#47 id=47 data-nosnippet>47</a> /// calls to `load_deletes` (e.g., across multiple file scan tasks).
+<a href=#48 id=48 data-nosnippet>48</a> </span>delete_filter: DeleteFilter,
+<a href=#49 id=49 data-nosnippet>49</a> runtime: Runtime,
+<a href=#50 id=50 data-nosnippet>50</a>}
+<a href=#51 id=51 data-nosnippet>51</a>
+<a href=#52 id=52 data-nosnippet>52</a><span class="comment">// Intermediate context during processing of a delete file task.
+<a href=#53 id=53 data-nosnippet>53</a></span><span class="kw">enum </span>DeleteFileContext {
+<a href=#54 id=54 data-nosnippet>54</a> <span class="comment">// TODO: Delete Vector loader from Puffin files
+<a href=#55 id=55 data-nosnippet>55</a> </span>ExistingEqDel,
+<a href=#56 id=56 data-nosnippet>56</a> ExistingPosDel,
+<a href=#57 id=57 data-nosnippet>57</a> PosDels {
+<a href=#58 id=58 data-nosnippet>58</a> file_path: String,
+<a href=#59 id=59 data-nosnippet>59</a> stream: ArrowRecordBatchStream,
+<a href=#60 id=60 data-nosnippet>60</a> },
+<a href=#61 id=61 data-nosnippet>61</a> FreshEqDel {
+<a href=#62 id=62 data-nosnippet>62</a> batch_stream: ArrowRecordBatchStream,
+<a href=#63 id=63 data-nosnippet>63</a> equality_ids: HashSet<i32>,
+<a href=#64 id=64 data-nosnippet>64</a> sender: tokio::sync::oneshot::Sender<Predicate>,
+<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>
+<a href=#68 id=68 data-nosnippet>68</a><span class="comment">// Final result of the processing of a delete file task before
+<a href=#69 id=69 data-nosnippet>69</a>// results are fully merged into the DeleteFileManager's state
+<a href=#70 id=70 data-nosnippet>70</a></span><span class="kw">enum </span>ParsedDeleteFileContext {
+<a href=#71 id=71 data-nosnippet>71</a> DelVecs {
+<a href=#72 id=72 data-nosnippet>72</a> file_path: String,
+<a href=#73 id=73 data-nosnippet>73</a> results: HashMap<String, DeleteVector>,
+<a href=#74 id=74 data-nosnippet>74</a> },
+<a href=#75 id=75 data-nosnippet>75</a> EqDel,
+<a href=#76 id=76 data-nosnippet>76</a> ExistingPosDel,
+<a href=#77 id=77 data-nosnippet>77</a>}
+<a href=#78 id=78 data-nosnippet>78</a>
+<a href=#79 id=79 data-nosnippet>79</a><span class="attr">#[allow(unused_variables)]
+<a href=#80 id=80 data-nosnippet>80</a></span><span class="kw">impl </span>CachingDeleteFileLoader {
+<a href=#81 id=81 data-nosnippet>81</a> <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>new(
+<a href=#82 id=82 data-nosnippet>82</a> file_io: FileIO,
+<a href=#83 id=83 data-nosnippet>83</a> concurrency_limit_data_files: usize,
+<a href=#84 id=84 data-nosnippet>84</a> runtime: Runtime,
+<a href=#85 id=85 data-nosnippet>85</a> ) -> <span class="self">Self </span>{
+<a href=#86 id=86 data-nosnippet>86</a> <span class="kw">let </span>scan_metrics = ScanMetrics::new();
+<a href=#87 id=87 data-nosnippet>87</a> CachingDeleteFileLoader {
+<a href=#88 id=88 data-nosnippet>88</a> basic_delete_file_loader: BasicDeleteFileLoader::new(file_io, scan_metrics),
+<a href=#89 id=89 data-nosnippet>89</a> concurrency_limit_data_files,
+<a href=#90 id=90 data-nosnippet>90</a> delete_filter: DeleteFilter::new(runtime.clone()),
+<a href=#91 id=91 data-nosnippet>91</a> runtime,
+<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>
+<a href=#95 id=95 data-nosnippet>95</a> <span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>with_scan_metrics(<span class="kw-2">mut </span><span class="self">self</span>, scan_metrics: ScanMetrics) -> <span class="self">Self </span>{
+<a href=#96 id=96 data-nosnippet>96</a> <span class="self">self</span>.basic_delete_file_loader = BasicDeleteFileLoader::new(
+<a href=#97 id=97 data-nosnippet>97</a> <span class="self">self</span>.basic_delete_file_loader.file_io().clone(),
+<a href=#98 id=98 data-nosnippet>98</a> scan_metrics,
+<a href=#99 id=99 data-nosnippet>99</a> );
+<a href=#100 id=100 data-nosnippet>100</a> <span class="self">self
+<a href=#101 id=101 data-nosnippet>101</a> </span>}
+<a href=#102 id=102 data-nosnippet>102</a>
+<a href=#103 id=103 data-nosnippet>103</a> <span class="doccomment">/// Initiates loading of all deletes for all the specified tasks
+<a href=#104 id=104 data-nosnippet>104</a> ///
+<a href=#105 id=105 data-nosnippet>105</a> /// Returned future completes once all positional deletes and delete vectors
+<a href=#106 id=106 data-nosnippet>106</a> /// have loaded. EQ deletes are not waited for in this method but the returned
+<a href=#107 id=107 data-nosnippet>107</a> /// DeleteFilter will await their loading when queried for them.
+<a href=#108 id=108 data-nosnippet>108</a> ///
+<a href=#109 id=109 data-nosnippet>109</a> /// * Create a single stream of all delete file tasks irrespective of type,
+<a href=#110 id=110 data-nosnippet>110</a> /// so that we can respect the combined concurrency limit
+<a href=#111 id=111 data-nosnippet>111</a> /// * We then process each in two phases: load and parse.
+<a href=#112 id=112 data-nosnippet>112</a> /// * for positional deletes the load phase instantiates an ArrowRecordBatchStream to
+<a href=#113 id=113 data-nosnippet>113</a> /// stream the file contents out
+<a href=#114 id=114 data-nosnippet>114</a> /// * for eq deletes, we first check if the EQ delete is already loaded or being loaded by
+<a href=#115 id=115 data-nosnippet>115</a> /// another concurrently processing data file scan task. If it is, we skip it.
+<a href=#116 id=116 data-nosnippet>116</a> /// If not, the DeleteFilter is updated to contain a notifier to prevent other data file
+<a href=#117 id=117 data-nosnippet>117</a> /// tasks from starting to load the same equality delete file. We spawn a task to load
+<a href=#118 id=118 data-nosnippet>118</a> /// the EQ delete's record batch stream, convert it to a predicate, update the delete filter,
+<a href=#119 id=119 data-nosnippet>119</a> /// and notify any task that was waiting for it.
+<a href=#120 id=120 data-nosnippet>120</a> /// * When this gets updated to add support for delete vectors, the load phase will return
+<a href=#121 id=121 data-nosnippet>121</a> /// a PuffinReader for them.
+<a href=#122 id=122 data-nosnippet>122</a> /// * The parse phase parses each record batch stream according to its associated data type.
+<a href=#123 id=123 data-nosnippet>123</a> /// The result of this is a map of data file paths to delete vectors for the positional
+<a href=#124 id=124 data-nosnippet>124</a> /// delete tasks (and in future for the delete vector tasks). For equality delete
+<a href=#125 id=125 data-nosnippet>125</a> /// file tasks, this results in an unbound Predicate.
+<a href=#126 id=126 data-nosnippet>126</a> /// * The unbound Predicates resulting from equality deletes are sent to their associated oneshot
+<a href=#127 id=127 data-nosnippet>127</a> /// channel to store them in the right place in the delete file managers state.
+<a href=#128 id=128 data-nosnippet>128</a> /// * The results of all of these futures are awaited on in parallel with the specified
+<a href=#129 id=129 data-nosnippet>129</a> /// level of concurrency and collected into a vec. We then combine all the delete
+<a href=#130 id=130 data-nosnippet>130</a> /// vector maps that resulted from any positional delete or delete vector files into a
+<a href=#131 id=131 data-nosnippet>131</a> /// single map and persist it in the state.
+<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> /// Conceptually, the data flow is like this:
-<a href=#136 id=136 data-nosnippet>136</a> /// ```none
-<a href=#137 id=137 data-nosnippet>137</a> /// FileScanTaskDeleteFile
-<a href=#138 id=138 data-nosnippet>138</a> /// |
-<a href=#139 id=139 data-nosnippet>139</a> /// Skip Started EQ Deletes
+<a href=#134 id=134 data-nosnippet>134</a> /// Conceptually, the data flow is like this:
+<a href=#135 id=135 data-nosnippet>135</a> /// ```none
+<a href=#136 id=136 data-nosnippet>136</a> /// FileScanTaskDeleteFile
+<a href=#137 id=137 data-nosnippet>137</a> /// |
+<a href=#138 id=138 data-nosnippet>138</a> /// Skip Started EQ Deletes
+<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> /// [load recordbatch stream / puffin]
-<a href=#143 id=143 data-nosnippet>143</a> /// DeleteFileContext
+<a href=#141 id=141 data-nosnippet>141</a> /// [load recordbatch stream / puffin]
+<a href=#142 id=142 data-nosnippet>142</a> /// DeleteFileContext
+<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> /// |
-<a href=#146 id=146 data-nosnippet>146</a> /// +-----------------------------+--------------------------+
-<a href=#147 id=147 data-nosnippet>147</a> /// Pos Del Del Vec (Not yet Implemented) EQ Del
-<a href=#148 id=148 data-nosnippet>148</a> /// | | |
-<a href=#149 id=149 data-nosnippet>149</a> /// [parse pos del stream] [parse del vec puffin] [parse eq del]
-<a href=#150 id=150 data-nosnippet>150</a> /// HashMap<String, RoaringTreeMap> HashMap<String, RoaringTreeMap> (Predicate, Sender)
-<a href=#151 id=151 data-nosnippet>151</a> /// | | |
-<a href=#152 id=152 data-nosnippet>152</a> /// | | [persist to state]
-<a href=#153 id=153 data-nosnippet>153</a> /// | | ()
-<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> /// |
-<a href=#157 id=157 data-nosnippet>157</a> /// [buffer unordered]
-<a href=#158 id=158 data-nosnippet>158</a> /// |
-<a href=#159 id=159 data-nosnippet>159</a> /// [combine del vectors]
-<a href=#160 id=160 data-nosnippet>160</a> /// HashMap<String, RoaringTreeMap>
-<a href=#161 id=161 data-nosnippet>161</a> /// |
-<a href=#162 id=162 data-nosnippet>162</a> /// [persist del vectors to state]
-<a href=#163 id=163 data-nosnippet>163</a> /// ()
+<a href=#145 id=145 data-nosnippet>145</a> /// +-----------------------------+--------------------------+
+<a href=#146 id=146 data-nosnippet>146</a> /// Pos Del Del Vec (Not yet Implemented) EQ Del
+<a href=#147 id=147 data-nosnippet>147</a> /// | | |
+<a href=#148 id=148 data-nosnippet>148</a> /// [parse pos del stream] [parse del vec puffin] [parse eq del]
+<a href=#149 id=149 data-nosnippet>149</a> /// HashMap<String, RoaringTreeMap> HashMap<String, RoaringTreeMap> (Predicate, Sender)
+<a href=#150 id=150 data-nosnippet>150</a> /// | | |
+<a href=#151 id=151 data-nosnippet>151</a> /// | | [persist to state]
+<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> /// +-----------------------------+--------------------------+
+<a href=#155 id=155 data-nosnippet>155</a> /// |
+<a href=#156 id=156 data-nosnippet>156</a> /// [buffer unordered]
+<a href=#157 id=157 data-nosnippet>157</a> /// |
+<a href=#158 id=158 data-nosnippet>158</a> /// [combine del vectors]
+<a href=#159 id=159 data-nosnippet>159</a> /// HashMap<String, RoaringTreeMap>
+<a href=#160 id=160 data-nosnippet>160</a> /// |
+<a href=#161 id=161 data-nosnippet>161</a> /// [persist del vectors to state]
+<a href=#162 id=162 data-nosnippet>162</a> /// ()
+<a href=#163 id=163 data-nosnippet>163</a> /// |
<a href=#164 id=164 data-nosnippet>164</a> /// |
-<a href=#165 id=165 data-nosnippet>165</a> /// |
-<a href=#166 id=166 data-nosnippet>166</a> /// [join!]
-<a href=#167 id=167 data-nosnippet>167</a> /// ```
-<a href=#168 id=168 data-nosnippet>168</a> </span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>load_deletes(
-<a href=#169 id=169 data-nosnippet>169</a> <span class="kw-2">&</span><span class="self">self</span>,
-<a href=#170 id=170 data-nosnippet>170</a> delete_file_entries: <span class="kw-2">&</span>[FileScanTaskDeleteFile],
-<a href=#171 id=171 data-nosnippet>171</a> schema: SchemaRef,
-<a href=#172 id=172 data-nosnippet>172</a> ) -> Receiver<<span class="prelude-ty">Result</span><DeleteFilter>> {
-<a href=#173 id=173 data-nosnippet>173</a> <span class="kw">let </span>(tx, rx) = channel();
-<a href=#174 id=174 data-nosnippet>174</a>
-<a href=#175 id=175 data-nosnippet>175</a> <span class="kw">let </span>stream_items = delete_file_entries
-<a href=#176 id=176 data-nosnippet>176</a> .iter()
-<a href=#177 id=177 data-nosnippet>177</a> .map(|t| {
-<a href=#178 id=178 data-nosnippet>178</a> (
-<a href=#179 id=179 data-nosnippet>179</a> t.clone(),
-<a href=#180 id=180 data-nosnippet>180</a> <span class="self">self</span>.basic_delete_file_loader.clone(),
-<a href=#181 id=181 data-nosnippet>181</a> <span class="self">self</span>.delete_filter.clone(),
-<a href=#182 id=182 data-nosnippet>182</a> schema.clone(),
-<a href=#183 id=183 data-nosnippet>183</a> )
-<a href=#184 id=184 data-nosnippet>184</a> })
-<a href=#185 id=185 data-nosnippet>185</a> .collect::<Vec<<span class="kw">_</span>>>();
-<a href=#186 id=186 data-nosnippet>186</a> <span class="kw">let </span>task_stream = futures::stream::iter(stream_items);
-<a href=#187 id=187 data-nosnippet>187</a>
-<a href=#188 id=188 data-nosnippet>188</a> <span class="kw">let </span>del_filter = <span class="self">self</span>.delete_filter.clone();
-<a href=#189 id=189 data-nosnippet>189</a> <span class="kw">let </span>concurrency_limit_data_files = <span class="self">self</span>.concurrency_limit_data_files;
-<a href=#190 id=190 data-nosnippet>190</a> <span class="kw">let </span>basic_delete_file_loader = <span class="self">self</span>.basic_delete_file_loader.clone();
-<a href=#191 id=191 data-nosnippet>191</a> <span class="self">self</span>.runtime.io().spawn(<span class="kw">async move </span>{
-<a href=#192 id=192 data-nosnippet>192</a> <span class="kw">let </span>result = <span class="kw">async move </span>{
-<a href=#193 id=193 data-nosnippet>193</a> <span class="kw">let </span><span class="kw-2">mut </span>del_filter = del_filter;
-<a href=#194 id=194 data-nosnippet>194</a> <span class="kw">let </span>basic_delete_file_loader = basic_delete_file_loader.clone();
-<a href=#195 id=195 data-nosnippet>195</a>
-<a href=#196 id=196 data-nosnippet>196</a> <span class="kw">let </span><span class="kw-2">mut </span>results_stream = task_stream
-<a href=#197 id=197 data-nosnippet>197</a> .map(<span class="kw">move </span>|(task, file_io, del_filter, schema)| {
-<a href=#198 id=198 data-nosnippet>198</a> <span class="kw">let </span>basic_delete_file_loader = basic_delete_file_loader.clone();
-<a href=#199 id=199 data-nosnippet>199</a> <span class="kw">async move </span>{
-<a href=#200 id=200 data-nosnippet>200</a> <span class="self">Self</span>::load_file_for_task(
-<a href=#201 id=201 data-nosnippet>201</a> <span class="kw-2">&</span>task,
-<a href=#202 id=202 data-nosnippet>202</a> basic_delete_file_loader.clone(),
-<a href=#203 id=203 data-nosnippet>203</a> del_filter,
-<a href=#204 id=204 data-nosnippet>204</a> schema,
-<a href=#205 id=205 data-nosnippet>205</a> )
-<a href=#206 id=206 data-nosnippet>206</a> .<span class="kw">await
-<a href=#207 id=207 data-nosnippet>207</a> </span>}
-<a href=#208 id=208 data-nosnippet>208</a> })
-<a href=#209 id=209 data-nosnippet>209</a> .map(<span class="kw">move </span>|ctx| {
-<a href=#210 id=210 data-nosnippet>210</a> <span class="prelude-val">Ok</span>(<span class="kw">async </span>{ <span class="self">Self</span>::parse_file_content_for_task(ctx.<span class="kw">await</span><span class="question-mark">?</span>).<span class="kw">await </span>})
-<a href=#211 id=211 data-nosnippet>211</a> })
-<a href=#212 id=212 data-nosnippet>212</a> .try_buffer_unordered(concurrency_limit_data_files);
-<a href=#213 id=213 data-nosnippet>213</a>
-<a href=#214 id=214 data-nosnippet>214</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(item) = results_stream.next().<span class="kw">await </span>{
-<a href=#215 id=215 data-nosnippet>215</a> <span class="kw">let </span>item = item<span class="question-mark">?</span>;
-<a href=#216 id=216 data-nosnippet>216</a> <span class="kw">if let </span>ParsedDeleteFileContext::DelVecs { file_path, results } = item {
-<a href=#217 id=217 data-nosnippet>217</a> <span class="kw">for </span>(data_file_path, delete_vector) <span class="kw">in </span>results.into_iter() {
-<a href=#218 id=218 data-nosnippet>218</a> del_filter.upsert_delete_vector(data_file_path, delete_vector);
-<a href=#219 id=219 data-nosnippet>219</a> }
-<a href=#220 id=220 data-nosnippet>220</a> <span class="comment">// Mark the positional delete file as fully loaded so waiters can proceed
-<a href=#221 id=221 data-nosnippet>221</a> </span>del_filter.finish_pos_del_load(<span class="kw-2">&</span>file_path);
-<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>
-<a href=#225 id=225 data-nosnippet>225</a> <span class="prelude-val">Ok</span>(del_filter)
-<a href=#226 id=226 data-nosnippet>226</a> }
-<a href=#227 id=227 data-nosnippet>227</a> .<span class="kw">await</span>;
-<a href=#228 id=228 data-nosnippet>228</a>
-<a href=#229 id=229 data-nosnippet>229</a> <span class="kw">let _ </span>= tx.send(result);
-<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> rx
-<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">async fn </span>load_file_for_task(
-<a href=#236 id=236 data-nosnippet>236</a> task: <span class="kw-2">&</span>FileScanTaskDeleteFile,
-<a href=#237 id=237 data-nosnippet>237</a> basic_delete_file_loader: BasicDeleteFileLoader,
-<a href=#238 id=238 data-nosnippet>238</a> del_filter: DeleteFilter,
-<a href=#239 id=239 data-nosnippet>239</a> schema: SchemaRef,
-<a href=#240 id=240 data-nosnippet>240</a> ) -> <span class="prelude-ty">Result</span><DeleteFileContext> {
-<a href=#241 id=241 data-nosnippet>241</a> <span class="kw">match </span>task.file_type {
-<a href=#242 id=242 data-nosnippet>242</a> DataContentType::PositionDeletes => {
-<a href=#243 id=243 data-nosnippet>243</a> <span class="kw">match </span>del_filter.try_start_pos_del_load(<span class="kw-2">&</span>task.file_path) {
-<a href=#244 id=244 data-nosnippet>244</a> PosDelLoadAction::AlreadyLoaded => <span class="prelude-val">Ok</span>(DeleteFileContext::ExistingPosDel),
-<a href=#245 id=245 data-nosnippet>245</a> PosDelLoadAction::WaitFor(notify) => {
-<a href=#246 id=246 data-nosnippet>246</a> <span class="comment">// Positional deletes are accessed synchronously by ArrowReader.
-<a href=#247 id=247 data-nosnippet>247</a> // We must wait here to ensure the data is ready before returning,
-<a href=#248 id=248 data-nosnippet>248</a> // otherwise ArrowReader might get an empty/partial result.
-<a href=#249 id=249 data-nosnippet>249</a> </span>notify.notified().<span class="kw">await</span>;
-<a href=#250 id=250 data-nosnippet>250</a> <span class="prelude-val">Ok</span>(DeleteFileContext::ExistingPosDel)
-<a href=#251 id=251 data-nosnippet>251</a> }
-<a href=#252 id=252 data-nosnippet>252</a> PosDelLoadAction::Load => <span class="prelude-val">Ok</span>(DeleteFileContext::PosDels {
-<a href=#253 id=253 data-nosnippet>253</a> file_path: task.file_path.clone(),
-<a href=#254 id=254 data-nosnippet>254</a> stream: basic_delete_file_loader
-<a href=#255 id=255 data-nosnippet>255</a> .parquet_to_batch_stream(<span class="kw-2">&</span>task.file_path, task.file_size_in_bytes)
-<a href=#256 id=256 data-nosnippet>256</a> .<span class="kw">await</span><span class="question-mark">?</span>,
-<a href=#257 id=257 data-nosnippet>257</a> }),
-<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> DataContentType::EqualityDeletes => {
-<a href=#262 id=262 data-nosnippet>262</a> <span class="kw">let </span><span class="prelude-val">Some</span>(notify) = del_filter.try_start_eq_del_load(<span class="kw-2">&</span>task.file_path) <span class="kw">else </span>{
-<a href=#263 id=263 data-nosnippet>263</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(DeleteFileContext::ExistingEqDel);
-<a href=#264 id=264 data-nosnippet>264</a> };
-<a href=#265 id=265 data-nosnippet>265</a>
-<a href=#266 id=266 data-nosnippet>266</a> <span class="kw">let </span>(sender, receiver) = channel();
-<a href=#267 id=267 data-nosnippet>267</a> del_filter.insert_equality_delete(<span class="kw-2">&</span>task.file_path, receiver);
-<a href=#268 id=268 data-nosnippet>268</a>
-<a href=#269 id=269 data-nosnippet>269</a> <span class="comment">// Per the Iceberg spec, evolve schema for equality deletes but only for the
-<a href=#270 id=270 data-nosnippet>270</a> // equality_ids columns, not all table columns.
-<a href=#271 id=271 data-nosnippet>271</a> </span><span class="kw">let </span>equality_ids_vec = task.equality_ids.clone().unwrap();
-<a href=#272 id=272 data-nosnippet>272</a> <span class="kw">let </span>evolved_stream = BasicDeleteFileLoader::evolve_schema(
-<a href=#273 id=273 data-nosnippet>273</a> basic_delete_file_loader
-<a href=#274 id=274 data-nosnippet>274</a> .parquet_to_batch_stream(<span class="kw-2">&</span>task.file_path, task.file_size_in_bytes)
-<a href=#275 id=275 data-nosnippet>275</a> .<span class="kw">await</span><span class="question-mark">?</span>,
-<a href=#276 id=276 data-nosnippet>276</a> schema,
-<a href=#277 id=277 data-nosnippet>277</a> <span class="kw-2">&</span>equality_ids_vec,
-<a href=#278 id=278 data-nosnippet>278</a> )
-<a href=#279 id=279 data-nosnippet>279</a> .<span class="kw">await</span><span class="question-mark">?</span>;
-<a href=#280 id=280 data-nosnippet>280</a>
-<a href=#281 id=281 data-nosnippet>281</a> <span class="prelude-val">Ok</span>(DeleteFileContext::FreshEqDel {
-<a href=#282 id=282 data-nosnippet>282</a> batch_stream: evolved_stream,
-<a href=#283 id=283 data-nosnippet>283</a> sender,
-<a href=#284 id=284 data-nosnippet>284</a> equality_ids: HashSet::from_iter(equality_ids_vec),
-<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>
-<a href=#288 id=288 data-nosnippet>288</a> DataContentType::Data => <span class="prelude-val">Err</span>(Error::new(
-<a href=#289 id=289 data-nosnippet>289</a> ErrorKind::Unexpected,
-<a href=#290 id=290 data-nosnippet>290</a> <span class="string">"tasks with files of type Data not expected here"</span>,
-<a href=#291 id=291 data-nosnippet>291</a> )),
-<a href=#292 id=292 data-nosnippet>292</a> }
-<a href=#293 id=293 data-nosnippet>293</a> }
-<a href=#294 id=294 data-nosnippet>294</a>
-<a href=#295 id=295 data-nosnippet>295</a> <span class="kw">async fn </span>parse_file_content_for_task(
-<a href=#296 id=296 data-nosnippet>296</a> ctx: DeleteFileContext,
-<a href=#297 id=297 data-nosnippet>297</a> ) -> <span class="prelude-ty">Result</span><ParsedDeleteFileContext> {
-<a href=#298 id=298 data-nosnippet>298</a> <span class="kw">match </span>ctx {
-<a href=#299 id=299 data-nosnippet>299</a> DeleteFileContext::ExistingEqDel => <span class="prelude-val">Ok</span>(ParsedDeleteFileContext::EqDel),
-<a href=#300 id=300 data-nosnippet>300</a> DeleteFileContext::ExistingPosDel => <span class="prelude-val">Ok</span>(ParsedDeleteFileContext::ExistingPosDel),
-<a href=#301 id=301 data-nosnippet>301</a> DeleteFileContext::PosDels { file_path, stream } => {
-<a href=#302 id=302 data-nosnippet>302</a> <span class="kw">let </span>del_vecs = <span class="self">Self</span>::parse_positional_deletes_record_batch_stream(stream).<span class="kw">await</span><span class="question-mark">?</span>;
-<a href=#303 id=303 data-nosnippet>303</a> <span class="prelude-val">Ok</span>(ParsedDeleteFileContext::DelVecs {
-<a href=#304 id=304 data-nosnippet>304</a> file_path,
-<a href=#305 id=305 data-nosnippet>305</a> results: del_vecs,
-<a href=#306 id=306 data-nosnippet>306</a> })
-<a href=#307 id=307 data-nosnippet>307</a> }
-<a href=#308 id=308 data-nosnippet>308</a> DeleteFileContext::FreshEqDel {
-<a href=#309 id=309 data-nosnippet>309</a> sender,
-<a href=#310 id=310 data-nosnippet>310</a> batch_stream,
-<a href=#311 id=311 data-nosnippet>311</a> equality_ids,
-<a href=#312 id=312 data-nosnippet>312</a> } => {
-<a href=#313 id=313 data-nosnippet>313</a> <span class="kw">let </span>predicate =
-<a href=#314 id=314 data-nosnippet>314</a> <span class="self">Self</span>::parse_equality_deletes_record_batch_stream(batch_stream, equality_ids)
-<a href=#315 id=315 data-nosnippet>315</a> .<span class="kw">await</span><span class="question-mark">?</span>;
-<a href=#316 id=316 data-nosnippet>316</a>
-<a href=#317 id=317 data-nosnippet>317</a> sender
-<a href=#318 id=318 data-nosnippet>318</a> .send(predicate)
-<a href=#319 id=319 data-nosnippet>319</a> .map_err(|err| {
-<a href=#320 id=320 data-nosnippet>320</a> Error::new(
-<a href=#321 id=321 data-nosnippet>321</a> ErrorKind::Unexpected,
-<a href=#322 id=322 data-nosnippet>322</a> <span class="string">"Could not send eq delete predicate to state"</span>,
-<a href=#323 id=323 data-nosnippet>323</a> )
-<a href=#324 id=324 data-nosnippet>324</a> })
-<a href=#325 id=325 data-nosnippet>325</a> .map(|<span class="kw">_</span>| ParsedDeleteFileContext::EqDel)
-<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> }
-<a href=#329 id=329 data-nosnippet>329</a>
-<a href=#330 id=330 data-nosnippet>330</a> <span class="doccomment">/// Parses a record batch stream coming from positional delete files
-<a href=#331 id=331 data-nosnippet>331</a> ///
-<a href=#332 id=332 data-nosnippet>332</a> /// Returns a map of data file path to a delete vector
-<a href=#333 id=333 data-nosnippet>333</a> </span><span class="kw">async fn </span>parse_positional_deletes_record_batch_stream(
-<a href=#334 id=334 data-nosnippet>334</a> <span class="kw-2">mut </span>stream: ArrowRecordBatchStream,
-<a href=#335 id=335 data-nosnippet>335</a> ) -> <span class="prelude-ty">Result</span><HashMap<String, DeleteVector>> {
-<a href=#336 id=336 data-nosnippet>336</a> <span class="kw">let </span><span class="kw-2">mut </span>result: HashMap<String, DeleteVector> = HashMap::default();
-<a href=#337 id=337 data-nosnippet>337</a>
-<a href=#338 id=338 data-nosnippet>338</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(batch) = stream.next().<span class="kw">await </span>{
-<a href=#339 id=339 data-nosnippet>339</a> <span class="kw">let </span>batch = batch<span class="question-mark">?</span>;
-<a href=#340 id=340 data-nosnippet>340</a> <span class="kw">let </span>schema = batch.schema();
-<a href=#341 id=341 data-nosnippet>341</a> <span class="kw">let </span>columns = batch.columns();
-<a href=#342 id=342 data-nosnippet>342</a>
-<a href=#343 id=343 data-nosnippet>343</a> <span class="kw">let </span><span class="prelude-val">Some</span>(file_paths) = columns[<span class="number">0</span>].as_any().downcast_ref::<StringArray>() <span class="kw">else </span>{
-<a href=#344 id=344 data-nosnippet>344</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#345 id=345 data-nosnippet>345</a> ErrorKind::DataInvalid,
-<a href=#346 id=346 data-nosnippet>346</a> <span class="string">"Could not downcast file paths array to StringArray"</span>,
-<a href=#347 id=347 data-nosnippet>347</a> ));
-<a href=#348 id=348 data-nosnippet>348</a> };
-<a href=#349 id=349 data-nosnippet>349</a> <span class="kw">let </span><span class="prelude-val">Some</span>(positions) = columns[<span class="number">1</span>].as_any().downcast_ref::<Int64Array>() <span class="kw">else </span>{
-<a href=#350 id=350 data-nosnippet>350</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#351 id=351 data-nosnippet>351</a> ErrorKind::DataInvalid,
-<a href=#352 id=352 data-nosnippet>352</a> <span class="string">"Could not downcast positions array to Int64Array"</span>,
-<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>
-<a href=#356 id=356 data-nosnippet>356</a> <span class="kw">for </span>(file_path, pos) <span class="kw">in </span>file_paths.iter().zip(positions.iter()) {
-<a href=#357 id=357 data-nosnippet>357</a> <span class="kw">let </span>(<span class="prelude-val">Some</span>(file_path), <span class="prelude-val">Some</span>(pos)) = (file_path, pos) <span class="kw">else </span>{
-<a href=#358 id=358 data-nosnippet>358</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#359 id=359 data-nosnippet>359</a> ErrorKind::DataInvalid,
-<a href=#360 id=360 data-nosnippet>360</a> <span class="string">"null values in delete file"</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> <span class="kw">if </span>pos < <span class="number">0 </span>{
-<a href=#364 id=364 data-nosnippet>364</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#365 id=365 data-nosnippet>365</a> ErrorKind::DataInvalid,
-<a href=#366 id=366 data-nosnippet>366</a> <span class="macro">format!</span>(<span class="string">"negative position in delete file {file_path}: {pos}"</span>),
-<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>
-<a href=#370 id=370 data-nosnippet>370</a> result
-<a href=#371 id=371 data-nosnippet>371</a> .entry(file_path.to_string())
-<a href=#372 id=372 data-nosnippet>372</a> .or_default()
-<a href=#373 id=373 data-nosnippet>373</a> .insert(pos <span class="kw">as </span>u64);
-<a href=#374 id=374 data-nosnippet>374</a> }
-<a href=#375 id=375 data-nosnippet>375</a> }
-<a href=#376 id=376 data-nosnippet>376</a>
-<a href=#377 id=377 data-nosnippet>377</a> <span class="prelude-val">Ok</span>(result)
-<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">async fn </span>parse_equality_deletes_record_batch_stream(
-<a href=#381 id=381 data-nosnippet>381</a> <span class="kw-2">mut </span>stream: ArrowRecordBatchStream,
-<a href=#382 id=382 data-nosnippet>382</a> equality_ids: HashSet<i32>,
-<a href=#383 id=383 data-nosnippet>383</a> ) -> <span class="prelude-ty">Result</span><Predicate> {
-<a href=#384 id=384 data-nosnippet>384</a> <span class="kw">let </span><span class="kw-2">mut </span>row_predicates = Vec::new();
-<a href=#385 id=385 data-nosnippet>385</a> <span class="kw">let </span><span class="kw-2">mut </span>batch_schema_iceberg: <span class="prelude-ty">Option</span><Schema> = <span class="prelude-val">None</span>;
-<a href=#386 id=386 data-nosnippet>386</a> <span class="kw">let </span>accessor = EqDelRecordBatchPartnerAccessor;
-<a href=#387 id=387 data-nosnippet>387</a>
-<a href=#388 id=388 data-nosnippet>388</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(record_batch) = stream.next().<span class="kw">await </span>{
-<a href=#389 id=389 data-nosnippet>389</a> <span class="kw">let </span>record_batch = record_batch<span class="question-mark">?</span>;
-<a href=#390 id=390 data-nosnippet>390</a>
-<a href=#391 id=391 data-nosnippet>391</a> <span class="kw">if </span>record_batch.num_columns() == <span class="number">0 </span>{
-<a href=#392 id=392 data-nosnippet>392</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(AlwaysTrue);
-<a href=#393 id=393 data-nosnippet>393</a> }
-<a href=#394 id=394 data-nosnippet>394</a>
-<a href=#395 id=395 data-nosnippet>395</a> <span class="kw">let </span>schema = <span class="kw">match </span><span class="kw-2">&</span>batch_schema_iceberg {
-<a href=#396 id=396 data-nosnippet>396</a> <span class="prelude-val">Some</span>(schema) => schema,
-<a href=#397 id=397 data-nosnippet>397</a> <span class="prelude-val">None </span>=> {
-<a href=#398 id=398 data-nosnippet>398</a> <span class="kw">let </span>schema = arrow_schema_to_schema(record_batch.schema().as_ref())<span class="question-mark">?</span>;
-<a href=#399 id=399 data-nosnippet>399</a> batch_schema_iceberg = <span class="prelude-val">Some</span>(schema);
-<a href=#400 id=400 data-nosnippet>400</a> batch_schema_iceberg.as_ref().unwrap()
-<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>
-<a href=#404 id=404 data-nosnippet>404</a> <span class="kw">let </span>root_array: ArrayRef = Arc::new(StructArray::from(record_batch));
-<a href=#405 id=405 data-nosnippet>405</a>
-<a href=#406 id=406 data-nosnippet>406</a> <span class="kw">let </span><span class="kw-2">mut </span>processor = EqDelColumnProcessor::new(<span class="kw-2">&</span>equality_ids);
-<a href=#407 id=407 data-nosnippet>407</a> visit_schema_with_partner(schema, <span class="kw-2">&</span>root_array, <span class="kw-2">&mut </span>processor, <span class="kw-2">&</span>accessor)<span class="question-mark">?</span>;
-<a href=#408 id=408 data-nosnippet>408</a>
-<a href=#409 id=409 data-nosnippet>409</a> <span class="kw">let </span><span class="kw-2">mut </span>datum_columns_with_names = processor.finish()<span class="question-mark">?</span>;
-<a href=#410 id=410 data-nosnippet>410</a> <span class="kw">if </span>datum_columns_with_names.is_empty() {
-<a href=#411 id=411 data-nosnippet>411</a> <span class="kw">continue</span>;
-<a href=#412 id=412 data-nosnippet>412</a> }
-<a href=#413 id=413 data-nosnippet>413</a>
-<a href=#414 id=414 data-nosnippet>414</a> <span class="comment">// Process the collected columns in lockstep
-<a href=#415 id=415 data-nosnippet>415</a> </span><span class="attr">#[allow(clippy::len_zero)]
-<a href=#416 id=416 data-nosnippet>416</a> </span><span class="kw">while </span>datum_columns_with_names[<span class="number">0</span>].<span class="number">0</span>.len() > <span class="number">0 </span>{
-<a href=#417 id=417 data-nosnippet>417</a> <span class="kw">let </span><span class="kw-2">mut </span>row_predicate = AlwaysTrue;
-<a href=#418 id=418 data-nosnippet>418</a> <span class="kw">for </span><span class="kw-2">&mut </span>(<span class="kw-2">ref mut </span>column, <span class="kw-2">ref </span>field_name) <span class="kw">in </span><span class="kw-2">&mut </span>datum_columns_with_names {
-<a href=#419 id=419 data-nosnippet>419</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(item) = column.next() {
-<a href=#420 id=420 data-nosnippet>420</a> <span class="kw">let </span>cell_predicate = <span class="kw">if let </span><span class="prelude-val">Some</span>(datum) = item<span class="question-mark">? </span>{
-<a href=#421 id=421 data-nosnippet>421</a> Reference::new(field_name.clone()).equal_to(datum.clone())
-<a href=#422 id=422 data-nosnippet>422</a> } <span class="kw">else </span>{
-<a href=#423 id=423 data-nosnippet>423</a> Reference::new(field_name.clone()).is_null()
-<a href=#424 id=424 data-nosnippet>424</a> };
-<a href=#425 id=425 data-nosnippet>425</a> row_predicate = row_predicate.and(cell_predicate)
-<a href=#426 id=426 data-nosnippet>426</a> }
-<a href=#427 id=427 data-nosnippet>427</a> }
-<a href=#428 id=428 data-nosnippet>428</a> row_predicates.push(row_predicate.not().rewrite_not());
-<a href=#429 id=429 data-nosnippet>429</a> }
-<a href=#430 id=430 data-nosnippet>430</a> }
-<a href=#431 id=431 data-nosnippet>431</a>
-<a href=#432 id=432 data-nosnippet>432</a> <span class="comment">// All row predicates are combined to a single predicate by creating a balanced binary tree.
-<a href=#433 id=433 data-nosnippet>433</a> // Using a simple fold would result in a deeply nested predicate that can cause a stack overflow.
-<a href=#434 id=434 data-nosnippet>434</a> </span><span class="kw">while </span>row_predicates.len() > <span class="number">1 </span>{
-<a href=#435 id=435 data-nosnippet>435</a> <span class="kw">let </span><span class="kw-2">mut </span>next_level = Vec::with_capacity(row_predicates.len().div_ceil(<span class="number">2</span>));
-<a href=#436 id=436 data-nosnippet>436</a> <span class="kw">let </span><span class="kw-2">mut </span>iter = row_predicates.into_iter();
-<a href=#437 id=437 data-nosnippet>437</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(p1) = iter.next() {
-<a href=#438 id=438 data-nosnippet>438</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(p2) = iter.next() {
-<a href=#439 id=439 data-nosnippet>439</a> next_level.push(p1.and(p2));
-<a href=#440 id=440 data-nosnippet>440</a> } <span class="kw">else </span>{
-<a href=#441 id=441 data-nosnippet>441</a> next_level.push(p1);
-<a href=#442 id=442 data-nosnippet>442</a> }
-<a href=#443 id=443 data-nosnippet>443</a> }
-<a href=#444 id=444 data-nosnippet>444</a> row_predicates = next_level;
-<a href=#445 id=445 data-nosnippet>445</a> }
-<a href=#446 id=446 data-nosnippet>446</a>
-<a href=#447 id=447 data-nosnippet>447</a> <span class="kw">match </span>row_predicates.pop() {
-<a href=#448 id=448 data-nosnippet>448</a> <span class="prelude-val">Some</span>(p) => <span class="prelude-val">Ok</span>(p),
-<a href=#449 id=449 data-nosnippet>449</a> <span class="prelude-val">None </span>=> <span class="prelude-val">Ok</span>(AlwaysTrue),
-<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>}
-<a href=#453 id=453 data-nosnippet>453</a>
-<a href=#454 id=454 data-nosnippet>454</a><span class="kw">struct </span>EqDelColumnProcessor<<span class="lifetime">'a</span>> {
-<a href=#455 id=455 data-nosnippet>455</a> equality_ids: <span class="kw-2">&</span><span class="lifetime">'a </span>HashSet<i32>,
-<a href=#456 id=456 data-nosnippet>456</a> collected_columns: Vec<(ArrayRef, String, Type)>,
-<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">impl</span><<span class="lifetime">'a</span>> EqDelColumnProcessor<<span class="lifetime">'a</span>> {
-<a href=#460 id=460 data-nosnippet>460</a> <span class="kw">fn </span>new(equality_ids: <span class="kw-2">&</span><span class="lifetime">'a </span>HashSet<i32>) -> <span class="self">Self </span>{
-<a href=#461 id=461 data-nosnippet>461</a> <span class="self">Self </span>{
-<a href=#462 id=462 data-nosnippet>462</a> equality_ids,
-<a href=#463 id=463 data-nosnippet>463</a> collected_columns: Vec::with_capacity(equality_ids.len()),
-<a href=#464 id=464 data-nosnippet>464</a> }
-<a href=#465 id=465 data-nosnippet>465</a> }
-<a href=#466 id=466 data-nosnippet>466</a>
-<a href=#467 id=467 data-nosnippet>467</a> <span class="attr">#[allow(clippy::type_complexity)]
-<a href=#468 id=468 data-nosnippet>468</a> </span><span class="kw">fn </span>finish(
-<a href=#469 id=469 data-nosnippet>469</a> <span class="self">self</span>,
-<a href=#470 id=470 data-nosnippet>470</a> ) -> <span class="prelude-ty">Result</span><
-<a href=#471 id=471 data-nosnippet>471</a> Vec<(
-<a href=#472 id=472 data-nosnippet>472</a> Box<<span class="kw">dyn </span>ExactSizeIterator<Item = <span class="prelude-ty">Result</span><<span class="prelude-ty">Option</span><Datum>>>>,
-<a href=#473 id=473 data-nosnippet>473</a> String,
-<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="self">self</span>.collected_columns
-<a href=#477 id=477 data-nosnippet>477</a> .into_iter()
-<a href=#478 id=478 data-nosnippet>478</a> .map(|(array, field_name, field_type)| {
-<a href=#479 id=479 data-nosnippet>479</a> <span class="kw">let </span>primitive_type = field_type
-<a href=#480 id=480 data-nosnippet>480</a> .as_primitive_type()
-<a href=#481 id=481 data-nosnippet>481</a> .ok_or_else(|| {
-<a href=#482 id=482 data-nosnippet>482</a> Error::new(ErrorKind::Unexpected, <span class="string">"field is not a primitive type"</span>)
-<a href=#483 id=483 data-nosnippet>483</a> })<span class="question-mark">?
-<a href=#484 id=484 data-nosnippet>484</a> </span>.clone();
-<a href=#485 id=485 data-nosnippet>485</a>
-<a href=#486 id=486 data-nosnippet>486</a> <span class="kw">let </span>lit_vec = arrow_primitive_to_literal(<span class="kw-2">&</span>array, <span class="kw-2">&</span>field_type)<span class="question-mark">?</span>;
-<a href=#487 id=487 data-nosnippet>487</a> <span class="kw">let </span>datum_iterator: Box<<span class="kw">dyn </span>ExactSizeIterator<Item = <span class="prelude-ty">Result</span><<span class="prelude-ty">Option</span><Datum>>>> =
-<a href=#488 id=488 data-nosnippet>488</a> Box::new(lit_vec.into_iter().map(<span class="kw">move </span>|c| {
-<a href=#489 id=489 data-nosnippet>489</a> c.map(|literal| {
-<a href=#490 id=490 data-nosnippet>490</a> literal
-<a href=#491 id=491 data-nosnippet>491</a> .as_primitive_literal()
-<a href=#492 id=492 data-nosnippet>492</a> .map(|primitive_literal| {
-<a href=#493 id=493 data-nosnippet>493</a> Datum::new(primitive_type.clone(), primitive_literal)
-<a href=#494 id=494 data-nosnippet>494</a> })
-<a href=#495 id=495 data-nosnippet>495</a> .ok_or(Error::new(
-<a href=#496 id=496 data-nosnippet>496</a> ErrorKind::Unexpected,
-<a href=#497 id=497 data-nosnippet>497</a> <span class="string">"failed to convert to primitive literal"</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> .transpose()
-<a href=#501 id=501 data-nosnippet>501</a> }));
-<a href=#502 id=502 data-nosnippet>502</a>
-<a href=#503 id=503 data-nosnippet>503</a> <span class="prelude-val">Ok</span>((datum_iterator, field_name))
-<a href=#504 id=504 data-nosnippet>504</a> })
-<a href=#505 id=505 data-nosnippet>505</a> .collect::<<span class="prelude-ty">Result</span><Vec<<span class="kw">_</span>>>>()
-<a href=#506 id=506 data-nosnippet>506</a> }
-<a href=#507 id=507 data-nosnippet>507</a>}
+<a href=#165 id=165 data-nosnippet>165</a> /// [join!]
+<a href=#166 id=166 data-nosnippet>166</a> /// ```
+<a href=#167 id=167 data-nosnippet>167</a> </span><span class="kw">pub</span>(<span class="kw">crate</span>) <span class="kw">fn </span>load_deletes(
+<a href=#168 id=168 data-nosnippet>168</a> <span class="kw-2">&</span><span class="self">self</span>,
+<a href=#169 id=169 data-nosnippet>169</a> delete_file_entries: <span class="kw-2">&</span>[FileScanTaskDeleteFile],
+<a href=#170 id=170 data-nosnippet>170</a> schema: SchemaRef,
+<a href=#171 id=171 data-nosnippet>171</a> ) -> Receiver<<span class="prelude-ty">Result</span><DeleteFilter>> {
+<a href=#172 id=172 data-nosnippet>172</a> <span class="kw">let </span>(tx, rx) = channel();
+<a href=#173 id=173 data-nosnippet>173</a>
+<a href=#174 id=174 data-nosnippet>174</a> <span class="kw">let </span>stream_items = delete_file_entries
+<a href=#175 id=175 data-nosnippet>175</a> .iter()
+<a href=#176 id=176 data-nosnippet>176</a> .map(|t| {
+<a href=#177 id=177 data-nosnippet>177</a> (
+<a href=#178 id=178 data-nosnippet>178</a> t.clone(),
+<a href=#179 id=179 data-nosnippet>179</a> <span class="self">self</span>.basic_delete_file_loader.clone(),
+<a href=#180 id=180 data-nosnippet>180</a> <span class="self">self</span>.delete_filter.clone(),
+<a href=#181 id=181 data-nosnippet>181</a> schema.clone(),
+<a href=#182 id=182 data-nosnippet>182</a> )
+<a href=#183 id=183 data-nosnippet>183</a> })
+<a href=#184 id=184 data-nosnippet>184</a> .collect::<Vec<<span class="kw">_</span>>>();
+<a href=#185 id=185 data-nosnippet>185</a> <span class="kw">let </span>task_stream = futures::stream::iter(stream_items);
+<a href=#186 id=186 data-nosnippet>186</a>
+<a href=#187 id=187 data-nosnippet>187</a> <span class="kw">let </span>del_filter = <span class="self">self</span>.delete_filter.clone();
+<a href=#188 id=188 data-nosnippet>188</a> <span class="kw">let </span>concurrency_limit_data_files = <span class="self">self</span>.concurrency_limit_data_files;
+<a href=#189 id=189 data-nosnippet>189</a> <span class="kw">let </span>basic_delete_file_loader = <span class="self">self</span>.basic_delete_file_loader.clone();
+<a href=#190 id=190 data-nosnippet>190</a> <span class="self">self</span>.runtime.io().spawn(<span class="kw">async move </span>{
+<a href=#191 id=191 data-nosnippet>191</a> <span class="kw">let </span>result = <span class="kw">async move </span>{
+<a href=#192 id=192 data-nosnippet>192</a> <span class="kw">let </span><span class="kw-2">mut </span>del_filter = del_filter;
+<a href=#193 id=193 data-nosnippet>193</a> <span class="kw">let </span>basic_delete_file_loader = basic_delete_file_loader.clone();
+<a href=#194 id=194 data-nosnippet>194</a>
+<a href=#195 id=195 data-nosnippet>195</a> <span class="kw">let </span><span class="kw-2">mut </span>results_stream = task_stream
+<a href=#196 id=196 data-nosnippet>196</a> .map(<span class="kw">move </span>|(task, file_io, del_filter, schema)| {
+<a href=#197 id=197 data-nosnippet>197</a> <span class="kw">let </span>basic_delete_file_loader = basic_delete_file_loader.clone();
+<a href=#198 id=198 data-nosnippet>198</a> <span class="kw">async move </span>{
+<a href=#199 id=199 data-nosnippet>199</a> <span class="self">Self</span>::load_file_for_task(
+<a href=#200 id=200 data-nosnippet>200</a> <span class="kw-2">&</span>task,
+<a href=#201 id=201 data-nosnippet>201</a> basic_delete_file_loader.clone(),
+<a href=#202 id=202 data-nosnippet>202</a> del_filter,
+<a href=#203 id=203 data-nosnippet>203</a> schema,
+<a href=#204 id=204 data-nosnippet>204</a> )
+<a href=#205 id=205 data-nosnippet>205</a> .<span class="kw">await
+<a href=#206 id=206 data-nosnippet>206</a> </span>}
+<a href=#207 id=207 data-nosnippet>207</a> })
+<a href=#208 id=208 data-nosnippet>208</a> .map(<span class="kw">move </span>|ctx| {
+<a href=#209 id=209 data-nosnippet>209</a> <span class="prelude-val">Ok</span>(<span class="kw">async </span>{ <span class="self">Self</span>::parse_file_content_for_task(ctx.<span class="kw">await</span><span class="question-mark">?</span>).<span class="kw">await </span>})
+<a href=#210 id=210 data-nosnippet>210</a> })
+<a href=#211 id=211 data-nosnippet>211</a> .try_buffer_unordered(concurrency_limit_data_files);
+<a href=#212 id=212 data-nosnippet>212</a>
+<a href=#213 id=213 data-nosnippet>213</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(item) = results_stream.next().<span class="kw">await </span>{
+<a href=#214 id=214 data-nosnippet>214</a> <span class="kw">let </span>item = item<span class="question-mark">?</span>;
+<a href=#215 id=215 data-nosnippet>215</a> <span class="kw">if let </span>ParsedDeleteFileContext::DelVecs { file_path, results } = item {
+<a href=#216 id=216 data-nosnippet>216</a> <span class="kw">for </span>(data_file_path, delete_vector) <span class="kw">in </span>results.into_iter() {
+<a href=#217 id=217 data-nosnippet>217</a> del_filter.upsert_delete_vector(data_file_path, delete_vector);
+<a href=#218 id=218 data-nosnippet>218</a> }
+<a href=#219 id=219 data-nosnippet>219</a> <span class="comment">// Mark the positional delete file as fully loaded so waiters can proceed
+<a href=#220 id=220 data-nosnippet>220</a> </span>del_filter.finish_pos_del_load(<span class="kw-2">&</span>file_path);
+<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>
+<a href=#224 id=224 data-nosnippet>224</a> <span class="prelude-val">Ok</span>(del_filter)
+<a href=#225 id=225 data-nosnippet>225</a> }
+<a href=#226 id=226 data-nosnippet>226</a> .<span class="kw">await</span>;
+<a href=#227 id=227 data-nosnippet>227</a>
+<a href=#228 id=228 data-nosnippet>228</a> <span class="kw">let _ </span>= tx.send(result);
+<a href=#229 id=229 data-nosnippet>229</a> });
+<a href=#230 id=230 data-nosnippet>230</a>
+<a href=#231 id=231 data-nosnippet>231</a> rx
+<a href=#232 id=232 data-nosnippet>232</a> }
+<a href=#233 id=233 data-nosnippet>233</a>
+<a href=#234 id=234 data-nosnippet>234</a> <span class="kw">async fn </span>load_file_for_task(
+<a href=#235 id=235 data-nosnippet>235</a> task: <span class="kw-2">&</span>FileScanTaskDeleteFile,
+<a href=#236 id=236 data-nosnippet>236</a> basic_delete_file_loader: BasicDeleteFileLoader,
+<a href=#237 id=237 data-nosnippet>237</a> del_filter: DeleteFilter,
+<a href=#238 id=238 data-nosnippet>238</a> schema: SchemaRef,
+<a href=#239 id=239 data-nosnippet>239</a> ) -> <span class="prelude-ty">Result</span><DeleteFileContext> {
+<a href=#240 id=240 data-nosnippet>240</a> <span class="kw">match </span>task.file_type {
+<a href=#241 id=241 data-nosnippet>241</a> DataContentType::PositionDeletes => {
+<a href=#242 id=242 data-nosnippet>242</a> <span class="kw">match </span>del_filter.try_start_pos_del_load(<span class="kw-2">&</span>task.file_path) {
+<a href=#243 id=243 data-nosnippet>243</a> PosDelLoadAction::AlreadyLoaded => <span class="prelude-val">Ok</span>(DeleteFileContext::ExistingPosDel),
+<a href=#244 id=244 data-nosnippet>244</a> PosDelLoadAction::WaitFor(notify) => {
+<a href=#245 id=245 data-nosnippet>245</a> <span class="comment">// Positional deletes are accessed synchronously by ArrowReader.
+<a href=#246 id=246 data-nosnippet>246</a> // We must wait here to ensure the data is ready before returning,
+<a href=#247 id=247 data-nosnippet>247</a> // otherwise ArrowReader might get an empty/partial result.
+<a href=#248 id=248 data-nosnippet>248</a> </span>notify.notified().<span class="kw">await</span>;
+<a href=#249 id=249 data-nosnippet>249</a> <span class="prelude-val">Ok</span>(DeleteFileContext::ExistingPosDel)
+<a href=#250 id=250 data-nosnippet>250</a> }
+<a href=#251 id=251 data-nosnippet>251</a> PosDelLoadAction::Load => <span class="prelude-val">Ok</span>(DeleteFileContext::PosDels {
+<a href=#252 id=252 data-nosnippet>252</a> file_path: task.file_path.clone(),
+<a href=#253 id=253 data-nosnippet>253</a> stream: basic_delete_file_loader
+<a href=#254 id=254 data-nosnippet>254</a> .parquet_to_batch_stream(<span class="kw-2">&</span>task.file_path, task.file_size_in_bytes)
+<a href=#255 id=255 data-nosnippet>255</a> .<span class="kw">await</span><span class="question-mark">?</span>,
+<a href=#256 id=256 data-nosnippet>256</a> }),
+<a href=#257 id=257 data-nosnippet>257</a> }
+<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> DataContentType::EqualityDeletes => {
+<a href=#261 id=261 data-nosnippet>261</a> <span class="kw">let </span><span class="prelude-val">Some</span>(notify) = del_filter.try_start_eq_del_load(<span class="kw-2">&</span>task.file_path) <span class="kw">else </span>{
+<a href=#262 id=262 data-nosnippet>262</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(DeleteFileContext::ExistingEqDel);
+<a href=#263 id=263 data-nosnippet>263</a> };
+<a href=#264 id=264 data-nosnippet>264</a>
+<a href=#265 id=265 data-nosnippet>265</a> <span class="kw">let </span>(sender, receiver) = channel();
+<a href=#266 id=266 data-nosnippet>266</a> del_filter.insert_equality_delete(<span class="kw-2">&</span>task.file_path, receiver);
+<a href=#267 id=267 data-nosnippet>267</a>
+<a href=#268 id=268 data-nosnippet>268</a> <span class="comment">// Per the Iceberg spec, evolve schema for equality deletes but only for the
+<a href=#269 id=269 data-nosnippet>269</a> // equality_ids columns, not all table columns.
+<a href=#270 id=270 data-nosnippet>270</a> </span><span class="kw">let </span>equality_ids_vec = task.equality_ids.clone().unwrap();
+<a href=#271 id=271 data-nosnippet>271</a> <span class="kw">let </span>evolved_stream = BasicDeleteFileLoader::evolve_schema(
+<a href=#272 id=272 data-nosnippet>272</a> basic_delete_file_loader
+<a href=#273 id=273 data-nosnippet>273</a> .parquet_to_batch_stream(<span class="kw-2">&</span>task.file_path, task.file_size_in_bytes)
+<a href=#274 id=274 data-nosnippet>274</a> .<span class="kw">await</span><span class="question-mark">?</span>,
+<a href=#275 id=275 data-nosnippet>275</a> schema,
+<a href=#276 id=276 data-nosnippet>276</a> <span class="kw-2">&</span>equality_ids_vec,
+<a href=#277 id=277 data-nosnippet>277</a> )
+<a href=#278 id=278 data-nosnippet>278</a> .<span class="kw">await</span><span class="question-mark">?</span>;
+<a href=#279 id=279 data-nosnippet>279</a>
+<a href=#280 id=280 data-nosnippet>280</a> <span class="prelude-val">Ok</span>(DeleteFileContext::FreshEqDel {
+<a href=#281 id=281 data-nosnippet>281</a> batch_stream: evolved_stream,
+<a href=#282 id=282 data-nosnippet>282</a> sender,
+<a href=#283 id=283 data-nosnippet>283</a> equality_ids: HashSet::from_iter(equality_ids_vec),
+<a href=#284 id=284 data-nosnippet>284</a> })
+<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> DataContentType::Data => <span class="prelude-val">Err</span>(Error::new(
+<a href=#288 id=288 data-nosnippet>288</a> ErrorKind::Unexpected,
+<a href=#289 id=289 data-nosnippet>289</a> <span class="string">"tasks with files of type Data not expected here"</span>,
+<a href=#290 id=290 data-nosnippet>290</a> )),
+<a href=#291 id=291 data-nosnippet>291</a> }
+<a href=#292 id=292 data-nosnippet>292</a> }
+<a href=#293 id=293 data-nosnippet>293</a>
+<a href=#294 id=294 data-nosnippet>294</a> <span class="kw">async fn </span>parse_file_content_for_task(
+<a href=#295 id=295 data-nosnippet>295</a> ctx: DeleteFileContext,
+<a href=#296 id=296 data-nosnippet>296</a> ) -> <span class="prelude-ty">Result</span><ParsedDeleteFileContext> {
+<a href=#297 id=297 data-nosnippet>297</a> <span class="kw">match </span>ctx {
+<a href=#298 id=298 data-nosnippet>298</a> DeleteFileContext::ExistingEqDel => <span class="prelude-val">Ok</span>(ParsedDeleteFileContext::EqDel),
+<a href=#299 id=299 data-nosnippet>299</a> DeleteFileContext::ExistingPosDel => <span class="prelude-val">Ok</span>(ParsedDeleteFileContext::ExistingPosDel),
+<a href=#300 id=300 data-nosnippet>300</a> DeleteFileContext::PosDels { file_path, stream } => {
+<a href=#301 id=301 data-nosnippet>301</a> <span class="kw">let </span>del_vecs = <span class="self">Self</span>::parse_positional_deletes_record_batch_stream(stream).<span class="kw">await</span><span class="question-mark">?</span>;
+<a href=#302 id=302 data-nosnippet>302</a> <span class="prelude-val">Ok</span>(ParsedDeleteFileContext::DelVecs {
+<a href=#303 id=303 data-nosnippet>303</a> file_path,
+<a href=#304 id=304 data-nosnippet>304</a> results: del_vecs,
+<a href=#305 id=305 data-nosnippet>305</a> })
+<a href=#306 id=306 data-nosnippet>306</a> }
+<a href=#307 id=307 data-nosnippet>307</a> DeleteFileContext::FreshEqDel {
+<a href=#308 id=308 data-nosnippet>308</a> sender,
+<a href=#309 id=309 data-nosnippet>309</a> batch_stream,
+<a href=#310 id=310 data-nosnippet>310</a> equality_ids,
+<a href=#311 id=311 data-nosnippet>311</a> } => {
+<a href=#312 id=312 data-nosnippet>312</a> <span class="kw">let </span>predicate =
+<a href=#313 id=313 data-nosnippet>313</a> <span class="self">Self</span>::parse_equality_deletes_record_batch_stream(batch_stream, equality_ids)
+<a href=#314 id=314 data-nosnippet>314</a> .<span class="kw">await</span><span class="question-mark">?</span>;
+<a href=#315 id=315 data-nosnippet>315</a>
+<a href=#316 id=316 data-nosnippet>316</a> sender
+<a href=#317 id=317 data-nosnippet>317</a> .send(predicate)
+<a href=#318 id=318 data-nosnippet>318</a> .map_err(|err| {
+<a href=#319 id=319 data-nosnippet>319</a> Error::new(
+<a href=#320 id=320 data-nosnippet>320</a> ErrorKind::Unexpected,
+<a href=#321 id=321 data-nosnippet>321</a> <span class="string">"Could not send eq delete predicate to state"</span>,
+<a href=#322 id=322 data-nosnippet>322</a> )
+<a href=#323 id=323 data-nosnippet>323</a> })
+<a href=#324 id=324 data-nosnippet>324</a> .map(|<span class="kw">_</span>| ParsedDeleteFileContext::EqDel)
+<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> }
+<a href=#328 id=328 data-nosnippet>328</a>
+<a href=#329 id=329 data-nosnippet>329</a> <span class="doccomment">/// Parses a record batch stream coming from positional delete files
+<a href=#330 id=330 data-nosnippet>330</a> ///
+<a href=#331 id=331 data-nosnippet>331</a> /// Returns a map of data file path to a delete vector
+<a href=#332 id=332 data-nosnippet>332</a> </span><span class="kw">async fn </span>parse_positional_deletes_record_batch_stream(
+<a href=#333 id=333 data-nosnippet>333</a> <span class="kw-2">mut </span>stream: ArrowRecordBatchStream,
+<a href=#334 id=334 data-nosnippet>334</a> ) -> <span class="prelude-ty">Result</span><HashMap<String, DeleteVector>> {
+<a href=#335 id=335 data-nosnippet>335</a> <span class="kw">let </span><span class="kw-2">mut </span>result: HashMap<String, DeleteVector> = HashMap::default();
+<a href=#336 id=336 data-nosnippet>336</a>
+<a href=#337 id=337 data-nosnippet>337</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(batch) = stream.next().<span class="kw">await </span>{
+<a href=#338 id=338 data-nosnippet>338</a> <span class="kw">let </span>batch = batch<span class="question-mark">?</span>;
+<a href=#339 id=339 data-nosnippet>339</a> <span class="kw">let </span>schema = batch.schema();
+<a href=#340 id=340 data-nosnippet>340</a> <span class="kw">let </span>columns = batch.columns();
+<a href=#341 id=341 data-nosnippet>341</a>
+<a href=#342 id=342 data-nosnippet>342</a> <span class="kw">let </span><span class="prelude-val">Some</span>(file_paths) = columns[<span class="number">0</span>].as_any().downcast_ref::<StringArray>() <span class="kw">else </span>{
+<a href=#343 id=343 data-nosnippet>343</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#344 id=344 data-nosnippet>344</a> ErrorKind::DataInvalid,
+<a href=#345 id=345 data-nosnippet>345</a> <span class="string">"Could not downcast file paths array to StringArray"</span>,
+<a href=#346 id=346 data-nosnippet>346</a> ));
+<a href=#347 id=347 data-nosnippet>347</a> };
+<a href=#348 id=348 data-nosnippet>348</a> <span class="kw">let </span><span class="prelude-val">Some</span>(positions) = columns[<span class="number">1</span>].as_any().downcast_ref::<Int64Array>() <span class="kw">else </span>{
+<a href=#349 id=349 data-nosnippet>349</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#350 id=350 data-nosnippet>350</a> ErrorKind::DataInvalid,
+<a href=#351 id=351 data-nosnippet>351</a> <span class="string">"Could not downcast positions array to Int64Array"</span>,
+<a href=#352 id=352 data-nosnippet>352</a> ));
+<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">for </span>(file_path, pos) <span class="kw">in </span>file_paths.iter().zip(positions.iter()) {
+<a href=#356 id=356 data-nosnippet>356</a> <span class="kw">let </span>(<span class="prelude-val">Some</span>(file_path), <span class="prelude-val">Some</span>(pos)) = (file_path, pos) <span class="kw">else </span>{
+<a href=#357 id=357 data-nosnippet>357</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#358 id=358 data-nosnippet>358</a> ErrorKind::DataInvalid,
+<a href=#359 id=359 data-nosnippet>359</a> <span class="string">"null values in delete file"</span>,
+<a href=#360 id=360 data-nosnippet>360</a> ));
+<a href=#361 id=361 data-nosnippet>361</a> };
+<a href=#362 id=362 data-nosnippet>362</a> <span class="kw">if </span>pos < <span class="number">0 </span>{
+<a href=#363 id=363 data-nosnippet>363</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#364 id=364 data-nosnippet>364</a> ErrorKind::DataInvalid,
+<a href=#365 id=365 data-nosnippet>365</a> <span class="macro">format!</span>(<span class="string">"negative position in delete file {file_path}: {pos}"</span>),
+<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> result
+<a href=#370 id=370 data-nosnippet>370</a> .entry(file_path.to_string())
+<a href=#371 id=371 data-nosnippet>371</a> .or_default()
+<a href=#372 id=372 data-nosnippet>372</a> .insert(pos <span class="kw">as </span>u64);
+<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>
+<a href=#376 id=376 data-nosnippet>376</a> <span class="prelude-val">Ok</span>(result)
+<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> <span class="kw">async fn </span>parse_equality_deletes_record_batch_stream(
+<a href=#380 id=380 data-nosnippet>380</a> <span class="kw-2">mut </span>stream: ArrowRecordBatchStream,
+<a href=#381 id=381 data-nosnippet>381</a> equality_ids: HashSet<i32>,
+<a href=#382 id=382 data-nosnippet>382</a> ) -> <span class="prelude-ty">Result</span><Predicate> {
+<a href=#383 id=383 data-nosnippet>383</a> <span class="kw">let </span><span class="kw-2">mut </span>row_predicates = Vec::new();
+<a href=#384 id=384 data-nosnippet>384</a> <span class="kw">let </span><span class="kw-2">mut </span>batch_schema_iceberg: <span class="prelude-ty">Option</span><Schema> = <span class="prelude-val">None</span>;
+<a href=#385 id=385 data-nosnippet>385</a> <span class="kw">let </span>accessor = EqDelRecordBatchPartnerAccessor;
+<a href=#386 id=386 data-nosnippet>386</a>
+<a href=#387 id=387 data-nosnippet>387</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(record_batch) = stream.next().<span class="kw">await </span>{
+<a href=#388 id=388 data-nosnippet>388</a> <span class="kw">let </span>record_batch = record_batch<span class="question-mark">?</span>;
+<a href=#389 id=389 data-nosnippet>389</a>
+<a href=#390 id=390 data-nosnippet>390</a> <span class="kw">if </span>record_batch.num_columns() == <span class="number">0 </span>{
+<a href=#391 id=391 data-nosnippet>391</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(AlwaysTrue);
+<a href=#392 id=392 data-nosnippet>392</a> }
+<a href=#393 id=393 data-nosnippet>393</a>
+<a href=#394 id=394 data-nosnippet>394</a> <span class="kw">let </span>schema = <span class="kw">match </span><span class="kw-2">&</span>batch_schema_iceberg {
+<a href=#395 id=395 data-nosnippet>395</a> <span class="prelude-val">Some</span>(schema) => schema,
+<a href=#396 id=396 data-nosnippet>396</a> <span class="prelude-val">None </span>=> {
+<a href=#397 id=397 data-nosnippet>397</a> <span class="kw">let </span>schema = arrow_schema_to_schema(record_batch.schema().as_ref())<span class="question-mark">?</span>;
+<a href=#398 id=398 data-nosnippet>398</a> batch_schema_iceberg = <span class="prelude-val">Some</span>(schema);
+<a href=#399 id=399 data-nosnippet>399</a> batch_schema_iceberg.as_ref().unwrap()
+<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">let </span>root_array: ArrayRef = Arc::new(StructArray::from(record_batch));
+<a href=#404 id=404 data-nosnippet>404</a>
+<a href=#405 id=405 data-nosnippet>405</a> <span class="kw">let </span><span class="kw-2">mut </span>processor = EqDelColumnProcessor::new(<span class="kw-2">&</span>equality_ids);
+<a href=#406 id=406 data-nosnippet>406</a> visit_schema_with_partner(schema, <span class="kw-2">&</span>root_array, <span class="kw-2">&mut </span>processor, <span class="kw-2">&</span>accessor)<span class="question-mark">?</span>;
+<a href=#407 id=407 data-nosnippet>407</a>
+<a href=#408 id=408 data-nosnippet>408</a> <span class="kw">let </span><span class="kw-2">mut </span>datum_columns_with_names = processor.finish()<span class="question-mark">?</span>;
+<a href=#409 id=409 data-nosnippet>409</a> <span class="kw">if </span>datum_columns_with_names.is_empty() {
+<a href=#410 id=410 data-nosnippet>410</a> <span class="kw">continue</span>;
+<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">// Iceberg spec (Equality Delete Files): a null data value never equals a non-null
+<a href=#414 id=414 data-nosnippet>414</a> // delete value, so a row with a null equality column must be kept. Build the keep
+<a href=#415 id=415 data-nosnippet>415</a> // predicate as `col IS NULL OR col != v` (`col IS NOT NULL` for a null delete value);
+<a href=#416 id=416 data-nosnippet>416</a> // a bare `col != v` drops nulls.
+<a href=#417 id=417 data-nosnippet>417</a> </span><span class="attr">#[allow(clippy::len_zero)]
+<a href=#418 id=418 data-nosnippet>418</a> </span><span class="kw">while </span>datum_columns_with_names[<span class="number">0</span>].<span class="number">0</span>.len() > <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>row_keep_predicate = Predicate::AlwaysFalse;
+<a href=#420 id=420 data-nosnippet>420</a> <span class="kw">for </span><span class="kw-2">&mut </span>(<span class="kw-2">ref mut </span>column, <span class="kw-2">ref </span>field_name) <span class="kw">in </span><span class="kw-2">&mut </span>datum_columns_with_names {
+<a href=#421 id=421 data-nosnippet>421</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(item) = column.next() {
+<a href=#422 id=422 data-nosnippet>422</a> <span class="kw">let </span>reference = Reference::new(field_name.clone());
+<a href=#423 id=423 data-nosnippet>423</a> <span class="kw">let </span>cell_keep_predicate = <span class="kw">if let </span><span class="prelude-val">Some</span>(datum) = item<span class="question-mark">? </span>{
+<a href=#424 id=424 data-nosnippet>424</a> reference
+<a href=#425 id=425 data-nosnippet>425</a> .clone()
+<a href=#426 id=426 data-nosnippet>426</a> .is_null()
+<a href=#427 id=427 data-nosnippet>427</a> .or(reference.not_equal_to(datum.clone()))
+<a href=#428 id=428 data-nosnippet>428</a> } <span class="kw">else </span>{
+<a href=#429 id=429 data-nosnippet>429</a> reference.is_not_null()
+<a href=#430 id=430 data-nosnippet>430</a> };
+<a href=#431 id=431 data-nosnippet>431</a> row_keep_predicate = row_keep_predicate.or(cell_keep_predicate);
+<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> row_predicates.push(row_keep_predicate);
+<a href=#435 id=435 data-nosnippet>435</a> }
+<a href=#436 id=436 data-nosnippet>436</a> }
+<a href=#437 id=437 data-nosnippet>437</a>
+<a href=#438 id=438 data-nosnippet>438</a> <span class="comment">// All row predicates are combined to a single predicate by creating a balanced binary tree.
+<a href=#439 id=439 data-nosnippet>439</a> // Using a simple fold would result in a deeply nested predicate that can cause a stack overflow.
+<a href=#440 id=440 data-nosnippet>440</a> </span><span class="kw">while </span>row_predicates.len() > <span class="number">1 </span>{
+<a href=#441 id=441 data-nosnippet>441</a> <span class="kw">let </span><span class="kw-2">mut </span>next_level = Vec::with_capacity(row_predicates.len().div_ceil(<span class="number">2</span>));
+<a href=#442 id=442 data-nosnippet>442</a> <span class="kw">let </span><span class="kw-2">mut </span>iter = row_predicates.into_iter();
+<a href=#443 id=443 data-nosnippet>443</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(p1) = iter.next() {
+<a href=#444 id=444 data-nosnippet>444</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(p2) = iter.next() {
+<a href=#445 id=445 data-nosnippet>445</a> next_level.push(p1.and(p2));
+<a href=#446 id=446 data-nosnippet>446</a> } <span class="kw">else </span>{
+<a href=#447 id=447 data-nosnippet>447</a> next_level.push(p1);
+<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> row_predicates = next_level;
+<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> <span class="kw">match </span>row_predicates.pop() {
+<a href=#454 id=454 data-nosnippet>454</a> <span class="prelude-val">Some</span>(p) => <span class="prelude-val">Ok</span>(p),
+<a href=#455 id=455 data-nosnippet>455</a> <span class="prelude-val">None </span>=> <span class="prelude-val">Ok</span>(AlwaysTrue),
+<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>
+<a href=#460 id=460 data-nosnippet>460</a><span class="kw">struct </span>EqDelColumnProcessor<<span class="lifetime">'a</span>> {
+<a href=#461 id=461 data-nosnippet>461</a> equality_ids: <span class="kw-2">&</span><span class="lifetime">'a </span>HashSet<i32>,
+<a href=#462 id=462 data-nosnippet>462</a> collected_columns: Vec<(ArrayRef, String, Type)>,
+<a href=#463 id=463 data-nosnippet>463</a>}
+<a href=#464 id=464 data-nosnippet>464</a>
+<a href=#465 id=465 data-nosnippet>465</a><span class="kw">impl</span><<span class="lifetime">'a</span>> EqDelColumnProcessor<<span class="lifetime">'a</span>> {
+<a href=#466 id=466 data-nosnippet>466</a> <span class="kw">fn </span>new(equality_ids: <span class="kw-2">&</span><span class="lifetime">'a </span>HashSet<i32>) -> <span class="self">Self </span>{
+<a href=#467 id=467 data-nosnippet>467</a> <span class="self">Self </span>{
+<a href=#468 id=468 data-nosnippet>468</a> equality_ids,
+<a href=#469 id=469 data-nosnippet>469</a> collected_columns: Vec::with_capacity(equality_ids.len()),
+<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>
+<a href=#473 id=473 data-nosnippet>473</a> <span class="attr">#[allow(clippy::type_complexity)]
+<a href=#474 id=474 data-nosnippet>474</a> </span><span class="kw">fn </span>finish(
+<a href=#475 id=475 data-nosnippet>475</a> <span class="self">self</span>,
+<a href=#476 id=476 data-nosnippet>476</a> ) -> <span class="prelude-ty">Result</span><
+<a href=#477 id=477 data-nosnippet>477</a> Vec<(
+<a href=#478 id=478 data-nosnippet>478</a> Box<<span class="kw">dyn </span>ExactSizeIterator<Item = <span class="prelude-ty">Result</span><<span class="prelude-ty">Option</span><Datum>>>>,
+<a href=#479 id=479 data-nosnippet>479</a> String,
+<a href=#480 id=480 data-nosnippet>480</a> )>,
+<a href=#481 id=481 data-nosnippet>481</a> > {
+<a href=#482 id=482 data-nosnippet>482</a> <span class="self">self</span>.collected_columns
+<a href=#483 id=483 data-nosnippet>483</a> .into_iter()
+<a href=#484 id=484 data-nosnippet>484</a> .map(|(array, field_name, field_type)| {
+<a href=#485 id=485 data-nosnippet>485</a> <span class="kw">let </span>primitive_type = field_type
+<a href=#486 id=486 data-nosnippet>486</a> .as_primitive_type()
+<a href=#487 id=487 data-nosnippet>487</a> .ok_or_else(|| {
+<a href=#488 id=488 data-nosnippet>488</a> Error::new(ErrorKind::Unexpected, <span class="string">"field is not a primitive type"</span>)
+<a href=#489 id=489 data-nosnippet>489</a> })<span class="question-mark">?
+<a href=#490 id=490 data-nosnippet>490</a> </span>.clone();
+<a href=#491 id=491 data-nosnippet>491</a>
+<a href=#492 id=492 data-nosnippet>492</a> <span class="kw">let </span>lit_vec = arrow_primitive_to_literal(<span class="kw-2">&</span>array, <span class="kw-2">&</span>field_type)<span class="question-mark">?</span>;
+<a href=#493 id=493 data-nosnippet>493</a> <span class="kw">let </span>datum_iterator: Box<<span class="kw">dyn </span>ExactSizeIterator<Item = <span class="prelude-ty">Result</span><<span class="prelude-ty">Option</span><Datum>>>> =
+<a href=#494 id=494 data-nosnippet>494</a> Box::new(lit_vec.into_iter().map(<span class="kw">move </span>|c| {
+<a href=#495 id=495 data-nosnippet>495</a> c.map(|literal| {
+<a href=#496 id=496 data-nosnippet>496</a> literal
+<a href=#497 id=497 data-nosnippet>497</a> .as_primitive_literal()
+<a href=#498 id=498 data-nosnippet>498</a> .map(|primitive_literal| {
+<a href=#499 id=499 data-nosnippet>499</a> Datum::new(primitive_type.clone(), primitive_literal)
+<a href=#500 id=500 data-nosnippet>500</a> })
+<a href=#501 id=501 data-nosnippet>501</a> .ok_or(Error::new(
+<a href=#502 id=502 data-nosnippet>502</a> ErrorKind::Unexpected,
+<a href=#503 id=503 data-nosnippet>503</a> <span class="string">"failed to convert to primitive literal"</span>,
+<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> .transpose()
+<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">impl </span>SchemaWithPartnerVisitor<ArrayRef> <span class="kw">for </span>EqDelColumnProcessor<<span class="lifetime">'_</span>> {
-<a href=#510 id=510 data-nosnippet>510</a> <span class="kw">type </span>T = ();
-<a href=#511 id=511 data-nosnippet>511</a>
-<a href=#512 id=512 data-nosnippet>512</a> <span class="kw">fn </span>schema(<span class="kw-2">&mut </span><span class="self">self</span>, _schema: <span class="kw-2">&</span>Schema, _partner: <span class="kw-2">&</span>ArrayRef, _value: ()) -> <span class="prelude-ty">Result</span><()> {
-<a href=#513 id=513 data-nosnippet>513</a> <span class="prelude-val">Ok</span>(())
-<a href=#514 id=514 data-nosnippet>514</a> }
-<a href=#515 id=515 data-nosnippet>515</a>
-<a href=#516 id=516 data-nosnippet>516</a> <span class="kw">fn </span>field(<span class="kw-2">&mut </span><span class="self">self</span>, field: <span class="kw-2">&</span>NestedFieldRef, partner: <span class="kw-2">&</span>ArrayRef, _value: ()) -> <span class="prelude-ty">Result</span><()> {
-<a href=#517 id=517 data-nosnippet>517</a> <span class="kw">if </span><span class="self">self</span>.equality_ids.contains(<span class="kw-2">&</span>field.id) && field.field_type.as_primitive_type().is_some() {
-<a href=#518 id=518 data-nosnippet>518</a> <span class="self">self</span>.collected_columns.push((
-<a href=#519 id=519 data-nosnippet>519</a> partner.clone(),
-<a href=#520 id=520 data-nosnippet>520</a> field.name.clone(),
-<a href=#521 id=521 data-nosnippet>521</a> field.field_type.as_ref().clone(),
-<a href=#522 id=522 data-nosnippet>522</a> ));
-<a href=#523 id=523 data-nosnippet>523</a> }
-<a href=#524 id=524 data-nosnippet>524</a> <span class="prelude-val">Ok</span>(())
-<a href=#525 id=525 data-nosnippet>525</a> }
-<a href=#526 id=526 data-nosnippet>526</a>
-<a href=#527 id=527 data-nosnippet>527</a> <span class="kw">fn </span>r#struct(
-<a href=#528 id=528 data-nosnippet>528</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#529 id=529 data-nosnippet>529</a> _struct: <span class="kw-2">&</span>StructType,
-<a href=#530 id=530 data-nosnippet>530</a> _partner: <span class="kw-2">&</span>ArrayRef,
-<a href=#531 id=531 data-nosnippet>531</a> _results: Vec<()>,
-<a href=#532 id=532 data-nosnippet>532</a> ) -> <span class="prelude-ty">Result</span><()> {
-<a href=#533 id=533 data-nosnippet>533</a> <span class="prelude-val">Ok</span>(())
-<a href=#534 id=534 data-nosnippet>534</a> }
-<a href=#535 id=535 data-nosnippet>535</a>
-<a href=#536 id=536 data-nosnippet>536</a> <span class="kw">fn </span>list(<span class="kw-2">&mut </span><span class="self">self</span>, _list: <span class="kw-2">&</span>ListType, _partner: <span class="kw-2">&</span>ArrayRef, _value: ()) -> <span class="prelude-ty">Result</span><()> {
-<a href=#537 id=537 data-nosnippet>537</a> <span class="prelude-val">Ok</span>(())
-<a href=#538 id=538 data-nosnippet>538</a> }
-<a href=#539 id=539 data-nosnippet>539</a>
-<a href=#540 id=540 data-nosnippet>540</a> <span class="kw">fn </span>map(
-<a href=#541 id=541 data-nosnippet>541</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#542 id=542 data-nosnippet>542</a> _map: <span class="kw-2">&</span>MapType,
-<a href=#543 id=543 data-nosnippet>543</a> _partner: <span class="kw-2">&</span>ArrayRef,
-<a href=#544 id=544 data-nosnippet>544</a> _key_value: (),
-<a href=#545 id=545 data-nosnippet>545</a> _value: (),
-<a href=#546 id=546 data-nosnippet>546</a> ) -> <span class="prelude-ty">Result</span><()> {
-<a href=#547 id=547 data-nosnippet>547</a> <span class="prelude-val">Ok</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> <span class="kw">fn </span>primitive(<span class="kw-2">&mut </span><span class="self">self</span>, _primitive: <span class="kw-2">&</span>PrimitiveType, _partner: <span class="kw-2">&</span>ArrayRef) -> <span class="prelude-ty">Result</span><()> {
-<a href=#551 id=551 data-nosnippet>551</a> <span class="prelude-val">Ok</span>(())
-<a href=#552 id=552 data-nosnippet>552</a> }
-<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><span class="kw">struct </span>EqDelRecordBatchPartnerAccessor;
-<a href=#556 id=556 data-nosnippet>556</a>
-<a href=#557 id=557 data-nosnippet>557</a><span class="kw">impl </span>PartnerAccessor<ArrayRef> <span class="kw">for </span>EqDelRecordBatchPartnerAccessor {
-<a href=#558 id=558 data-nosnippet>558</a> <span class="kw">fn </span>struct_partner<<span class="lifetime">'a</span>>(<span class="kw-2">&</span><span class="self">self</span>, schema_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
-<a href=#559 id=559 data-nosnippet>559</a> <span class="prelude-val">Ok</span>(schema_partner)
-<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> <span class="kw">fn </span>field_partner<<span class="lifetime">'a</span>>(
-<a href=#563 id=563 data-nosnippet>563</a> <span class="kw-2">&</span><span class="self">self</span>,
-<a href=#564 id=564 data-nosnippet>564</a> struct_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef,
-<a href=#565 id=565 data-nosnippet>565</a> field: <span class="kw-2">&</span>NestedField,
-<a href=#566 id=566 data-nosnippet>566</a> ) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
-<a href=#567 id=567 data-nosnippet>567</a> <span class="kw">let </span><span class="prelude-val">Some</span>(struct_array) = struct_partner.as_any().downcast_ref::<StructArray>() <span class="kw">else </span>{
-<a href=#568 id=568 data-nosnippet>568</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#569 id=569 data-nosnippet>569</a> ErrorKind::Unexpected,
-<a href=#570 id=570 data-nosnippet>570</a> <span class="string">"Expected struct array for field extraction"</span>,
-<a href=#571 id=571 data-nosnippet>571</a> ));
-<a href=#572 id=572 data-nosnippet>572</a> };
-<a href=#573 id=573 data-nosnippet>573</a>
-<a href=#574 id=574 data-nosnippet>574</a> <span class="comment">// Find the field by name within the struct
-<a href=#575 id=575 data-nosnippet>575</a> </span><span class="kw">for </span>(i, field_def) <span class="kw">in </span>struct_array.fields().iter().enumerate() {
-<a href=#576 id=576 data-nosnippet>576</a> <span class="kw">if </span>field_def.name() == <span class="kw-2">&</span>field.name {
-<a href=#577 id=577 data-nosnippet>577</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(struct_array.column(i));
-<a href=#578 id=578 data-nosnippet>578</a> }
-<a href=#579 id=579 data-nosnippet>579</a> }
-<a href=#580 id=580 data-nosnippet>580</a>
-<a href=#581 id=581 data-nosnippet>581</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#582 id=582 data-nosnippet>582</a> ErrorKind::Unexpected,
-<a href=#583 id=583 data-nosnippet>583</a> <span class="macro">format!</span>(<span class="string">"Field {} not found in parent struct"</span>, field.name),
-<a href=#584 id=584 data-nosnippet>584</a> ))
-<a href=#585 id=585 data-nosnippet>585</a> }
+<a href=#509 id=509 data-nosnippet>509</a> <span class="prelude-val">Ok</span>((datum_iterator, field_name))
+<a href=#510 id=510 data-nosnippet>510</a> })
+<a href=#511 id=511 data-nosnippet>511</a> .collect::<<span class="prelude-ty">Result</span><Vec<<span class="kw">_</span>>>>()
+<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>
+<a href=#515 id=515 data-nosnippet>515</a><span class="kw">impl </span>SchemaWithPartnerVisitor<ArrayRef> <span class="kw">for </span>EqDelColumnProcessor<<span class="lifetime">'_</span>> {
+<a href=#516 id=516 data-nosnippet>516</a> <span class="kw">type </span>T = ();
+<a href=#517 id=517 data-nosnippet>517</a>
+<a href=#518 id=518 data-nosnippet>518</a> <span class="kw">fn </span>schema(<span class="kw-2">&mut </span><span class="self">self</span>, _schema: <span class="kw-2">&</span>Schema, _partner: <span class="kw-2">&</span>ArrayRef, _value: ()) -> <span class="prelude-ty">Result</span><()> {
+<a href=#519 id=519 data-nosnippet>519</a> <span class="prelude-val">Ok</span>(())
+<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">fn </span>field(<span class="kw-2">&mut </span><span class="self">self</span>, field: <span class="kw-2">&</span>NestedFieldRef, partner: <span class="kw-2">&</span>ArrayRef, _value: ()) -> <span class="prelude-ty">Result</span><()> {
+<a href=#523 id=523 data-nosnippet>523</a> <span class="kw">if </span><span class="self">self</span>.equality_ids.contains(<span class="kw-2">&</span>field.id) && field.field_type.as_primitive_type().is_some() {
+<a href=#524 id=524 data-nosnippet>524</a> <span class="self">self</span>.collected_columns.push((
+<a href=#525 id=525 data-nosnippet>525</a> partner.clone(),
+<a href=#526 id=526 data-nosnippet>526</a> field.name.clone(),
+<a href=#527 id=527 data-nosnippet>527</a> field.field_type.as_ref().clone(),
+<a href=#528 id=528 data-nosnippet>528</a> ));
+<a href=#529 id=529 data-nosnippet>529</a> }
+<a href=#530 id=530 data-nosnippet>530</a> <span class="prelude-val">Ok</span>(())
+<a href=#531 id=531 data-nosnippet>531</a> }
+<a href=#532 id=532 data-nosnippet>532</a>
+<a href=#533 id=533 data-nosnippet>533</a> <span class="kw">fn </span>r#struct(
+<a href=#534 id=534 data-nosnippet>534</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#535 id=535 data-nosnippet>535</a> _struct: <span class="kw-2">&</span>StructType,
+<a href=#536 id=536 data-nosnippet>536</a> _partner: <span class="kw-2">&</span>ArrayRef,
+<a href=#537 id=537 data-nosnippet>537</a> _results: Vec<()>,
+<a href=#538 id=538 data-nosnippet>538</a> ) -> <span class="prelude-ty">Result</span><()> {
+<a href=#539 id=539 data-nosnippet>539</a> <span class="prelude-val">Ok</span>(())
+<a href=#540 id=540 data-nosnippet>540</a> }
+<a href=#541 id=541 data-nosnippet>541</a>
+<a href=#542 id=542 data-nosnippet>542</a> <span class="kw">fn </span>list(<span class="kw-2">&mut </span><span class="self">self</span>, _list: <span class="kw-2">&</span>ListType, _partner: <span class="kw-2">&</span>ArrayRef, _value: ()) -> <span class="prelude-ty">Result</span><()> {
+<a href=#543 id=543 data-nosnippet>543</a> <span class="prelude-val">Ok</span>(())
+<a href=#544 id=544 data-nosnippet>544</a> }
+<a href=#545 id=545 data-nosnippet>545</a>
+<a href=#546 id=546 data-nosnippet>546</a> <span class="kw">fn </span>map(
+<a href=#547 id=547 data-nosnippet>547</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#548 id=548 data-nosnippet>548</a> _map: <span class="kw-2">&</span>MapType,
+<a href=#549 id=549 data-nosnippet>549</a> _partner: <span class="kw-2">&</span>ArrayRef,
+<a href=#550 id=550 data-nosnippet>550</a> _key_value: (),
+<a href=#551 id=551 data-nosnippet>551</a> _value: (),
+<a href=#552 id=552 data-nosnippet>552</a> ) -> <span class="prelude-ty">Result</span><()> {
+<a href=#553 id=553 data-nosnippet>553</a> <span class="prelude-val">Ok</span>(())
+<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">fn </span>primitive(<span class="kw-2">&mut </span><span class="self">self</span>, _primitive: <span class="kw-2">&</span>PrimitiveType, _partner: <span class="kw-2">&</span>ArrayRef) -> <span class="prelude-ty">Result</span><()> {
+<a href=#557 id=557 data-nosnippet>557</a> <span class="prelude-val">Ok</span>(())
+<a href=#558 id=558 data-nosnippet>558</a> }
+<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><span class="kw">struct </span>EqDelRecordBatchPartnerAccessor;
+<a href=#562 id=562 data-nosnippet>562</a>
+<a href=#563 id=563 data-nosnippet>563</a><span class="kw">impl </span>PartnerAccessor<ArrayRef> <span class="kw">for </span>EqDelRecordBatchPartnerAccessor {
+<a href=#564 id=564 data-nosnippet>564</a> <span class="kw">fn </span>struct_partner<<span class="lifetime">'a</span>>(<span class="kw-2">&</span><span class="self">self</span>, schema_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
+<a href=#565 id=565 data-nosnippet>565</a> <span class="prelude-val">Ok</span>(schema_partner)
+<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>field_partner<<span class="lifetime">'a</span>>(
+<a href=#569 id=569 data-nosnippet>569</a> <span class="kw-2">&</span><span class="self">self</span>,
+<a href=#570 id=570 data-nosnippet>570</a> struct_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef,
+<a href=#571 id=571 data-nosnippet>571</a> field: <span class="kw-2">&</span>NestedField,
+<a href=#572 id=572 data-nosnippet>572</a> ) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
+<a href=#573 id=573 data-nosnippet>573</a> <span class="kw">let </span><span class="prelude-val">Some</span>(struct_array) = struct_partner.as_any().downcast_ref::<StructArray>() <span class="kw">else </span>{
+<a href=#574 id=574 data-nosnippet>574</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#575 id=575 data-nosnippet>575</a> ErrorKind::Unexpected,
+<a href=#576 id=576 data-nosnippet>576</a> <span class="string">"Expected struct array for field extraction"</span>,
+<a href=#577 id=577 data-nosnippet>577</a> ));
+<a href=#578 id=578 data-nosnippet>578</a> };
+<a href=#579 id=579 data-nosnippet>579</a>
+<a href=#580 id=580 data-nosnippet>580</a> <span class="comment">// Find the field by name within the struct
+<a href=#581 id=581 data-nosnippet>581</a> </span><span class="kw">for </span>(i, field_def) <span class="kw">in </span>struct_array.fields().iter().enumerate() {
+<a href=#582 id=582 data-nosnippet>582</a> <span class="kw">if </span>field_def.name() == <span class="kw-2">&</span>field.name {
+<a href=#583 id=583 data-nosnippet>583</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(struct_array.column(i));
+<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>
-<a href=#587 id=587 data-nosnippet>587</a> <span class="kw">fn </span>list_element_partner<<span class="lifetime">'a</span>>(<span class="kw-2">&</span><span class="self">self</span>, _list_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
-<a href=#588 id=588 data-nosnippet>588</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#589 id=589 data-nosnippet>589</a> ErrorKind::FeatureUnsupported,
-<a href=#590 id=590 data-nosnippet>590</a> <span class="string">"List columns are unsupported in equality deletes"</span>,
-<a href=#591 id=591 data-nosnippet>591</a> ))
-<a href=#592 id=592 data-nosnippet>592</a> }
-<a href=#593 id=593 data-nosnippet>593</a>
-<a href=#594 id=594 data-nosnippet>594</a> <span class="kw">fn </span>map_key_partner<<span class="lifetime">'a</span>>(<span class="kw-2">&</span><span class="self">self</span>, _map_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
-<a href=#595 id=595 data-nosnippet>595</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#596 id=596 data-nosnippet>596</a> ErrorKind::FeatureUnsupported,
-<a href=#597 id=597 data-nosnippet>597</a> <span class="string">"Map columns are unsupported in equality deletes"</span>,
-<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>
-<a href=#601 id=601 data-nosnippet>601</a> <span class="kw">fn </span>map_value_partner<<span class="lifetime">'a</span>>(<span class="kw-2">&</span><span class="self">self</span>, _map_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
-<a href=#602 id=602 data-nosnippet>602</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#603 id=603 data-nosnippet>603</a> ErrorKind::FeatureUnsupported,
-<a href=#604 id=604 data-nosnippet>604</a> <span class="string">"Map columns are unsupported in equality deletes"</span>,
-<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>
-<a href=#609 id=609 data-nosnippet>609</a><span class="attr">#[cfg(test)]
-<a href=#610 id=610 data-nosnippet>610</a></span><span class="kw">mod </span>tests {
-<a href=#611 id=611 data-nosnippet>611</a> <span class="kw">use </span>std::collections::HashMap;
-<a href=#612 id=612 data-nosnippet>612</a> <span class="kw">use </span>std::fs::File;
-<a href=#613 id=613 data-nosnippet>613</a> <span class="kw">use </span>std::sync::Arc;
+<a href=#587 id=587 data-nosnippet>587</a> <span class="prelude-val">Err</span>(Error::new(
+<a href=#588 id=588 data-nosnippet>588</a> ErrorKind::Unexpected,
+<a href=#589 id=589 data-nosnippet>589</a> <span class="macro">format!</span>(<span class="string">"Field {} not found in parent struct"</span>, field.name),
+<a href=#590 id=590 data-nosnippet>590</a> ))
+<a href=#591 id=591 data-nosnippet>591</a> }
+<a href=#592 id=592 data-nosnippet>592</a>
+<a href=#593 id=593 data-nosnippet>593</a> <span class="kw">fn </span>list_element_partner<<span class="lifetime">'a</span>>(<span class="kw-2">&</span><span class="self">self</span>, _list_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
+<a href=#594 id=594 data-nosnippet>594</a> <span class="prelude-val">Err</span>(Error::new(
+<a href=#595 id=595 data-nosnippet>595</a> ErrorKind::FeatureUnsupported,
+<a href=#596 id=596 data-nosnippet>596</a> <span class="string">"List columns are unsupported in equality deletes"</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">fn </span>map_key_partner<<span class="lifetime">'a</span>>(<span class="kw-2">&</span><span class="self">self</span>, _map_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
+<a href=#601 id=601 data-nosnippet>601</a> <span class="prelude-val">Err</span>(Error::new(
+<a href=#602 id=602 data-nosnippet>602</a> ErrorKind::FeatureUnsupported,
+<a href=#603 id=603 data-nosnippet>603</a> <span class="string">"Map columns are unsupported in equality deletes"</span>,
+<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> <span class="kw">fn </span>map_value_partner<<span class="lifetime">'a</span>>(<span class="kw-2">&</span><span class="self">self</span>, _map_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
+<a href=#608 id=608 data-nosnippet>608</a> <span class="prelude-val">Err</span>(Error::new(
+<a href=#609 id=609 data-nosnippet>609</a> ErrorKind::FeatureUnsupported,
+<a href=#610 id=610 data-nosnippet>610</a> <span class="string">"Map columns are unsupported in equality deletes"</span>,
+<a href=#611 id=611 data-nosnippet>611</a> ))
+<a href=#612 id=612 data-nosnippet>612</a> }
+<a href=#613 id=613 data-nosnippet>613</a>}
<a href=#614 id=614 data-nosnippet>614</a>
-<a href=#615 id=615 data-nosnippet>615</a> <span class="kw">use </span>arrow_array::cast::AsArray;
-<a href=#616 id=616 data-nosnippet>616</a> <span class="kw">use </span>arrow_array::{
-<a href=#617 id=617 data-nosnippet>617</a> ArrayRef, BinaryArray, Int32Array, Int64Array, RecordBatch, StringArray, StructArray,
-<a href=#618 id=618 data-nosnippet>618</a> };
-<a href=#619 id=619 data-nosnippet>619</a> <span class="kw">use </span>arrow_schema::{DataType, Field, Fields};
-<a href=#620 id=620 data-nosnippet>620</a> <span class="kw">use </span>parquet::arrow::{ArrowWriter, PARQUET_FIELD_ID_META_KEY};
-<a href=#621 id=621 data-nosnippet>621</a> <span class="kw">use </span>parquet::basic::Compression;
-<a href=#622 id=622 data-nosnippet>622</a> <span class="kw">use </span>parquet::file::properties::WriterProperties;
-<a href=#623 id=623 data-nosnippet>623</a> <span class="kw">use </span>tempfile::TempDir;
-<a href=#624 id=624 data-nosnippet>624</a>
-<a href=#625 id=625 data-nosnippet>625</a> <span class="kw">use </span>super::<span class="kw-2">*</span>;
-<a href=#626 id=626 data-nosnippet>626</a> <span class="kw">use </span><span class="kw">crate</span>::arrow::delete_filter::tests::setup;
-<a href=#627 id=627 data-nosnippet>627</a> <span class="kw">use </span><span class="kw">crate</span>::scan::FileScanTaskDeleteFile;
-<a href=#628 id=628 data-nosnippet>628</a> <span class="kw">use </span><span class="kw">crate</span>::spec::{DataContentType, Schema};
-<a href=#629 id=629 data-nosnippet>629</a>
-<a href=#630 id=630 data-nosnippet>630</a> <span class="attr">#[tokio::test]
-<a href=#631 id=631 data-nosnippet>631</a> </span><span class="kw">async fn </span>test_delete_file_loader_parse_equality_deletes() {
-<a href=#632 id=632 data-nosnippet>632</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#633 id=633 data-nosnippet>633</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
-<a href=#634 id=634 data-nosnippet>634</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<a href=#615 id=615 data-nosnippet>615</a><span class="attr">#[cfg(test)]
+<a href=#616 id=616 data-nosnippet>616</a></span><span class="kw">mod </span>tests {
+<a href=#617 id=617 data-nosnippet>617</a> <span class="kw">use </span>std::collections::HashMap;
+<a href=#618 id=618 data-nosnippet>618</a> <span class="kw">use </span>std::fs::File;
+<a href=#619 id=619 data-nosnippet>619</a> <span class="kw">use </span>std::sync::Arc;
+<a href=#620 id=620 data-nosnippet>620</a>
+<a href=#621 id=621 data-nosnippet>621</a> <span class="kw">use </span>arrow_array::cast::AsArray;
+<a href=#622 id=622 data-nosnippet>622</a> <span class="kw">use </span>arrow_array::{
+<a href=#623 id=623 data-nosnippet>623</a> ArrayRef, BinaryArray, Int32Array, Int64Array, RecordBatch, StringArray, StructArray,
+<a href=#624 id=624 data-nosnippet>624</a> };
+<a href=#625 id=625 data-nosnippet>625</a> <span class="kw">use </span>arrow_schema::{DataType, Field, Fields};
+<a href=#626 id=626 data-nosnippet>626</a> <span class="kw">use </span>parquet::arrow::{ArrowWriter, PARQUET_FIELD_ID_META_KEY};
+<a href=#627 id=627 data-nosnippet>627</a> <span class="kw">use </span>parquet::basic::Compression;
+<a href=#628 id=628 data-nosnippet>628</a> <span class="kw">use </span>parquet::file::properties::WriterProperties;
+<a href=#629 id=629 data-nosnippet>629</a> <span class="kw">use </span>tempfile::TempDir;
+<a href=#630 id=630 data-nosnippet>630</a>
+<a href=#631 id=631 data-nosnippet>631</a> <span class="kw">use </span>super::<span class="kw-2">*</span>;
+<a href=#632 id=632 data-nosnippet>632</a> <span class="kw">use </span><span class="kw">crate</span>::arrow::delete_filter::tests::setup;
+<a href=#633 id=633 data-nosnippet>633</a> <span class="kw">use </span><span class="kw">crate</span>::scan::FileScanTaskDeleteFile;
+<a href=#634 id=634 data-nosnippet>634</a> <span class="kw">use </span><span class="kw">crate</span>::spec::{DataContentType, Schema};
<a href=#635 id=635 data-nosnippet>635</a>
-<a href=#636 id=636 data-nosnippet>636</a> <span class="kw">let </span>eq_delete_file_path = setup_write_equality_delete_file_1(table_location);
-<a href=#637 id=637 data-nosnippet>637</a>
-<a href=#638 id=638 data-nosnippet>638</a> <span class="kw">let </span>basic_delete_file_loader =
-<a href=#639 id=639 data-nosnippet>639</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
-<a href=#640 id=640 data-nosnippet>640</a> <span class="kw">let </span>record_batch_stream = basic_delete_file_loader
-<a href=#641 id=641 data-nosnippet>641</a> .parquet_to_batch_stream(
-<a href=#642 id=642 data-nosnippet>642</a> <span class="kw-2">&</span>eq_delete_file_path,
-<a href=#643 id=643 data-nosnippet>643</a> std::fs::metadata(<span class="kw-2">&</span>eq_delete_file_path).unwrap().len(),
-<a href=#644 id=644 data-nosnippet>644</a> )
-<a href=#645 id=645 data-nosnippet>645</a> .<span class="kw">await
-<a href=#646 id=646 data-nosnippet>646</a> </span>.expect(<span class="string">"could not get batch stream"</span>);
-<a href=#647 id=647 data-nosnippet>647</a>
-<a href=#648 id=648 data-nosnippet>648</a> <span class="kw">let </span>eq_ids = HashSet::from_iter(<span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">3</span>, <span class="number">4</span>, <span class="number">6</span>, <span class="number">8</span>]);
-<a href=#649 id=649 data-nosnippet>649</a>
-<a href=#650 id=650 data-nosnippet>650</a> <span class="kw">let </span>parsed_eq_delete = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
-<a href=#651 id=651 data-nosnippet>651</a> record_batch_stream,
-<a href=#652 id=652 data-nosnippet>652</a> eq_ids,
-<a href=#653 id=653 data-nosnippet>653</a> )
-<a href=#654 id=654 data-nosnippet>654</a> .<span class="kw">await
-<a href=#655 id=655 data-nosnippet>655</a> </span>.expect(<span class="string">"error parsing batch stream"</span>);
-<a href=#656 id=656 data-nosnippet>656</a> <span class="macro">println!</span>(<span class="string">"{parsed_eq_delete}"</span>);
-<a href=#657 id=657 data-nosnippet>657</a>
-<a href=#658 id=658 data-nosnippet>658</a> <span class="kw">let </span>expected = <span class="string">"(((((y != 1) OR (z != 100)) OR (a != \"HELP\")) OR (sa != 4)) OR (b != 62696E6172795F64617461)) AND (((((y != 2) OR (z IS NOT NULL)) OR (a IS NOT NULL)) OR (sa != 5)) OR (b IS NOT NULL))"</span>.to_string();
-<a href=#659 id=659 data-nosnippet>659</a>
-<a href=#660 id=660 data-nosnippet>660</a> <span class="macro">assert_eq!</span>(parsed_eq_delete.to_string(), expected);
-<a href=#661 id=661 data-nosnippet>661</a> }
+<a href=#636 id=636 data-nosnippet>636</a> <span class="attr">#[tokio::test]
+<a href=#637 id=637 data-nosnippet>637</a> </span><span class="kw">async fn </span>test_delete_file_loader_parse_equality_deletes() {
+<a href=#638 id=638 data-nosnippet>638</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#639 id=639 data-nosnippet>639</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
+<a href=#640 id=640 data-nosnippet>640</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<a href=#641 id=641 data-nosnippet>641</a>
+<a href=#642 id=642 data-nosnippet>642</a> <span class="kw">let </span>eq_delete_file_path = setup_write_equality_delete_file_1(table_location);
+<a href=#643 id=643 data-nosnippet>643</a>
+<a href=#644 id=644 data-nosnippet>644</a> <span class="kw">let </span>basic_delete_file_loader =
+<a href=#645 id=645 data-nosnippet>645</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
+<a href=#646 id=646 data-nosnippet>646</a> <span class="kw">let </span>record_batch_stream = basic_delete_file_loader
+<a href=#647 id=647 data-nosnippet>647</a> .parquet_to_batch_stream(
+<a href=#648 id=648 data-nosnippet>648</a> <span class="kw-2">&</span>eq_delete_file_path,
+<a href=#649 id=649 data-nosnippet>649</a> std::fs::metadata(<span class="kw-2">&</span>eq_delete_file_path).unwrap().len(),
+<a href=#650 id=650 data-nosnippet>650</a> )
+<a href=#651 id=651 data-nosnippet>651</a> .<span class="kw">await
+<a href=#652 id=652 data-nosnippet>652</a> </span>.expect(<span class="string">"could not get batch stream"</span>);
+<a href=#653 id=653 data-nosnippet>653</a>
+<a href=#654 id=654 data-nosnippet>654</a> <span class="kw">let </span>eq_ids = HashSet::from_iter(<span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">3</span>, <span class="number">4</span>, <span class="number">6</span>, <span class="number">8</span>]);
+<a href=#655 id=655 data-nosnippet>655</a>
+<a href=#656 id=656 data-nosnippet>656</a> <span class="kw">let </span>parsed_eq_delete = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
+<a href=#657 id=657 data-nosnippet>657</a> record_batch_stream,
+<a href=#658 id=658 data-nosnippet>658</a> eq_ids,
+<a href=#659 id=659 data-nosnippet>659</a> )
+<a href=#660 id=660 data-nosnippet>660</a> .<span class="kw">await
+<a href=#661 id=661 data-nosnippet>661</a> </span>.expect(<span class="string">"error parsing batch stream"</span>);
<a href=#662 id=662 data-nosnippet>662</a>
-<a href=#663 id=663 data-nosnippet>663</a> <span class="doccomment">/// Create a simple field with metadata.
-<a href=#664 id=664 data-nosnippet>664</a> </span><span class="kw">fn </span>simple_field(name: <span class="kw-2">&</span>str, ty: DataType, nullable: bool, value: <span class="kw-2">&</span>str) -> Field {
-<a href=#665 id=665 data-nosnippet>665</a> arrow_schema::Field::new(name, ty, nullable).with_metadata(HashMap::from([(
-<a href=#666 id=666 data-nosnippet>666</a> PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#667 id=667 data-nosnippet>667</a> value.to_string(),
-<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="kw">fn </span>setup_write_equality_delete_file_1(table_location: <span class="kw-2">&</span>str) -> String {
-<a href=#672 id=672 data-nosnippet>672</a> <span class="kw">let </span>col_y_vals = <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>];
-<a href=#673 id=673 data-nosnippet>673</a> <span class="kw">let </span>col_y = Arc::new(Int64Array::from(col_y_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#674 id=674 data-nosnippet>674</a>
-<a href=#675 id=675 data-nosnippet>675</a> <span class="kw">let </span>col_z_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">100</span>), <span class="prelude-val">None</span>];
-<a href=#676 id=676 data-nosnippet>676</a> <span class="kw">let </span>col_z = Arc::new(Int64Array::from(col_z_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#677 id=677 data-nosnippet>677</a>
-<a href=#678 id=678 data-nosnippet>678</a> <span class="kw">let </span>col_a_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"HELP"</span>), <span class="prelude-val">None</span>];
-<a href=#679 id=679 data-nosnippet>679</a> <span class="kw">let </span>col_a = Arc::new(StringArray::from(col_a_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#680 id=680 data-nosnippet>680</a>
-<a href=#681 id=681 data-nosnippet>681</a> <span class="kw">let </span>col_s = Arc::new(StructArray::from(<span class="macro">vec!</span>[
-<a href=#682 id=682 data-nosnippet>682</a> (
-<a href=#683 id=683 data-nosnippet>683</a> Arc::new(simple_field(<span class="string">"sa"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"6"</span>)),
-<a href=#684 id=684 data-nosnippet>684</a> Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">4</span>, <span class="number">5</span>])) <span class="kw">as </span>ArrayRef,
-<a href=#685 id=685 data-nosnippet>685</a> ),
-<a href=#686 id=686 data-nosnippet>686</a> (
-<a href=#687 id=687 data-nosnippet>687</a> Arc::new(simple_field(<span class="string">"sb"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"7"</span>)),
-<a href=#688 id=688 data-nosnippet>688</a> Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"x"</span>), <span class="prelude-val">None</span>])) <span class="kw">as </span>ArrayRef,
-<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="kw">let </span>col_b_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="kw-2">&</span><span class="string">b"binary_data"</span>[..]), <span class="prelude-val">None</span>];
-<a href=#693 id=693 data-nosnippet>693</a> <span class="kw">let </span>col_b = Arc::new(BinaryArray::from(col_b_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#694 id=694 data-nosnippet>694</a>
-<a href=#695 id=695 data-nosnippet>695</a> <span class="kw">let </span>equality_delete_schema = {
-<a href=#696 id=696 data-nosnippet>696</a> <span class="kw">let </span>struct_field = DataType::Struct(Fields::from(<span class="macro">vec!</span>[
-<a href=#697 id=697 data-nosnippet>697</a> simple_field(<span class="string">"sa"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"6"</span>),
-<a href=#698 id=698 data-nosnippet>698</a> simple_field(<span class="string">"sb"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"7"</span>),
-<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> <span class="kw">let </span>fields = <span class="macro">vec!</span>[
-<a href=#702 id=702 data-nosnippet>702</a> Field::new(<span class="string">"y"</span>, arrow_schema::DataType::Int64, <span class="bool-val">true</span>).with_metadata(HashMap::from(
-<a href=#703 id=703 data-nosnippet>703</a> [(PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"2"</span>.to_string())],
-<a href=#704 id=704 data-nosnippet>704</a> )),
-<a href=#705 id=705 data-nosnippet>705</a> Field::new(<span class="string">"z"</span>, arrow_schema::DataType::Int64, <span class="bool-val">true</span>).with_metadata(HashMap::from(
-<a href=#706 id=706 data-nosnippet>706</a> [(PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"3"</span>.to_string())],
-<a href=#707 id=707 data-nosnippet>707</a> )),
-<a href=#708 id=708 data-nosnippet>708</a> Field::new(<span class="string">"a"</span>, arrow_schema::DataType::Utf8, <span class="bool-val">true</span>).with_metadata(HashMap::from([
-<a href=#709 id=709 data-nosnippet>709</a> (PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"4"</span>.to_string()),
-<a href=#710 id=710 data-nosnippet>710</a> ])),
-<a href=#711 id=711 data-nosnippet>711</a> simple_field(<span class="string">"s"</span>, struct_field, <span class="bool-val">false</span>, <span class="string">"5"</span>),
-<a href=#712 id=712 data-nosnippet>712</a> simple_field(<span class="string">"b"</span>, DataType::Binary, <span class="bool-val">true</span>, <span class="string">"8"</span>),
-<a href=#713 id=713 data-nosnippet>713</a> ];
-<a href=#714 id=714 data-nosnippet>714</a> Arc::new(arrow_schema::Schema::new(fields))
-<a href=#715 id=715 data-nosnippet>715</a> };
-<a href=#716 id=716 data-nosnippet>716</a>
-<a href=#717 id=717 data-nosnippet>717</a> <span class="kw">let </span>equality_deletes_to_write = RecordBatch::try_new(equality_delete_schema.clone(), <span class="macro">vec!</span>[
-<a href=#718 id=718 data-nosnippet>718</a> col_y, col_z, col_a, col_s, col_b,
-<a href=#719 id=719 data-nosnippet>719</a> ])
-<a href=#720 id=720 data-nosnippet>720</a> .unwrap();
-<a href=#721 id=721 data-nosnippet>721</a>
-<a href=#722 id=722 data-nosnippet>722</a> <span class="kw">let </span>path = <span class="macro">format!</span>(<span class="string">"{}/equality-deletes-1.parquet"</span>, <span class="kw-2">&</span>table_location);
-<a href=#723 id=723 data-nosnippet>723</a>
-<a href=#724 id=724 data-nosnippet>724</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
+<a href=#663 id=663 data-nosnippet>663</a> <span class="kw">let </span>expected = <span class="string">"((((((y IS NULL) OR (y != 1)) OR ((z IS NULL) OR (z != 100))) OR ((a IS NULL) OR (a != \"HELP\"))) OR ((sa IS NULL) OR (sa != 4))) OR ((b IS NULL) OR (b != 62696E6172795F64617461))) AND ((((((y IS NULL) OR (y != 2)) OR (z IS NOT NULL)) OR (a IS NOT NULL)) OR ((sa IS NULL) OR (sa != 5))) OR (b IS NOT NULL))"</span>.to_string();
+<a href=#664 id=664 data-nosnippet>664</a>
+<a href=#665 id=665 data-nosnippet>665</a> <span class="macro">assert_eq!</span>(parsed_eq_delete.to_string(), expected);
+<a href=#666 id=666 data-nosnippet>666</a> }
+<a href=#667 id=667 data-nosnippet>667</a>
+<a href=#668 id=668 data-nosnippet>668</a> <span class="comment">// An equality delete keyed on a nullable column must not delete rows whose value in that
+<a href=#669 id=669 data-nosnippet>669</a> // column is null: per the Iceberg spec (Equality Delete Files), a null matches only a null
+<a href=#670 id=670 data-nosnippet>670</a> // delete value. Mirrors Iceberg-Java's
+<a href=#671 id=671 data-nosnippet>671</a> // TestSparkReaderDeletes.testEqualityDeleteWithSchemaEvolution.
+<a href=#672 id=672 data-nosnippet>672</a> </span><span class="attr">#[tokio::test]
+<a href=#673 id=673 data-nosnippet>673</a> </span><span class="kw">async fn </span>test_equality_delete_predicate_preserves_null_rows() {
+<a href=#674 id=674 data-nosnippet>674</a> <span class="kw">let </span>schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[simple_field(
+<a href=#675 id=675 data-nosnippet>675</a> <span class="string">"status"</span>,
+<a href=#676 id=676 data-nosnippet>676</a> DataType::Utf8,
+<a href=#677 id=677 data-nosnippet>677</a> <span class="bool-val">true</span>,
+<a href=#678 id=678 data-nosnippet>678</a> <span class="string">"3"</span>,
+<a href=#679 id=679 data-nosnippet>679</a> )]));
+<a href=#680 id=680 data-nosnippet>680</a> <span class="kw">let </span>batch =
+<a href=#681 id=681 data-nosnippet>681</a> RecordBatch::try_new(schema, <span class="macro">vec!</span>[
+<a href=#682 id=682 data-nosnippet>682</a> Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"INACTIVE"</span>)])) <span class="kw">as </span>ArrayRef,
+<a href=#683 id=683 data-nosnippet>683</a> ])
+<a href=#684 id=684 data-nosnippet>684</a> .unwrap();
+<a href=#685 id=685 data-nosnippet>685</a> <span class="kw">let </span>stream: ArrowRecordBatchStream = futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(batch)]).boxed();
+<a href=#686 id=686 data-nosnippet>686</a>
+<a href=#687 id=687 data-nosnippet>687</a> <span class="kw">let </span>predicate = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
+<a href=#688 id=688 data-nosnippet>688</a> stream,
+<a href=#689 id=689 data-nosnippet>689</a> HashSet::from_iter(<span class="macro">vec!</span>[<span class="number">3</span>]),
+<a href=#690 id=690 data-nosnippet>690</a> )
+<a href=#691 id=691 data-nosnippet>691</a> .<span class="kw">await
+<a href=#692 id=692 data-nosnippet>692</a> </span>.expect(<span class="string">"error parsing equality delete stream"</span>);
+<a href=#693 id=693 data-nosnippet>693</a>
+<a href=#694 id=694 data-nosnippet>694</a> <span class="macro">assert_eq!</span>(
+<a href=#695 id=695 data-nosnippet>695</a> predicate.to_string(),
+<a href=#696 id=696 data-nosnippet>696</a> <span class="string">"(status IS NULL) OR (status != \"INACTIVE\")"
+<a href=#697 id=697 data-nosnippet>697</a> </span>);
+<a href=#698 id=698 data-nosnippet>698</a> }
+<a href=#699 id=699 data-nosnippet>699</a>
+<a href=#700 id=700 data-nosnippet>700</a> <span class="comment">// A delete row with a null value in the column matches only rows whose value is null (Iceberg
+<a href=#701 id=701 data-nosnippet>701</a> // spec, Equality Delete Files), so the keep predicate is `col IS NOT NULL`.
+<a href=#702 id=702 data-nosnippet>702</a> </span><span class="attr">#[tokio::test]
+<a href=#703 id=703 data-nosnippet>703</a> </span><span class="kw">async fn </span>test_equality_delete_predicate_matches_null_delete_value() {
+<a href=#704 id=704 data-nosnippet>704</a> <span class="kw">let </span>schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[simple_field(
+<a href=#705 id=705 data-nosnippet>705</a> <span class="string">"status"</span>,
+<a href=#706 id=706 data-nosnippet>706</a> DataType::Utf8,
+<a href=#707 id=707 data-nosnippet>707</a> <span class="bool-val">true</span>,
+<a href=#708 id=708 data-nosnippet>708</a> <span class="string">"3"</span>,
+<a href=#709 id=709 data-nosnippet>709</a> )]));
+<a href=#710 id=710 data-nosnippet>710</a> <span class="kw">let </span>batch = RecordBatch::try_new(schema, <span class="macro">vec!</span>[Arc::new(StringArray::from(<span class="macro">vec!</span>[
+<a href=#711 id=711 data-nosnippet>711</a> <span class="prelude-val">None </span><span class="kw">as </span><span class="prelude-ty">Option</span><<span class="kw-2">&</span>str>,
+<a href=#712 id=712 data-nosnippet>712</a> ])) <span class="kw">as </span>ArrayRef])
+<a href=#713 id=713 data-nosnippet>713</a> .unwrap();
+<a href=#714 id=714 data-nosnippet>714</a> <span class="kw">let </span>stream: ArrowRecordBatchStream = futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(batch)]).boxed();
+<a href=#715 id=715 data-nosnippet>715</a>
+<a href=#716 id=716 data-nosnippet>716</a> <span class="kw">let </span>predicate = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
+<a href=#717 id=717 data-nosnippet>717</a> stream,
+<a href=#718 id=718 data-nosnippet>718</a> HashSet::from_iter(<span class="macro">vec!</span>[<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">await
+<a href=#721 id=721 data-nosnippet>721</a> </span>.expect(<span class="string">"error parsing equality delete stream"</span>);
+<a href=#722 id=722 data-nosnippet>722</a>
+<a href=#723 id=723 data-nosnippet>723</a> <span class="macro">assert_eq!</span>(predicate.to_string(), <span class="string">"status IS NOT NULL"</span>);
+<a href=#724 id=724 data-nosnippet>724</a> }
<a href=#725 id=725 data-nosnippet>725</a>
-<a href=#726 id=726 data-nosnippet>726</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#727 id=727 data-nosnippet>727</a> .set_compression(Compression::SNAPPY)
-<a href=#728 id=728 data-nosnippet>728</a> .build();
-<a href=#729 id=729 data-nosnippet>729</a>
-<a href=#730 id=730 data-nosnippet>730</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(
-<a href=#731 id=731 data-nosnippet>731</a> file,
-<a href=#732 id=732 data-nosnippet>732</a> equality_deletes_to_write.schema(),
-<a href=#733 id=733 data-nosnippet>733</a> <span class="prelude-val">Some</span>(props.clone()),
-<a href=#734 id=734 data-nosnippet>734</a> )
-<a href=#735 id=735 data-nosnippet>735</a> .unwrap();
-<a href=#736 id=736 data-nosnippet>736</a>
-<a href=#737 id=737 data-nosnippet>737</a> writer
-<a href=#738 id=738 data-nosnippet>738</a> .write(<span class="kw-2">&</span>equality_deletes_to_write)
-<a href=#739 id=739 data-nosnippet>739</a> .expect(<span class="string">"Writing batch"</span>);
+<a href=#726 id=726 data-nosnippet>726</a> <span class="comment">// A delete row with several equality columns keeps a data row that differs in any one of them,
+<a href=#727 id=727 data-nosnippet>727</a> // so the per-column keep predicates are OR-ed.
+<a href=#728 id=728 data-nosnippet>728</a> </span><span class="attr">#[tokio::test]
+<a href=#729 id=729 data-nosnippet>729</a> </span><span class="kw">async fn </span>test_equality_delete_predicate_multiple_columns() {
+<a href=#730 id=730 data-nosnippet>730</a> <span class="kw">let </span>schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[
+<a href=#731 id=731 data-nosnippet>731</a> simple_field(<span class="string">"id"</span>, DataType::Int64, <span class="bool-val">true</span>, <span class="string">"1"</span>),
+<a href=#732 id=732 data-nosnippet>732</a> simple_field(<span class="string">"status"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"3"</span>),
+<a href=#733 id=733 data-nosnippet>733</a> ]));
+<a href=#734 id=734 data-nosnippet>734</a> <span class="kw">let </span>batch = RecordBatch::try_new(schema, <span class="macro">vec!</span>[
+<a href=#735 id=735 data-nosnippet>735</a> Arc::new(Int64Array::from(<span class="macro">vec!</span>[<span class="number">1</span>])) <span class="kw">as </span>ArrayRef,
+<a href=#736 id=736 data-nosnippet>736</a> Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"X"</span>)])) <span class="kw">as </span>ArrayRef,
+<a href=#737 id=737 data-nosnippet>737</a> ])
+<a href=#738 id=738 data-nosnippet>738</a> .unwrap();
+<a href=#739 id=739 data-nosnippet>739</a> <span class="kw">let </span>stream: ArrowRecordBatchStream = futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(batch)]).boxed();
<a href=#740 id=740 data-nosnippet>740</a>
-<a href=#741 id=741 data-nosnippet>741</a> <span class="comment">// writer must be closed to write footer
-<a href=#742 id=742 data-nosnippet>742</a> </span>writer.close().unwrap();
-<a href=#743 id=743 data-nosnippet>743</a>
-<a href=#744 id=744 data-nosnippet>744</a> path
-<a href=#745 id=745 data-nosnippet>745</a> }
-<a href=#746 id=746 data-nosnippet>746</a>
-<a href=#747 id=747 data-nosnippet>747</a> <span class="attr">#[tokio::test]
-<a href=#748 id=748 data-nosnippet>748</a> </span><span class="kw">async fn </span>test_caching_delete_file_loader_load_deletes() {
-<a href=#749 id=749 data-nosnippet>749</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#750 id=750 data-nosnippet>750</a> <span class="kw">let </span>table_location = tmp_dir.path();
-<a href=#751 id=751 data-nosnippet>751</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
-<a href=#752 id=752 data-nosnippet>752</a>
-<a href=#753 id=753 data-nosnippet>753</a> <span class="kw">let </span>delete_file_loader =
-<a href=#754 id=754 data-nosnippet>754</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
-<a href=#755 id=755 data-nosnippet>755</a>
-<a href=#756 id=756 data-nosnippet>756</a> <span class="kw">let </span>file_scan_tasks = setup(table_location);
-<a href=#757 id=757 data-nosnippet>757</a>
-<a href=#758 id=758 data-nosnippet>758</a> <span class="kw">let </span>delete_filter = delete_file_loader
-<a href=#759 id=759 data-nosnippet>759</a> .load_deletes(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>].deletes, file_scan_tasks[<span class="number">0</span>].schema_ref())
-<a href=#760 id=760 data-nosnippet>760</a> .<span class="kw">await
-<a href=#761 id=761 data-nosnippet>761</a> </span>.unwrap()
-<a href=#762 id=762 data-nosnippet>762</a> .unwrap();
-<a href=#763 id=763 data-nosnippet>763</a>
-<a href=#764 id=764 data-nosnippet>764</a> <span class="kw">let </span>result = delete_filter
-<a href=#765 id=765 data-nosnippet>765</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
-<a href=#766 id=766 data-nosnippet>766</a> .unwrap();
-<a href=#767 id=767 data-nosnippet>767</a>
-<a href=#768 id=768 data-nosnippet>768</a> <span class="comment">// union of pos dels from pos del file 1 and 2, ie
-<a href=#769 id=769 data-nosnippet>769</a> // [0, 1, 3, 5, 6, 8, 1022, 1023] | [0, 1, 3, 5, 20, 21, 22, 23]
-<a href=#770 id=770 data-nosnippet>770</a> // = [0, 1, 3, 5, 6, 8, 20, 21, 22, 23, 1022, 1023]
-<a href=#771 id=771 data-nosnippet>771</a> </span><span class="macro">assert_eq!</span>(result.lock().unwrap().len(), <span class="number">12</span>);
-<a href=#772 id=772 data-nosnippet>772</a>
-<a href=#773 id=773 data-nosnippet>773</a> <span class="kw">let </span>result = delete_filter.get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">1</span>]);
-<a href=#774 id=774 data-nosnippet>774</a> <span class="macro">assert!</span>(result.is_none()); <span class="comment">// no pos dels for file 3
-<a href=#775 id=775 data-nosnippet>775</a> </span>}
-<a href=#776 id=776 data-nosnippet>776</a>
-<a href=#777 id=777 data-nosnippet>777</a> <span class="attr">#[tokio::test]
-<a href=#778 id=778 data-nosnippet>778</a> </span><span class="kw">async fn </span>test_parse_positional_deletes_rejects_negative_positions() {
-<a href=#779 id=779 data-nosnippet>779</a> <span class="kw">let </span>schema = <span class="kw">crate</span>::arrow::delete_filter::tests::create_pos_del_schema();
-<a href=#780 id=780 data-nosnippet>780</a> <span class="kw">let </span>file_path_col = Arc::new(StringArray::from_iter_values(<span class="macro">vec!</span>[<span class="string">"data.parquet"</span>]));
-<a href=#781 id=781 data-nosnippet>781</a> <span class="kw">let </span>pos_col = Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[-<span class="number">1i64</span>]));
-<a href=#782 id=782 data-nosnippet>782</a> <span class="kw">let </span>batch = RecordBatch::try_new(schema, <span class="macro">vec!</span>[file_path_col, pos_col]).unwrap();
-<a href=#783 id=783 data-nosnippet>783</a> <span class="kw">let </span>stream = futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(batch)]).boxed();
-<a href=#784 id=784 data-nosnippet>784</a>
-<a href=#785 id=785 data-nosnippet>785</a> <span class="kw">let </span>err = CachingDeleteFileLoader::parse_positional_deletes_record_batch_stream(stream)
-<a href=#786 id=786 data-nosnippet>786</a> .<span class="kw">await
-<a href=#787 id=787 data-nosnippet>787</a> </span>.unwrap_err();
-<a href=#788 id=788 data-nosnippet>788</a>
-<a href=#789 id=789 data-nosnippet>789</a> <span class="macro">assert_eq!</span>(err.kind(), ErrorKind::DataInvalid);
-<a href=#790 id=790 data-nosnippet>790</a> <span class="macro">assert!</span>(err.message().contains(<span class="string">"negative position"</span>));
-<a href=#791 id=791 data-nosnippet>791</a> }
-<a href=#792 id=792 data-nosnippet>792</a>
-<a href=#793 id=793 data-nosnippet>793</a> <span class="doccomment">/// Verifies that evolve_schema on partial-schema equality deletes works correctly
-<a href=#794 id=794 data-nosnippet>794</a> /// when only equality_ids columns are evolved, not all table columns.
-<a href=#795 id=795 data-nosnippet>795</a> ///
-<a href=#796 id=796 data-nosnippet>796</a> /// Per the [Iceberg spec](https://iceberg.apache.org/spec/#equality-delete-files),
-<a href=#797 id=797 data-nosnippet>797</a> /// equality delete files can contain only a subset of columns.
-<a href=#798 id=798 data-nosnippet>798</a> </span><span class="attr">#[tokio::test]
-<a href=#799 id=799 data-nosnippet>799</a> </span><span class="kw">async fn </span>test_partial_schema_equality_deletes_evolve_succeeds() {
-<a href=#800 id=800 data-nosnippet>800</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#801 id=801 data-nosnippet>801</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
-<a href=#802 id=802 data-nosnippet>802</a>
-<a href=#803 id=803 data-nosnippet>803</a> <span class="comment">// Create table schema with REQUIRED fields
-<a href=#804 id=804 data-nosnippet>804</a> </span><span class="kw">let </span>table_schema = Arc::new(
-<a href=#805 id=805 data-nosnippet>805</a> Schema::builder()
-<a href=#806 id=806 data-nosnippet>806</a> .with_schema_id(<span class="number">1</span>)
-<a href=#807 id=807 data-nosnippet>807</a> .with_fields(<span class="macro">vec!</span>[
-<a href=#808 id=808 data-nosnippet>808</a> <span class="kw">crate</span>::spec::NestedField::required(
-<a href=#809 id=809 data-nosnippet>809</a> <span class="number">1</span>,
-<a href=#810 id=810 data-nosnippet>810</a> <span class="string">"id"</span>,
-<a href=#811 id=811 data-nosnippet>811</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Int),
-<a href=#812 id=812 data-nosnippet>812</a> )
-<a href=#813 id=813 data-nosnippet>813</a> .into(),
-<a href=#814 id=814 data-nosnippet>814</a> <span class="kw">crate</span>::spec::NestedField::required(
-<a href=#815 id=815 data-nosnippet>815</a> <span class="number">2</span>,
-<a href=#816 id=816 data-nosnippet>816</a> <span class="string">"data"</span>,
-<a href=#817 id=817 data-nosnippet>817</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::String),
-<a href=#818 id=818 data-nosnippet>818</a> )
-<a href=#819 id=819 data-nosnippet>819</a> .into(),
-<a href=#820 id=820 data-nosnippet>820</a> ])
-<a href=#821 id=821 data-nosnippet>821</a> .build()
-<a href=#822 id=822 data-nosnippet>822</a> .unwrap(),
-<a href=#823 id=823 data-nosnippet>823</a> );
-<a href=#824 id=824 data-nosnippet>824</a>
-<a href=#825 id=825 data-nosnippet>825</a> <span class="comment">// Write equality delete file with PARTIAL schema (only 'data' column)
-<a href=#826 id=826 data-nosnippet>826</a> </span><span class="kw">let </span>delete_file_path = {
-<a href=#827 id=827 data-nosnippet>827</a> <span class="kw">let </span>data_vals = <span class="macro">vec!</span>[<span class="string">"a"</span>, <span class="string">"d"</span>, <span class="string">"g"</span>];
-<a href=#828 id=828 data-nosnippet>828</a> <span class="kw">let </span>data_col = Arc::new(StringArray::from(data_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#829 id=829 data-nosnippet>829</a>
-<a href=#830 id=830 data-nosnippet>830</a> <span class="kw">let </span>delete_schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[simple_field(
-<a href=#831 id=831 data-nosnippet>831</a> <span class="string">"data"</span>,
-<a href=#832 id=832 data-nosnippet>832</a> DataType::Utf8,
-<a href=#833 id=833 data-nosnippet>833</a> <span class="bool-val">false</span>,
-<a href=#834 id=834 data-nosnippet>834</a> <span class="string">"2"</span>, <span class="comment">// field ID
-<a href=#835 id=835 data-nosnippet>835</a> </span>)]));
-<a href=#836 id=836 data-nosnippet>836</a>
-<a href=#837 id=837 data-nosnippet>837</a> <span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[data_col]).unwrap();
-<a href=#838 id=838 data-nosnippet>838</a>
-<a href=#839 id=839 data-nosnippet>839</a> <span class="kw">let </span>path = <span class="macro">format!</span>(<span class="string">"{}/partial-eq-deletes.parquet"</span>, <span class="kw-2">&</span>table_location);
-<a href=#840 id=840 data-nosnippet>840</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
-<a href=#841 id=841 data-nosnippet>841</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#842 id=842 data-nosnippet>842</a> .set_compression(Compression::SNAPPY)
-<a href=#843 id=843 data-nosnippet>843</a> .build();
-<a href=#844 id=844 data-nosnippet>844</a> <span class="kw">let </span><span class="kw-2">mut </span>writer =
-<a href=#845 id=845 data-nosnippet>845</a> ArrowWriter::try_new(file, delete_batch.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#846 id=846 data-nosnippet>846</a> writer.write(<span class="kw-2">&</span>delete_batch).expect(<span class="string">"Writing batch"</span>);
-<a href=#847 id=847 data-nosnippet>847</a> writer.close().unwrap();
-<a href=#848 id=848 data-nosnippet>848</a> path
-<a href=#849 id=849 data-nosnippet>849</a> };
+<a href=#741 id=741 data-nosnippet>741</a> <span class="kw">let </span>predicate = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
+<a href=#742 id=742 data-nosnippet>742</a> stream,
+<a href=#743 id=743 data-nosnippet>743</a> HashSet::from_iter(<span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">3</span>]),
+<a href=#744 id=744 data-nosnippet>744</a> )
+<a href=#745 id=745 data-nosnippet>745</a> .<span class="kw">await
+<a href=#746 id=746 data-nosnippet>746</a> </span>.expect(<span class="string">"error parsing equality delete stream"</span>);
+<a href=#747 id=747 data-nosnippet>747</a>
+<a href=#748 id=748 data-nosnippet>748</a> <span class="macro">assert_eq!</span>(
+<a href=#749 id=749 data-nosnippet>749</a> predicate.to_string(),
+<a href=#750 id=750 data-nosnippet>750</a> <span class="string">"((id IS NULL) OR (id != 1)) OR ((status IS NULL) OR (status != \"X\"))"
+<a href=#751 id=751 data-nosnippet>751</a> </span>);
+<a href=#752 id=752 data-nosnippet>752</a> }
+<a href=#753 id=753 data-nosnippet>753</a>
+<a href=#754 id=754 data-nosnippet>754</a> <span class="comment">// A data row is kept only if it matches none of the delete rows, so the per-row keep
+<a href=#755 id=755 data-nosnippet>755</a> // predicates are AND-ed.
+<a href=#756 id=756 data-nosnippet>756</a> </span><span class="attr">#[tokio::test]
+<a href=#757 id=757 data-nosnippet>757</a> </span><span class="kw">async fn </span>test_equality_delete_predicate_multiple_delete_rows() {
+<a href=#758 id=758 data-nosnippet>758</a> <span class="kw">let </span>schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[simple_field(
+<a href=#759 id=759 data-nosnippet>759</a> <span class="string">"status"</span>,
+<a href=#760 id=760 data-nosnippet>760</a> DataType::Utf8,
+<a href=#761 id=761 data-nosnippet>761</a> <span class="bool-val">true</span>,
+<a href=#762 id=762 data-nosnippet>762</a> <span class="string">"3"</span>,
+<a href=#763 id=763 data-nosnippet>763</a> )]));
+<a href=#764 id=764 data-nosnippet>764</a> <span class="kw">let </span>batch = RecordBatch::try_new(schema, <span class="macro">vec!</span>[Arc::new(StringArray::from(<span class="macro">vec!</span>[
+<a href=#765 id=765 data-nosnippet>765</a> <span class="prelude-val">Some</span>(<span class="string">"A"</span>),
+<a href=#766 id=766 data-nosnippet>766</a> <span class="prelude-val">Some</span>(<span class="string">"B"</span>),
+<a href=#767 id=767 data-nosnippet>767</a> ])) <span class="kw">as </span>ArrayRef])
+<a href=#768 id=768 data-nosnippet>768</a> .unwrap();
+<a href=#769 id=769 data-nosnippet>769</a> <span class="kw">let </span>stream: ArrowRecordBatchStream = futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(batch)]).boxed();
+<a href=#770 id=770 data-nosnippet>770</a>
+<a href=#771 id=771 data-nosnippet>771</a> <span class="kw">let </span>predicate = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
+<a href=#772 id=772 data-nosnippet>772</a> stream,
+<a href=#773 id=773 data-nosnippet>773</a> HashSet::from_iter(<span class="macro">vec!</span>[<span class="number">3</span>]),
+<a href=#774 id=774 data-nosnippet>774</a> )
+<a href=#775 id=775 data-nosnippet>775</a> .<span class="kw">await
+<a href=#776 id=776 data-nosnippet>776</a> </span>.expect(<span class="string">"error parsing equality delete stream"</span>);
+<a href=#777 id=777 data-nosnippet>777</a>
+<a href=#778 id=778 data-nosnippet>778</a> <span class="macro">assert_eq!</span>(
+<a href=#779 id=779 data-nosnippet>779</a> predicate.to_string(),
+<a href=#780 id=780 data-nosnippet>780</a> <span class="string">"((status IS NULL) OR (status != \"A\")) AND ((status IS NULL) OR (status != \"B\"))"
+<a href=#781 id=781 data-nosnippet>781</a> </span>);
+<a href=#782 id=782 data-nosnippet>782</a> }
+<a href=#783 id=783 data-nosnippet>783</a>
+<a href=#784 id=784 data-nosnippet>784</a> <span class="doccomment">/// Create a simple field with metadata.
+<a href=#785 id=785 data-nosnippet>785</a> </span><span class="kw">fn </span>simple_field(name: <span class="kw-2">&</span>str, ty: DataType, nullable: bool, value: <span class="kw-2">&</span>str) -> Field {
+<a href=#786 id=786 data-nosnippet>786</a> arrow_schema::Field::new(name, ty, nullable).with_metadata(HashMap::from([(
+<a href=#787 id=787 data-nosnippet>787</a> PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#788 id=788 data-nosnippet>788</a> value.to_string(),
+<a href=#789 id=789 data-nosnippet>789</a> )]))
+<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="kw">fn </span>setup_write_equality_delete_file_1(table_location: <span class="kw-2">&</span>str) -> String {
+<a href=#793 id=793 data-nosnippet>793</a> <span class="kw">let </span>col_y_vals = <span class="macro">vec!</span>[<span class="number">1</span>, <span class="number">2</span>];
+<a href=#794 id=794 data-nosnippet>794</a> <span class="kw">let </span>col_y = Arc::new(Int64Array::from(col_y_vals)) <span class="kw">as </span>ArrayRef;
+<a href=#795 id=795 data-nosnippet>795</a>
+<a href=#796 id=796 data-nosnippet>796</a> <span class="kw">let </span>col_z_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="number">100</span>), <span class="prelude-val">None</span>];
+<a href=#797 id=797 data-nosnippet>797</a> <span class="kw">let </span>col_z = Arc::new(Int64Array::from(col_z_vals)) <span class="kw">as </span>ArrayRef;
+<a href=#798 id=798 data-nosnippet>798</a>
+<a href=#799 id=799 data-nosnippet>799</a> <span class="kw">let </span>col_a_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"HELP"</span>), <span class="prelude-val">None</span>];
+<a href=#800 id=800 data-nosnippet>800</a> <span class="kw">let </span>col_a = Arc::new(StringArray::from(col_a_vals)) <span class="kw">as </span>ArrayRef;
+<a href=#801 id=801 data-nosnippet>801</a>
+<a href=#802 id=802 data-nosnippet>802</a> <span class="kw">let </span>col_s = Arc::new(StructArray::from(<span class="macro">vec!</span>[
+<a href=#803 id=803 data-nosnippet>803</a> (
+<a href=#804 id=804 data-nosnippet>804</a> Arc::new(simple_field(<span class="string">"sa"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"6"</span>)),
+<a href=#805 id=805 data-nosnippet>805</a> Arc::new(Int32Array::from(<span class="macro">vec!</span>[<span class="number">4</span>, <span class="number">5</span>])) <span class="kw">as </span>ArrayRef,
+<a href=#806 id=806 data-nosnippet>806</a> ),
+<a href=#807 id=807 data-nosnippet>807</a> (
+<a href=#808 id=808 data-nosnippet>808</a> Arc::new(simple_field(<span class="string">"sb"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"7"</span>)),
+<a href=#809 id=809 data-nosnippet>809</a> Arc::new(StringArray::from(<span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"x"</span>), <span class="prelude-val">None</span>])) <span class="kw">as </span>ArrayRef,
+<a href=#810 id=810 data-nosnippet>810</a> ),
+<a href=#811 id=811 data-nosnippet>811</a> ]));
+<a href=#812 id=812 data-nosnippet>812</a>
+<a href=#813 id=813 data-nosnippet>813</a> <span class="kw">let </span>col_b_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="kw-2">&</span><span class="string">b"binary_data"</span>[..]), <span class="prelude-val">None</span>];
+<a href=#814 id=814 data-nosnippet>814</a> <span class="kw">let </span>col_b = Arc::new(BinaryArray::from(col_b_vals)) <span class="kw">as </span>ArrayRef;
+<a href=#815 id=815 data-nosnippet>815</a>
+<a href=#816 id=816 data-nosnippet>816</a> <span class="kw">let </span>equality_delete_schema = {
+<a href=#817 id=817 data-nosnippet>817</a> <span class="kw">let </span>struct_field = DataType::Struct(Fields::from(<span class="macro">vec!</span>[
+<a href=#818 id=818 data-nosnippet>818</a> simple_field(<span class="string">"sa"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"6"</span>),
+<a href=#819 id=819 data-nosnippet>819</a> simple_field(<span class="string">"sb"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"7"</span>),
+<a href=#820 id=820 data-nosnippet>820</a> ]));
+<a href=#821 id=821 data-nosnippet>821</a>
+<a href=#822 id=822 data-nosnippet>822</a> <span class="kw">let </span>fields = <span class="macro">vec!</span>[
+<a href=#823 id=823 data-nosnippet>823</a> Field::new(<span class="string">"y"</span>, arrow_schema::DataType::Int64, <span class="bool-val">true</span>).with_metadata(HashMap::from(
+<a href=#824 id=824 data-nosnippet>824</a> [(PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"2"</span>.to_string())],
+<a href=#825 id=825 data-nosnippet>825</a> )),
+<a href=#826 id=826 data-nosnippet>826</a> Field::new(<span class="string">"z"</span>, arrow_schema::DataType::Int64, <span class="bool-val">true</span>).with_metadata(HashMap::from(
+<a href=#827 id=827 data-nosnippet>827</a> [(PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"3"</span>.to_string())],
+<a href=#828 id=828 data-nosnippet>828</a> )),
+<a href=#829 id=829 data-nosnippet>829</a> Field::new(<span class="string">"a"</span>, arrow_schema::DataType::Utf8, <span class="bool-val">true</span>).with_metadata(HashMap::from([
+<a href=#830 id=830 data-nosnippet>830</a> (PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"4"</span>.to_string()),
+<a href=#831 id=831 data-nosnippet>831</a> ])),
+<a href=#832 id=832 data-nosnippet>832</a> simple_field(<span class="string">"s"</span>, struct_field, <span class="bool-val">false</span>, <span class="string">"5"</span>),
+<a href=#833 id=833 data-nosnippet>833</a> simple_field(<span class="string">"b"</span>, DataType::Binary, <span class="bool-val">true</span>, <span class="string">"8"</span>),
+<a href=#834 id=834 data-nosnippet>834</a> ];
+<a href=#835 id=835 data-nosnippet>835</a> Arc::new(arrow_schema::Schema::new(fields))
+<a href=#836 id=836 data-nosnippet>836</a> };
+<a href=#837 id=837 data-nosnippet>837</a>
+<a href=#838 id=838 data-nosnippet>838</a> <span class="kw">let </span>equality_deletes_to_write = RecordBatch::try_new(equality_delete_schema.clone(), <span class="macro">vec!</span>[
+<a href=#839 id=839 data-nosnippet>839</a> col_y, col_z, col_a, col_s, col_b,
+<a href=#840 id=840 data-nosnippet>840</a> ])
+<a href=#841 id=841 data-nosnippet>841</a> .unwrap();
+<a href=#842 id=842 data-nosnippet>842</a>
+<a href=#843 id=843 data-nosnippet>843</a> <span class="kw">let </span>path = <span class="macro">format!</span>(<span class="string">"{}/equality-deletes-1.parquet"</span>, <span class="kw-2">&</span>table_location);
+<a href=#844 id=844 data-nosnippet>844</a>
+<a href=#845 id=845 data-nosnippet>845</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
+<a href=#846 id=846 data-nosnippet>846</a>
+<a href=#847 id=847 data-nosnippet>847</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#848 id=848 data-nosnippet>848</a> .set_compression(Compression::SNAPPY)
+<a href=#849 id=849 data-nosnippet>849</a> .build();
<a href=#850 id=850 data-nosnippet>850</a>
-<a href=#851 id=851 data-nosnippet>851</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
-<a href=#852 id=852 data-nosnippet>852</a> <span class="kw">let </span>basic_delete_file_loader =
-<a href=#853 id=853 data-nosnippet>853</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
-<a href=#854 id=854 data-nosnippet>854</a>
-<a href=#855 id=855 data-nosnippet>855</a> <span class="kw">let </span>batch_stream = basic_delete_file_loader
-<a href=#856 id=856 data-nosnippet>856</a> .parquet_to_batch_stream(
-<a href=#857 id=857 data-nosnippet>857</a> <span class="kw-2">&</span>delete_file_path,
-<a href=#858 id=858 data-nosnippet>858</a> std::fs::metadata(<span class="kw-2">&</span>delete_file_path).unwrap().len(),
-<a href=#859 id=859 data-nosnippet>859</a> )
-<a href=#860 id=860 data-nosnippet>860</a> .<span class="kw">await
-<a href=#861 id=861 data-nosnippet>861</a> </span>.unwrap();
-<a href=#862 id=862 data-nosnippet>862</a>
-<a href=#863 id=863 data-nosnippet>863</a> <span class="comment">// Only evolve the equality_ids columns (field 2), not all table columns
-<a href=#864 id=864 data-nosnippet>864</a> </span><span class="kw">let </span>equality_ids = <span class="macro">vec!</span>[<span class="number">2</span>];
-<a href=#865 id=865 data-nosnippet>865</a> <span class="kw">let </span>evolved_stream =
-<a href=#866 id=866 data-nosnippet>866</a> BasicDeleteFileLoader::evolve_schema(batch_stream, table_schema, <span class="kw-2">&</span>equality_ids)
-<a href=#867 id=867 data-nosnippet>867</a> .<span class="kw">await
-<a href=#868 id=868 data-nosnippet>868</a> </span>.unwrap();
-<a href=#869 id=869 data-nosnippet>869</a>
-<a href=#870 id=870 data-nosnippet>870</a> <span class="kw">let </span>result = evolved_stream.try_collect::<Vec<<span class="kw">_</span>>>().<span class="kw">await</span>;
-<a href=#871 id=871 data-nosnippet>871</a>
-<a href=#872 id=872 data-nosnippet>872</a> <span class="macro">assert!</span>(
-<a href=#873 id=873 data-nosnippet>873</a> result.is_ok(),
-<a href=#874 id=874 data-nosnippet>874</a> <span class="string">"Expected success when evolving only equality_ids columns, got error: {:?}"</span>,
-<a href=#875 id=875 data-nosnippet>875</a> result.err()
-<a href=#876 id=876 data-nosnippet>876</a> );
-<a href=#877 id=877 data-nosnippet>877</a>
-<a href=#878 id=878 data-nosnippet>878</a> <span class="kw">let </span>batches = result.unwrap();
-<a href=#879 id=879 data-nosnippet>879</a> <span class="macro">assert_eq!</span>(batches.len(), <span class="number">1</span>);
-<a href=#880 id=880 data-nosnippet>880</a>
-<a href=#881 id=881 data-nosnippet>881</a> <span class="kw">let </span>batch = <span class="kw-2">&</span>batches[<span class="number">0</span>];
-<a href=#882 id=882 data-nosnippet>882</a> <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">3</span>);
-<a href=#883 id=883 data-nosnippet>883</a> <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">1</span>); <span class="comment">// Only 'data' column
+<a href=#851 id=851 data-nosnippet>851</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(
+<a href=#852 id=852 data-nosnippet>852</a> file,
+<a href=#853 id=853 data-nosnippet>853</a> equality_deletes_to_write.schema(),
+<a href=#854 id=854 data-nosnippet>854</a> <span class="prelude-val">Some</span>(props.clone()),
+<a href=#855 id=855 data-nosnippet>855</a> )
+<a href=#856 id=856 data-nosnippet>856</a> .unwrap();
+<a href=#857 id=857 data-nosnippet>857</a>
+<a href=#858 id=858 data-nosnippet>858</a> writer
+<a href=#859 id=859 data-nosnippet>859</a> .write(<span class="kw-2">&</span>equality_deletes_to_write)
+<a href=#860 id=860 data-nosnippet>860</a> .expect(<span class="string">"Writing batch"</span>);
+<a href=#861 id=861 data-nosnippet>861</a>
+<a href=#862 id=862 data-nosnippet>862</a> <span class="comment">// writer must be closed to write footer
+<a href=#863 id=863 data-nosnippet>863</a> </span>writer.close().unwrap();
+<a href=#864 id=864 data-nosnippet>864</a>
+<a href=#865 id=865 data-nosnippet>865</a> path
+<a href=#866 id=866 data-nosnippet>866</a> }
+<a href=#867 id=867 data-nosnippet>867</a>
+<a href=#868 id=868 data-nosnippet>868</a> <span class="attr">#[tokio::test]
+<a href=#869 id=869 data-nosnippet>869</a> </span><span class="kw">async fn </span>test_caching_delete_file_loader_load_deletes() {
+<a href=#870 id=870 data-nosnippet>870</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#871 id=871 data-nosnippet>871</a> <span class="kw">let </span>table_location = tmp_dir.path();
+<a href=#872 id=872 data-nosnippet>872</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<a href=#873 id=873 data-nosnippet>873</a>
+<a href=#874 id=874 data-nosnippet>874</a> <span class="kw">let </span>delete_file_loader =
+<a href=#875 id=875 data-nosnippet>875</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
+<a href=#876 id=876 data-nosnippet>876</a>
+<a href=#877 id=877 data-nosnippet>877</a> <span class="kw">let </span>file_scan_tasks = setup(table_location);
+<a href=#878 id=878 data-nosnippet>878</a>
+<a href=#879 id=879 data-nosnippet>879</a> <span class="kw">let </span>delete_filter = delete_file_loader
+<a href=#880 id=880 data-nosnippet>880</a> .load_deletes(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>].deletes, file_scan_tasks[<span class="number">0</span>].schema_ref())
+<a href=#881 id=881 data-nosnippet>881</a> .<span class="kw">await
+<a href=#882 id=882 data-nosnippet>882</a> </span>.unwrap()
+<a href=#883 id=883 data-nosnippet>883</a> .unwrap();
<a href=#884 id=884 data-nosnippet>884</a>
-<a href=#885 id=885 data-nosnippet>885</a> // Verify the actual values are preserved after schema evolution
-<a href=#886 id=886 data-nosnippet>886</a> </span><span class="kw">let </span>data_col = batch.column(<span class="number">0</span>).as_string::<i32>();
-<a href=#887 id=887 data-nosnippet>887</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">0</span>), <span class="string">"a"</span>);
-<a href=#888 id=888 data-nosnippet>888</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">1</span>), <span class="string">"d"</span>);
-<a href=#889 id=889 data-nosnippet>889</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">2</span>), <span class="string">"g"</span>);
-<a href=#890 id=890 data-nosnippet>890</a> }
-<a href=#891 id=891 data-nosnippet>891</a>
-<a href=#892 id=892 data-nosnippet>892</a> <span class="doccomment">/// Test loading a FileScanTask with BOTH positional and equality deletes.
-<a href=#893 id=893 data-nosnippet>893</a> /// Verifies the fix for the inverted condition that caused "Missing predicate for equality delete file" errors.
-<a href=#894 id=894 data-nosnippet>894</a> </span><span class="attr">#[tokio::test]
-<a href=#895 id=895 data-nosnippet>895</a> </span><span class="kw">async fn </span>test_load_deletes_with_mixed_types() {
-<a href=#896 id=896 data-nosnippet>896</a> <span class="kw">use </span><span class="kw">crate</span>::scan::FileScanTask;
-<a href=#897 id=897 data-nosnippet>897</a> <span class="kw">use </span><span class="kw">crate</span>::spec::{DataFileFormat, Schema};
-<a href=#898 id=898 data-nosnippet>898</a>
-<a href=#899 id=899 data-nosnippet>899</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#900 id=900 data-nosnippet>900</a> <span class="kw">let </span>table_location = tmp_dir.path();
-<a href=#901 id=901 data-nosnippet>901</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
-<a href=#902 id=902 data-nosnippet>902</a>
-<a href=#903 id=903 data-nosnippet>903</a> <span class="comment">// Create the data file schema
-<a href=#904 id=904 data-nosnippet>904</a> </span><span class="kw">let </span>data_file_schema = Arc::new(
-<a href=#905 id=905 data-nosnippet>905</a> Schema::builder()
-<a href=#906 id=906 data-nosnippet>906</a> .with_fields(<span class="macro">vec!</span>[
-<a href=#907 id=907 data-nosnippet>907</a> <span class="kw">crate</span>::spec::NestedField::optional(
-<a href=#908 id=908 data-nosnippet>908</a> <span class="number">2</span>,
-<a href=#909 id=909 data-nosnippet>909</a> <span class="string">"y"</span>,
-<a href=#910 id=910 data-nosnippet>910</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Long),
-<a href=#911 id=911 data-nosnippet>911</a> )
-<a href=#912 id=912 data-nosnippet>912</a> .into(),
-<a href=#913 id=913 data-nosnippet>913</a> <span class="kw">crate</span>::spec::NestedField::optional(
-<a href=#914 id=914 data-nosnippet>914</a> <span class="number">3</span>,
-<a href=#915 id=915 data-nosnippet>915</a> <span class="string">"z"</span>,
-<a href=#916 id=916 data-nosnippet>916</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Long),
-<a href=#917 id=917 data-nosnippet>917</a> )
-<a href=#918 id=918 data-nosnippet>918</a> .into(),
-<a href=#919 id=919 data-nosnippet>919</a> ])
-<a href=#920 id=920 data-nosnippet>920</a> .build()
-<a href=#921 id=921 data-nosnippet>921</a> .unwrap(),
-<a href=#922 id=922 data-nosnippet>922</a> );
+<a href=#885 id=885 data-nosnippet>885</a> <span class="kw">let </span>result = delete_filter
+<a href=#886 id=886 data-nosnippet>886</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
+<a href=#887 id=887 data-nosnippet>887</a> .unwrap();
+<a href=#888 id=888 data-nosnippet>888</a>
+<a href=#889 id=889 data-nosnippet>889</a> <span class="comment">// union of pos dels from pos del file 1 and 2, ie
+<a href=#890 id=890 data-nosnippet>890</a> // [0, 1, 3, 5, 6, 8, 1022, 1023] | [0, 1, 3, 5, 20, 21, 22, 23]
+<a href=#891 id=891 data-nosnippet>891</a> // = [0, 1, 3, 5, 6, 8, 20, 21, 22, 23, 1022, 1023]
+<a href=#892 id=892 data-nosnippet>892</a> </span><span class="macro">assert_eq!</span>(result.lock().unwrap().len(), <span class="number">12</span>);
+<a href=#893 id=893 data-nosnippet>893</a>
+<a href=#894 id=894 data-nosnippet>894</a> <span class="kw">let </span>result = delete_filter.get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">1</span>]);
+<a href=#895 id=895 data-nosnippet>895</a> <span class="macro">assert!</span>(result.is_none()); <span class="comment">// no pos dels for file 3
+<a href=#896 id=896 data-nosnippet>896</a> </span>}
+<a href=#897 id=897 data-nosnippet>897</a>
+<a href=#898 id=898 data-nosnippet>898</a> <span class="attr">#[tokio::test]
+<a href=#899 id=899 data-nosnippet>899</a> </span><span class="kw">async fn </span>test_parse_positional_deletes_rejects_negative_positions() {
+<a href=#900 id=900 data-nosnippet>900</a> <span class="kw">let </span>schema = <span class="kw">crate</span>::arrow::delete_filter::tests::create_pos_del_schema();
+<a href=#901 id=901 data-nosnippet>901</a> <span class="kw">let </span>file_path_col = Arc::new(StringArray::from_iter_values(<span class="macro">vec!</span>[<span class="string">"data.parquet"</span>]));
+<a href=#902 id=902 data-nosnippet>902</a> <span class="kw">let </span>pos_col = Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[-<span class="number">1i64</span>]));
+<a href=#903 id=903 data-nosnippet>903</a> <span class="kw">let </span>batch = RecordBatch::try_new(schema, <span class="macro">vec!</span>[file_path_col, pos_col]).unwrap();
+<a href=#904 id=904 data-nosnippet>904</a> <span class="kw">let </span>stream = futures::stream::iter(<span class="macro">vec!</span>[<span class="prelude-val">Ok</span>(batch)]).boxed();
+<a href=#905 id=905 data-nosnippet>905</a>
+<a href=#906 id=906 data-nosnippet>906</a> <span class="kw">let </span>err = CachingDeleteFileLoader::parse_positional_deletes_record_batch_stream(stream)
+<a href=#907 id=907 data-nosnippet>907</a> .<span class="kw">await
+<a href=#908 id=908 data-nosnippet>908</a> </span>.unwrap_err();
+<a href=#909 id=909 data-nosnippet>909</a>
+<a href=#910 id=910 data-nosnippet>910</a> <span class="macro">assert_eq!</span>(err.kind(), ErrorKind::DataInvalid);
+<a href=#911 id=911 data-nosnippet>911</a> <span class="macro">assert!</span>(err.message().contains(<span class="string">"negative position"</span>));
+<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> <span class="doccomment">/// Verifies that evolve_schema on partial-schema equality deletes works correctly
+<a href=#915 id=915 data-nosnippet>915</a> /// when only equality_ids columns are evolved, not all table columns.
+<a href=#916 id=916 data-nosnippet>916</a> ///
+<a href=#917 id=917 data-nosnippet>917</a> /// Per the [Iceberg spec](https://iceberg.apache.org/spec/#equality-delete-files),
+<a href=#918 id=918 data-nosnippet>918</a> /// equality delete files can contain only a subset of columns.
+<a href=#919 id=919 data-nosnippet>919</a> </span><span class="attr">#[tokio::test]
+<a href=#920 id=920 data-nosnippet>920</a> </span><span class="kw">async fn </span>test_partial_schema_equality_deletes_evolve_succeeds() {
+<a href=#921 id=921 data-nosnippet>921</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#922 id=922 data-nosnippet>922</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
<a href=#923 id=923 data-nosnippet>923</a>
-<a href=#924 id=924 data-nosnippet>924</a> <span class="comment">// Write positional delete file
-<a href=#925 id=925 data-nosnippet>925</a> </span><span class="kw">let </span>positional_delete_schema = <span class="kw">crate</span>::arrow::delete_filter::tests::create_pos_del_schema();
-<a href=#926 id=926 data-nosnippet>926</a> <span class="kw">let </span>file_path_values =
-<a href=#927 id=927 data-nosnippet>927</a> <span class="macro">vec!</span>[<span class="macro">format!</span>(<span class="string">"{}/data-1.parquet"</span>, table_location.to_str().unwrap()); <span class="number">4</span>];
-<a href=#928 id=928 data-nosnippet>928</a> <span class="kw">let </span>file_path_col = Arc::new(StringArray::from_iter_values(<span class="kw-2">&</span>file_path_values));
-<a href=#929 id=929 data-nosnippet>929</a> <span class="kw">let </span>pos_col = Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[<span class="number">0i64</span>, <span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>]));
-<a href=#930 id=930 data-nosnippet>930</a>
-<a href=#931 id=931 data-nosnippet>931</a> <span class="kw">let </span>positional_deletes_to_write =
-<a href=#932 id=932 data-nosnippet>932</a> RecordBatch::try_new(positional_delete_schema.clone(), <span class="macro">vec!</span>[
-<a href=#933 id=933 data-nosnippet>933</a> file_path_col,
-<a href=#934 id=934 data-nosnippet>934</a> pos_col,
-<a href=#935 id=935 data-nosnippet>935</a> ])
-<a href=#936 id=936 data-nosnippet>936</a> .unwrap();
-<a href=#937 id=937 data-nosnippet>937</a>
-<a href=#938 id=938 data-nosnippet>938</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#939 id=939 data-nosnippet>939</a> .set_compression(Compression::SNAPPY)
-<a href=#940 id=940 data-nosnippet>940</a> .build();
-<a href=#941 id=941 data-nosnippet>941</a>
-<a href=#942 id=942 data-nosnippet>942</a> <span class="kw">let </span>pos_del_path = <span class="macro">format!</span>(<span class="string">"{}/pos-del-mixed.parquet"</span>, table_location.to_str().unwrap());
-<a href=#943 id=943 data-nosnippet>943</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>pos_del_path).unwrap();
-<a href=#944 id=944 data-nosnippet>944</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(
-<a href=#945 id=945 data-nosnippet>945</a> file,
-<a href=#946 id=946 data-nosnippet>946</a> positional_deletes_to_write.schema(),
-<a href=#947 id=947 data-nosnippet>947</a> <span class="prelude-val">Some</span>(props.clone()),
-<a href=#948 id=948 data-nosnippet>948</a> )
-<a href=#949 id=949 data-nosnippet>949</a> .unwrap();
-<a href=#950 id=950 data-nosnippet>950</a> writer.write(<span class="kw-2">&</span>positional_deletes_to_write).unwrap();
-<a href=#951 id=951 data-nosnippet>951</a> writer.close().unwrap();
-<a href=#952 id=952 data-nosnippet>952</a>
-<a href=#953 id=953 data-nosnippet>953</a> <span class="comment">// Write equality delete file
-<a href=#954 id=954 data-nosnippet>954</a> </span><span class="kw">let </span>eq_delete_path = setup_write_equality_delete_file_1(table_location.to_str().unwrap());
-<a href=#955 id=955 data-nosnippet>955</a>
-<a href=#956 id=956 data-nosnippet>956</a> <span class="comment">// Create FileScanTask with BOTH positional and equality deletes
-<a href=#957 id=957 data-nosnippet>957</a> </span><span class="kw">let </span>pos_del = FileScanTaskDeleteFile::builder()
-<a href=#958 id=958 data-nosnippet>958</a> .with_file_path(pos_del_path.clone())
-<a href=#959 id=959 data-nosnippet>959</a> .with_file_size_in_bytes(std::fs::metadata(<span class="kw-2">&</span>pos_del_path).unwrap().len())
-<a href=#960 id=960 data-nosnippet>960</a> .with_file_type(DataContentType::PositionDeletes)
-<a href=#961 id=961 data-nosnippet>961</a> .with_partition_spec_id(<span class="number">0</span>)
-<a href=#962 id=962 data-nosnippet>962</a> .build();
-<a href=#963 id=963 data-nosnippet>963</a>
-<a href=#964 id=964 data-nosnippet>964</a> <span class="kw">let </span>eq_del = FileScanTaskDeleteFile::builder()
-<a href=#965 id=965 data-nosnippet>965</a> .with_file_path(eq_delete_path.clone())
-<a href=#966 id=966 data-nosnippet>966</a> .with_file_size_in_bytes(std::fs::metadata(<span class="kw-2">&</span>eq_delete_path).unwrap().len())
-<a href=#967 id=967 data-nosnippet>967</a> .with_file_type(DataContentType::EqualityDeletes)
-<a href=#968 id=968 data-nosnippet>968</a> .with_partition_spec_id(<span class="number">0</span>)
-<a href=#969 id=969 data-nosnippet>969</a> .with_equality_ids(<span class="prelude-val">Some</span>(<span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">3</span>])) <span class="comment">// Only use field IDs that exist in both schemas
-<a href=#970 id=970 data-nosnippet>970</a> </span>.build();
+<a href=#924 id=924 data-nosnippet>924</a> <span class="comment">// Create table schema with REQUIRED fields
+<a href=#925 id=925 data-nosnippet>925</a> </span><span class="kw">let </span>table_schema = Arc::new(
+<a href=#926 id=926 data-nosnippet>926</a> Schema::builder()
+<a href=#927 id=927 data-nosnippet>927</a> .with_schema_id(<span class="number">1</span>)
+<a href=#928 id=928 data-nosnippet>928</a> .with_fields(<span class="macro">vec!</span>[
+<a href=#929 id=929 data-nosnippet>929</a> <span class="kw">crate</span>::spec::NestedField::required(
+<a href=#930 id=930 data-nosnippet>930</a> <span class="number">1</span>,
+<a href=#931 id=931 data-nosnippet>931</a> <span class="string">"id"</span>,
+<a href=#932 id=932 data-nosnippet>932</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Int),
+<a href=#933 id=933 data-nosnippet>933</a> )
+<a href=#934 id=934 data-nosnippet>934</a> .into(),
+<a href=#935 id=935 data-nosnippet>935</a> <span class="kw">crate</span>::spec::NestedField::required(
+<a href=#936 id=936 data-nosnippet>936</a> <span class="number">2</span>,
+<a href=#937 id=937 data-nosnippet>937</a> <span class="string">"data"</span>,
+<a href=#938 id=938 data-nosnippet>938</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::String),
+<a href=#939 id=939 data-nosnippet>939</a> )
+<a href=#940 id=940 data-nosnippet>940</a> .into(),
+<a href=#941 id=941 data-nosnippet>941</a> ])
+<a href=#942 id=942 data-nosnippet>942</a> .build()
+<a href=#943 id=943 data-nosnippet>943</a> .unwrap(),
+<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> <span class="comment">// Write equality delete file with PARTIAL schema (only 'data' column)
+<a href=#947 id=947 data-nosnippet>947</a> </span><span class="kw">let </span>delete_file_path = {
+<a href=#948 id=948 data-nosnippet>948</a> <span class="kw">let </span>data_vals = <span class="macro">vec!</span>[<span class="string">"a"</span>, <span class="string">"d"</span>, <span class="string">"g"</span>];
+<a href=#949 id=949 data-nosnippet>949</a> <span class="kw">let </span>data_col = Arc::new(StringArray::from(data_vals)) <span class="kw">as </span>ArrayRef;
+<a href=#950 id=950 data-nosnippet>950</a>
+<a href=#951 id=951 data-nosnippet>951</a> <span class="kw">let </span>delete_schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[simple_field(
+<a href=#952 id=952 data-nosnippet>952</a> <span class="string">"data"</span>,
+<a href=#953 id=953 data-nosnippet>953</a> DataType::Utf8,
+<a href=#954 id=954 data-nosnippet>954</a> <span class="bool-val">false</span>,
+<a href=#955 id=955 data-nosnippet>955</a> <span class="string">"2"</span>, <span class="comment">// field ID
+<a href=#956 id=956 data-nosnippet>956</a> </span>)]));
+<a href=#957 id=957 data-nosnippet>957</a>
+<a href=#958 id=958 data-nosnippet>958</a> <span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[data_col]).unwrap();
+<a href=#959 id=959 data-nosnippet>959</a>
+<a href=#960 id=960 data-nosnippet>960</a> <span class="kw">let </span>path = <span class="macro">format!</span>(<span class="string">"{}/partial-eq-deletes.parquet"</span>, <span class="kw-2">&</span>table_location);
+<a href=#961 id=961 data-nosnippet>961</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
+<a href=#962 id=962 data-nosnippet>962</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#963 id=963 data-nosnippet>963</a> .set_compression(Compression::SNAPPY)
+<a href=#964 id=964 data-nosnippet>964</a> .build();
+<a href=#965 id=965 data-nosnippet>965</a> <span class="kw">let </span><span class="kw-2">mut </span>writer =
+<a href=#966 id=966 data-nosnippet>966</a> ArrowWriter::try_new(file, delete_batch.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#967 id=967 data-nosnippet>967</a> writer.write(<span class="kw-2">&</span>delete_batch).expect(<span class="string">"Writing batch"</span>);
+<a href=#968 id=968 data-nosnippet>968</a> writer.close().unwrap();
+<a href=#969 id=969 data-nosnippet>969</a> path
+<a href=#970 id=970 data-nosnippet>970</a> };
<a href=#971 id=971 data-nosnippet>971</a>
-<a href=#972 id=972 data-nosnippet>972</a> <span class="kw">let </span>file_scan_task = FileScanTask::builder()
-<a href=#973 id=973 data-nosnippet>973</a> .with_file_size_in_bytes(<span class="number">0</span>)
-<a href=#974 id=974 data-nosnippet>974</a> .with_start(<span class="number">0</span>)
-<a href=#975 id=975 data-nosnippet>975</a> .with_length(<span class="number">0</span>)
-<a href=#976 id=976 data-nosnippet>976</a> .with_data_file_path(<span class="macro">format!</span>(
-<a href=#977 id=977 data-nosnippet>977</a> <span class="string">"{}/data-1.parquet"</span>,
-<a href=#978 id=978 data-nosnippet>978</a> table_location.to_str().unwrap()
-<a href=#979 id=979 data-nosnippet>979</a> ))
-<a href=#980 id=980 data-nosnippet>980</a> .with_data_file_format(DataFileFormat::Parquet)
-<a href=#981 id=981 data-nosnippet>981</a> .with_schema(data_file_schema.clone())
-<a href=#982 id=982 data-nosnippet>982</a> .with_project_field_ids(<span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">3</span>])
-<a href=#983 id=983 data-nosnippet>983</a> .with_deletes(<span class="macro">vec!</span>[pos_del, eq_del])
-<a href=#984 id=984 data-nosnippet>984</a> .with_case_sensitive(<span class="bool-val">false</span>)
-<a href=#985 id=985 data-nosnippet>985</a> .build();
-<a href=#986 id=986 data-nosnippet>986</a>
-<a href=#987 id=987 data-nosnippet>987</a> <span class="comment">// Load the deletes - should handle both types without error
-<a href=#988 id=988 data-nosnippet>988</a> </span><span class="kw">let </span>delete_file_loader =
-<a href=#989 id=989 data-nosnippet>989</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
-<a href=#990 id=990 data-nosnippet>990</a> <span class="kw">let </span>delete_filter = delete_file_loader
-<a href=#991 id=991 data-nosnippet>991</a> .load_deletes(<span class="kw-2">&</span>file_scan_task.deletes, file_scan_task.schema_ref())
-<a href=#992 id=992 data-nosnippet>992</a> .<span class="kw">await
-<a href=#993 id=993 data-nosnippet>993</a> </span>.unwrap()
-<a href=#994 id=994 data-nosnippet>994</a> .unwrap();
-<a href=#995 id=995 data-nosnippet>995</a>
-<a href=#996 id=996 data-nosnippet>996</a> <span class="comment">// Verify both delete types can be processed together
-<a href=#997 id=997 data-nosnippet>997</a> </span><span class="kw">let </span>result = delete_filter
-<a href=#998 id=998 data-nosnippet>998</a> .build_equality_delete_predicate(<span class="kw-2">&</span>file_scan_task)
-<a href=#999 id=999 data-nosnippet>999</a> .<span class="kw">await</span>;
-<a href=#1000 id=1000 data-nosnippet>1000</a> <span class="macro">assert!</span>(
-<a href=#1001 id=1001 data-nosnippet>1001</a> result.is_ok(),
-<a href=#1002 id=1002 data-nosnippet>1002</a> <span class="string">"Failed to build equality delete predicate: {:?}"</span>,
-<a href=#1003 id=1003 data-nosnippet>1003</a> result.err()
-<a href=#1004 id=1004 data-nosnippet>1004</a> );
-<a href=#1005 id=1005 data-nosnippet>1005</a> }
-<a href=#1006 id=1006 data-nosnippet>1006</a>
-<a href=#1007 id=1007 data-nosnippet>1007</a> <span class="attr">#[tokio::test]
-<a href=#1008 id=1008 data-nosnippet>1008</a> </span><span class="kw">async fn </span>test_large_equality_delete_batch_stack_overflow() {
-<a href=#1009 id=1009 data-nosnippet>1009</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#1010 id=1010 data-nosnippet>1010</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
-<a href=#1011 id=1011 data-nosnippet>1011</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<a href=#972 id=972 data-nosnippet>972</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<a href=#973 id=973 data-nosnippet>973</a> <span class="kw">let </span>basic_delete_file_loader =
+<a href=#974 id=974 data-nosnippet>974</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
+<a href=#975 id=975 data-nosnippet>975</a>
+<a href=#976 id=976 data-nosnippet>976</a> <span class="kw">let </span>batch_stream = basic_delete_file_loader
+<a href=#977 id=977 data-nosnippet>977</a> .parquet_to_batch_stream(
+<a href=#978 id=978 data-nosnippet>978</a> <span class="kw-2">&</span>delete_file_path,
+<a href=#979 id=979 data-nosnippet>979</a> std::fs::metadata(<span class="kw-2">&</span>delete_file_path).unwrap().len(),
+<a href=#980 id=980 data-nosnippet>980</a> )
+<a href=#981 id=981 data-nosnippet>981</a> .<span class="kw">await
+<a href=#982 id=982 data-nosnippet>982</a> </span>.unwrap();
+<a href=#983 id=983 data-nosnippet>983</a>
+<a href=#984 id=984 data-nosnippet>984</a> <span class="comment">// Only evolve the equality_ids columns (field 2), not all table columns
+<a href=#985 id=985 data-nosnippet>985</a> </span><span class="kw">let </span>equality_ids = <span class="macro">vec!</span>[<span class="number">2</span>];
+<a href=#986 id=986 data-nosnippet>986</a> <span class="kw">let </span>evolved_stream =
+<a href=#987 id=987 data-nosnippet>987</a> BasicDeleteFileLoader::evolve_schema(batch_stream, table_schema, <span class="kw-2">&</span>equality_ids)
+<a href=#988 id=988 data-nosnippet>988</a> .<span class="kw">await
+<a href=#989 id=989 data-nosnippet>989</a> </span>.unwrap();
+<a href=#990 id=990 data-nosnippet>990</a>
+<a href=#991 id=991 data-nosnippet>991</a> <span class="kw">let </span>result = evolved_stream.try_collect::<Vec<<span class="kw">_</span>>>().<span class="kw">await</span>;
+<a href=#992 id=992 data-nosnippet>992</a>
+<a href=#993 id=993 data-nosnippet>993</a> <span class="macro">assert!</span>(
+<a href=#994 id=994 data-nosnippet>994</a> result.is_ok(),
+<a href=#995 id=995 data-nosnippet>995</a> <span class="string">"Expected success when evolving only equality_ids columns, got error: {:?}"</span>,
+<a href=#996 id=996 data-nosnippet>996</a> result.err()
+<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">let </span>batches = result.unwrap();
+<a href=#1000 id=1000 data-nosnippet>1000</a> <span class="macro">assert_eq!</span>(batches.len(), <span class="number">1</span>);
+<a href=#1001 id=1001 data-nosnippet>1001</a>
+<a href=#1002 id=1002 data-nosnippet>1002</a> <span class="kw">let </span>batch = <span class="kw-2">&</span>batches[<span class="number">0</span>];
+<a href=#1003 id=1003 data-nosnippet>1003</a> <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">3</span>);
+<a href=#1004 id=1004 data-nosnippet>1004</a> <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">1</span>); <span class="comment">// Only 'data' column
+<a href=#1005 id=1005 data-nosnippet>1005</a>
+<a href=#1006 id=1006 data-nosnippet>1006</a> // Verify the actual values are preserved after schema evolution
+<a href=#1007 id=1007 data-nosnippet>1007</a> </span><span class="kw">let </span>data_col = batch.column(<span class="number">0</span>).as_string::<i32>();
+<a href=#1008 id=1008 data-nosnippet>1008</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">0</span>), <span class="string">"a"</span>);
+<a href=#1009 id=1009 data-nosnippet>1009</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">1</span>), <span class="string">"d"</span>);
+<a href=#1010 id=1010 data-nosnippet>1010</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">2</span>), <span class="string">"g"</span>);
+<a href=#1011 id=1011 data-nosnippet>1011</a> }
<a href=#1012 id=1012 data-nosnippet>1012</a>
-<a href=#1013 id=1013 data-nosnippet>1013</a> <span class="comment">// Create a large batch of equality deletes
-<a href=#1014 id=1014 data-nosnippet>1014</a> </span><span class="kw">let </span>num_rows = <span class="number">20_000</span>;
-<a href=#1015 id=1015 data-nosnippet>1015</a> <span class="kw">let </span>col_y_vals: Vec<i64> = (<span class="number">0</span>..num_rows).collect();
-<a href=#1016 id=1016 data-nosnippet>1016</a> <span class="kw">let </span>col_y = Arc::new(Int64Array::from(col_y_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#1017 id=1017 data-nosnippet>1017</a>
-<a href=#1018 id=1018 data-nosnippet>1018</a> <span class="kw">let </span>schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[
-<a href=#1019 id=1019 data-nosnippet>1019</a> Field::new(<span class="string">"y"</span>, arrow_schema::DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#1020 id=1020 data-nosnippet>1020</a> PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#1021 id=1021 data-nosnippet>1021</a> <span class="string">"2"</span>.to_string(),
-<a href=#1022 id=1022 data-nosnippet>1022</a> )])),
-<a href=#1023 id=1023 data-nosnippet>1023</a> ]));
-<a href=#1024 id=1024 data-nosnippet>1024</a>
-<a href=#1025 id=1025 data-nosnippet>1025</a> <span class="kw">let </span>record_batch = RecordBatch::try_new(schema.clone(), <span class="macro">vec!</span>[col_y]).unwrap();
-<a href=#1026 id=1026 data-nosnippet>1026</a>
-<a href=#1027 id=1027 data-nosnippet>1027</a> <span class="comment">// Write to file
-<a href=#1028 id=1028 data-nosnippet>1028</a> </span><span class="kw">let </span>path = <span class="macro">format!</span>(<span class="string">"{}/large-eq-deletes.parquet"</span>, <span class="kw-2">&</span>table_location);
-<a href=#1029 id=1029 data-nosnippet>1029</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
-<a href=#1030 id=1030 data-nosnippet>1030</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#1031 id=1031 data-nosnippet>1031</a> .set_compression(Compression::SNAPPY)
-<a href=#1032 id=1032 data-nosnippet>1032</a> .build();
-<a href=#1033 id=1033 data-nosnippet>1033</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, schema, <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#1034 id=1034 data-nosnippet>1034</a> writer.write(<span class="kw-2">&</span>record_batch).unwrap();
-<a href=#1035 id=1035 data-nosnippet>1035</a> writer.close().unwrap();
-<a href=#1036 id=1036 data-nosnippet>1036</a>
-<a href=#1037 id=1037 data-nosnippet>1037</a> <span class="kw">let </span>basic_delete_file_loader =
-<a href=#1038 id=1038 data-nosnippet>1038</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
-<a href=#1039 id=1039 data-nosnippet>1039</a> <span class="kw">let </span>record_batch_stream = basic_delete_file_loader
-<a href=#1040 id=1040 data-nosnippet>1040</a> .parquet_to_batch_stream(<span class="kw-2">&</span>path, std::fs::metadata(<span class="kw-2">&</span>path).unwrap().len())
-<a href=#1041 id=1041 data-nosnippet>1041</a> .<span class="kw">await
-<a href=#1042 id=1042 data-nosnippet>1042</a> </span>.expect(<span class="string">"could not get batch stream"</span>);
-<a href=#1043 id=1043 data-nosnippet>1043</a>
-<a href=#1044 id=1044 data-nosnippet>1044</a> <span class="kw">let </span>eq_ids = HashSet::from_iter(<span class="macro">vec!</span>[<span class="number">2</span>]);
-<a href=#1045 id=1045 data-nosnippet>1045</a>
-<a href=#1046 id=1046 data-nosnippet>1046</a> <span class="kw">let </span>result = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
-<a href=#1047 id=1047 data-nosnippet>1047</a> record_batch_stream,
-<a href=#1048 id=1048 data-nosnippet>1048</a> eq_ids,
-<a href=#1049 id=1049 data-nosnippet>1049</a> )
-<a href=#1050 id=1050 data-nosnippet>1050</a> .<span class="kw">await</span>;
+<a href=#1013 id=1013 data-nosnippet>1013</a> <span class="doccomment">/// Test loading a FileScanTask with BOTH positional and equality deletes.
+<a href=#1014 id=1014 data-nosnippet>1014</a> /// Verifies the fix for the inverted condition that caused "Missing predicate for equality delete file" errors.
+<a href=#1015 id=1015 data-nosnippet>1015</a> </span><span class="attr">#[tokio::test]
+<a href=#1016 id=1016 data-nosnippet>1016</a> </span><span class="kw">async fn </span>test_load_deletes_with_mixed_types() {
+<a href=#1017 id=1017 data-nosnippet>1017</a> <span class="kw">use </span><span class="kw">crate</span>::scan::FileScanTask;
+<a href=#1018 id=1018 data-nosnippet>1018</a> <span class="kw">use </span><span class="kw">crate</span>::spec::{DataFileFormat, Schema};
+<a href=#1019 id=1019 data-nosnippet>1019</a>
+<a href=#1020 id=1020 data-nosnippet>1020</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#1021 id=1021 data-nosnippet>1021</a> <span class="kw">let </span>table_location = tmp_dir.path();
+<a href=#1022 id=1022 data-nosnippet>1022</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<a href=#1023 id=1023 data-nosnippet>1023</a>
+<a href=#1024 id=1024 data-nosnippet>1024</a> <span class="comment">// Create the data file schema
+<a href=#1025 id=1025 data-nosnippet>1025</a> </span><span class="kw">let </span>data_file_schema = Arc::new(
+<a href=#1026 id=1026 data-nosnippet>1026</a> Schema::builder()
+<a href=#1027 id=1027 data-nosnippet>1027</a> .with_fields(<span class="macro">vec!</span>[
+<a href=#1028 id=1028 data-nosnippet>1028</a> <span class="kw">crate</span>::spec::NestedField::optional(
+<a href=#1029 id=1029 data-nosnippet>1029</a> <span class="number">2</span>,
+<a href=#1030 id=1030 data-nosnippet>1030</a> <span class="string">"y"</span>,
+<a href=#1031 id=1031 data-nosnippet>1031</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Long),
+<a href=#1032 id=1032 data-nosnippet>1032</a> )
+<a href=#1033 id=1033 data-nosnippet>1033</a> .into(),
+<a href=#1034 id=1034 data-nosnippet>1034</a> <span class="kw">crate</span>::spec::NestedField::optional(
+<a href=#1035 id=1035 data-nosnippet>1035</a> <span class="number">3</span>,
+<a href=#1036 id=1036 data-nosnippet>1036</a> <span class="string">"z"</span>,
+<a href=#1037 id=1037 data-nosnippet>1037</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Long),
+<a href=#1038 id=1038 data-nosnippet>1038</a> )
+<a href=#1039 id=1039 data-nosnippet>1039</a> .into(),
+<a href=#1040 id=1040 data-nosnippet>1040</a> ])
+<a href=#1041 id=1041 data-nosnippet>1041</a> .build()
+<a href=#1042 id=1042 data-nosnippet>1042</a> .unwrap(),
+<a href=#1043 id=1043 data-nosnippet>1043</a> );
+<a href=#1044 id=1044 data-nosnippet>1044</a>
+<a href=#1045 id=1045 data-nosnippet>1045</a> <span class="comment">// Write positional delete file
+<a href=#1046 id=1046 data-nosnippet>1046</a> </span><span class="kw">let </span>positional_delete_schema = <span class="kw">crate</span>::arrow::delete_filter::tests::create_pos_del_schema();
+<a href=#1047 id=1047 data-nosnippet>1047</a> <span class="kw">let </span>file_path_values =
+<a href=#1048 id=1048 data-nosnippet>1048</a> <span class="macro">vec!</span>[<span class="macro">format!</span>(<span class="string">"{}/data-1.parquet"</span>, table_location.to_str().unwrap()); <span class="number">4</span>];
+<a href=#1049 id=1049 data-nosnippet>1049</a> <span class="kw">let </span>file_path_col = Arc::new(StringArray::from_iter_values(<span class="kw-2">&</span>file_path_values));
+<a href=#1050 id=1050 data-nosnippet>1050</a> <span class="kw">let </span>pos_col = Arc::new(Int64Array::from_iter_values(<span class="macro">vec!</span>[<span class="number">0i64</span>, <span class="number">1</span>, <span class="number">2</span>, <span class="number">3</span>]));
<a href=#1051 id=1051 data-nosnippet>1051</a>
-<a href=#1052 id=1052 data-nosnippet>1052</a> <span class="macro">assert!</span>(result.is_ok());
-<a href=#1053 id=1053 data-nosnippet>1053</a> }
-<a href=#1054 id=1054 data-nosnippet>1054</a>
-<a href=#1055 id=1055 data-nosnippet>1055</a> <span class="attr">#[tokio::test]
-<a href=#1056 id=1056 data-nosnippet>1056</a> </span><span class="kw">async fn </span>test_caching_delete_file_loader_caches_results() {
-<a href=#1057 id=1057 data-nosnippet>1057</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#1058 id=1058 data-nosnippet>1058</a> <span class="kw">let </span>table_location = tmp_dir.path();
-<a href=#1059 id=1059 data-nosnippet>1059</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
-<a href=#1060 id=1060 data-nosnippet>1060</a>
-<a href=#1061 id=1061 data-nosnippet>1061</a> <span class="kw">let </span>delete_file_loader =
-<a href=#1062 id=1062 data-nosnippet>1062</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
-<a href=#1063 id=1063 data-nosnippet>1063</a>
-<a href=#1064 id=1064 data-nosnippet>1064</a> <span class="kw">let </span>file_scan_tasks = setup(table_location);
-<a href=#1065 id=1065 data-nosnippet>1065</a>
-<a href=#1066 id=1066 data-nosnippet>1066</a> <span class="comment">// Load deletes for the first time
-<a href=#1067 id=1067 data-nosnippet>1067</a> </span><span class="kw">let </span>delete_filter_1 = delete_file_loader
-<a href=#1068 id=1068 data-nosnippet>1068</a> .load_deletes(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>].deletes, file_scan_tasks[<span class="number">0</span>].schema_ref())
-<a href=#1069 id=1069 data-nosnippet>1069</a> .<span class="kw">await
-<a href=#1070 id=1070 data-nosnippet>1070</a> </span>.unwrap()
-<a href=#1071 id=1071 data-nosnippet>1071</a> .unwrap();
-<a href=#1072 id=1072 data-nosnippet>1072</a>
-<a href=#1073 id=1073 data-nosnippet>1073</a> <span class="comment">// Load deletes for the second time (same task/files)
-<a href=#1074 id=1074 data-nosnippet>1074</a> </span><span class="kw">let </span>delete_filter_2 = delete_file_loader
-<a href=#1075 id=1075 data-nosnippet>1075</a> .load_deletes(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>].deletes, file_scan_tasks[<span class="number">0</span>].schema_ref())
-<a href=#1076 id=1076 data-nosnippet>1076</a> .<span class="kw">await
-<a href=#1077 id=1077 data-nosnippet>1077</a> </span>.unwrap()
-<a href=#1078 id=1078 data-nosnippet>1078</a> .unwrap();
-<a href=#1079 id=1079 data-nosnippet>1079</a>
-<a href=#1080 id=1080 data-nosnippet>1080</a> <span class="kw">let </span>dv1 = delete_filter_1
-<a href=#1081 id=1081 data-nosnippet>1081</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
-<a href=#1082 id=1082 data-nosnippet>1082</a> .unwrap();
-<a href=#1083 id=1083 data-nosnippet>1083</a> <span class="kw">let </span>dv2 = delete_filter_2
-<a href=#1084 id=1084 data-nosnippet>1084</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
-<a href=#1085 id=1085 data-nosnippet>1085</a> .unwrap();
-<a href=#1086 id=1086 data-nosnippet>1086</a>
-<a href=#1087 id=1087 data-nosnippet>1087</a> <span class="comment">// Verify that the delete vectors point to the same memory location,
-<a href=#1088 id=1088 data-nosnippet>1088</a> // confirming that the second load reused the result from the first.
-<a href=#1089 id=1089 data-nosnippet>1089</a> </span><span class="macro">assert!</span>(Arc::ptr_eq(<span class="kw-2">&</span>dv1, <span class="kw-2">&</span>dv2));
-<a href=#1090 id=1090 data-nosnippet>1090</a> }
-<a href=#1091 id=1091 data-nosnippet>1091</a>}
+<a href=#1052 id=1052 data-nosnippet>1052</a> <span class="kw">let </span>positional_deletes_to_write =
+<a href=#1053 id=1053 data-nosnippet>1053</a> RecordBatch::try_new(positional_delete_schema.clone(), <span class="macro">vec!</span>[
+<a href=#1054 id=1054 data-nosnippet>1054</a> file_path_col,
+<a href=#1055 id=1055 data-nosnippet>1055</a> pos_col,
+<a href=#1056 id=1056 data-nosnippet>1056</a> ])
+<a href=#1057 id=1057 data-nosnippet>1057</a> .unwrap();
+<a href=#1058 id=1058 data-nosnippet>1058</a>
+<a href=#1059 id=1059 data-nosnippet>1059</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#1060 id=1060 data-nosnippet>1060</a> .set_compression(Compression::SNAPPY)
+<a href=#1061 id=1061 data-nosnippet>1061</a> .build();
+<a href=#1062 id=1062 data-nosnippet>1062</a>
+<a href=#1063 id=1063 data-nosnippet>1063</a> <span class="kw">let </span>pos_del_path = <span class="macro">format!</span>(<span class="string">"{}/pos-del-mixed.parquet"</span>, table_location.to_str().unwrap());
+<a href=#1064 id=1064 data-nosnippet>1064</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>pos_del_path).unwrap();
+<a href=#1065 id=1065 data-nosnippet>1065</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(
+<a href=#1066 id=1066 data-nosnippet>1066</a> file,
+<a href=#1067 id=1067 data-nosnippet>1067</a> positional_deletes_to_write.schema(),
+<a href=#1068 id=1068 data-nosnippet>1068</a> <span class="prelude-val">Some</span>(props.clone()),
+<a href=#1069 id=1069 data-nosnippet>1069</a> )
+<a href=#1070 id=1070 data-nosnippet>1070</a> .unwrap();
+<a href=#1071 id=1071 data-nosnippet>1071</a> writer.write(<span class="kw-2">&</span>positional_deletes_to_write).unwrap();
+<a href=#1072 id=1072 data-nosnippet>1072</a> writer.close().unwrap();
+<a href=#1073 id=1073 data-nosnippet>1073</a>
+<a href=#1074 id=1074 data-nosnippet>1074</a> <span class="comment">// Write equality delete file
+<a href=#1075 id=1075 data-nosnippet>1075</a> </span><span class="kw">let </span>eq_delete_path = setup_write_equality_delete_file_1(table_location.to_str().unwrap());
+<a href=#1076 id=1076 data-nosnippet>1076</a>
+<a href=#1077 id=1077 data-nosnippet>1077</a> <span class="comment">// Create FileScanTask with BOTH positional and equality deletes
+<a href=#1078 id=1078 data-nosnippet>1078</a> </span><span class="kw">let </span>pos_del = FileScanTaskDeleteFile::builder()
+<a href=#1079 id=1079 data-nosnippet>1079</a> .with_file_path(pos_del_path.clone())
+<a href=#1080 id=1080 data-nosnippet>1080</a> .with_file_size_in_bytes(std::fs::metadata(<span class="kw-2">&</span>pos_del_path).unwrap().len())
+<a href=#1081 id=1081 data-nosnippet>1081</a> .with_file_type(DataContentType::PositionDeletes)
+<a href=#1082 id=1082 data-nosnippet>1082</a> .with_partition_spec_id(<span class="number">0</span>)
+<a href=#1083 id=1083 data-nosnippet>1083</a> .build();
+<a href=#1084 id=1084 data-nosnippet>1084</a>
+<a href=#1085 id=1085 data-nosnippet>1085</a> <span class="kw">let </span>eq_del = FileScanTaskDeleteFile::builder()
+<a href=#1086 id=1086 data-nosnippet>1086</a> .with_file_path(eq_delete_path.clone())
+<a href=#1087 id=1087 data-nosnippet>1087</a> .with_file_size_in_bytes(std::fs::metadata(<span class="kw-2">&</span>eq_delete_path).unwrap().len())
+<a href=#1088 id=1088 data-nosnippet>1088</a> .with_file_type(DataContentType::EqualityDeletes)
+<a href=#1089 id=1089 data-nosnippet>1089</a> .with_partition_spec_id(<span class="number">0</span>)
+<a href=#1090 id=1090 data-nosnippet>1090</a> .with_equality_ids(<span class="prelude-val">Some</span>(<span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">3</span>])) <span class="comment">// Only use field IDs that exist in both schemas
+<a href=#1091 id=1091 data-nosnippet>1091</a> </span>.build();
+<a href=#1092 id=1092 data-nosnippet>1092</a>
+<a href=#1093 id=1093 data-nosnippet>1093</a> <span class="kw">let </span>file_scan_task = FileScanTask::builder()
+<a href=#1094 id=1094 data-nosnippet>1094</a> .with_file_size_in_bytes(<span class="number">0</span>)
+<a href=#1095 id=1095 data-nosnippet>1095</a> .with_start(<span class="number">0</span>)
+<a href=#1096 id=1096 data-nosnippet>1096</a> .with_length(<span class="number">0</span>)
+<a href=#1097 id=1097 data-nosnippet>1097</a> .with_data_file_path(<span class="macro">format!</span>(
+<a href=#1098 id=1098 data-nosnippet>1098</a> <span class="string">"{}/data-1.parquet"</span>,
+<a href=#1099 id=1099 data-nosnippet>1099</a> table_location.to_str().unwrap()
+<a href=#1100 id=1100 data-nosnippet>1100</a> ))
+<a href=#1101 id=1101 data-nosnippet>1101</a> .with_data_file_format(DataFileFormat::Parquet)
+<a href=#1102 id=1102 data-nosnippet>1102</a> .with_schema(data_file_schema.clone())
+<a href=#1103 id=1103 data-nosnippet>1103</a> .with_project_field_ids(<span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">3</span>])
+<a href=#1104 id=1104 data-nosnippet>1104</a> .with_deletes(<span class="macro">vec!</span>[pos_del, eq_del])
+<a href=#1105 id=1105 data-nosnippet>1105</a> .with_case_sensitive(<span class="bool-val">false</span>)
+<a href=#1106 id=1106 data-nosnippet>1106</a> .build();
+<a href=#1107 id=1107 data-nosnippet>1107</a>
+<a href=#1108 id=1108 data-nosnippet>1108</a> <span class="comment">// Load the deletes - should handle both types without error
+<a href=#1109 id=1109 data-nosnippet>1109</a> </span><span class="kw">let </span>delete_file_loader =
+<a href=#1110 id=1110 data-nosnippet>1110</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
+<a href=#1111 id=1111 data-nosnippet>1111</a> <span class="kw">let </span>delete_filter = delete_file_loader
+<a href=#1112 id=1112 data-nosnippet>1112</a> .load_deletes(<span class="kw-2">&</span>file_scan_task.deletes, file_scan_task.schema_ref())
+<a href=#1113 id=1113 data-nosnippet>1113</a> .<span class="kw">await
+<a href=#1114 id=1114 data-nosnippet>1114</a> </span>.unwrap()
+<a href=#1115 id=1115 data-nosnippet>1115</a> .unwrap();
+<a href=#1116 id=1116 data-nosnippet>1116</a>
+<a href=#1117 id=1117 data-nosnippet>1117</a> <span class="comment">// Verify both delete types can be processed together
+<a href=#1118 id=1118 data-nosnippet>1118</a> </span><span class="kw">let </span>result = delete_filter
+<a href=#1119 id=1119 data-nosnippet>1119</a> .build_equality_delete_predicate(<span class="kw-2">&</span>file_scan_task)
+<a href=#1120 id=1120 data-nosnippet>1120</a> .<span class="kw">await</span>;
+<a href=#1121 id=1121 data-nosnippet>1121</a> <span class="macro">assert!</span>(
+<a href=#1122 id=1122 data-nosnippet>1122</a> result.is_ok(),
+<a href=#1123 id=1123 data-nosnippet>1123</a> <span class="string">"Failed to build equality delete predicate: {:?}"</span>,
+<a href=#1124 id=1124 data-nosnippet>1124</a> result.err()
+<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>
+<a href=#1128 id=1128 data-nosnippet>1128</a> <span class="attr">#[tokio::test]
+<a href=#1129 id=1129 data-nosnippet>1129</a> </span><span class="kw">async fn </span>test_large_equality_delete_batch_stack_overflow() {
+<a href=#1130 id=1130 data-nosnippet>1130</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#1131 id=1131 data-nosnippet>1131</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
+<a href=#1132 id=1132 data-nosnippet>1132</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<a href=#1133 id=1133 data-nosnippet>1133</a>
+<a href=#1134 id=1134 data-nosnippet>1134</a> <span class="comment">// Create a large batch of equality deletes
+<a href=#1135 id=1135 data-nosnippet>1135</a> </span><span class="kw">let </span>num_rows = <span class="number">20_000</span>;
+<a href=#1136 id=1136 data-nosnippet>1136</a> <span class="kw">let </span>col_y_vals: Vec<i64> = (<span class="number">0</span>..num_rows).collect();
+<a href=#1137 id=1137 data-nosnippet>1137</a> <span class="kw">let </span>col_y = Arc::new(Int64Array::from(col_y_vals)) <span class="kw">as </span>ArrayRef;
+<a href=#1138 id=1138 data-nosnippet>1138</a>
+<a href=#1139 id=1139 data-nosnippet>1139</a> <span class="kw">let </span>schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[
+<a href=#1140 id=1140 data-nosnippet>1140</a> Field::new(<span class="string">"y"</span>, arrow_schema::DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
+<a href=#1141 id=1141 data-nosnippet>1141</a> PARQUET_FIELD_ID_META_KEY.to_string(),
+<a href=#1142 id=1142 data-nosnippet>1142</a> <span class="string">"2"</span>.to_string(),
+<a href=#1143 id=1143 data-nosnippet>1143</a> )])),
+<a href=#1144 id=1144 data-nosnippet>1144</a> ]));
+<a href=#1145 id=1145 data-nosnippet>1145</a>
+<a href=#1146 id=1146 data-nosnippet>1146</a> <span class="kw">let </span>record_batch = RecordBatch::try_new(schema.clone(), <span class="macro">vec!</span>[col_y]).unwrap();
+<a href=#1147 id=1147 data-nosnippet>1147</a>
+<a href=#1148 id=1148 data-nosnippet>1148</a> <span class="comment">// Write to file
+<a href=#1149 id=1149 data-nosnippet>1149</a> </span><span class="kw">let </span>path = <span class="macro">format!</span>(<span class="string">"{}/large-eq-deletes.parquet"</span>, <span class="kw-2">&</span>table_location);
+<a href=#1150 id=1150 data-nosnippet>1150</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
+<a href=#1151 id=1151 data-nosnippet>1151</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#1152 id=1152 data-nosnippet>1152</a> .set_compression(Compression::SNAPPY)
+<a href=#1153 id=1153 data-nosnippet>1153</a> .build();
+<a href=#1154 id=1154 data-nosnippet>1154</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, schema, <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#1155 id=1155 data-nosnippet>1155</a> writer.write(<span class="kw-2">&</span>record_batch).unwrap();
+<a href=#1156 id=1156 data-nosnippet>1156</a> writer.close().unwrap();
+<a href=#1157 id=1157 data-nosnippet>1157</a>
+<a href=#1158 id=1158 data-nosnippet>1158</a> <span class="kw">let </span>basic_delete_file_loader =
+<a href=#1159 id=1159 data-nosnippet>1159</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
+<a href=#1160 id=1160 data-nosnippet>1160</a> <span class="kw">let </span>record_batch_stream = basic_delete_file_loader
+<a href=#1161 id=1161 data-nosnippet>1161</a> .parquet_to_batch_stream(<span class="kw-2">&</span>path, std::fs::metadata(<span class="kw-2">&</span>path).unwrap().len())
+<a href=#1162 id=1162 data-nosnippet>1162</a> .<span class="kw">await
+<a href=#1163 id=1163 data-nosnippet>1163</a> </span>.expect(<span class="string">"could not get batch stream"</span>);
+<a href=#1164 id=1164 data-nosnippet>1164</a>
+<a href=#1165 id=1165 data-nosnippet>1165</a> <span class="kw">let </span>eq_ids = HashSet::from_iter(<span class="macro">vec!</span>[<span class="number">2</span>]);
+<a href=#1166 id=1166 data-nosnippet>1166</a>
+<a href=#1167 id=1167 data-nosnippet>1167</a> <span class="kw">let </span>result = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
+<a href=#1168 id=1168 data-nosnippet>1168</a> record_batch_stream,
+<a href=#1169 id=1169 data-nosnippet>1169</a> eq_ids,
+<a href=#1170 id=1170 data-nosnippet>1170</a> )
+<a href=#1171 id=1171 data-nosnippet>1171</a> .<span class="kw">await</span>;
+<a href=#1172 id=1172 data-nosnippet>1172</a>
+<a href=#1173 id=1173 data-nosnippet>1173</a> <span class="macro">assert!</span>(result.is_ok());
+<a href=#1174 id=1174 data-nosnippet>1174</a> }
+<a href=#1175 id=1175 data-nosnippet>1175</a>
+<a href=#1176 id=1176 data-nosnippet>1176</a> <span class="attr">#[tokio::test]
+<a href=#1177 id=1177 data-nosnippet>1177</a> </span><span class="kw">async fn </span>test_caching_delete_file_loader_caches_results() {
+<a href=#1178 id=1178 data-nosnippet>1178</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
+<a href=#1179 id=1179 data-nosnippet>1179</a> <span class="kw">let </span>table_location = tmp_dir.path();
+<a href=#1180 id=1180 data-nosnippet>1180</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<a href=#1181 id=1181 data-nosnippet>1181</a>
+<a href=#1182 id=1182 data-nosnippet>1182</a> <span class="kw">let </span>delete_file_loader =
+<a href=#1183 id=1183 data-nosnippet>1183</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
+<a href=#1184 id=1184 data-nosnippet>1184</a>
+<a href=#1185 id=1185 data-nosnippet>1185</a> <span class="kw">let </span>file_scan_tasks = setup(table_location);
+<a href=#1186 id=1186 data-nosnippet>1186</a>
+<a href=#1187 id=1187 data-nosnippet>1187</a> <span class="comment">// Load deletes for the first time
+<a href=#1188 id=1188 data-nosnippet>1188</a> </span><span class="kw">let </span>delete_filter_1 = delete_file_loader
+<a href=#1189 id=1189 data-nosnippet>1189</a> .load_deletes(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>].deletes, file_scan_tasks[<span class="number">0</span>].schema_ref())
+<a href=#1190 id=1190 data-nosnippet>1190</a> .<span class="kw">await
+<a href=#1191 id=1191 data-nosnippet>1191</a> </span>.unwrap()
+<a href=#1192 id=1192 data-nosnippet>1192</a> .unwrap();
+<a href=#1193 id=1193 data-nosnippet>1193</a>
+<a href=#1194 id=1194 data-nosnippet>1194</a> <span class="comment">// Load deletes for the second time (same task/files)
+<a href=#1195 id=1195 data-nosnippet>1195</a> </span><span class="kw">let </span>delete_filter_2 = delete_file_loader
+<a href=#1196 id=1196 data-nosnippet>1196</a> .load_deletes(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>].deletes, file_scan_tasks[<span class="number">0</span>].schema_ref())
+<a href=#1197 id=1197 data-nosnippet>1197</a> .<span class="kw">await
+<a href=#1198 id=1198 data-nosnippet>1198</a> </span>.unwrap()
+<a href=#1199 id=1199 data-nosnippet>1199</a> .unwrap();
+<a href=#1200 id=1200 data-nosnippet>1200</a>
+<a href=#1201 id=1201 data-nosnippet>1201</a> <span class="kw">let </span>dv1 = delete_filter_1
+<a href=#1202 id=1202 data-nosnippet>1202</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
+<a href=#1203 id=1203 data-nosnippet>1203</a> .unwrap();
+<a href=#1204 id=1204 data-nosnippet>1204</a> <span class="kw">let </span>dv2 = delete_filter_2
+<a href=#1205 id=1205 data-nosnippet>1205</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
+<a href=#1206 id=1206 data-nosnippet>1206</a> .unwrap();
+<a href=#1207 id=1207 data-nosnippet>1207</a>
+<a href=#1208 id=1208 data-nosnippet>1208</a> <span class="comment">// Verify that the delete vectors point to the same memory location,
+<a href=#1209 id=1209 data-nosnippet>1209</a> // confirming that the second load reused the result from the first.
+<a href=#1210 id=1210 data-nosnippet>1210</a> </span><span class="macro">assert!</span>(Arc::ptr_eq(<span class="kw-2">&</span>dv1, <span class="kw-2">&</span>dv2));
+<a href=#1211 id=1211 data-nosnippet>1211</a> }
+<a href=#1212 id=1212 data-nosnippet>1212</a>}
</code></pre></div></section></main></body></html>
\ No newline at end of file
diff --git a/api/trait.impl/iceberg/catalog/trait.Catalog.js b/api/trait.impl/iceberg/catalog/trait.Catalog.js
index 7265f76e4..0b6c0381e 100644
--- a/api/trait.impl/iceberg/catalog/trait.Catalog.js
+++ b/api/trait.impl/iceberg/catalog/trait.Catalog.js
@@ -1,9 +1,9 @@
(function() {
- const implementors = Object.fromEntries([["iceberg",[]],["iceberg_catalog_glue",[["impl <a class=\"trait\" href=\"iceberg/catalog/trait.Catalog.html\" title=\"trait iceberg::catalog::Catalog\">Catalog</a> for <a class=\"struct\" href=\"iceberg_catalog_glue/struct.GlueCatalog.html\" title=\"struct iceberg_catalog_glue::GlueCatalog\">GlueCatalog</a>",0]]],["iceberg_catalog_hms",[["impl Catalog for <a class=\"struct\" href=\"iceberg_catalog_hms/struct.HmsCatalog.html\" title=\"struct ic [...]
+ const implementors = Object.fromEntries([["iceberg",[]],["iceberg_catalog_glue",[["impl <a class=\"trait\" href=\"iceberg/catalog/trait.Catalog.html\" title=\"trait iceberg::catalog::Catalog\">Catalog</a> for <a class=\"struct\" href=\"iceberg_catalog_glue/struct.GlueCatalog.html\" title=\"struct iceberg_catalog_glue::GlueCatalog\">GlueCatalog</a>",0]]],["iceberg_catalog_hms",[["impl <a class=\"trait\" href=\"iceberg/catalog/trait.Catalog.html\" title=\"trait iceberg::catalog::Catalo [...]
if (window.register_implementors) {
window.register_implementors(implementors);
} else {
window.pending_implementors = implementors;
}
})()
-//{"start":59,"fragment_lengths":[14,300,185,191,215,185]}
\ No newline at end of file
+//{"start":59,"fragment_lengths":[14,300,294,191,215,185]}
\ No newline at end of file
diff --git a/api/trait.impl/iceberg/catalog/trait.CatalogBuilder.js b/api/trait.impl/iceberg/catalog/trait.CatalogBuilder.js
index 40a6eeb28..7cec58fdc 100644
--- a/api/trait.impl/iceberg/catalog/trait.CatalogBuilder.js
+++ b/api/trait.impl/iceberg/catalog/trait.CatalogBuilder.js
@@ -1,9 +1,9 @@
(function() {
- const implementors = Object.fromEntries([["iceberg",[]],["iceberg_catalog_glue",[["impl <a class=\"trait\" href=\"iceberg/catalog/trait.CatalogBuilder.html\" title=\"trait iceberg::catalog::CatalogBuilder\">CatalogBuilder</a> for <a class=\"struct\" href=\"iceberg_catalog_glue/struct.GlueCatalogBuilder.html\" title=\"struct iceberg_catalog_glue::GlueCatalogBuilder\">GlueCatalogBuilder</a>",0]]],["iceberg_catalog_hms",[["impl CatalogBuilder for <a class=\"struct\" href=\"iceberg_catal [...]
+ const implementors = Object.fromEntries([["iceberg",[]],["iceberg_catalog_glue",[["impl <a class=\"trait\" href=\"iceberg/catalog/trait.CatalogBuilder.html\" title=\"trait iceberg::catalog::CatalogBuilder\">CatalogBuilder</a> for <a class=\"struct\" href=\"iceberg_catalog_glue/struct.GlueCatalogBuilder.html\" title=\"struct iceberg_catalog_glue::GlueCatalogBuilder\">GlueCatalogBuilder</a>",0]]],["iceberg_catalog_hms",[["impl <a class=\"trait\" href=\"iceberg/catalog/trait.CatalogBuil [...]
if (window.register_implementors) {
window.register_implementors(implementors);
} else {
window.pending_implementors = implementors;
}
})()
-//{"start":59,"fragment_lengths":[14,342,213,219,243,213]}
\ No newline at end of file
+//{"start":59,"fragment_lengths":[14,342,336,219,243,213]}
\ No newline at end of file