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 2026/07/07 05:12:13 UTC
(iceberg-rust) branch gh-pages updated: deploy: f12772deed34b4181605d79b6dc20fb487684ae8
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 e79414939 deploy: f12772deed34b4181605d79b6dc20fb487684ae8
e79414939 is described below
commit e7941493962063da78267549c80db733d6aa1e21
Author: kevinjqliu <ke...@users.noreply.github.com>
AuthorDate: Tue Jul 7 05:12:08 2026 +0000
deploy: f12772deed34b4181605d79b6dc20fb487684ae8
---
.../path/{1525f4ce3f7d.js => 748e9d70caa8.js} | 2 +-
api/search.index/root.js | 2 +-
.../arrow/caching_delete_file_loader.rs.html | 1398 ++++++++++----------
3 files changed, 712 insertions(+), 690 deletions(-)
diff --git a/api/search.index/path/1525f4ce3f7d.js b/api/search.index/path/748e9d70caa8.js
similarity index 65%
rename from api/search.index/path/1525f4ce3f7d.js
rename to api/search.index/path/748e9d70caa8.js
index 16eca02bb..7a3b6eebd 100644
--- a/api/search.index/path/1525f4ce3f7d.js
+++ b/api/search.index/path/748e9d70caa8.js
@@ -1 +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
+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 2cba1d274..91ad4f95f 100644
--- a/api/search.index/root.js
+++ b/api/search.index/root.js
@@ -1 +1 @@
-rr_('{"normalizedName":{"I":"RQBAAAAEkRJlcQGrBqMLBg4qDhUAQAAAIFQDMQSQBDSOBAgOFQBAAAAACA42TBUOIDFAAACDAVQDMQSQBAUBwAAAAq8aBbMcYWUxQAAAqwajCwYOKg5jAASgEAAADgiwcB7QAH41OGRt9iAAAABTAeAB3QBfAHgJVQJAAAAxqwajCwYOKg4AjgQBBw4yNjkeAB8AcgG3D7gdvx8FAcAAAAOpHQCdGmFzAQEBoGAAABoTZDswAAABAADaAAMAmgQAALAEAACuCNgABQHDAACXjB+S5B5kczUAQgAABfMYY4sJjAmNCb0JBABCdGQAAAZlZmF1bHRwGmSzBFYLAMHIoAAAAAGDIh5PPAnKoFAAAB4NsCAPrwrvAAAAAQhOHAAA0hUZAACxBQBSAAAGGyBkBSAFAFQAAAYcIGQJIPODAnB0NQBEAACWECBkah9rH+kf6h9bA6BgAAAbUWF [...]
\ No newline at end of file
+rr_('{"normalizedName":{"I":"RQBAAAAEkRJlcQGrBqMLBg4qDhUAQAAAIFQDMQSQBDSOBAgOFQBAAAAACA42TBUOIDFAAACDAVQDMQSQBAUBwAAAAq8aBbMcYWUxQAAAqwajCwYOKg5jAASgEAAADgiwcB7QAH41OGRt9iAAAABTAeAB3QBfAHgJVQJAAAAxqwajCwYOKg4AjgQBBw4yNjkeAB8AcgG3D7gdvx8FAcAAAAOpHQCdGmFzAQEBoGAAABoTZDswAAABAADaAAMAmgQAALAEAACuCNgABQHDAACXjB+S5B5kczUAQgAABfMYY4sJjAmNCb0JBABCdGQAAAZlZmF1bHRwGmSzBFYLAMHIoAAAAAGDIh5PPAnKoFAAAB4NsCAPrwrvAAAAAQhOHAAA0hUZAACxBQBSAAAGGyBkBSAFAFQAAAYcIGQJIPODAnB0NQBEAACWECBkah9rH+kf6h9bA6BgAAAbUWF [...]
\ 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 9c48ab484..54b1112f0 100644
--- a/api/src/iceberg/arrow/caching_delete_file_loader.rs.html
+++ b/api/src/iceberg/arrow/caching_delete_file_loader.rs.html
@@ -360,711 +360,733 @@
<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>
-<a href=#364 id=364 data-nosnippet>364</a> result
-<a href=#365 id=365 data-nosnippet>365</a> .entry(file_path.to_string())
-<a href=#366 id=366 data-nosnippet>366</a> .or_default()
-<a href=#367 id=367 data-nosnippet>367</a> .insert(pos <span class="kw">as </span>u64);
-<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>
-<a href=#371 id=371 data-nosnippet>371</a> <span class="prelude-val">Ok</span>(result)
-<a href=#372 id=372 data-nosnippet>372</a> }
-<a href=#373 id=373 data-nosnippet>373</a>
-<a href=#374 id=374 data-nosnippet>374</a> <span class="kw">async fn </span>parse_equality_deletes_record_batch_stream(
-<a href=#375 id=375 data-nosnippet>375</a> <span class="kw-2">mut </span>stream: ArrowRecordBatchStream,
-<a href=#376 id=376 data-nosnippet>376</a> equality_ids: HashSet<i32>,
-<a href=#377 id=377 data-nosnippet>377</a> ) -> <span class="prelude-ty">Result</span><Predicate> {
-<a href=#378 id=378 data-nosnippet>378</a> <span class="kw">let </span><span class="kw-2">mut </span>row_predicates = Vec::new();
-<a href=#379 id=379 data-nosnippet>379</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=#380 id=380 data-nosnippet>380</a> <span class="kw">let </span>accessor = EqDelRecordBatchPartnerAccessor;
-<a href=#381 id=381 data-nosnippet>381</a>
-<a href=#382 id=382 data-nosnippet>382</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(record_batch) = stream.next().<span class="kw">await </span>{
-<a href=#383 id=383 data-nosnippet>383</a> <span class="kw">let </span>record_batch = record_batch<span class="question-mark">?</span>;
-<a href=#384 id=384 data-nosnippet>384</a>
-<a href=#385 id=385 data-nosnippet>385</a> <span class="kw">if </span>record_batch.num_columns() == <span class="number">0 </span>{
-<a href=#386 id=386 data-nosnippet>386</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(AlwaysTrue);
-<a href=#387 id=387 data-nosnippet>387</a> }
-<a href=#388 id=388 data-nosnippet>388</a>
-<a href=#389 id=389 data-nosnippet>389</a> <span class="kw">let </span>schema = <span class="kw">match </span><span class="kw-2">&</span>batch_schema_iceberg {
-<a href=#390 id=390 data-nosnippet>390</a> <span class="prelude-val">Some</span>(schema) => schema,
-<a href=#391 id=391 data-nosnippet>391</a> <span class="prelude-val">None </span>=> {
-<a href=#392 id=392 data-nosnippet>392</a> <span class="kw">let </span>schema = arrow_schema_to_schema(record_batch.schema().as_ref())<span class="question-mark">?</span>;
-<a href=#393 id=393 data-nosnippet>393</a> batch_schema_iceberg = <span class="prelude-val">Some</span>(schema);
-<a href=#394 id=394 data-nosnippet>394</a> batch_schema_iceberg.as_ref().unwrap()
-<a href=#395 id=395 data-nosnippet>395</a> }
-<a href=#396 id=396 data-nosnippet>396</a> };
-<a href=#397 id=397 data-nosnippet>397</a>
-<a href=#398 id=398 data-nosnippet>398</a> <span class="kw">let </span>root_array: ArrayRef = Arc::new(StructArray::from(record_batch));
-<a href=#399 id=399 data-nosnippet>399</a>
-<a href=#400 id=400 data-nosnippet>400</a> <span class="kw">let </span><span class="kw-2">mut </span>processor = EqDelColumnProcessor::new(<span class="kw-2">&</span>equality_ids);
-<a href=#401 id=401 data-nosnippet>401</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=#402 id=402 data-nosnippet>402</a>
-<a href=#403 id=403 data-nosnippet>403</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=#404 id=404 data-nosnippet>404</a> <span class="kw">if </span>datum_columns_with_names.is_empty() {
-<a href=#405 id=405 data-nosnippet>405</a> <span class="kw">continue</span>;
-<a href=#406 id=406 data-nosnippet>406</a> }
-<a href=#407 id=407 data-nosnippet>407</a>
-<a href=#408 id=408 data-nosnippet>408</a> <span class="comment">// Process the collected columns in lockstep
-<a href=#409 id=409 data-nosnippet>409</a> </span><span class="attr">#[allow(clippy::len_zero)]
-<a href=#410 id=410 data-nosnippet>410</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=#411 id=411 data-nosnippet>411</a> <span class="kw">let </span><span class="kw-2">mut </span>row_predicate = AlwaysTrue;
-<a href=#412 id=412 data-nosnippet>412</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=#413 id=413 data-nosnippet>413</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(item) = column.next() {
-<a href=#414 id=414 data-nosnippet>414</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=#415 id=415 data-nosnippet>415</a> Reference::new(field_name.clone()).equal_to(datum.clone())
-<a href=#416 id=416 data-nosnippet>416</a> } <span class="kw">else </span>{
-<a href=#417 id=417 data-nosnippet>417</a> Reference::new(field_name.clone()).is_null()
-<a href=#418 id=418 data-nosnippet>418</a> };
-<a href=#419 id=419 data-nosnippet>419</a> row_predicate = row_predicate.and(cell_predicate)
-<a href=#420 id=420 data-nosnippet>420</a> }
-<a href=#421 id=421 data-nosnippet>421</a> }
-<a href=#422 id=422 data-nosnippet>422</a> row_predicates.push(row_predicate.not().rewrite_not());
-<a href=#423 id=423 data-nosnippet>423</a> }
-<a href=#424 id=424 data-nosnippet>424</a> }
-<a href=#425 id=425 data-nosnippet>425</a>
-<a href=#426 id=426 data-nosnippet>426</a> <span class="comment">// All row predicates are combined to a single predicate by creating a balanced binary tree.
-<a href=#427 id=427 data-nosnippet>427</a> // Using a simple fold would result in a deeply nested predicate that can cause a stack overflow.
-<a href=#428 id=428 data-nosnippet>428</a> </span><span class="kw">while </span>row_predicates.len() > <span class="number">1 </span>{
-<a href=#429 id=429 data-nosnippet>429</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=#430 id=430 data-nosnippet>430</a> <span class="kw">let </span><span class="kw-2">mut </span>iter = row_predicates.into_iter();
-<a href=#431 id=431 data-nosnippet>431</a> <span class="kw">while let </span><span class="prelude-val">Some</span>(p1) = iter.next() {
-<a href=#432 id=432 data-nosnippet>432</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(p2) = iter.next() {
-<a href=#433 id=433 data-nosnippet>433</a> next_level.push(p1.and(p2));
-<a href=#434 id=434 data-nosnippet>434</a> } <span class="kw">else </span>{
-<a href=#435 id=435 data-nosnippet>435</a> next_level.push(p1);
-<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> row_predicates = next_level;
-<a href=#439 id=439 data-nosnippet>439</a> }
-<a href=#440 id=440 data-nosnippet>440</a>
-<a href=#441 id=441 data-nosnippet>441</a> <span class="kw">match </span>row_predicates.pop() {
-<a href=#442 id=442 data-nosnippet>442</a> <span class="prelude-val">Some</span>(p) => <span class="prelude-val">Ok</span>(p),
-<a href=#443 id=443 data-nosnippet>443</a> <span class="prelude-val">None </span>=> <span class="prelude-val">Ok</span>(AlwaysTrue),
-<a href=#444 id=444 data-nosnippet>444</a> }
-<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>
-<a href=#448 id=448 data-nosnippet>448</a><span class="kw">struct </span>EqDelColumnProcessor<<span class="lifetime">'a</span>> {
-<a href=#449 id=449 data-nosnippet>449</a> equality_ids: <span class="kw-2">&</span><span class="lifetime">'a </span>HashSet<i32>,
-<a href=#450 id=450 data-nosnippet>450</a> collected_columns: Vec<(ArrayRef, String, Type)>,
-<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">impl</span><<span class="lifetime">'a</span>> EqDelColumnProcessor<<span class="lifetime">'a</span>> {
-<a href=#454 id=454 data-nosnippet>454</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=#455 id=455 data-nosnippet>455</a> <span class="self">Self </span>{
-<a href=#456 id=456 data-nosnippet>456</a> equality_ids,
-<a href=#457 id=457 data-nosnippet>457</a> collected_columns: Vec::with_capacity(equality_ids.len()),
-<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>
-<a href=#461 id=461 data-nosnippet>461</a> <span class="attr">#[allow(clippy::type_complexity)]
-<a href=#462 id=462 data-nosnippet>462</a> </span><span class="kw">fn </span>finish(
-<a href=#463 id=463 data-nosnippet>463</a> <span class="self">self</span>,
-<a href=#464 id=464 data-nosnippet>464</a> ) -> <span class="prelude-ty">Result</span><
-<a href=#465 id=465 data-nosnippet>465</a> Vec<(
-<a href=#466 id=466 data-nosnippet>466</a> Box<<span class="kw">dyn </span>ExactSizeIterator<Item = <span class="prelude-ty">Result</span><<span class="prelude-ty">Option</span><Datum>>>>,
-<a href=#467 id=467 data-nosnippet>467</a> String,
-<a href=#468 id=468 data-nosnippet>468</a> )>,
-<a href=#469 id=469 data-nosnippet>469</a> > {
-<a href=#470 id=470 data-nosnippet>470</a> <span class="self">self</span>.collected_columns
-<a href=#471 id=471 data-nosnippet>471</a> .into_iter()
-<a href=#472 id=472 data-nosnippet>472</a> .map(|(array, field_name, field_type)| {
-<a href=#473 id=473 data-nosnippet>473</a> <span class="kw">let </span>primitive_type = field_type
-<a href=#474 id=474 data-nosnippet>474</a> .as_primitive_type()
-<a href=#475 id=475 data-nosnippet>475</a> .ok_or_else(|| {
-<a href=#476 id=476 data-nosnippet>476</a> Error::new(ErrorKind::Unexpected, <span class="string">"field is not a primitive type"</span>)
-<a href=#477 id=477 data-nosnippet>477</a> })<span class="question-mark">?
-<a href=#478 id=478 data-nosnippet>478</a> </span>.clone();
-<a href=#479 id=479 data-nosnippet>479</a>
-<a href=#480 id=480 data-nosnippet>480</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=#481 id=481 data-nosnippet>481</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=#482 id=482 data-nosnippet>482</a> Box::new(lit_vec.into_iter().map(<span class="kw">move </span>|c| {
-<a href=#483 id=483 data-nosnippet>483</a> c.map(|literal| {
-<a href=#484 id=484 data-nosnippet>484</a> literal
-<a href=#485 id=485 data-nosnippet>485</a> .as_primitive_literal()
-<a href=#486 id=486 data-nosnippet>486</a> .map(|primitive_literal| {
-<a href=#487 id=487 data-nosnippet>487</a> Datum::new(primitive_type.clone(), primitive_literal)
-<a href=#488 id=488 data-nosnippet>488</a> })
-<a href=#489 id=489 data-nosnippet>489</a> .ok_or(Error::new(
-<a href=#490 id=490 data-nosnippet>490</a> ErrorKind::Unexpected,
-<a href=#491 id=491 data-nosnippet>491</a> <span class="string">"failed to convert to primitive literal"</span>,
-<a href=#492 id=492 data-nosnippet>492</a> ))
-<a href=#493 id=493 data-nosnippet>493</a> })
-<a href=#494 id=494 data-nosnippet>494</a> .transpose()
-<a href=#495 id=495 data-nosnippet>495</a> }));
-<a href=#496 id=496 data-nosnippet>496</a>
-<a href=#497 id=497 data-nosnippet>497</a> <span class="prelude-val">Ok</span>((datum_iterator, field_name))
-<a href=#498 id=498 data-nosnippet>498</a> })
-<a href=#499 id=499 data-nosnippet>499</a> .collect::<<span class="prelude-ty">Result</span><Vec<<span class="kw">_</span>>>>()
-<a href=#500 id=500 data-nosnippet>500</a> }
-<a href=#501 id=501 data-nosnippet>501</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="kw">impl </span>SchemaWithPartnerVisitor<ArrayRef> <span class="kw">for </span>EqDelColumnProcessor<<span class="lifetime">'_</span>> {
-<a href=#504 id=504 data-nosnippet>504</a> <span class="kw">type </span>T = ();
-<a href=#505 id=505 data-nosnippet>505</a>
-<a href=#506 id=506 data-nosnippet>506</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=#507 id=507 data-nosnippet>507</a> <span class="prelude-val">Ok</span>(())
-<a href=#508 id=508 data-nosnippet>508</a> }
-<a href=#509 id=509 data-nosnippet>509</a>
-<a href=#510 id=510 data-nosnippet>510</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=#511 id=511 data-nosnippet>511</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=#512 id=512 data-nosnippet>512</a> <span class="self">self</span>.collected_columns.push((
-<a href=#513 id=513 data-nosnippet>513</a> partner.clone(),
-<a href=#514 id=514 data-nosnippet>514</a> field.name.clone(),
-<a href=#515 id=515 data-nosnippet>515</a> field.field_type.as_ref().clone(),
-<a href=#516 id=516 data-nosnippet>516</a> ));
-<a href=#517 id=517 data-nosnippet>517</a> }
-<a href=#518 id=518 data-nosnippet>518</a> <span class="prelude-val">Ok</span>(())
-<a href=#519 id=519 data-nosnippet>519</a> }
-<a href=#520 id=520 data-nosnippet>520</a>
-<a href=#521 id=521 data-nosnippet>521</a> <span class="kw">fn </span>r#struct(
-<a href=#522 id=522 data-nosnippet>522</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#523 id=523 data-nosnippet>523</a> _struct: <span class="kw-2">&</span>StructType,
-<a href=#524 id=524 data-nosnippet>524</a> _partner: <span class="kw-2">&</span>ArrayRef,
-<a href=#525 id=525 data-nosnippet>525</a> _results: Vec<()>,
-<a href=#526 id=526 data-nosnippet>526</a> ) -> <span class="prelude-ty">Result</span><()> {
-<a href=#527 id=527 data-nosnippet>527</a> <span class="prelude-val">Ok</span>(())
-<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="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=#531 id=531 data-nosnippet>531</a> <span class="prelude-val">Ok</span>(())
-<a href=#532 id=532 data-nosnippet>532</a> }
-<a href=#533 id=533 data-nosnippet>533</a>
-<a href=#534 id=534 data-nosnippet>534</a> <span class="kw">fn </span>map(
-<a href=#535 id=535 data-nosnippet>535</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#536 id=536 data-nosnippet>536</a> _map: <span class="kw-2">&</span>MapType,
-<a href=#537 id=537 data-nosnippet>537</a> _partner: <span class="kw-2">&</span>ArrayRef,
-<a href=#538 id=538 data-nosnippet>538</a> _key_value: (),
-<a href=#539 id=539 data-nosnippet>539</a> _value: (),
-<a href=#540 id=540 data-nosnippet>540</a> ) -> <span class="prelude-ty">Result</span><()> {
-<a href=#541 id=541 data-nosnippet>541</a> <span class="prelude-val">Ok</span>(())
-<a href=#542 id=542 data-nosnippet>542</a> }
-<a href=#543 id=543 data-nosnippet>543</a>
-<a href=#544 id=544 data-nosnippet>544</a> <span class="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=#545 id=545 data-nosnippet>545</a> <span class="prelude-val">Ok</span>(())
-<a href=#546 id=546 data-nosnippet>546</a> }
-<a href=#547 id=547 data-nosnippet>547</a>}
-<a href=#548 id=548 data-nosnippet>548</a>
-<a href=#549 id=549 data-nosnippet>549</a><span class="kw">struct </span>EqDelRecordBatchPartnerAccessor;
-<a href=#550 id=550 data-nosnippet>550</a>
-<a href=#551 id=551 data-nosnippet>551</a><span class="kw">impl </span>PartnerAccessor<ArrayRef> <span class="kw">for </span>EqDelRecordBatchPartnerAccessor {
-<a href=#552 id=552 data-nosnippet>552</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=#553 id=553 data-nosnippet>553</a> <span class="prelude-val">Ok</span>(schema_partner)
-<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>field_partner<<span class="lifetime">'a</span>>(
-<a href=#557 id=557 data-nosnippet>557</a> <span class="kw-2">&</span><span class="self">self</span>,
-<a href=#558 id=558 data-nosnippet>558</a> struct_partner: <span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef,
-<a href=#559 id=559 data-nosnippet>559</a> field: <span class="kw-2">&</span>NestedField,
-<a href=#560 id=560 data-nosnippet>560</a> ) -> <span class="prelude-ty">Result</span><<span class="kw-2">&</span><span class="lifetime">'a </span>ArrayRef> {
-<a href=#561 id=561 data-nosnippet>561</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=#562 id=562 data-nosnippet>562</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#563 id=563 data-nosnippet>563</a> ErrorKind::Unexpected,
-<a href=#564 id=564 data-nosnippet>564</a> <span class="string">"Expected struct array for field extraction"</span>,
-<a href=#565 id=565 data-nosnippet>565</a> ));
-<a href=#566 id=566 data-nosnippet>566</a> };
-<a href=#567 id=567 data-nosnippet>567</a>
-<a href=#568 id=568 data-nosnippet>568</a> <span class="comment">// Find the field by name within the struct
-<a href=#569 id=569 data-nosnippet>569</a> </span><span class="kw">for </span>(i, field_def) <span class="kw">in </span>struct_array.fields().iter().enumerate() {
-<a href=#570 id=570 data-nosnippet>570</a> <span class="kw">if </span>field_def.name() == <span class="kw-2">&</span>field.name {
-<a href=#571 id=571 data-nosnippet>571</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(struct_array.column(i));
-<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>
-<a href=#575 id=575 data-nosnippet>575</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#576 id=576 data-nosnippet>576</a> ErrorKind::Unexpected,
-<a href=#577 id=577 data-nosnippet>577</a> <span class="macro">format!</span>(<span class="string">"Field {} not found in parent struct"</span>, field.name),
-<a href=#578 id=578 data-nosnippet>578</a> ))
-<a href=#579 id=579 data-nosnippet>579</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=#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="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=#582 id=582 data-nosnippet>582</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#583 id=583 data-nosnippet>583</a> ErrorKind::FeatureUnsupported,
-<a href=#584 id=584 data-nosnippet>584</a> <span class="string">"List columns are unsupported in equality deletes"</span>,
-<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>
-<a href=#588 id=588 data-nosnippet>588</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=#589 id=589 data-nosnippet>589</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#590 id=590 data-nosnippet>590</a> ErrorKind::FeatureUnsupported,
-<a href=#591 id=591 data-nosnippet>591</a> <span class="string">"Map columns are unsupported in equality deletes"</span>,
-<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>
-<a href=#595 id=595 data-nosnippet>595</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=#596 id=596 data-nosnippet>596</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#597 id=597 data-nosnippet>597</a> ErrorKind::FeatureUnsupported,
-<a href=#598 id=598 data-nosnippet>598</a> <span class="string">"Map columns are unsupported in equality deletes"</span>,
-<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>}
-<a href=#602 id=602 data-nosnippet>602</a>
-<a href=#603 id=603 data-nosnippet>603</a><span class="attr">#[cfg(test)]
-<a href=#604 id=604 data-nosnippet>604</a></span><span class="kw">mod </span>tests {
-<a href=#605 id=605 data-nosnippet>605</a> <span class="kw">use </span>std::collections::HashMap;
-<a href=#606 id=606 data-nosnippet>606</a> <span class="kw">use </span>std::fs::File;
-<a href=#607 id=607 data-nosnippet>607</a> <span class="kw">use </span>std::sync::Arc;
+<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=#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="kw">use </span>arrow_array::cast::AsArray;
-<a href=#610 id=610 data-nosnippet>610</a> <span class="kw">use </span>arrow_array::{
-<a href=#611 id=611 data-nosnippet>611</a> ArrayRef, BinaryArray, Int32Array, Int64Array, RecordBatch, StringArray, StructArray,
-<a href=#612 id=612 data-nosnippet>612</a> };
-<a href=#613 id=613 data-nosnippet>613</a> <span class="kw">use </span>arrow_schema::{DataType, Field, Fields};
-<a href=#614 id=614 data-nosnippet>614</a> <span class="kw">use </span>parquet::arrow::{ArrowWriter, PARQUET_FIELD_ID_META_KEY};
-<a href=#615 id=615 data-nosnippet>615</a> <span class="kw">use </span>parquet::basic::Compression;
-<a href=#616 id=616 data-nosnippet>616</a> <span class="kw">use </span>parquet::file::properties::WriterProperties;
-<a href=#617 id=617 data-nosnippet>617</a> <span class="kw">use </span>tempfile::TempDir;
-<a href=#618 id=618 data-nosnippet>618</a>
-<a href=#619 id=619 data-nosnippet>619</a> <span class="kw">use </span>super::<span class="kw-2">*</span>;
-<a href=#620 id=620 data-nosnippet>620</a> <span class="kw">use </span><span class="kw">crate</span>::arrow::delete_filter::tests::setup;
-<a href=#621 id=621 data-nosnippet>621</a> <span class="kw">use </span><span class="kw">crate</span>::scan::FileScanTaskDeleteFile;
-<a href=#622 id=622 data-nosnippet>622</a> <span class="kw">use </span><span class="kw">crate</span>::spec::{DataContentType, Schema};
-<a href=#623 id=623 data-nosnippet>623</a>
-<a href=#624 id=624 data-nosnippet>624</a> <span class="attr">#[tokio::test]
-<a href=#625 id=625 data-nosnippet>625</a> </span><span class="kw">async fn </span>test_delete_file_loader_parse_equality_deletes() {
-<a href=#626 id=626 data-nosnippet>626</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#627 id=627 data-nosnippet>627</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
-<a href=#628 id=628 data-nosnippet>628</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<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=#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="kw">let </span>eq_delete_file_path = setup_write_equality_delete_file_1(table_location);
-<a href=#631 id=631 data-nosnippet>631</a>
-<a href=#632 id=632 data-nosnippet>632</a> <span class="kw">let </span>basic_delete_file_loader =
-<a href=#633 id=633 data-nosnippet>633</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
-<a href=#634 id=634 data-nosnippet>634</a> <span class="kw">let </span>record_batch_stream = basic_delete_file_loader
-<a href=#635 id=635 data-nosnippet>635</a> .parquet_to_batch_stream(
-<a href=#636 id=636 data-nosnippet>636</a> <span class="kw-2">&</span>eq_delete_file_path,
-<a href=#637 id=637 data-nosnippet>637</a> std::fs::metadata(<span class="kw-2">&</span>eq_delete_file_path).unwrap().len(),
-<a href=#638 id=638 data-nosnippet>638</a> )
-<a href=#639 id=639 data-nosnippet>639</a> .<span class="kw">await
-<a href=#640 id=640 data-nosnippet>640</a> </span>.expect(<span class="string">"could not get batch stream"</span>);
-<a href=#641 id=641 data-nosnippet>641</a>
-<a href=#642 id=642 data-nosnippet>642</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=#643 id=643 data-nosnippet>643</a>
-<a href=#644 id=644 data-nosnippet>644</a> <span class="kw">let </span>parsed_eq_delete = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
-<a href=#645 id=645 data-nosnippet>645</a> record_batch_stream,
-<a href=#646 id=646 data-nosnippet>646</a> eq_ids,
-<a href=#647 id=647 data-nosnippet>647</a> )
-<a href=#648 id=648 data-nosnippet>648</a> .<span class="kw">await
-<a href=#649 id=649 data-nosnippet>649</a> </span>.expect(<span class="string">"error parsing batch stream"</span>);
-<a href=#650 id=650 data-nosnippet>650</a> <span class="macro">println!</span>(<span class="string">"{parsed_eq_delete}"</span>);
-<a href=#651 id=651 data-nosnippet>651</a>
-<a href=#652 id=652 data-nosnippet>652</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=#653 id=653 data-nosnippet>653</a>
-<a href=#654 id=654 data-nosnippet>654</a> <span class="macro">assert_eq!</span>(parsed_eq_delete.to_string(), expected);
-<a href=#655 id=655 data-nosnippet>655</a> }
-<a href=#656 id=656 data-nosnippet>656</a>
-<a href=#657 id=657 data-nosnippet>657</a> <span class="doccomment">/// Create a simple field with metadata.
-<a href=#658 id=658 data-nosnippet>658</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=#659 id=659 data-nosnippet>659</a> arrow_schema::Field::new(name, ty, nullable).with_metadata(HashMap::from([(
-<a href=#660 id=660 data-nosnippet>660</a> PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#661 id=661 data-nosnippet>661</a> value.to_string(),
-<a href=#662 id=662 data-nosnippet>662</a> )]))
-<a href=#663 id=663 data-nosnippet>663</a> }
-<a href=#664 id=664 data-nosnippet>664</a>
-<a href=#665 id=665 data-nosnippet>665</a> <span class="kw">fn </span>setup_write_equality_delete_file_1(table_location: <span class="kw-2">&</span>str) -> String {
-<a href=#666 id=666 data-nosnippet>666</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=#667 id=667 data-nosnippet>667</a> <span class="kw">let </span>col_y = Arc::new(Int64Array::from(col_y_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#668 id=668 data-nosnippet>668</a>
-<a href=#669 id=669 data-nosnippet>669</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=#670 id=670 data-nosnippet>670</a> <span class="kw">let </span>col_z = Arc::new(Int64Array::from(col_z_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#671 id=671 data-nosnippet>671</a>
-<a href=#672 id=672 data-nosnippet>672</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=#673 id=673 data-nosnippet>673</a> <span class="kw">let </span>col_a = Arc::new(StringArray::from(col_a_vals)) <span class="kw">as </span>ArrayRef;
+<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=#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=#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_s = Arc::new(StructArray::from(<span class="macro">vec!</span>[
-<a href=#676 id=676 data-nosnippet>676</a> (
-<a href=#677 id=677 data-nosnippet>677</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=#678 id=678 data-nosnippet>678</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=#679 id=679 data-nosnippet>679</a> ),
-<a href=#680 id=680 data-nosnippet>680</a> (
-<a href=#681 id=681 data-nosnippet>681</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=#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">"x"</span>), <span class="prelude-val">None</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> ]));
-<a href=#685 id=685 data-nosnippet>685</a>
-<a href=#686 id=686 data-nosnippet>686</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=#687 id=687 data-nosnippet>687</a> <span class="kw">let </span>col_b = Arc::new(BinaryArray::from(col_b_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#688 id=688 data-nosnippet>688</a>
-<a href=#689 id=689 data-nosnippet>689</a> <span class="kw">let </span>equality_delete_schema = {
-<a href=#690 id=690 data-nosnippet>690</a> <span class="kw">let </span>struct_field = DataType::Struct(Fields::from(<span class="macro">vec!</span>[
-<a href=#691 id=691 data-nosnippet>691</a> simple_field(<span class="string">"sa"</span>, DataType::Int32, <span class="bool-val">false</span>, <span class="string">"6"</span>),
-<a href=#692 id=692 data-nosnippet>692</a> simple_field(<span class="string">"sb"</span>, DataType::Utf8, <span class="bool-val">true</span>, <span class="string">"7"</span>),
-<a href=#693 id=693 data-nosnippet>693</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>fields = <span class="macro">vec!</span>[
-<a href=#696 id=696 data-nosnippet>696</a> Field::new(<span class="string">"y"</span>, arrow_schema::DataType::Int64, <span class="bool-val">true</span>).with_metadata(HashMap::from(
-<a href=#697 id=697 data-nosnippet>697</a> [(PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"2"</span>.to_string())],
-<a href=#698 id=698 data-nosnippet>698</a> )),
-<a href=#699 id=699 data-nosnippet>699</a> Field::new(<span class="string">"z"</span>, arrow_schema::DataType::Int64, <span class="bool-val">true</span>).with_metadata(HashMap::from(
-<a href=#700 id=700 data-nosnippet>700</a> [(PARQUET_FIELD_ID_META_KEY.to_string(), <span class="string">"3"</span>.to_string())],
-<a href=#701 id=701 data-nosnippet>701</a> )),
-<a href=#702 id=702 data-nosnippet>702</a> Field::new(<span class="string">"a"</span>, arrow_schema::DataType::Utf8, <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">"4"</span>.to_string()),
-<a href=#704 id=704 data-nosnippet>704</a> ])),
-<a href=#705 id=705 data-nosnippet>705</a> simple_field(<span class="string">"s"</span>, struct_field, <span class="bool-val">false</span>, <span class="string">"5"</span>),
-<a href=#706 id=706 data-nosnippet>706</a> simple_field(<span class="string">"b"</span>, DataType::Binary, <span class="bool-val">true</span>, <span class="string">"8"</span>),
-<a href=#707 id=707 data-nosnippet>707</a> ];
-<a href=#708 id=708 data-nosnippet>708</a> Arc::new(arrow_schema::Schema::new(fields))
-<a href=#709 id=709 data-nosnippet>709</a> };
-<a href=#710 id=710 data-nosnippet>710</a>
-<a href=#711 id=711 data-nosnippet>711</a> <span class="kw">let </span>equality_deletes_to_write = RecordBatch::try_new(equality_delete_schema.clone(), <span class="macro">vec!</span>[
-<a href=#712 id=712 data-nosnippet>712</a> col_y, col_z, col_a, col_s, col_b,
-<a href=#713 id=713 data-nosnippet>713</a> ])
-<a href=#714 id=714 data-nosnippet>714</a> .unwrap();
-<a href=#715 id=715 data-nosnippet>715</a>
-<a href=#716 id=716 data-nosnippet>716</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=#717 id=717 data-nosnippet>717</a>
-<a href=#718 id=718 data-nosnippet>718</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
-<a href=#719 id=719 data-nosnippet>719</a>
-<a href=#720 id=720 data-nosnippet>720</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#721 id=721 data-nosnippet>721</a> .set_compression(Compression::SNAPPY)
-<a href=#722 id=722 data-nosnippet>722</a> .build();
+<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><span class="kw-2">mut </span>writer = ArrowWriter::try_new(
-<a href=#725 id=725 data-nosnippet>725</a> file,
-<a href=#726 id=726 data-nosnippet>726</a> equality_deletes_to_write.schema(),
-<a href=#727 id=727 data-nosnippet>727</a> <span class="prelude-val">Some</span>(props.clone()),
-<a href=#728 id=728 data-nosnippet>728</a> )
-<a href=#729 id=729 data-nosnippet>729</a> .unwrap();
-<a href=#730 id=730 data-nosnippet>730</a>
-<a href=#731 id=731 data-nosnippet>731</a> writer
-<a href=#732 id=732 data-nosnippet>732</a> .write(<span class="kw-2">&</span>equality_deletes_to_write)
-<a href=#733 id=733 data-nosnippet>733</a> .expect(<span class="string">"Writing batch"</span>);
-<a href=#734 id=734 data-nosnippet>734</a>
-<a href=#735 id=735 data-nosnippet>735</a> <span class="comment">// writer must be closed to write footer
-<a href=#736 id=736 data-nosnippet>736</a> </span>writer.close().unwrap();
-<a href=#737 id=737 data-nosnippet>737</a>
-<a href=#738 id=738 data-nosnippet>738</a> path
-<a href=#739 id=739 data-nosnippet>739</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=#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=#740 id=740 data-nosnippet>740</a>
-<a href=#741 id=741 data-nosnippet>741</a> <span class="attr">#[tokio::test]
-<a href=#742 id=742 data-nosnippet>742</a> </span><span class="kw">async fn </span>test_caching_delete_file_loader_load_deletes() {
-<a href=#743 id=743 data-nosnippet>743</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#744 id=744 data-nosnippet>744</a> <span class="kw">let </span>table_location = tmp_dir.path();
-<a href=#745 id=745 data-nosnippet>745</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<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="kw">let </span>delete_file_loader =
-<a href=#748 id=748 data-nosnippet>748</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
-<a href=#749 id=749 data-nosnippet>749</a>
-<a href=#750 id=750 data-nosnippet>750</a> <span class="kw">let </span>file_scan_tasks = setup(table_location);
-<a href=#751 id=751 data-nosnippet>751</a>
-<a href=#752 id=752 data-nosnippet>752</a> <span class="kw">let </span>delete_filter = delete_file_loader
-<a href=#753 id=753 data-nosnippet>753</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=#754 id=754 data-nosnippet>754</a> .<span class="kw">await
-<a href=#755 id=755 data-nosnippet>755</a> </span>.unwrap()
-<a href=#756 id=756 data-nosnippet>756</a> .unwrap();
+<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>result = delete_filter
-<a href=#759 id=759 data-nosnippet>759</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
-<a href=#760 id=760 data-nosnippet>760</a> .unwrap();
-<a href=#761 id=761 data-nosnippet>761</a>
-<a href=#762 id=762 data-nosnippet>762</a> <span class="comment">// union of pos dels from pos del file 1 and 2, ie
-<a href=#763 id=763 data-nosnippet>763</a> // [0, 1, 3, 5, 6, 8, 1022, 1023] | [0, 1, 3, 5, 20, 21, 22, 23]
-<a href=#764 id=764 data-nosnippet>764</a> // = [0, 1, 3, 5, 6, 8, 20, 21, 22, 23, 1022, 1023]
-<a href=#765 id=765 data-nosnippet>765</a> </span><span class="macro">assert_eq!</span>(result.lock().unwrap().len(), <span class="number">12</span>);
-<a href=#766 id=766 data-nosnippet>766</a>
-<a href=#767 id=767 data-nosnippet>767</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=#768 id=768 data-nosnippet>768</a> <span class="macro">assert!</span>(result.is_none()); <span class="comment">// no pos dels for file 3
-<a href=#769 id=769 data-nosnippet>769</a> </span>}
-<a href=#770 id=770 data-nosnippet>770</a>
-<a href=#771 id=771 data-nosnippet>771</a> <span class="doccomment">/// Verifies that evolve_schema on partial-schema equality deletes works correctly
-<a href=#772 id=772 data-nosnippet>772</a> /// when only equality_ids columns are evolved, not all table columns.
-<a href=#773 id=773 data-nosnippet>773</a> ///
-<a href=#774 id=774 data-nosnippet>774</a> /// Per the [Iceberg spec](https://iceberg.apache.org/spec/#equality-delete-files),
-<a href=#775 id=775 data-nosnippet>775</a> /// equality delete files can contain only a subset of columns.
-<a href=#776 id=776 data-nosnippet>776</a> </span><span class="attr">#[tokio::test]
-<a href=#777 id=777 data-nosnippet>777</a> </span><span class="kw">async fn </span>test_partial_schema_equality_deletes_evolve_succeeds() {
-<a href=#778 id=778 data-nosnippet>778</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#779 id=779 data-nosnippet>779</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
-<a href=#780 id=780 data-nosnippet>780</a>
-<a href=#781 id=781 data-nosnippet>781</a> <span class="comment">// Create table schema with REQUIRED fields
-<a href=#782 id=782 data-nosnippet>782</a> </span><span class="kw">let </span>table_schema = Arc::new(
-<a href=#783 id=783 data-nosnippet>783</a> Schema::builder()
-<a href=#784 id=784 data-nosnippet>784</a> .with_schema_id(<span class="number">1</span>)
-<a href=#785 id=785 data-nosnippet>785</a> .with_fields(<span class="macro">vec!</span>[
-<a href=#786 id=786 data-nosnippet>786</a> <span class="kw">crate</span>::spec::NestedField::required(
-<a href=#787 id=787 data-nosnippet>787</a> <span class="number">1</span>,
-<a href=#788 id=788 data-nosnippet>788</a> <span class="string">"id"</span>,
-<a href=#789 id=789 data-nosnippet>789</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Int),
-<a href=#790 id=790 data-nosnippet>790</a> )
-<a href=#791 id=791 data-nosnippet>791</a> .into(),
-<a href=#792 id=792 data-nosnippet>792</a> <span class="kw">crate</span>::spec::NestedField::required(
-<a href=#793 id=793 data-nosnippet>793</a> <span class="number">2</span>,
-<a href=#794 id=794 data-nosnippet>794</a> <span class="string">"data"</span>,
-<a href=#795 id=795 data-nosnippet>795</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::String),
-<a href=#796 id=796 data-nosnippet>796</a> )
-<a href=#797 id=797 data-nosnippet>797</a> .into(),
-<a href=#798 id=798 data-nosnippet>798</a> ])
-<a href=#799 id=799 data-nosnippet>799</a> .build()
-<a href=#800 id=800 data-nosnippet>800</a> .unwrap(),
-<a href=#801 id=801 data-nosnippet>801</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">// Write equality delete file with PARTIAL schema (only 'data' column)
-<a href=#804 id=804 data-nosnippet>804</a> </span><span class="kw">let </span>delete_file_path = {
-<a href=#805 id=805 data-nosnippet>805</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=#806 id=806 data-nosnippet>806</a> <span class="kw">let </span>data_col = Arc::new(StringArray::from(data_vals)) <span class="kw">as </span>ArrayRef;
-<a href=#807 id=807 data-nosnippet>807</a>
-<a href=#808 id=808 data-nosnippet>808</a> <span class="kw">let </span>delete_schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[simple_field(
-<a href=#809 id=809 data-nosnippet>809</a> <span class="string">"data"</span>,
-<a href=#810 id=810 data-nosnippet>810</a> DataType::Utf8,
-<a href=#811 id=811 data-nosnippet>811</a> <span class="bool-val">false</span>,
-<a href=#812 id=812 data-nosnippet>812</a> <span class="string">"2"</span>, <span class="comment">// field ID
-<a href=#813 id=813 data-nosnippet>813</a> </span>)]));
-<a href=#814 id=814 data-nosnippet>814</a>
-<a href=#815 id=815 data-nosnippet>815</a> <span class="kw">let </span>delete_batch = RecordBatch::try_new(delete_schema.clone(), <span class="macro">vec!</span>[data_col]).unwrap();
-<a href=#816 id=816 data-nosnippet>816</a>
-<a href=#817 id=817 data-nosnippet>817</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=#818 id=818 data-nosnippet>818</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
-<a href=#819 id=819 data-nosnippet>819</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#820 id=820 data-nosnippet>820</a> .set_compression(Compression::SNAPPY)
-<a href=#821 id=821 data-nosnippet>821</a> .build();
-<a href=#822 id=822 data-nosnippet>822</a> <span class="kw">let </span><span class="kw-2">mut </span>writer =
-<a href=#823 id=823 data-nosnippet>823</a> ArrowWriter::try_new(file, delete_batch.schema(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#824 id=824 data-nosnippet>824</a> writer.write(<span class="kw-2">&</span>delete_batch).expect(<span class="string">"Writing batch"</span>);
-<a href=#825 id=825 data-nosnippet>825</a> writer.close().unwrap();
-<a href=#826 id=826 data-nosnippet>826</a> path
-<a href=#827 id=827 data-nosnippet>827</a> };
-<a href=#828 id=828 data-nosnippet>828</a>
-<a href=#829 id=829 data-nosnippet>829</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
-<a href=#830 id=830 data-nosnippet>830</a> <span class="kw">let </span>basic_delete_file_loader =
-<a href=#831 id=831 data-nosnippet>831</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
-<a href=#832 id=832 data-nosnippet>832</a>
-<a href=#833 id=833 data-nosnippet>833</a> <span class="kw">let </span>batch_stream = basic_delete_file_loader
-<a href=#834 id=834 data-nosnippet>834</a> .parquet_to_batch_stream(
-<a href=#835 id=835 data-nosnippet>835</a> <span class="kw-2">&</span>delete_file_path,
-<a href=#836 id=836 data-nosnippet>836</a> std::fs::metadata(<span class="kw-2">&</span>delete_file_path).unwrap().len(),
-<a href=#837 id=837 data-nosnippet>837</a> )
-<a href=#838 id=838 data-nosnippet>838</a> .<span class="kw">await
-<a href=#839 id=839 data-nosnippet>839</a> </span>.unwrap();
-<a href=#840 id=840 data-nosnippet>840</a>
-<a href=#841 id=841 data-nosnippet>841</a> <span class="comment">// Only evolve the equality_ids columns (field 2), not all table columns
-<a href=#842 id=842 data-nosnippet>842</a> </span><span class="kw">let </span>equality_ids = <span class="macro">vec!</span>[<span class="number">2</span>];
-<a href=#843 id=843 data-nosnippet>843</a> <span class="kw">let </span>evolved_stream =
-<a href=#844 id=844 data-nosnippet>844</a> BasicDeleteFileLoader::evolve_schema(batch_stream, table_schema, <span class="kw-2">&</span>equality_ids)
-<a href=#845 id=845 data-nosnippet>845</a> .<span class="kw">await
-<a href=#846 id=846 data-nosnippet>846</a> </span>.unwrap();
-<a href=#847 id=847 data-nosnippet>847</a>
-<a href=#848 id=848 data-nosnippet>848</a> <span class="kw">let </span>result = evolved_stream.try_collect::<Vec<<span class="kw">_</span>>>().<span class="kw">await</span>;
-<a href=#849 id=849 data-nosnippet>849</a>
-<a href=#850 id=850 data-nosnippet>850</a> <span class="macro">assert!</span>(
-<a href=#851 id=851 data-nosnippet>851</a> result.is_ok(),
-<a href=#852 id=852 data-nosnippet>852</a> <span class="string">"Expected success when evolving only equality_ids columns, got error: {:?}"</span>,
-<a href=#853 id=853 data-nosnippet>853</a> result.err()
-<a href=#854 id=854 data-nosnippet>854</a> );
-<a href=#855 id=855 data-nosnippet>855</a>
-<a href=#856 id=856 data-nosnippet>856</a> <span class="kw">let </span>batches = result.unwrap();
-<a href=#857 id=857 data-nosnippet>857</a> <span class="macro">assert_eq!</span>(batches.len(), <span class="number">1</span>);
-<a href=#858 id=858 data-nosnippet>858</a>
-<a href=#859 id=859 data-nosnippet>859</a> <span class="kw">let </span>batch = <span class="kw-2">&</span>batches[<span class="number">0</span>];
-<a href=#860 id=860 data-nosnippet>860</a> <span class="macro">assert_eq!</span>(batch.num_rows(), <span class="number">3</span>);
-<a href=#861 id=861 data-nosnippet>861</a> <span class="macro">assert_eq!</span>(batch.num_columns(), <span class="number">1</span>); <span class="comment">// Only 'data' column
+<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=#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> // Verify the actual values are preserved after schema evolution
-<a href=#864 id=864 data-nosnippet>864</a> </span><span class="kw">let </span>data_col = batch.column(<span class="number">0</span>).as_string::<i32>();
-<a href=#865 id=865 data-nosnippet>865</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">0</span>), <span class="string">"a"</span>);
-<a href=#866 id=866 data-nosnippet>866</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">1</span>), <span class="string">"d"</span>);
-<a href=#867 id=867 data-nosnippet>867</a> <span class="macro">assert_eq!</span>(data_col.value(<span class="number">2</span>), <span class="string">"g"</span>);
-<a href=#868 id=868 data-nosnippet>868</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="doccomment">/// Test loading a FileScanTask with BOTH positional and equality deletes.
-<a href=#871 id=871 data-nosnippet>871</a> /// Verifies the fix for the inverted condition that caused "Missing predicate for equality delete file" errors.
-<a href=#872 id=872 data-nosnippet>872</a> </span><span class="attr">#[tokio::test]
-<a href=#873 id=873 data-nosnippet>873</a> </span><span class="kw">async fn </span>test_load_deletes_with_mixed_types() {
-<a href=#874 id=874 data-nosnippet>874</a> <span class="kw">use </span><span class="kw">crate</span>::scan::FileScanTask;
-<a href=#875 id=875 data-nosnippet>875</a> <span class="kw">use </span><span class="kw">crate</span>::spec::{DataFileFormat, Schema};
-<a href=#876 id=876 data-nosnippet>876</a>
-<a href=#877 id=877 data-nosnippet>877</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#878 id=878 data-nosnippet>878</a> <span class="kw">let </span>table_location = tmp_dir.path();
-<a href=#879 id=879 data-nosnippet>879</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
+<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="comment">// Create the data file schema
-<a href=#882 id=882 data-nosnippet>882</a> </span><span class="kw">let </span>data_file_schema = Arc::new(
-<a href=#883 id=883 data-nosnippet>883</a> Schema::builder()
-<a href=#884 id=884 data-nosnippet>884</a> .with_fields(<span class="macro">vec!</span>[
-<a href=#885 id=885 data-nosnippet>885</a> <span class="kw">crate</span>::spec::NestedField::optional(
-<a href=#886 id=886 data-nosnippet>886</a> <span class="number">2</span>,
-<a href=#887 id=887 data-nosnippet>887</a> <span class="string">"y"</span>,
-<a href=#888 id=888 data-nosnippet>888</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Long),
-<a href=#889 id=889 data-nosnippet>889</a> )
-<a href=#890 id=890 data-nosnippet>890</a> .into(),
-<a href=#891 id=891 data-nosnippet>891</a> <span class="kw">crate</span>::spec::NestedField::optional(
-<a href=#892 id=892 data-nosnippet>892</a> <span class="number">3</span>,
-<a href=#893 id=893 data-nosnippet>893</a> <span class="string">"z"</span>,
-<a href=#894 id=894 data-nosnippet>894</a> <span class="kw">crate</span>::spec::Type::Primitive(<span class="kw">crate</span>::spec::PrimitiveType::Long),
-<a href=#895 id=895 data-nosnippet>895</a> )
-<a href=#896 id=896 data-nosnippet>896</a> .into(),
-<a href=#897 id=897 data-nosnippet>897</a> ])
-<a href=#898 id=898 data-nosnippet>898</a> .build()
-<a href=#899 id=899 data-nosnippet>899</a> .unwrap(),
-<a href=#900 id=900 data-nosnippet>900</a> );
-<a href=#901 id=901 data-nosnippet>901</a>
-<a href=#902 id=902 data-nosnippet>902</a> <span class="comment">// Write positional delete file
-<a href=#903 id=903 data-nosnippet>903</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=#904 id=904 data-nosnippet>904</a> <span class="kw">let </span>file_path_values =
-<a href=#905 id=905 data-nosnippet>905</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=#906 id=906 data-nosnippet>906</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=#907 id=907 data-nosnippet>907</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=#908 id=908 data-nosnippet>908</a>
-<a href=#909 id=909 data-nosnippet>909</a> <span class="kw">let </span>positional_deletes_to_write =
-<a href=#910 id=910 data-nosnippet>910</a> RecordBatch::try_new(positional_delete_schema.clone(), <span class="macro">vec!</span>[
-<a href=#911 id=911 data-nosnippet>911</a> file_path_col,
-<a href=#912 id=912 data-nosnippet>912</a> pos_col,
-<a href=#913 id=913 data-nosnippet>913</a> ])
-<a href=#914 id=914 data-nosnippet>914</a> .unwrap();
-<a href=#915 id=915 data-nosnippet>915</a>
-<a href=#916 id=916 data-nosnippet>916</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#917 id=917 data-nosnippet>917</a> .set_compression(Compression::SNAPPY)
-<a href=#918 id=918 data-nosnippet>918</a> .build();
-<a href=#919 id=919 data-nosnippet>919</a>
-<a href=#920 id=920 data-nosnippet>920</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=#921 id=921 data-nosnippet>921</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>pos_del_path).unwrap();
-<a href=#922 id=922 data-nosnippet>922</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(
-<a href=#923 id=923 data-nosnippet>923</a> file,
-<a href=#924 id=924 data-nosnippet>924</a> positional_deletes_to_write.schema(),
-<a href=#925 id=925 data-nosnippet>925</a> <span class="prelude-val">Some</span>(props.clone()),
-<a href=#926 id=926 data-nosnippet>926</a> )
-<a href=#927 id=927 data-nosnippet>927</a> .unwrap();
-<a href=#928 id=928 data-nosnippet>928</a> writer.write(<span class="kw-2">&</span>positional_deletes_to_write).unwrap();
-<a href=#929 id=929 data-nosnippet>929</a> writer.close().unwrap();
+<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=#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=#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="comment">// Write equality delete file
-<a href=#932 id=932 data-nosnippet>932</a> </span><span class="kw">let </span>eq_delete_path = setup_write_equality_delete_file_1(table_location.to_str().unwrap());
-<a href=#933 id=933 data-nosnippet>933</a>
-<a href=#934 id=934 data-nosnippet>934</a> <span class="comment">// Create FileScanTask with BOTH positional and equality deletes
-<a href=#935 id=935 data-nosnippet>935</a> </span><span class="kw">let </span>pos_del = FileScanTaskDeleteFile::builder()
-<a href=#936 id=936 data-nosnippet>936</a> .with_file_path(pos_del_path.clone())
-<a href=#937 id=937 data-nosnippet>937</a> .with_file_size_in_bytes(std::fs::metadata(<span class="kw-2">&</span>pos_del_path).unwrap().len())
-<a href=#938 id=938 data-nosnippet>938</a> .with_file_type(DataContentType::PositionDeletes)
-<a href=#939 id=939 data-nosnippet>939</a> .with_partition_spec_id(<span class="number">0</span>)
+<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>eq_del = FileScanTaskDeleteFile::builder()
-<a href=#943 id=943 data-nosnippet>943</a> .with_file_path(eq_delete_path.clone())
-<a href=#944 id=944 data-nosnippet>944</a> .with_file_size_in_bytes(std::fs::metadata(<span class="kw-2">&</span>eq_delete_path).unwrap().len())
-<a href=#945 id=945 data-nosnippet>945</a> .with_file_type(DataContentType::EqualityDeletes)
-<a href=#946 id=946 data-nosnippet>946</a> .with_partition_spec_id(<span class="number">0</span>)
-<a href=#947 id=947 data-nosnippet>947</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=#948 id=948 data-nosnippet>948</a> </span>.build();
-<a href=#949 id=949 data-nosnippet>949</a>
-<a href=#950 id=950 data-nosnippet>950</a> <span class="kw">let </span>file_scan_task = FileScanTask::builder()
-<a href=#951 id=951 data-nosnippet>951</a> .with_file_size_in_bytes(<span class="number">0</span>)
-<a href=#952 id=952 data-nosnippet>952</a> .with_start(<span class="number">0</span>)
-<a href=#953 id=953 data-nosnippet>953</a> .with_length(<span class="number">0</span>)
-<a href=#954 id=954 data-nosnippet>954</a> .with_data_file_path(<span class="macro">format!</span>(
-<a href=#955 id=955 data-nosnippet>955</a> <span class="string">"{}/data-1.parquet"</span>,
-<a href=#956 id=956 data-nosnippet>956</a> table_location.to_str().unwrap()
-<a href=#957 id=957 data-nosnippet>957</a> ))
-<a href=#958 id=958 data-nosnippet>958</a> .with_data_file_format(DataFileFormat::Parquet)
-<a href=#959 id=959 data-nosnippet>959</a> .with_schema(data_file_schema.clone())
-<a href=#960 id=960 data-nosnippet>960</a> .with_project_field_ids(<span class="macro">vec!</span>[<span class="number">2</span>, <span class="number">3</span>])
-<a href=#961 id=961 data-nosnippet>961</a> .with_deletes(<span class="macro">vec!</span>[pos_del, eq_del])
-<a href=#962 id=962 data-nosnippet>962</a> .with_case_sensitive(<span class="bool-val">false</span>)
-<a href=#963 id=963 data-nosnippet>963</a> .build();
-<a href=#964 id=964 data-nosnippet>964</a>
-<a href=#965 id=965 data-nosnippet>965</a> <span class="comment">// Load the deletes - should handle both types without error
-<a href=#966 id=966 data-nosnippet>966</a> </span><span class="kw">let </span>delete_file_loader =
-<a href=#967 id=967 data-nosnippet>967</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
-<a href=#968 id=968 data-nosnippet>968</a> <span class="kw">let </span>delete_filter = delete_file_loader
-<a href=#969 id=969 data-nosnippet>969</a> .load_deletes(<span class="kw-2">&</span>file_scan_task.deletes, file_scan_task.schema_ref())
-<a href=#970 id=970 data-nosnippet>970</a> .<span class="kw">await
-<a href=#971 id=971 data-nosnippet>971</a> </span>.unwrap()
-<a href=#972 id=972 data-nosnippet>972</a> .unwrap();
-<a href=#973 id=973 data-nosnippet>973</a>
-<a href=#974 id=974 data-nosnippet>974</a> <span class="comment">// Verify both delete types can be processed together
-<a href=#975 id=975 data-nosnippet>975</a> </span><span class="kw">let </span>result = delete_filter
-<a href=#976 id=976 data-nosnippet>976</a> .build_equality_delete_predicate(<span class="kw-2">&</span>file_scan_task)
-<a href=#977 id=977 data-nosnippet>977</a> .<span class="kw">await</span>;
-<a href=#978 id=978 data-nosnippet>978</a> <span class="macro">assert!</span>(
-<a href=#979 id=979 data-nosnippet>979</a> result.is_ok(),
-<a href=#980 id=980 data-nosnippet>980</a> <span class="string">"Failed to build equality delete predicate: {:?}"</span>,
-<a href=#981 id=981 data-nosnippet>981</a> result.err()
-<a href=#982 id=982 data-nosnippet>982</a> );
-<a href=#983 id=983 data-nosnippet>983</a> }
-<a href=#984 id=984 data-nosnippet>984</a>
-<a href=#985 id=985 data-nosnippet>985</a> <span class="attr">#[tokio::test]
-<a href=#986 id=986 data-nosnippet>986</a> </span><span class="kw">async fn </span>test_large_equality_delete_batch_stack_overflow() {
-<a href=#987 id=987 data-nosnippet>987</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#988 id=988 data-nosnippet>988</a> <span class="kw">let </span>table_location = tmp_dir.path().as_os_str().to_str().unwrap();
-<a href=#989 id=989 data-nosnippet>989</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
-<a href=#990 id=990 data-nosnippet>990</a>
-<a href=#991 id=991 data-nosnippet>991</a> <span class="comment">// Create a large batch of equality deletes
-<a href=#992 id=992 data-nosnippet>992</a> </span><span class="kw">let </span>num_rows = <span class="number">20_000</span>;
-<a href=#993 id=993 data-nosnippet>993</a> <span class="kw">let </span>col_y_vals: Vec<i64> = (<span class="number">0</span>..num_rows).collect();
-<a href=#994 id=994 data-nosnippet>994</a> <span class="kw">let </span>col_y = Arc::new(Int64Array::from(col_y_vals)) <span class="kw">as </span>ArrayRef;
+<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=#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="kw">let </span>schema = Arc::new(arrow_schema::Schema::new(<span class="macro">vec!</span>[
-<a href=#997 id=997 data-nosnippet>997</a> Field::new(<span class="string">"y"</span>, arrow_schema::DataType::Int64, <span class="bool-val">false</span>).with_metadata(HashMap::from([(
-<a href=#998 id=998 data-nosnippet>998</a> PARQUET_FIELD_ID_META_KEY.to_string(),
-<a href=#999 id=999 data-nosnippet>999</a> <span class="string">"2"</span>.to_string(),
-<a href=#1000 id=1000 data-nosnippet>1000</a> )])),
-<a href=#1001 id=1001 data-nosnippet>1001</a> ]));
-<a href=#1002 id=1002 data-nosnippet>1002</a>
-<a href=#1003 id=1003 data-nosnippet>1003</a> <span class="kw">let </span>record_batch = RecordBatch::try_new(schema.clone(), <span class="macro">vec!</span>[col_y]).unwrap();
-<a href=#1004 id=1004 data-nosnippet>1004</a>
-<a href=#1005 id=1005 data-nosnippet>1005</a> <span class="comment">// Write to file
-<a href=#1006 id=1006 data-nosnippet>1006</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=#1007 id=1007 data-nosnippet>1007</a> <span class="kw">let </span>file = File::create(<span class="kw-2">&</span>path).unwrap();
-<a href=#1008 id=1008 data-nosnippet>1008</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#1009 id=1009 data-nosnippet>1009</a> .set_compression(Compression::SNAPPY)
-<a href=#1010 id=1010 data-nosnippet>1010</a> .build();
-<a href=#1011 id=1011 data-nosnippet>1011</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=#1012 id=1012 data-nosnippet>1012</a> writer.write(<span class="kw-2">&</span>record_batch).unwrap();
-<a href=#1013 id=1013 data-nosnippet>1013</a> writer.close().unwrap();
-<a href=#1014 id=1014 data-nosnippet>1014</a>
-<a href=#1015 id=1015 data-nosnippet>1015</a> <span class="kw">let </span>basic_delete_file_loader =
-<a href=#1016 id=1016 data-nosnippet>1016</a> BasicDeleteFileLoader::new(file_io.clone(), ScanMetrics::new());
-<a href=#1017 id=1017 data-nosnippet>1017</a> <span class="kw">let </span>record_batch_stream = basic_delete_file_loader
-<a href=#1018 id=1018 data-nosnippet>1018</a> .parquet_to_batch_stream(<span class="kw-2">&</span>path, std::fs::metadata(<span class="kw-2">&</span>path).unwrap().len())
-<a href=#1019 id=1019 data-nosnippet>1019</a> .<span class="kw">await
-<a href=#1020 id=1020 data-nosnippet>1020</a> </span>.expect(<span class="string">"could not get batch stream"</span>);
-<a href=#1021 id=1021 data-nosnippet>1021</a>
-<a href=#1022 id=1022 data-nosnippet>1022</a> <span class="kw">let </span>eq_ids = HashSet::from_iter(<span class="macro">vec!</span>[<span class="number">2</span>]);
-<a href=#1023 id=1023 data-nosnippet>1023</a>
-<a href=#1024 id=1024 data-nosnippet>1024</a> <span class="kw">let </span>result = CachingDeleteFileLoader::parse_equality_deletes_record_batch_stream(
-<a href=#1025 id=1025 data-nosnippet>1025</a> record_batch_stream,
-<a href=#1026 id=1026 data-nosnippet>1026</a> eq_ids,
-<a href=#1027 id=1027 data-nosnippet>1027</a> )
-<a href=#1028 id=1028 data-nosnippet>1028</a> .<span class="kw">await</span>;
-<a href=#1029 id=1029 data-nosnippet>1029</a>
-<a href=#1030 id=1030 data-nosnippet>1030</a> <span class="macro">assert!</span>(result.is_ok());
-<a href=#1031 id=1031 data-nosnippet>1031</a> }
-<a href=#1032 id=1032 data-nosnippet>1032</a>
-<a href=#1033 id=1033 data-nosnippet>1033</a> <span class="attr">#[tokio::test]
-<a href=#1034 id=1034 data-nosnippet>1034</a> </span><span class="kw">async fn </span>test_caching_delete_file_loader_caches_results() {
-<a href=#1035 id=1035 data-nosnippet>1035</a> <span class="kw">let </span>tmp_dir = TempDir::new().unwrap();
-<a href=#1036 id=1036 data-nosnippet>1036</a> <span class="kw">let </span>table_location = tmp_dir.path();
-<a href=#1037 id=1037 data-nosnippet>1037</a> <span class="kw">let </span>file_io = FileIO::new_with_fs();
-<a href=#1038 id=1038 data-nosnippet>1038</a>
-<a href=#1039 id=1039 data-nosnippet>1039</a> <span class="kw">let </span>delete_file_loader =
-<a href=#1040 id=1040 data-nosnippet>1040</a> CachingDeleteFileLoader::new(file_io.clone(), <span class="number">10</span>, Runtime::current());
-<a href=#1041 id=1041 data-nosnippet>1041</a>
-<a href=#1042 id=1042 data-nosnippet>1042</a> <span class="kw">let </span>file_scan_tasks = setup(table_location);
+<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=#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="comment">// Load deletes for the first time
-<a href=#1045 id=1045 data-nosnippet>1045</a> </span><span class="kw">let </span>delete_filter_1 = delete_file_loader
-<a href=#1046 id=1046 data-nosnippet>1046</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=#1047 id=1047 data-nosnippet>1047</a> .<span class="kw">await
-<a href=#1048 id=1048 data-nosnippet>1048</a> </span>.unwrap()
-<a href=#1049 id=1049 data-nosnippet>1049</a> .unwrap();
-<a href=#1050 id=1050 data-nosnippet>1050</a>
-<a href=#1051 id=1051 data-nosnippet>1051</a> <span class="comment">// Load deletes for the second time (same task/files)
-<a href=#1052 id=1052 data-nosnippet>1052</a> </span><span class="kw">let </span>delete_filter_2 = delete_file_loader
-<a href=#1053 id=1053 data-nosnippet>1053</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=#1054 id=1054 data-nosnippet>1054</a> .<span class="kw">await
-<a href=#1055 id=1055 data-nosnippet>1055</a> </span>.unwrap()
-<a href=#1056 id=1056 data-nosnippet>1056</a> .unwrap();
-<a href=#1057 id=1057 data-nosnippet>1057</a>
-<a href=#1058 id=1058 data-nosnippet>1058</a> <span class="kw">let </span>dv1 = delete_filter_1
-<a href=#1059 id=1059 data-nosnippet>1059</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
-<a href=#1060 id=1060 data-nosnippet>1060</a> .unwrap();
-<a href=#1061 id=1061 data-nosnippet>1061</a> <span class="kw">let </span>dv2 = delete_filter_2
-<a href=#1062 id=1062 data-nosnippet>1062</a> .get_delete_vector(<span class="kw-2">&</span>file_scan_tasks[<span class="number">0</span>])
-<a href=#1063 id=1063 data-nosnippet>1063</a> .unwrap();
-<a href=#1064 id=1064 data-nosnippet>1064</a>
-<a href=#1065 id=1065 data-nosnippet>1065</a> <span class="comment">// Verify that the delete vectors point to the same memory location,
-<a href=#1066 id=1066 data-nosnippet>1066</a> // confirming that the second load reused the result from the first.
-<a href=#1067 id=1067 data-nosnippet>1067</a> </span><span class="macro">assert!</span>(Arc::ptr_eq(<span class="kw-2">&</span>dv1, <span class="kw-2">&</span>dv2));
-<a href=#1068 id=1068 data-nosnippet>1068</a> }
-<a href=#1069 id=1069 data-nosnippet>1069</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=#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>}
</code></pre></div></section></main></body></html>
\ No newline at end of file