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 2026/10/11 10:47:00 UTC
(iceberg-rust) branch gh-pages updated: deploy: e42840bc98cbf91a67c014162ed68036a8452d7a
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 0f8d5ac3d deploy: e42840bc98cbf91a67c014162ed68036a8452d7a
0f8d5ac3d is described below
commit 0f8d5ac3d7cb26ce6b750dbaabd2c947118443fb
Author: github-merge-queue[bot] <gi...@users.noreply.github.com>
AuthorDate: Sun Oct 11 10:46:56 2026 +0000
deploy: e42840bc98cbf91a67c014162ed68036a8452d7a
---
.../expr/visitors/page_index_evaluator.rs.html | 3640 ++++++++++----------
1 file changed, 1883 insertions(+), 1757 deletions(-)
diff --git a/api/src/iceberg/expr/visitors/page_index_evaluator.rs.html b/api/src/iceberg/expr/visitors/page_index_evaluator.rs.html
index aad555fe7..4addebb65 100644
--- a/api/src/iceberg/expr/visitors/page_index_evaluator.rs.html
+++ b/api/src/iceberg/expr/visitors/page_index_evaluator.rs.html
@@ -438,1800 +438,1926 @@
<a href=#438 id=438 data-nosnippet>438</a> .enumerate()
<a href=#439 id=439 data-nosnippet>439</a> .zip(row_counts.iter())
<a href=#440 id=440 data-nosnippet>440</a> {
-<a href=#441 id=441 data-nosnippet>441</a> <span class="comment">// A bound that won't decode (e.g. a min/max stat truncated
-<a href=#442 id=442 data-nosnippet>442</a> // by column_index_truncate_length, whose byte prefix no
-<a href=#443 id=443 data-nosnippet>443</a> // longer preserves the two's-complement decimal ordering)
-<a href=#444 id=444 data-nosnippet>444</a> // means this column's page index can't be trusted, so skip
-<a href=#445 id=445 data-nosnippet>445</a> // pruning for the whole column rather than abort the scan.
-<a href=#446 id=446 data-nosnippet>446</a> </span><span class="kw">let </span>(min, max) = <span class="kw">match </span>(
-<a href=#447 id=447 data-nosnippet>447</a> min.map(|val| {
-<a href=#448 id=448 data-nosnippet>448</a> <span class="self">Self</span>::fixed_len_byte_array_decimal_bound_to_datum(
-<a href=#449 id=449 data-nosnippet>449</a> field_type,
-<a href=#450 id=450 data-nosnippet>450</a> val,
-<a href=#451 id=451 data-nosnippet>451</a> type_length,
-<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> .transpose(),
-<a href=#455 id=455 data-nosnippet>455</a> max.map(|val| {
-<a href=#456 id=456 data-nosnippet>456</a> <span class="self">Self</span>::fixed_len_byte_array_decimal_bound_to_datum(
-<a href=#457 id=457 data-nosnippet>457</a> field_type,
-<a href=#458 id=458 data-nosnippet>458</a> val,
-<a href=#459 id=459 data-nosnippet>459</a> type_length,
-<a href=#460 id=460 data-nosnippet>460</a> )
-<a href=#461 id=461 data-nosnippet>461</a> })
-<a href=#462 id=462 data-nosnippet>462</a> .transpose(),
-<a href=#463 id=463 data-nosnippet>463</a> ) {
-<a href=#464 id=464 data-nosnippet>464</a> (<span class="prelude-val">Ok</span>(min), <span class="prelude-val">Ok</span>(max)) => (min, max),
-<a href=#465 id=465 data-nosnippet>465</a> (<span class="prelude-val">Err</span>(err), <span class="kw">_</span>) | (<span class="kw">_</span>, <span class="prelude-val">Err</span>(err)) => {
-<a href=#466 id=466 data-nosnippet>466</a> <span class="macro">tracing::debug!</span>(
-<a href=#467 id=467 data-nosnippet>467</a> field_id,
-<a href=#468 id=468 data-nosnippet>468</a> %field_type,
-<a href=#469 id=469 data-nosnippet>469</a> page_index = i,
-<a href=#470 id=470 data-nosnippet>470</a> %err,
-<a href=#471 id=471 data-nosnippet>471</a> <span class="string">"Skipping page-index pruning: undecodable FIXED_LEN_BYTE_ARRAY decimal page bound"
-<a href=#472 id=472 data-nosnippet>472</a> </span>);
-<a href=#473 id=473 data-nosnippet>473</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>);
-<a href=#474 id=474 data-nosnippet>474</a> }
-<a href=#475 id=475 data-nosnippet>475</a> };
-<a href=#476 id=476 data-nosnippet>476</a>
-<a href=#477 id=477 data-nosnippet>477</a> page_filter.push(predicate(
-<a href=#478 id=478 data-nosnippet>478</a> min,
-<a href=#479 id=479 data-nosnippet>479</a> max,
-<a href=#480 id=480 data-nosnippet>480</a> PageNullCount::from_row_and_null_counts(row_count, idx.null_count(i)),
-<a href=#481 id=481 data-nosnippet>481</a> )<span class="question-mark">?</span>);
-<a href=#482 id=482 data-nosnippet>482</a> }
-<a href=#483 id=483 data-nosnippet>483</a>
-<a href=#484 id=484 data-nosnippet>484</a> <span class="prelude-val">Ok</span>(page_filter)
-<a href=#485 id=485 data-nosnippet>485</a> }
-<a href=#486 id=486 data-nosnippet>486</a> <span class="comment">// Column index types we can't interpret: skip page pruning rather
-<a href=#487 id=487 data-nosnippet>487</a> // than abort the scan. Row-group filtering and the Arrow row filter
-<a href=#488 id=488 data-nosnippet>488</a> // still apply the predicate.
-<a href=#489 id=489 data-nosnippet>489</a> </span>ColumnIndexMetaData::INT96(<span class="kw">_</span>) => {
-<a href=#490 id=490 data-nosnippet>490</a> <span class="macro">tracing::debug!</span>(
-<a href=#491 id=491 data-nosnippet>491</a> field_id,
-<a href=#492 id=492 data-nosnippet>492</a> %field_type,
-<a href=#493 id=493 data-nosnippet>493</a> <span class="string">"Skipping page-index pruning: unsupported INT96 column index"
-<a href=#494 id=494 data-nosnippet>494</a> </span>);
-<a href=#495 id=495 data-nosnippet>495</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>);
-<a href=#496 id=496 data-nosnippet>496</a> }
-<a href=#497 id=497 data-nosnippet>497</a> };
-<a href=#498 id=498 data-nosnippet>498</a>
-<a href=#499 id=499 data-nosnippet>499</a> <span class="prelude-val">Ok</span>(<span class="prelude-val">Some</span>(result<span class="question-mark">?</span>))
-<a href=#500 id=500 data-nosnippet>500</a> }
-<a href=#501 id=501 data-nosnippet>501</a>
-<a href=#502 id=502 data-nosnippet>502</a> <span class="doccomment">/// Converts an `INT32` page bound into a [`Datum`] according to the field's
-<a href=#503 id=503 data-nosnippet>503</a> /// primitive type.
-<a href=#504 id=504 data-nosnippet>504</a> </span><span class="kw">fn </span>int32_bound_to_datum(field_type: <span class="kw-2">&</span>PrimitiveType, val: i32) -> Datum {
-<a href=#505 id=505 data-nosnippet>505</a> <span class="kw">match </span>field_type {
-<a href=#506 id=506 data-nosnippet>506</a> PrimitiveType::Decimal { .. } => Datum::new(
-<a href=#507 id=507 data-nosnippet>507</a> field_type.clone(),
-<a href=#508 id=508 data-nosnippet>508</a> PrimitiveLiteral::Int128(i128::from(val)),
-<a href=#509 id=509 data-nosnippet>509</a> ),
-<a href=#510 id=510 data-nosnippet>510</a> <span class="kw">_ </span>=> Datum::new(field_type.clone(), PrimitiveLiteral::Int(val)),
-<a href=#511 id=511 data-nosnippet>511</a> }
-<a href=#512 id=512 data-nosnippet>512</a> }
-<a href=#513 id=513 data-nosnippet>513</a>
-<a href=#514 id=514 data-nosnippet>514</a> <span class="doccomment">/// Converts an `INT64` page bound into a [`Datum`] according to the field's
-<a href=#515 id=515 data-nosnippet>515</a> /// primitive type.
-<a href=#516 id=516 data-nosnippet>516</a> </span><span class="kw">fn </span>int64_bound_to_datum(field_type: <span class="kw-2">&</span>PrimitiveType, val: i64) -> Datum {
-<a href=#517 id=517 data-nosnippet>517</a> <span class="kw">match </span>field_type {
-<a href=#518 id=518 data-nosnippet>518</a> PrimitiveType::Decimal { .. } => Datum::new(
-<a href=#519 id=519 data-nosnippet>519</a> field_type.clone(),
-<a href=#520 id=520 data-nosnippet>520</a> PrimitiveLiteral::Int128(i128::from(val)),
-<a href=#521 id=521 data-nosnippet>521</a> ),
-<a href=#522 id=522 data-nosnippet>522</a> <span class="kw">_ </span>=> Datum::new(field_type.clone(), PrimitiveLiteral::Long(val)),
-<a href=#523 id=523 data-nosnippet>523</a> }
-<a href=#524 id=524 data-nosnippet>524</a> }
-<a href=#525 id=525 data-nosnippet>525</a>
-<a href=#526 id=526 data-nosnippet>526</a> <span class="doccomment">/// Converts a `BYTE_ARRAY` page bound into a [`Datum`] according to the
-<a href=#527 id=527 data-nosnippet>527</a> /// field's primitive type. Parquet stores Iceberg `string` and `binary`
-<a href=#528 id=528 data-nosnippet>528</a> /// bounds as `BYTE_ARRAY`.
-<a href=#529 id=529 data-nosnippet>529</a> </span><span class="kw">fn </span>byte_array_bound_to_datum(field_type: <span class="kw-2">&</span>PrimitiveType, bytes: <span class="kw-2">&</span>[u8]) -> <span class="prelude-ty">Result</span><Datum> {
-<a href=#530 id=530 data-nosnippet>530</a> <span class="kw">match </span>field_type {
-<a href=#531 id=531 data-nosnippet>531</a> PrimitiveType::String => {
-<a href=#532 id=532 data-nosnippet>532</a> <span class="kw">let </span>value = std::str::from_utf8(bytes).map_err(|err| {
-<a href=#533 id=533 data-nosnippet>533</a> Error::new(ErrorKind::DataInvalid, <span class="string">"Invalid UTF-8 in string page bound"</span>)
-<a href=#534 id=534 data-nosnippet>534</a> .with_source(err)
-<a href=#535 id=535 data-nosnippet>535</a> })<span class="question-mark">?</span>;
-<a href=#536 id=536 data-nosnippet>536</a> <span class="prelude-val">Ok</span>(Datum::string(value))
-<a href=#537 id=537 data-nosnippet>537</a> }
-<a href=#538 id=538 data-nosnippet>538</a> PrimitiveType::Binary => <span class="prelude-val">Ok</span>(Datum::binary(bytes.to_vec())),
-<a href=#539 id=539 data-nosnippet>539</a> <span class="comment">// Defensive: production callers only invoke this for String/Binary.
-<a href=#540 id=540 data-nosnippet>540</a> </span><span class="kw">_ </span>=> <span class="prelude-val">Err</span>(Error::new(
-<a href=#541 id=541 data-nosnippet>541</a> ErrorKind::Unexpected,
-<a href=#542 id=542 data-nosnippet>542</a> <span class="macro">format!</span>(<span class="string">"Unsupported primitive type for BYTE_ARRAY page bound: {field_type}"</span>),
-<a href=#543 id=543 data-nosnippet>543</a> )),
-<a href=#544 id=544 data-nosnippet>544</a> }
-<a href=#545 id=545 data-nosnippet>545</a> }
-<a href=#546 id=546 data-nosnippet>546</a>
-<a href=#547 id=547 data-nosnippet>547</a> <span class="doccomment">/// Converts a `FIXED_LEN_BYTE_ARRAY` decimal page bound into a [`Datum`].
-<a href=#548 id=548 data-nosnippet>548</a> /// Parquet stores Iceberg decimals with precision > 18 as big-endian
-<a href=#549 id=549 data-nosnippet>549</a> /// two's-complement `FIXED_LEN_BYTE_ARRAY` of width `type_length`.
-<a href=#550 id=550 data-nosnippet>550</a> ///
-<a href=#551 id=551 data-nosnippet>551</a> /// Returns an error for a bound whose width differs from `type_length`: that
-<a href=#552 id=552 data-nosnippet>552</a> /// signals a `column_index_truncate_length` truncation, and a truncated byte
-<a href=#553 id=553 data-nosnippet>553</a> /// prefix decodes to the wrong number under two's-complement.
-<a href=#554 id=554 data-nosnippet>554</a> </span><span class="kw">fn </span>fixed_len_byte_array_decimal_bound_to_datum(
-<a href=#555 id=555 data-nosnippet>555</a> field_type: <span class="kw-2">&</span>PrimitiveType,
-<a href=#556 id=556 data-nosnippet>556</a> bytes: <span class="kw-2">&</span>[u8],
-<a href=#557 id=557 data-nosnippet>557</a> type_length: usize,
-<a href=#558 id=558 data-nosnippet>558</a> ) -> <span class="prelude-ty">Result</span><Datum> {
-<a href=#559 id=559 data-nosnippet>559</a> <span class="kw">if </span>bytes.len() != type_length {
-<a href=#560 id=560 data-nosnippet>560</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#561 id=561 data-nosnippet>561</a> ErrorKind::DataInvalid,
-<a href=#562 id=562 data-nosnippet>562</a> <span class="macro">format!</span>(
-<a href=#563 id=563 data-nosnippet>563</a> <span class="string">"Truncated FIXED_LEN_BYTE_ARRAY decimal page bound: {} bytes (expected {type_length})"</span>,
-<a href=#564 id=564 data-nosnippet>564</a> bytes.len()
-<a href=#565 id=565 data-nosnippet>565</a> ),
-<a href=#566 id=566 data-nosnippet>566</a> ));
-<a href=#567 id=567 data-nosnippet>567</a> }
-<a href=#568 id=568 data-nosnippet>568</a>
-<a href=#569 id=569 data-nosnippet>569</a> <span class="kw">let </span>value = i128_from_be_bytes(bytes).ok_or_else(|| {
-<a href=#570 id=570 data-nosnippet>570</a> Error::new(
-<a href=#571 id=571 data-nosnippet>571</a> ErrorKind::DataInvalid,
-<a href=#572 id=572 data-nosnippet>572</a> <span class="macro">format!</span>(<span class="string">"Can't convert bytes to i128: {bytes:?}"</span>),
-<a href=#573 id=573 data-nosnippet>573</a> )
-<a href=#574 id=574 data-nosnippet>574</a> })<span class="question-mark">?</span>;
-<a href=#575 id=575 data-nosnippet>575</a> <span class="prelude-val">Ok</span>(Datum::new(
-<a href=#576 id=576 data-nosnippet>576</a> field_type.clone(),
-<a href=#577 id=577 data-nosnippet>577</a> PrimitiveLiteral::Int128(value),
-<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>visit_inequality(
-<a href=#582 id=582 data-nosnippet>582</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#583 id=583 data-nosnippet>583</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#584 id=584 data-nosnippet>584</a> datum: <span class="kw-2">&</span>Datum,
-<a href=#585 id=585 data-nosnippet>585</a> cmp_fn: <span class="kw">fn</span>(<span class="kw-2">&</span>Datum, <span class="kw-2">&</span>Datum) -> bool,
-<a href=#586 id=586 data-nosnippet>586</a> use_lower_bound: bool,
-<a href=#587 id=587 data-nosnippet>587</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#588 id=588 data-nosnippet>588</a> <span class="kw">let </span>field_id = reference.field().id;
-<a href=#589 id=589 data-nosnippet>589</a>
-<a href=#590 id=590 data-nosnippet>590</a> <span class="self">self</span>.calc_row_selection(
-<a href=#591 id=591 data-nosnippet>591</a> field_id,
-<a href=#592 id=592 data-nosnippet>592</a> |min, max, null_count| {
-<a href=#593 id=593 data-nosnippet>593</a> <span class="kw">if </span><span class="macro">matches!</span>(null_count, PageNullCount::AllNull) {
-<a href=#594 id=594 data-nosnippet>594</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#595 id=595 data-nosnippet>595</a> }
-<a href=#596 id=596 data-nosnippet>596</a>
-<a href=#597 id=597 data-nosnippet>597</a> <span class="kw">if </span>datum.is_nan() {
-<a href=#598 id=598 data-nosnippet>598</a> <span class="comment">// NaN indicates unreliable bounds.
-<a href=#599 id=599 data-nosnippet>599</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
-<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> <span class="kw">let </span>bound = <span class="kw">if </span>use_lower_bound { min } <span class="kw">else </span>{ max };
-<a href=#603 id=603 data-nosnippet>603</a>
-<a href=#604 id=604 data-nosnippet>604</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(bound) = bound {
-<a href=#605 id=605 data-nosnippet>605</a> <span class="kw">if </span>cmp_fn(<span class="kw-2">&</span>bound, datum) {
-<a href=#606 id=606 data-nosnippet>606</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
-<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">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#610 id=610 data-nosnippet>610</a> }
-<a href=#611 id=611 data-nosnippet>611</a>
-<a href=#612 id=612 data-nosnippet>612</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
-<a href=#613 id=613 data-nosnippet>613</a> },
-<a href=#614 id=614 data-nosnippet>614</a> MissingColBehavior::MightMatch,
-<a href=#615 id=615 data-nosnippet>615</a> )
-<a href=#616 id=616 data-nosnippet>616</a> }
-<a href=#617 id=617 data-nosnippet>617</a>}
-<a href=#618 id=618 data-nosnippet>618</a>
-<a href=#619 id=619 data-nosnippet>619</a><span class="kw">impl </span>BoundPredicateVisitor <span class="kw">for </span>PageIndexEvaluator<<span class="lifetime">'_</span>> {
-<a href=#620 id=620 data-nosnippet>620</a> <span class="kw">type </span>T = RowSelection;
-<a href=#621 id=621 data-nosnippet>621</a>
-<a href=#622 id=622 data-nosnippet>622</a> <span class="kw">fn </span>always_true(<span class="kw-2">&mut </span><span class="self">self</span>) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#623 id=623 data-nosnippet>623</a> <span class="self">self</span>.select_all_rows()
-<a href=#624 id=624 data-nosnippet>624</a> }
-<a href=#625 id=625 data-nosnippet>625</a>
-<a href=#626 id=626 data-nosnippet>626</a> <span class="kw">fn </span>always_false(<span class="kw-2">&mut </span><span class="self">self</span>) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#627 id=627 data-nosnippet>627</a> <span class="self">self</span>.skip_all_rows()
-<a href=#628 id=628 data-nosnippet>628</a> }
-<a href=#629 id=629 data-nosnippet>629</a>
-<a href=#630 id=630 data-nosnippet>630</a> <span class="kw">fn </span>and(<span class="kw-2">&mut </span><span class="self">self</span>, lhs: RowSelection, rhs: RowSelection) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#631 id=631 data-nosnippet>631</a> <span class="prelude-val">Ok</span>(lhs.intersection(<span class="kw-2">&</span>rhs))
-<a href=#632 id=632 data-nosnippet>632</a> }
-<a href=#633 id=633 data-nosnippet>633</a>
-<a href=#634 id=634 data-nosnippet>634</a> <span class="kw">fn </span>or(<span class="kw-2">&mut </span><span class="self">self</span>, lhs: RowSelection, rhs: RowSelection) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#635 id=635 data-nosnippet>635</a> <span class="prelude-val">Ok</span>(lhs.union(<span class="kw-2">&</span>rhs))
-<a href=#636 id=636 data-nosnippet>636</a> }
-<a href=#637 id=637 data-nosnippet>637</a>
-<a href=#638 id=638 data-nosnippet>638</a> <span class="kw">fn </span>not(<span class="kw-2">&mut </span><span class="self">self</span>, <span class="kw">_</span>: RowSelection) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#639 id=639 data-nosnippet>639</a> <span class="prelude-val">Err</span>(Error::new(
-<a href=#640 id=640 data-nosnippet>640</a> ErrorKind::Unexpected,
-<a href=#641 id=641 data-nosnippet>641</a> <span class="string">"NOT unsupported at this point. NOT-rewrite should be performed first"</span>,
-<a href=#642 id=642 data-nosnippet>642</a> ))
-<a href=#643 id=643 data-nosnippet>643</a> }
-<a href=#644 id=644 data-nosnippet>644</a>
-<a href=#645 id=645 data-nosnippet>645</a> <span class="kw">fn </span>is_null(
-<a href=#646 id=646 data-nosnippet>646</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#647 id=647 data-nosnippet>647</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#648 id=648 data-nosnippet>648</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#649 id=649 data-nosnippet>649</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#650 id=650 data-nosnippet>650</a> <span class="kw">let </span>field_id = reference.field().id;
-<a href=#651 id=651 data-nosnippet>651</a>
-<a href=#652 id=652 data-nosnippet>652</a> <span class="self">self</span>.calc_row_selection(
-<a href=#653 id=653 data-nosnippet>653</a> field_id,
-<a href=#654 id=654 data-nosnippet>654</a> |_max, _min, null_count| <span class="prelude-val">Ok</span>(!<span class="macro">matches!</span>(null_count, PageNullCount::NoneNull)),
-<a href=#655 id=655 data-nosnippet>655</a> MissingColBehavior::MightMatch,
-<a href=#656 id=656 data-nosnippet>656</a> )
-<a href=#657 id=657 data-nosnippet>657</a> }
-<a href=#658 id=658 data-nosnippet>658</a>
-<a href=#659 id=659 data-nosnippet>659</a> <span class="kw">fn </span>not_null(
-<a href=#660 id=660 data-nosnippet>660</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#661 id=661 data-nosnippet>661</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#662 id=662 data-nosnippet>662</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#663 id=663 data-nosnippet>663</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#664 id=664 data-nosnippet>664</a> <span class="kw">let </span>field_id = reference.field().id;
-<a href=#665 id=665 data-nosnippet>665</a>
-<a href=#666 id=666 data-nosnippet>666</a> <span class="self">self</span>.calc_row_selection(
-<a href=#667 id=667 data-nosnippet>667</a> field_id,
-<a href=#668 id=668 data-nosnippet>668</a> |_max, _min, null_count| <span class="prelude-val">Ok</span>(!<span class="macro">matches!</span>(null_count, PageNullCount::AllNull)),
-<a href=#669 id=669 data-nosnippet>669</a> MissingColBehavior::CantMatch,
-<a href=#670 id=670 data-nosnippet>670</a> )
-<a href=#671 id=671 data-nosnippet>671</a> }
-<a href=#672 id=672 data-nosnippet>672</a>
-<a href=#673 id=673 data-nosnippet>673</a> <span class="kw">fn </span>is_nan(
-<a href=#674 id=674 data-nosnippet>674</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#675 id=675 data-nosnippet>675</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#676 id=676 data-nosnippet>676</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#677 id=677 data-nosnippet>677</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#678 id=678 data-nosnippet>678</a> <span class="comment">// NaN counts not present in ColumnChunkMetadata Statistics.
-<a href=#679 id=679 data-nosnippet>679</a> // Only float columns can be NaN.
-<a href=#680 id=680 data-nosnippet>680</a> </span><span class="kw">if </span>reference.field().field_type.is_floating_type() {
-<a href=#681 id=681 data-nosnippet>681</a> <span class="self">self</span>.select_all_rows()
-<a href=#682 id=682 data-nosnippet>682</a> } <span class="kw">else </span>{
-<a href=#683 id=683 data-nosnippet>683</a> <span class="self">self</span>.skip_all_rows()
-<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>
-<a href=#687 id=687 data-nosnippet>687</a> <span class="kw">fn </span>not_nan(
-<a href=#688 id=688 data-nosnippet>688</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#689 id=689 data-nosnippet>689</a> _reference: <span class="kw-2">&</span>BoundReference,
-<a href=#690 id=690 data-nosnippet>690</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#691 id=691 data-nosnippet>691</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#692 id=692 data-nosnippet>692</a> <span class="comment">// NaN counts not present in ColumnChunkMetadata Statistics
-<a href=#693 id=693 data-nosnippet>693</a> </span><span class="self">self</span>.select_all_rows()
-<a href=#694 id=694 data-nosnippet>694</a> }
-<a href=#695 id=695 data-nosnippet>695</a>
-<a href=#696 id=696 data-nosnippet>696</a> <span class="kw">fn </span>less_than(
-<a href=#697 id=697 data-nosnippet>697</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#698 id=698 data-nosnippet>698</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#699 id=699 data-nosnippet>699</a> datum: <span class="kw-2">&</span>Datum,
-<a href=#700 id=700 data-nosnippet>700</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#701 id=701 data-nosnippet>701</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#702 id=702 data-nosnippet>702</a> <span class="self">self</span>.visit_inequality(reference, datum, PartialOrd::lt, <span class="bool-val">true</span>)
-<a href=#703 id=703 data-nosnippet>703</a> }
-<a href=#704 id=704 data-nosnippet>704</a>
-<a href=#705 id=705 data-nosnippet>705</a> <span class="kw">fn </span>less_than_or_eq(
-<a href=#706 id=706 data-nosnippet>706</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#707 id=707 data-nosnippet>707</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#708 id=708 data-nosnippet>708</a> datum: <span class="kw-2">&</span>Datum,
-<a href=#709 id=709 data-nosnippet>709</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#710 id=710 data-nosnippet>710</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#711 id=711 data-nosnippet>711</a> <span class="self">self</span>.visit_inequality(reference, datum, PartialOrd::le, <span class="bool-val">true</span>)
-<a href=#712 id=712 data-nosnippet>712</a> }
-<a href=#713 id=713 data-nosnippet>713</a>
-<a href=#714 id=714 data-nosnippet>714</a> <span class="kw">fn </span>greater_than(
-<a href=#715 id=715 data-nosnippet>715</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#716 id=716 data-nosnippet>716</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#717 id=717 data-nosnippet>717</a> datum: <span class="kw-2">&</span>Datum,
-<a href=#718 id=718 data-nosnippet>718</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#719 id=719 data-nosnippet>719</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#720 id=720 data-nosnippet>720</a> <span class="self">self</span>.visit_inequality(reference, datum, PartialOrd::gt, <span class="bool-val">false</span>)
-<a href=#721 id=721 data-nosnippet>721</a> }
-<a href=#722 id=722 data-nosnippet>722</a>
-<a href=#723 id=723 data-nosnippet>723</a> <span class="kw">fn </span>greater_than_or_eq(
-<a href=#724 id=724 data-nosnippet>724</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#725 id=725 data-nosnippet>725</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#726 id=726 data-nosnippet>726</a> datum: <span class="kw-2">&</span>Datum,
-<a href=#727 id=727 data-nosnippet>727</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#728 id=728 data-nosnippet>728</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#729 id=729 data-nosnippet>729</a> <span class="self">self</span>.visit_inequality(reference, datum, PartialOrd::ge, <span class="bool-val">false</span>)
-<a href=#730 id=730 data-nosnippet>730</a> }
-<a href=#731 id=731 data-nosnippet>731</a>
-<a href=#732 id=732 data-nosnippet>732</a> <span class="kw">fn </span>eq(
-<a href=#733 id=733 data-nosnippet>733</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#734 id=734 data-nosnippet>734</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#735 id=735 data-nosnippet>735</a> datum: <span class="kw-2">&</span>Datum,
-<a href=#736 id=736 data-nosnippet>736</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#737 id=737 data-nosnippet>737</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#738 id=738 data-nosnippet>738</a> <span class="kw">let </span>field_id = reference.field().id;
-<a href=#739 id=739 data-nosnippet>739</a>
-<a href=#740 id=740 data-nosnippet>740</a> <span class="self">self</span>.calc_row_selection(
-<a href=#741 id=741 data-nosnippet>741</a> field_id,
-<a href=#742 id=742 data-nosnippet>742</a> |min, max, nulls| {
-<a href=#743 id=743 data-nosnippet>743</a> <span class="kw">if </span><span class="macro">matches!</span>(nulls, PageNullCount::AllNull) {
+<a href=#441 id=441 data-nosnippet>441</a> <span class="comment">// An undecodable bound (only a non-conforming writer
+<a href=#442 id=442 data-nosnippet>442</a> // produces one, see fixed_len_byte_array_decimal_bound_to_datum)
+<a href=#443 id=443 data-nosnippet>443</a> // means this column's page index can't be trusted, so skip
+<a href=#444 id=444 data-nosnippet>444</a> // pruning for the whole column rather than abort the scan.
+<a href=#445 id=445 data-nosnippet>445</a> </span><span class="kw">let </span>decode = |val: <span class="kw-2">&</span>[u8]| {
+<a href=#446 id=446 data-nosnippet>446</a> <span class="self">Self</span>::fixed_len_byte_array_decimal_bound_to_datum(
+<a href=#447 id=447 data-nosnippet>447</a> field_type,
+<a href=#448 id=448 data-nosnippet>448</a> val,
+<a href=#449 id=449 data-nosnippet>449</a> type_length,
+<a href=#450 id=450 data-nosnippet>450</a> )
+<a href=#451 id=451 data-nosnippet>451</a> };
+<a href=#452 id=452 data-nosnippet>452</a> <span class="kw">let </span>(min, max) = <span class="kw">match </span>(
+<a href=#453 id=453 data-nosnippet>453</a> min.map(decode).transpose(),
+<a href=#454 id=454 data-nosnippet>454</a> max.map(decode).transpose(),
+<a href=#455 id=455 data-nosnippet>455</a> ) {
+<a href=#456 id=456 data-nosnippet>456</a> (<span class="prelude-val">Ok</span>(min), <span class="prelude-val">Ok</span>(max)) => (min, max),
+<a href=#457 id=457 data-nosnippet>457</a> (<span class="prelude-val">Err</span>(err), <span class="kw">_</span>) | (<span class="kw">_</span>, <span class="prelude-val">Err</span>(err)) => {
+<a href=#458 id=458 data-nosnippet>458</a> <span class="macro">tracing::debug!</span>(
+<a href=#459 id=459 data-nosnippet>459</a> field_id,
+<a href=#460 id=460 data-nosnippet>460</a> %field_type,
+<a href=#461 id=461 data-nosnippet>461</a> page_index = i,
+<a href=#462 id=462 data-nosnippet>462</a> %err,
+<a href=#463 id=463 data-nosnippet>463</a> <span class="string">"Skipping page-index pruning: undecodable FIXED_LEN_BYTE_ARRAY decimal page bound"
+<a href=#464 id=464 data-nosnippet>464</a> </span>);
+<a href=#465 id=465 data-nosnippet>465</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>);
+<a href=#466 id=466 data-nosnippet>466</a> }
+<a href=#467 id=467 data-nosnippet>467</a> };
+<a href=#468 id=468 data-nosnippet>468</a>
+<a href=#469 id=469 data-nosnippet>469</a> page_filter.push(predicate(
+<a href=#470 id=470 data-nosnippet>470</a> min,
+<a href=#471 id=471 data-nosnippet>471</a> max,
+<a href=#472 id=472 data-nosnippet>472</a> PageNullCount::from_row_and_null_counts(row_count, idx.null_count(i)),
+<a href=#473 id=473 data-nosnippet>473</a> )<span class="question-mark">?</span>);
+<a href=#474 id=474 data-nosnippet>474</a> }
+<a href=#475 id=475 data-nosnippet>475</a>
+<a href=#476 id=476 data-nosnippet>476</a> <span class="prelude-val">Ok</span>(page_filter)
+<a href=#477 id=477 data-nosnippet>477</a> }
+<a href=#478 id=478 data-nosnippet>478</a> <span class="comment">// Column index types we can't interpret: skip page pruning rather
+<a href=#479 id=479 data-nosnippet>479</a> // than abort the scan. Row-group filtering and the Arrow row filter
+<a href=#480 id=480 data-nosnippet>480</a> // still apply the predicate.
+<a href=#481 id=481 data-nosnippet>481</a> </span>ColumnIndexMetaData::INT96(<span class="kw">_</span>) => {
+<a href=#482 id=482 data-nosnippet>482</a> <span class="macro">tracing::debug!</span>(
+<a href=#483 id=483 data-nosnippet>483</a> field_id,
+<a href=#484 id=484 data-nosnippet>484</a> %field_type,
+<a href=#485 id=485 data-nosnippet>485</a> <span class="string">"Skipping page-index pruning: unsupported INT96 column index"
+<a href=#486 id=486 data-nosnippet>486</a> </span>);
+<a href=#487 id=487 data-nosnippet>487</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="prelude-val">None</span>);
+<a href=#488 id=488 data-nosnippet>488</a> }
+<a href=#489 id=489 data-nosnippet>489</a> };
+<a href=#490 id=490 data-nosnippet>490</a>
+<a href=#491 id=491 data-nosnippet>491</a> <span class="prelude-val">Ok</span>(<span class="prelude-val">Some</span>(result<span class="question-mark">?</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> <span class="doccomment">/// Converts an `INT32` page bound into a [`Datum`] according to the field's
+<a href=#495 id=495 data-nosnippet>495</a> /// primitive type.
+<a href=#496 id=496 data-nosnippet>496</a> </span><span class="kw">fn </span>int32_bound_to_datum(field_type: <span class="kw-2">&</span>PrimitiveType, val: i32) -> Datum {
+<a href=#497 id=497 data-nosnippet>497</a> <span class="kw">match </span>field_type {
+<a href=#498 id=498 data-nosnippet>498</a> PrimitiveType::Decimal { .. } => Datum::new(
+<a href=#499 id=499 data-nosnippet>499</a> field_type.clone(),
+<a href=#500 id=500 data-nosnippet>500</a> PrimitiveLiteral::Int128(i128::from(val)),
+<a href=#501 id=501 data-nosnippet>501</a> ),
+<a href=#502 id=502 data-nosnippet>502</a> <span class="kw">_ </span>=> Datum::new(field_type.clone(), PrimitiveLiteral::Int(val)),
+<a href=#503 id=503 data-nosnippet>503</a> }
+<a href=#504 id=504 data-nosnippet>504</a> }
+<a href=#505 id=505 data-nosnippet>505</a>
+<a href=#506 id=506 data-nosnippet>506</a> <span class="doccomment">/// Converts an `INT64` page bound into a [`Datum`] according to the field's
+<a href=#507 id=507 data-nosnippet>507</a> /// primitive type.
+<a href=#508 id=508 data-nosnippet>508</a> </span><span class="kw">fn </span>int64_bound_to_datum(field_type: <span class="kw-2">&</span>PrimitiveType, val: i64) -> Datum {
+<a href=#509 id=509 data-nosnippet>509</a> <span class="kw">match </span>field_type {
+<a href=#510 id=510 data-nosnippet>510</a> PrimitiveType::Decimal { .. } => Datum::new(
+<a href=#511 id=511 data-nosnippet>511</a> field_type.clone(),
+<a href=#512 id=512 data-nosnippet>512</a> PrimitiveLiteral::Int128(i128::from(val)),
+<a href=#513 id=513 data-nosnippet>513</a> ),
+<a href=#514 id=514 data-nosnippet>514</a> <span class="kw">_ </span>=> Datum::new(field_type.clone(), PrimitiveLiteral::Long(val)),
+<a href=#515 id=515 data-nosnippet>515</a> }
+<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="doccomment">/// Converts a `BYTE_ARRAY` page bound into a [`Datum`] according to the
+<a href=#519 id=519 data-nosnippet>519</a> /// field's primitive type. Parquet stores Iceberg `string` and `binary`
+<a href=#520 id=520 data-nosnippet>520</a> /// bounds as `BYTE_ARRAY`.
+<a href=#521 id=521 data-nosnippet>521</a> </span><span class="kw">fn </span>byte_array_bound_to_datum(field_type: <span class="kw-2">&</span>PrimitiveType, bytes: <span class="kw-2">&</span>[u8]) -> <span class="prelude-ty">Result</span><Datum> {
+<a href=#522 id=522 data-nosnippet>522</a> <span class="kw">match </span>field_type {
+<a href=#523 id=523 data-nosnippet>523</a> PrimitiveType::String => {
+<a href=#524 id=524 data-nosnippet>524</a> <span class="kw">let </span>value = std::str::from_utf8(bytes).map_err(|err| {
+<a href=#525 id=525 data-nosnippet>525</a> Error::new(ErrorKind::DataInvalid, <span class="string">"Invalid UTF-8 in string page bound"</span>)
+<a href=#526 id=526 data-nosnippet>526</a> .with_source(err)
+<a href=#527 id=527 data-nosnippet>527</a> })<span class="question-mark">?</span>;
+<a href=#528 id=528 data-nosnippet>528</a> <span class="prelude-val">Ok</span>(Datum::string(value))
+<a href=#529 id=529 data-nosnippet>529</a> }
+<a href=#530 id=530 data-nosnippet>530</a> PrimitiveType::Binary => <span class="prelude-val">Ok</span>(Datum::binary(bytes.to_vec())),
+<a href=#531 id=531 data-nosnippet>531</a> <span class="comment">// Defensive: production callers only invoke this for String/Binary.
+<a href=#532 id=532 data-nosnippet>532</a> </span><span class="kw">_ </span>=> <span class="prelude-val">Err</span>(Error::new(
+<a href=#533 id=533 data-nosnippet>533</a> ErrorKind::Unexpected,
+<a href=#534 id=534 data-nosnippet>534</a> <span class="macro">format!</span>(<span class="string">"Unsupported primitive type for BYTE_ARRAY page bound: {field_type}"</span>),
+<a href=#535 id=535 data-nosnippet>535</a> )),
+<a href=#536 id=536 data-nosnippet>536</a> }
+<a href=#537 id=537 data-nosnippet>537</a> }
+<a href=#538 id=538 data-nosnippet>538</a>
+<a href=#539 id=539 data-nosnippet>539</a> <span class="doccomment">/// Converts a `FIXED_LEN_BYTE_ARRAY` decimal page bound into a [`Datum`].
+<a href=#540 id=540 data-nosnippet>540</a> /// Parquet stores Iceberg decimals with precision > 18 as big-endian
+<a href=#541 id=541 data-nosnippet>541</a> /// two's-complement `FIXED_LEN_BYTE_ARRAY` of width `type_length`.
+<a href=#542 id=542 data-nosnippet>542</a> ///
+<a href=#543 id=543 data-nosnippet>543</a> /// Returns an error for a bound whose width differs from `type_length`.
+<a href=#544 id=544 data-nosnippet>544</a> /// Conforming writers never truncate a decimal column index (a truncated
+<a href=#545 id=545 data-nosnippet>545</a> /// two's-complement prefix would not preserve numeric order), so a short
+<a href=#546 id=546 data-nosnippet>546</a> /// bound comes from a non-conforming writer and would decode to the wrong
+<a href=#547 id=547 data-nosnippet>547</a> /// value. Reject it rather than prune on a bad bound.
+<a href=#548 id=548 data-nosnippet>548</a> </span><span class="kw">fn </span>fixed_len_byte_array_decimal_bound_to_datum(
+<a href=#549 id=549 data-nosnippet>549</a> field_type: <span class="kw-2">&</span>PrimitiveType,
+<a href=#550 id=550 data-nosnippet>550</a> bytes: <span class="kw-2">&</span>[u8],
+<a href=#551 id=551 data-nosnippet>551</a> type_length: usize,
+<a href=#552 id=552 data-nosnippet>552</a> ) -> <span class="prelude-ty">Result</span><Datum> {
+<a href=#553 id=553 data-nosnippet>553</a> <span class="kw">if </span>bytes.len() != type_length {
+<a href=#554 id=554 data-nosnippet>554</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#555 id=555 data-nosnippet>555</a> ErrorKind::DataInvalid,
+<a href=#556 id=556 data-nosnippet>556</a> <span class="macro">format!</span>(
+<a href=#557 id=557 data-nosnippet>557</a> <span class="string">"Truncated FIXED_LEN_BYTE_ARRAY decimal page bound: {} bytes (expected {type_length})"</span>,
+<a href=#558 id=558 data-nosnippet>558</a> bytes.len()
+<a href=#559 id=559 data-nosnippet>559</a> ),
+<a href=#560 id=560 data-nosnippet>560</a> ));
+<a href=#561 id=561 data-nosnippet>561</a> }
+<a href=#562 id=562 data-nosnippet>562</a>
+<a href=#563 id=563 data-nosnippet>563</a> <span class="kw">let </span>value = i128_from_be_bytes(bytes).ok_or_else(|| {
+<a href=#564 id=564 data-nosnippet>564</a> Error::new(
+<a href=#565 id=565 data-nosnippet>565</a> ErrorKind::DataInvalid,
+<a href=#566 id=566 data-nosnippet>566</a> <span class="macro">format!</span>(<span class="string">"Can't convert bytes to i128: {bytes:?}"</span>),
+<a href=#567 id=567 data-nosnippet>567</a> )
+<a href=#568 id=568 data-nosnippet>568</a> })<span class="question-mark">?</span>;
+<a href=#569 id=569 data-nosnippet>569</a> <span class="prelude-val">Ok</span>(Datum::new(
+<a href=#570 id=570 data-nosnippet>570</a> field_type.clone(),
+<a href=#571 id=571 data-nosnippet>571</a> PrimitiveLiteral::Int128(value),
+<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="kw">fn </span>visit_inequality(
+<a href=#576 id=576 data-nosnippet>576</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#577 id=577 data-nosnippet>577</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#578 id=578 data-nosnippet>578</a> datum: <span class="kw-2">&</span>Datum,
+<a href=#579 id=579 data-nosnippet>579</a> cmp_fn: <span class="kw">fn</span>(<span class="kw-2">&</span>Datum, <span class="kw-2">&</span>Datum) -> bool,
+<a href=#580 id=580 data-nosnippet>580</a> use_lower_bound: bool,
+<a href=#581 id=581 data-nosnippet>581</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#582 id=582 data-nosnippet>582</a> <span class="kw">let </span>field_id = reference.field().id;
+<a href=#583 id=583 data-nosnippet>583</a>
+<a href=#584 id=584 data-nosnippet>584</a> <span class="self">self</span>.calc_row_selection(
+<a href=#585 id=585 data-nosnippet>585</a> field_id,
+<a href=#586 id=586 data-nosnippet>586</a> |min, max, null_count| {
+<a href=#587 id=587 data-nosnippet>587</a> <span class="kw">if </span><span class="macro">matches!</span>(null_count, PageNullCount::AllNull) {
+<a href=#588 id=588 data-nosnippet>588</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#589 id=589 data-nosnippet>589</a> }
+<a href=#590 id=590 data-nosnippet>590</a>
+<a href=#591 id=591 data-nosnippet>591</a> <span class="kw">if </span>datum.is_nan() {
+<a href=#592 id=592 data-nosnippet>592</a> <span class="comment">// NaN indicates unreliable bounds.
+<a href=#593 id=593 data-nosnippet>593</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
+<a href=#594 id=594 data-nosnippet>594</a> }
+<a href=#595 id=595 data-nosnippet>595</a>
+<a href=#596 id=596 data-nosnippet>596</a> <span class="kw">let </span>bound = <span class="kw">if </span>use_lower_bound { min } <span class="kw">else </span>{ max };
+<a href=#597 id=597 data-nosnippet>597</a>
+<a href=#598 id=598 data-nosnippet>598</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(bound) = bound {
+<a href=#599 id=599 data-nosnippet>599</a> <span class="kw">if </span>cmp_fn(<span class="kw-2">&</span>bound, datum) {
+<a href=#600 id=600 data-nosnippet>600</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
+<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="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#604 id=604 data-nosnippet>604</a> }
+<a href=#605 id=605 data-nosnippet>605</a>
+<a href=#606 id=606 data-nosnippet>606</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
+<a href=#607 id=607 data-nosnippet>607</a> },
+<a href=#608 id=608 data-nosnippet>608</a> MissingColBehavior::MightMatch,
+<a href=#609 id=609 data-nosnippet>609</a> )
+<a href=#610 id=610 data-nosnippet>610</a> }
+<a href=#611 id=611 data-nosnippet>611</a>}
+<a href=#612 id=612 data-nosnippet>612</a>
+<a href=#613 id=613 data-nosnippet>613</a><span class="kw">impl </span>BoundPredicateVisitor <span class="kw">for </span>PageIndexEvaluator<<span class="lifetime">'_</span>> {
+<a href=#614 id=614 data-nosnippet>614</a> <span class="kw">type </span>T = RowSelection;
+<a href=#615 id=615 data-nosnippet>615</a>
+<a href=#616 id=616 data-nosnippet>616</a> <span class="kw">fn </span>always_true(<span class="kw-2">&mut </span><span class="self">self</span>) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#617 id=617 data-nosnippet>617</a> <span class="self">self</span>.select_all_rows()
+<a href=#618 id=618 data-nosnippet>618</a> }
+<a href=#619 id=619 data-nosnippet>619</a>
+<a href=#620 id=620 data-nosnippet>620</a> <span class="kw">fn </span>always_false(<span class="kw-2">&mut </span><span class="self">self</span>) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#621 id=621 data-nosnippet>621</a> <span class="self">self</span>.skip_all_rows()
+<a href=#622 id=622 data-nosnippet>622</a> }
+<a href=#623 id=623 data-nosnippet>623</a>
+<a href=#624 id=624 data-nosnippet>624</a> <span class="kw">fn </span>and(<span class="kw-2">&mut </span><span class="self">self</span>, lhs: RowSelection, rhs: RowSelection) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#625 id=625 data-nosnippet>625</a> <span class="prelude-val">Ok</span>(lhs.intersection(<span class="kw-2">&</span>rhs))
+<a href=#626 id=626 data-nosnippet>626</a> }
+<a href=#627 id=627 data-nosnippet>627</a>
+<a href=#628 id=628 data-nosnippet>628</a> <span class="kw">fn </span>or(<span class="kw-2">&mut </span><span class="self">self</span>, lhs: RowSelection, rhs: RowSelection) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#629 id=629 data-nosnippet>629</a> <span class="prelude-val">Ok</span>(lhs.union(<span class="kw-2">&</span>rhs))
+<a href=#630 id=630 data-nosnippet>630</a> }
+<a href=#631 id=631 data-nosnippet>631</a>
+<a href=#632 id=632 data-nosnippet>632</a> <span class="kw">fn </span>not(<span class="kw-2">&mut </span><span class="self">self</span>, <span class="kw">_</span>: RowSelection) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#633 id=633 data-nosnippet>633</a> <span class="prelude-val">Err</span>(Error::new(
+<a href=#634 id=634 data-nosnippet>634</a> ErrorKind::Unexpected,
+<a href=#635 id=635 data-nosnippet>635</a> <span class="string">"NOT unsupported at this point. NOT-rewrite should be performed first"</span>,
+<a href=#636 id=636 data-nosnippet>636</a> ))
+<a href=#637 id=637 data-nosnippet>637</a> }
+<a href=#638 id=638 data-nosnippet>638</a>
+<a href=#639 id=639 data-nosnippet>639</a> <span class="kw">fn </span>is_null(
+<a href=#640 id=640 data-nosnippet>640</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#641 id=641 data-nosnippet>641</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#642 id=642 data-nosnippet>642</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#643 id=643 data-nosnippet>643</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#644 id=644 data-nosnippet>644</a> <span class="kw">let </span>field_id = reference.field().id;
+<a href=#645 id=645 data-nosnippet>645</a>
+<a href=#646 id=646 data-nosnippet>646</a> <span class="self">self</span>.calc_row_selection(
+<a href=#647 id=647 data-nosnippet>647</a> field_id,
+<a href=#648 id=648 data-nosnippet>648</a> |_max, _min, null_count| <span class="prelude-val">Ok</span>(!<span class="macro">matches!</span>(null_count, PageNullCount::NoneNull)),
+<a href=#649 id=649 data-nosnippet>649</a> MissingColBehavior::MightMatch,
+<a href=#650 id=650 data-nosnippet>650</a> )
+<a href=#651 id=651 data-nosnippet>651</a> }
+<a href=#652 id=652 data-nosnippet>652</a>
+<a href=#653 id=653 data-nosnippet>653</a> <span class="kw">fn </span>not_null(
+<a href=#654 id=654 data-nosnippet>654</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#655 id=655 data-nosnippet>655</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#656 id=656 data-nosnippet>656</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#657 id=657 data-nosnippet>657</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#658 id=658 data-nosnippet>658</a> <span class="kw">let </span>field_id = reference.field().id;
+<a href=#659 id=659 data-nosnippet>659</a>
+<a href=#660 id=660 data-nosnippet>660</a> <span class="self">self</span>.calc_row_selection(
+<a href=#661 id=661 data-nosnippet>661</a> field_id,
+<a href=#662 id=662 data-nosnippet>662</a> |_max, _min, null_count| <span class="prelude-val">Ok</span>(!<span class="macro">matches!</span>(null_count, PageNullCount::AllNull)),
+<a href=#663 id=663 data-nosnippet>663</a> MissingColBehavior::CantMatch,
+<a href=#664 id=664 data-nosnippet>664</a> )
+<a href=#665 id=665 data-nosnippet>665</a> }
+<a href=#666 id=666 data-nosnippet>666</a>
+<a href=#667 id=667 data-nosnippet>667</a> <span class="kw">fn </span>is_nan(
+<a href=#668 id=668 data-nosnippet>668</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#669 id=669 data-nosnippet>669</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#670 id=670 data-nosnippet>670</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#671 id=671 data-nosnippet>671</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#672 id=672 data-nosnippet>672</a> <span class="comment">// NaN counts not present in ColumnChunkMetadata Statistics.
+<a href=#673 id=673 data-nosnippet>673</a> // Only float columns can be NaN.
+<a href=#674 id=674 data-nosnippet>674</a> </span><span class="kw">if </span>reference.field().field_type.is_floating_type() {
+<a href=#675 id=675 data-nosnippet>675</a> <span class="self">self</span>.select_all_rows()
+<a href=#676 id=676 data-nosnippet>676</a> } <span class="kw">else </span>{
+<a href=#677 id=677 data-nosnippet>677</a> <span class="self">self</span>.skip_all_rows()
+<a href=#678 id=678 data-nosnippet>678</a> }
+<a href=#679 id=679 data-nosnippet>679</a> }
+<a href=#680 id=680 data-nosnippet>680</a>
+<a href=#681 id=681 data-nosnippet>681</a> <span class="kw">fn </span>not_nan(
+<a href=#682 id=682 data-nosnippet>682</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#683 id=683 data-nosnippet>683</a> _reference: <span class="kw-2">&</span>BoundReference,
+<a href=#684 id=684 data-nosnippet>684</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#685 id=685 data-nosnippet>685</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#686 id=686 data-nosnippet>686</a> <span class="comment">// NaN counts not present in ColumnChunkMetadata Statistics
+<a href=#687 id=687 data-nosnippet>687</a> </span><span class="self">self</span>.select_all_rows()
+<a href=#688 id=688 data-nosnippet>688</a> }
+<a href=#689 id=689 data-nosnippet>689</a>
+<a href=#690 id=690 data-nosnippet>690</a> <span class="kw">fn </span>less_than(
+<a href=#691 id=691 data-nosnippet>691</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#692 id=692 data-nosnippet>692</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#693 id=693 data-nosnippet>693</a> datum: <span class="kw-2">&</span>Datum,
+<a href=#694 id=694 data-nosnippet>694</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#695 id=695 data-nosnippet>695</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#696 id=696 data-nosnippet>696</a> <span class="self">self</span>.visit_inequality(reference, datum, PartialOrd::lt, <span class="bool-val">true</span>)
+<a href=#697 id=697 data-nosnippet>697</a> }
+<a href=#698 id=698 data-nosnippet>698</a>
+<a href=#699 id=699 data-nosnippet>699</a> <span class="kw">fn </span>less_than_or_eq(
+<a href=#700 id=700 data-nosnippet>700</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#701 id=701 data-nosnippet>701</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#702 id=702 data-nosnippet>702</a> datum: <span class="kw-2">&</span>Datum,
+<a href=#703 id=703 data-nosnippet>703</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#704 id=704 data-nosnippet>704</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#705 id=705 data-nosnippet>705</a> <span class="self">self</span>.visit_inequality(reference, datum, PartialOrd::le, <span class="bool-val">true</span>)
+<a href=#706 id=706 data-nosnippet>706</a> }
+<a href=#707 id=707 data-nosnippet>707</a>
+<a href=#708 id=708 data-nosnippet>708</a> <span class="kw">fn </span>greater_than(
+<a href=#709 id=709 data-nosnippet>709</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#710 id=710 data-nosnippet>710</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#711 id=711 data-nosnippet>711</a> datum: <span class="kw-2">&</span>Datum,
+<a href=#712 id=712 data-nosnippet>712</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#713 id=713 data-nosnippet>713</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#714 id=714 data-nosnippet>714</a> <span class="self">self</span>.visit_inequality(reference, datum, PartialOrd::gt, <span class="bool-val">false</span>)
+<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">fn </span>greater_than_or_eq(
+<a href=#718 id=718 data-nosnippet>718</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#719 id=719 data-nosnippet>719</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#720 id=720 data-nosnippet>720</a> datum: <span class="kw-2">&</span>Datum,
+<a href=#721 id=721 data-nosnippet>721</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#722 id=722 data-nosnippet>722</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#723 id=723 data-nosnippet>723</a> <span class="self">self</span>.visit_inequality(reference, datum, PartialOrd::ge, <span class="bool-val">false</span>)
+<a href=#724 id=724 data-nosnippet>724</a> }
+<a href=#725 id=725 data-nosnippet>725</a>
+<a href=#726 id=726 data-nosnippet>726</a> <span class="kw">fn </span>eq(
+<a href=#727 id=727 data-nosnippet>727</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#728 id=728 data-nosnippet>728</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#729 id=729 data-nosnippet>729</a> datum: <span class="kw-2">&</span>Datum,
+<a href=#730 id=730 data-nosnippet>730</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#731 id=731 data-nosnippet>731</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#732 id=732 data-nosnippet>732</a> <span class="kw">let </span>field_id = reference.field().id;
+<a href=#733 id=733 data-nosnippet>733</a>
+<a href=#734 id=734 data-nosnippet>734</a> <span class="self">self</span>.calc_row_selection(
+<a href=#735 id=735 data-nosnippet>735</a> field_id,
+<a href=#736 id=736 data-nosnippet>736</a> |min, max, nulls| {
+<a href=#737 id=737 data-nosnippet>737</a> <span class="kw">if </span><span class="macro">matches!</span>(nulls, PageNullCount::AllNull) {
+<a href=#738 id=738 data-nosnippet>738</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#739 id=739 data-nosnippet>739</a> }
+<a href=#740 id=740 data-nosnippet>740</a>
+<a href=#741 id=741 data-nosnippet>741</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(min) = min
+<a href=#742 id=742 data-nosnippet>742</a> && min.gt(datum)
+<a href=#743 id=743 data-nosnippet>743</a> {
<a href=#744 id=744 data-nosnippet>744</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
<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">if let </span><span class="prelude-val">Some</span>(min) = min
-<a href=#748 id=748 data-nosnippet>748</a> && min.gt(datum)
+<a href=#747 id=747 data-nosnippet>747</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(max) = max
+<a href=#748 id=748 data-nosnippet>748</a> && max.lt(datum)
<a href=#749 id=749 data-nosnippet>749</a> {
<a href=#750 id=750 data-nosnippet>750</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
<a href=#751 id=751 data-nosnippet>751</a> }
<a href=#752 id=752 data-nosnippet>752</a>
-<a href=#753 id=753 data-nosnippet>753</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(max) = max
-<a href=#754 id=754 data-nosnippet>754</a> && max.lt(datum)
-<a href=#755 id=755 data-nosnippet>755</a> {
-<a href=#756 id=756 data-nosnippet>756</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#757 id=757 data-nosnippet>757</a> }
+<a href=#753 id=753 data-nosnippet>753</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
+<a href=#754 id=754 data-nosnippet>754</a> },
+<a href=#755 id=755 data-nosnippet>755</a> MissingColBehavior::CantMatch,
+<a href=#756 id=756 data-nosnippet>756</a> )
+<a href=#757 id=757 data-nosnippet>757</a> }
<a href=#758 id=758 data-nosnippet>758</a>
-<a href=#759 id=759 data-nosnippet>759</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
-<a href=#760 id=760 data-nosnippet>760</a> },
-<a href=#761 id=761 data-nosnippet>761</a> MissingColBehavior::CantMatch,
-<a href=#762 id=762 data-nosnippet>762</a> )
-<a href=#763 id=763 data-nosnippet>763</a> }
-<a href=#764 id=764 data-nosnippet>764</a>
-<a href=#765 id=765 data-nosnippet>765</a> <span class="kw">fn </span>not_eq(
-<a href=#766 id=766 data-nosnippet>766</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#767 id=767 data-nosnippet>767</a> _reference: <span class="kw-2">&</span>BoundReference,
-<a href=#768 id=768 data-nosnippet>768</a> _datum: <span class="kw-2">&</span>Datum,
-<a href=#769 id=769 data-nosnippet>769</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#770 id=770 data-nosnippet>770</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#771 id=771 data-nosnippet>771</a> <span class="comment">// Because the bounds are not necessarily a min or max value,
-<a href=#772 id=772 data-nosnippet>772</a> // this cannot be answered using them. notEq(col, X) with (X, Y)
-<a href=#773 id=773 data-nosnippet>773</a> // doesn't guarantee that X is a value in col.
-<a href=#774 id=774 data-nosnippet>774</a> </span><span class="self">self</span>.select_all_rows()
-<a href=#775 id=775 data-nosnippet>775</a> }
-<a href=#776 id=776 data-nosnippet>776</a>
-<a href=#777 id=777 data-nosnippet>777</a> <span class="kw">fn </span>starts_with(
-<a href=#778 id=778 data-nosnippet>778</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#779 id=779 data-nosnippet>779</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#780 id=780 data-nosnippet>780</a> datum: <span class="kw-2">&</span>Datum,
-<a href=#781 id=781 data-nosnippet>781</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#782 id=782 data-nosnippet>782</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#783 id=783 data-nosnippet>783</a> <span class="kw">let </span>field_id = reference.field().id;
-<a href=#784 id=784 data-nosnippet>784</a>
-<a href=#785 id=785 data-nosnippet>785</a> <span class="kw">let </span>PrimitiveLiteral::String(datum) = datum.literal() <span class="kw">else </span>{
-<a href=#786 id=786 data-nosnippet>786</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#787 id=787 data-nosnippet>787</a> ErrorKind::Unexpected,
-<a href=#788 id=788 data-nosnippet>788</a> <span class="string">"Cannot use StartsWith operator on non-string values"</span>,
-<a href=#789 id=789 data-nosnippet>789</a> ));
-<a href=#790 id=790 data-nosnippet>790</a> };
-<a href=#791 id=791 data-nosnippet>791</a>
-<a href=#792 id=792 data-nosnippet>792</a> <span class="self">self</span>.calc_row_selection(
-<a href=#793 id=793 data-nosnippet>793</a> field_id,
-<a href=#794 id=794 data-nosnippet>794</a> |min, max, nulls| {
-<a href=#795 id=795 data-nosnippet>795</a> <span class="kw">if </span><span class="macro">matches!</span>(nulls, PageNullCount::AllNull) {
-<a href=#796 id=796 data-nosnippet>796</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#797 id=797 data-nosnippet>797</a> }
-<a href=#798 id=798 data-nosnippet>798</a>
-<a href=#799 id=799 data-nosnippet>799</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(lower_bound) = min {
-<a href=#800 id=800 data-nosnippet>800</a> <span class="kw">let </span>PrimitiveLiteral::String(lower_bound) = lower_bound.literal() <span class="kw">else </span>{
-<a href=#801 id=801 data-nosnippet>801</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#802 id=802 data-nosnippet>802</a> ErrorKind::Unexpected,
-<a href=#803 id=803 data-nosnippet>803</a> <span class="string">"Cannot use StartsWith operator on non-string lower_bound value"</span>,
-<a href=#804 id=804 data-nosnippet>804</a> ));
-<a href=#805 id=805 data-nosnippet>805</a> };
-<a href=#806 id=806 data-nosnippet>806</a>
-<a href=#807 id=807 data-nosnippet>807</a> <span class="kw">let </span>prefix_length = lower_bound.chars().count().min(datum.chars().count());
-<a href=#808 id=808 data-nosnippet>808</a>
-<a href=#809 id=809 data-nosnippet>809</a> <span class="comment">// truncate lower bound so that its length
-<a href=#810 id=810 data-nosnippet>810</a> // is not greater than the length of prefix
-<a href=#811 id=811 data-nosnippet>811</a> </span><span class="kw">let </span>truncated_lower_bound =
-<a href=#812 id=812 data-nosnippet>812</a> lower_bound.chars().take(prefix_length).collect::<String>();
-<a href=#813 id=813 data-nosnippet>813</a> <span class="kw">if </span>datum < <span class="kw-2">&</span>truncated_lower_bound {
-<a href=#814 id=814 data-nosnippet>814</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#815 id=815 data-nosnippet>815</a> }
-<a href=#816 id=816 data-nosnippet>816</a> }
-<a href=#817 id=817 data-nosnippet>817</a>
-<a href=#818 id=818 data-nosnippet>818</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(upper_bound) = max {
-<a href=#819 id=819 data-nosnippet>819</a> <span class="kw">let </span>PrimitiveLiteral::String(upper_bound) = upper_bound.literal() <span class="kw">else </span>{
-<a href=#820 id=820 data-nosnippet>820</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#821 id=821 data-nosnippet>821</a> ErrorKind::Unexpected,
-<a href=#822 id=822 data-nosnippet>822</a> <span class="string">"Cannot use StartsWith operator on non-string upper_bound value"</span>,
-<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>
-<a href=#826 id=826 data-nosnippet>826</a> <span class="kw">let </span>prefix_length = upper_bound.chars().count().min(datum.chars().count());
-<a href=#827 id=827 data-nosnippet>827</a>
-<a href=#828 id=828 data-nosnippet>828</a> <span class="comment">// truncate upper bound so that its length
-<a href=#829 id=829 data-nosnippet>829</a> // is not greater than the length of prefix
-<a href=#830 id=830 data-nosnippet>830</a> </span><span class="kw">let </span>truncated_upper_bound =
-<a href=#831 id=831 data-nosnippet>831</a> upper_bound.chars().take(prefix_length).collect::<String>();
-<a href=#832 id=832 data-nosnippet>832</a> <span class="kw">if </span>datum > <span class="kw-2">&</span>truncated_upper_bound {
-<a href=#833 id=833 data-nosnippet>833</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#834 id=834 data-nosnippet>834</a> }
-<a href=#835 id=835 data-nosnippet>835</a> }
+<a href=#759 id=759 data-nosnippet>759</a> <span class="kw">fn </span>not_eq(
+<a href=#760 id=760 data-nosnippet>760</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#761 id=761 data-nosnippet>761</a> _reference: <span class="kw-2">&</span>BoundReference,
+<a href=#762 id=762 data-nosnippet>762</a> _datum: <span class="kw-2">&</span>Datum,
+<a href=#763 id=763 data-nosnippet>763</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#764 id=764 data-nosnippet>764</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#765 id=765 data-nosnippet>765</a> <span class="comment">// Because the bounds are not necessarily a min or max value,
+<a href=#766 id=766 data-nosnippet>766</a> // this cannot be answered using them. notEq(col, X) with (X, Y)
+<a href=#767 id=767 data-nosnippet>767</a> // doesn't guarantee that X is a value in col.
+<a href=#768 id=768 data-nosnippet>768</a> </span><span class="self">self</span>.select_all_rows()
+<a href=#769 id=769 data-nosnippet>769</a> }
+<a href=#770 id=770 data-nosnippet>770</a>
+<a href=#771 id=771 data-nosnippet>771</a> <span class="kw">fn </span>starts_with(
+<a href=#772 id=772 data-nosnippet>772</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#773 id=773 data-nosnippet>773</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#774 id=774 data-nosnippet>774</a> datum: <span class="kw-2">&</span>Datum,
+<a href=#775 id=775 data-nosnippet>775</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#776 id=776 data-nosnippet>776</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#777 id=777 data-nosnippet>777</a> <span class="kw">let </span>field_id = reference.field().id;
+<a href=#778 id=778 data-nosnippet>778</a>
+<a href=#779 id=779 data-nosnippet>779</a> <span class="kw">let </span>PrimitiveLiteral::String(datum) = datum.literal() <span class="kw">else </span>{
+<a href=#780 id=780 data-nosnippet>780</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#781 id=781 data-nosnippet>781</a> ErrorKind::Unexpected,
+<a href=#782 id=782 data-nosnippet>782</a> <span class="string">"Cannot use StartsWith operator on non-string values"</span>,
+<a href=#783 id=783 data-nosnippet>783</a> ));
+<a href=#784 id=784 data-nosnippet>784</a> };
+<a href=#785 id=785 data-nosnippet>785</a>
+<a href=#786 id=786 data-nosnippet>786</a> <span class="self">self</span>.calc_row_selection(
+<a href=#787 id=787 data-nosnippet>787</a> field_id,
+<a href=#788 id=788 data-nosnippet>788</a> |min, max, nulls| {
+<a href=#789 id=789 data-nosnippet>789</a> <span class="kw">if </span><span class="macro">matches!</span>(nulls, PageNullCount::AllNull) {
+<a href=#790 id=790 data-nosnippet>790</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</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="kw">if let </span><span class="prelude-val">Some</span>(lower_bound) = min {
+<a href=#794 id=794 data-nosnippet>794</a> <span class="kw">let </span>PrimitiveLiteral::String(lower_bound) = lower_bound.literal() <span class="kw">else </span>{
+<a href=#795 id=795 data-nosnippet>795</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#796 id=796 data-nosnippet>796</a> ErrorKind::Unexpected,
+<a href=#797 id=797 data-nosnippet>797</a> <span class="string">"Cannot use StartsWith operator on non-string lower_bound value"</span>,
+<a href=#798 id=798 data-nosnippet>798</a> ));
+<a href=#799 id=799 data-nosnippet>799</a> };
+<a href=#800 id=800 data-nosnippet>800</a>
+<a href=#801 id=801 data-nosnippet>801</a> <span class="kw">let </span>prefix_length = lower_bound.chars().count().min(datum.chars().count());
+<a href=#802 id=802 data-nosnippet>802</a>
+<a href=#803 id=803 data-nosnippet>803</a> <span class="comment">// truncate lower bound so that its length
+<a href=#804 id=804 data-nosnippet>804</a> // is not greater than the length of prefix
+<a href=#805 id=805 data-nosnippet>805</a> </span><span class="kw">let </span>truncated_lower_bound =
+<a href=#806 id=806 data-nosnippet>806</a> lower_bound.chars().take(prefix_length).collect::<String>();
+<a href=#807 id=807 data-nosnippet>807</a> <span class="kw">if </span>datum < <span class="kw-2">&</span>truncated_lower_bound {
+<a href=#808 id=808 data-nosnippet>808</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#809 id=809 data-nosnippet>809</a> }
+<a href=#810 id=810 data-nosnippet>810</a> }
+<a href=#811 id=811 data-nosnippet>811</a>
+<a href=#812 id=812 data-nosnippet>812</a> <span class="kw">if let </span><span class="prelude-val">Some</span>(upper_bound) = max {
+<a href=#813 id=813 data-nosnippet>813</a> <span class="kw">let </span>PrimitiveLiteral::String(upper_bound) = upper_bound.literal() <span class="kw">else </span>{
+<a href=#814 id=814 data-nosnippet>814</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#815 id=815 data-nosnippet>815</a> ErrorKind::Unexpected,
+<a href=#816 id=816 data-nosnippet>816</a> <span class="string">"Cannot use StartsWith operator on non-string upper_bound value"</span>,
+<a href=#817 id=817 data-nosnippet>817</a> ));
+<a href=#818 id=818 data-nosnippet>818</a> };
+<a href=#819 id=819 data-nosnippet>819</a>
+<a href=#820 id=820 data-nosnippet>820</a> <span class="kw">let </span>prefix_length = upper_bound.chars().count().min(datum.chars().count());
+<a href=#821 id=821 data-nosnippet>821</a>
+<a href=#822 id=822 data-nosnippet>822</a> <span class="comment">// truncate upper bound so that its length
+<a href=#823 id=823 data-nosnippet>823</a> // is not greater than the length of prefix
+<a href=#824 id=824 data-nosnippet>824</a> </span><span class="kw">let </span>truncated_upper_bound =
+<a href=#825 id=825 data-nosnippet>825</a> upper_bound.chars().take(prefix_length).collect::<String>();
+<a href=#826 id=826 data-nosnippet>826</a> <span class="kw">if </span>datum > <span class="kw-2">&</span>truncated_upper_bound {
+<a href=#827 id=827 data-nosnippet>827</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#828 id=828 data-nosnippet>828</a> }
+<a href=#829 id=829 data-nosnippet>829</a> }
+<a href=#830 id=830 data-nosnippet>830</a>
+<a href=#831 id=831 data-nosnippet>831</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
+<a href=#832 id=832 data-nosnippet>832</a> },
+<a href=#833 id=833 data-nosnippet>833</a> MissingColBehavior::CantMatch,
+<a href=#834 id=834 data-nosnippet>834</a> )
+<a href=#835 id=835 data-nosnippet>835</a> }
<a href=#836 id=836 data-nosnippet>836</a>
-<a href=#837 id=837 data-nosnippet>837</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
-<a href=#838 id=838 data-nosnippet>838</a> },
-<a href=#839 id=839 data-nosnippet>839</a> MissingColBehavior::CantMatch,
-<a href=#840 id=840 data-nosnippet>840</a> )
-<a href=#841 id=841 data-nosnippet>841</a> }
-<a href=#842 id=842 data-nosnippet>842</a>
-<a href=#843 id=843 data-nosnippet>843</a> <span class="kw">fn </span>not_starts_with(
-<a href=#844 id=844 data-nosnippet>844</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#845 id=845 data-nosnippet>845</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#846 id=846 data-nosnippet>846</a> datum: <span class="kw-2">&</span>Datum,
-<a href=#847 id=847 data-nosnippet>847</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#848 id=848 data-nosnippet>848</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#849 id=849 data-nosnippet>849</a> <span class="kw">let </span>field_id = reference.field().id;
-<a href=#850 id=850 data-nosnippet>850</a>
-<a href=#851 id=851 data-nosnippet>851</a> <span class="comment">// notStartsWith will match unless all values must start with the prefix.
-<a href=#852 id=852 data-nosnippet>852</a> // This happens when the lower and upper bounds both start with the prefix.
-<a href=#853 id=853 data-nosnippet>853</a>
-<a href=#854 id=854 data-nosnippet>854</a> </span><span class="kw">let </span>PrimitiveLiteral::String(prefix) = datum.literal() <span class="kw">else </span>{
-<a href=#855 id=855 data-nosnippet>855</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#856 id=856 data-nosnippet>856</a> ErrorKind::Unexpected,
-<a href=#857 id=857 data-nosnippet>857</a> <span class="string">"Cannot use StartsWith operator on non-string values"</span>,
-<a href=#858 id=858 data-nosnippet>858</a> ));
-<a href=#859 id=859 data-nosnippet>859</a> };
-<a href=#860 id=860 data-nosnippet>860</a>
-<a href=#861 id=861 data-nosnippet>861</a> <span class="self">self</span>.calc_row_selection(
-<a href=#862 id=862 data-nosnippet>862</a> field_id,
-<a href=#863 id=863 data-nosnippet>863</a> |min, max, nulls| {
-<a href=#864 id=864 data-nosnippet>864</a> <span class="kw">if </span>!<span class="macro">matches!</span>(nulls, PageNullCount::NoneNull) {
-<a href=#865 id=865 data-nosnippet>865</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
-<a href=#866 id=866 data-nosnippet>866</a> }
-<a href=#867 id=867 data-nosnippet>867</a>
-<a href=#868 id=868 data-nosnippet>868</a> <span class="kw">let </span><span class="prelude-val">Some</span>(lower_bound) = min <span class="kw">else </span>{
-<a href=#869 id=869 data-nosnippet>869</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
-<a href=#870 id=870 data-nosnippet>870</a> };
-<a href=#871 id=871 data-nosnippet>871</a>
-<a href=#872 id=872 data-nosnippet>872</a> <span class="kw">let </span>PrimitiveLiteral::String(lower_bound_str) = lower_bound.literal() <span class="kw">else </span>{
-<a href=#873 id=873 data-nosnippet>873</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#874 id=874 data-nosnippet>874</a> ErrorKind::Unexpected,
-<a href=#875 id=875 data-nosnippet>875</a> <span class="string">"Cannot use NotStartsWith operator on non-string lower_bound value"</span>,
-<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>
-<a href=#879 id=879 data-nosnippet>879</a> <span class="kw">if </span>lower_bound_str < prefix {
-<a href=#880 id=880 data-nosnippet>880</a> <span class="comment">// if lower is shorter than the prefix then lower doesn't start with the prefix
-<a href=#881 id=881 data-nosnippet>881</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
-<a href=#882 id=882 data-nosnippet>882</a> }
-<a href=#883 id=883 data-nosnippet>883</a>
-<a href=#884 id=884 data-nosnippet>884</a> <span class="kw">let </span>prefix_len = prefix.chars().count();
-<a href=#885 id=885 data-nosnippet>885</a>
-<a href=#886 id=886 data-nosnippet>886</a> <span class="kw">if </span>lower_bound_str.chars().take(prefix_len).collect::<String>() == <span class="kw-2">*</span>prefix {
-<a href=#887 id=887 data-nosnippet>887</a> <span class="comment">// lower bound matches the prefix
-<a href=#888 id=888 data-nosnippet>888</a>
-<a href=#889 id=889 data-nosnippet>889</a> </span><span class="kw">let </span><span class="prelude-val">Some</span>(upper_bound) = max <span class="kw">else </span>{
-<a href=#890 id=890 data-nosnippet>890</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
-<a href=#891 id=891 data-nosnippet>891</a> };
-<a href=#892 id=892 data-nosnippet>892</a>
-<a href=#893 id=893 data-nosnippet>893</a> <span class="kw">let </span>PrimitiveLiteral::String(upper_bound) = upper_bound.literal() <span class="kw">else </span>{
-<a href=#894 id=894 data-nosnippet>894</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
-<a href=#895 id=895 data-nosnippet>895</a> ErrorKind::Unexpected,
-<a href=#896 id=896 data-nosnippet>896</a> <span class="string">"Cannot use NotStartsWith operator on non-string upper_bound value"</span>,
-<a href=#897 id=897 data-nosnippet>897</a> ));
-<a href=#898 id=898 data-nosnippet>898</a> };
-<a href=#899 id=899 data-nosnippet>899</a>
-<a href=#900 id=900 data-nosnippet>900</a> <span class="comment">// if upper is shorter than the prefix then upper can't start with the prefix
-<a href=#901 id=901 data-nosnippet>901</a> </span><span class="kw">if </span>upper_bound.chars().count() < prefix_len {
-<a href=#902 id=902 data-nosnippet>902</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
+<a href=#837 id=837 data-nosnippet>837</a> <span class="kw">fn </span>not_starts_with(
+<a href=#838 id=838 data-nosnippet>838</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#839 id=839 data-nosnippet>839</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#840 id=840 data-nosnippet>840</a> datum: <span class="kw-2">&</span>Datum,
+<a href=#841 id=841 data-nosnippet>841</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#842 id=842 data-nosnippet>842</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#843 id=843 data-nosnippet>843</a> <span class="kw">let </span>field_id = reference.field().id;
+<a href=#844 id=844 data-nosnippet>844</a>
+<a href=#845 id=845 data-nosnippet>845</a> <span class="comment">// notStartsWith will match unless all values must start with the prefix.
+<a href=#846 id=846 data-nosnippet>846</a> // This happens when the lower and upper bounds both start with the prefix.
+<a href=#847 id=847 data-nosnippet>847</a>
+<a href=#848 id=848 data-nosnippet>848</a> </span><span class="kw">let </span>PrimitiveLiteral::String(prefix) = datum.literal() <span class="kw">else </span>{
+<a href=#849 id=849 data-nosnippet>849</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#850 id=850 data-nosnippet>850</a> ErrorKind::Unexpected,
+<a href=#851 id=851 data-nosnippet>851</a> <span class="string">"Cannot use StartsWith operator on non-string values"</span>,
+<a href=#852 id=852 data-nosnippet>852</a> ));
+<a href=#853 id=853 data-nosnippet>853</a> };
+<a href=#854 id=854 data-nosnippet>854</a>
+<a href=#855 id=855 data-nosnippet>855</a> <span class="self">self</span>.calc_row_selection(
+<a href=#856 id=856 data-nosnippet>856</a> field_id,
+<a href=#857 id=857 data-nosnippet>857</a> |min, max, nulls| {
+<a href=#858 id=858 data-nosnippet>858</a> <span class="kw">if </span>!<span class="macro">matches!</span>(nulls, PageNullCount::NoneNull) {
+<a href=#859 id=859 data-nosnippet>859</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
+<a href=#860 id=860 data-nosnippet>860</a> }
+<a href=#861 id=861 data-nosnippet>861</a>
+<a href=#862 id=862 data-nosnippet>862</a> <span class="kw">let </span><span class="prelude-val">Some</span>(lower_bound) = min <span class="kw">else </span>{
+<a href=#863 id=863 data-nosnippet>863</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
+<a href=#864 id=864 data-nosnippet>864</a> };
+<a href=#865 id=865 data-nosnippet>865</a>
+<a href=#866 id=866 data-nosnippet>866</a> <span class="kw">let </span>PrimitiveLiteral::String(lower_bound_str) = lower_bound.literal() <span class="kw">else </span>{
+<a href=#867 id=867 data-nosnippet>867</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#868 id=868 data-nosnippet>868</a> ErrorKind::Unexpected,
+<a href=#869 id=869 data-nosnippet>869</a> <span class="string">"Cannot use NotStartsWith operator on non-string lower_bound value"</span>,
+<a href=#870 id=870 data-nosnippet>870</a> ));
+<a href=#871 id=871 data-nosnippet>871</a> };
+<a href=#872 id=872 data-nosnippet>872</a>
+<a href=#873 id=873 data-nosnippet>873</a> <span class="kw">if </span>lower_bound_str < prefix {
+<a href=#874 id=874 data-nosnippet>874</a> <span class="comment">// if lower is shorter than the prefix then lower doesn't start with the prefix
+<a href=#875 id=875 data-nosnippet>875</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
+<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>prefix_len = prefix.chars().count();
+<a href=#879 id=879 data-nosnippet>879</a>
+<a href=#880 id=880 data-nosnippet>880</a> <span class="kw">if </span>lower_bound_str.chars().take(prefix_len).collect::<String>() == <span class="kw-2">*</span>prefix {
+<a href=#881 id=881 data-nosnippet>881</a> <span class="comment">// lower bound matches the prefix
+<a href=#882 id=882 data-nosnippet>882</a>
+<a href=#883 id=883 data-nosnippet>883</a> </span><span class="kw">let </span><span class="prelude-val">Some</span>(upper_bound) = max <span class="kw">else </span>{
+<a href=#884 id=884 data-nosnippet>884</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
+<a href=#885 id=885 data-nosnippet>885</a> };
+<a href=#886 id=886 data-nosnippet>886</a>
+<a href=#887 id=887 data-nosnippet>887</a> <span class="kw">let </span>PrimitiveLiteral::String(upper_bound) = upper_bound.literal() <span class="kw">else </span>{
+<a href=#888 id=888 data-nosnippet>888</a> <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(
+<a href=#889 id=889 data-nosnippet>889</a> ErrorKind::Unexpected,
+<a href=#890 id=890 data-nosnippet>890</a> <span class="string">"Cannot use NotStartsWith operator on non-string upper_bound value"</span>,
+<a href=#891 id=891 data-nosnippet>891</a> ));
+<a href=#892 id=892 data-nosnippet>892</a> };
+<a href=#893 id=893 data-nosnippet>893</a>
+<a href=#894 id=894 data-nosnippet>894</a> <span class="comment">// if upper is shorter than the prefix then upper can't start with the prefix
+<a href=#895 id=895 data-nosnippet>895</a> </span><span class="kw">if </span>upper_bound.chars().count() < prefix_len {
+<a href=#896 id=896 data-nosnippet>896</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">true</span>);
+<a href=#897 id=897 data-nosnippet>897</a> }
+<a href=#898 id=898 data-nosnippet>898</a>
+<a href=#899 id=899 data-nosnippet>899</a> <span class="kw">if </span>upper_bound.chars().take(prefix_len).collect::<String>() == <span class="kw-2">*</span>prefix {
+<a href=#900 id=900 data-nosnippet>900</a> <span class="comment">// both bounds match the prefix, so all rows must match the
+<a href=#901 id=901 data-nosnippet>901</a> // prefix and therefore do not satisfy the predicate
+<a href=#902 id=902 data-nosnippet>902</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
<a href=#903 id=903 data-nosnippet>903</a> }
-<a href=#904 id=904 data-nosnippet>904</a>
-<a href=#905 id=905 data-nosnippet>905</a> <span class="kw">if </span>upper_bound.chars().take(prefix_len).collect::<String>() == <span class="kw-2">*</span>prefix {
-<a href=#906 id=906 data-nosnippet>906</a> <span class="comment">// both bounds match the prefix, so all rows must match the
-<a href=#907 id=907 data-nosnippet>907</a> // prefix and therefore do not satisfy the predicate
-<a href=#908 id=908 data-nosnippet>908</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#909 id=909 data-nosnippet>909</a> }
-<a href=#910 id=910 data-nosnippet>910</a> }
+<a href=#904 id=904 data-nosnippet>904</a> }
+<a href=#905 id=905 data-nosnippet>905</a>
+<a href=#906 id=906 data-nosnippet>906</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
+<a href=#907 id=907 data-nosnippet>907</a> },
+<a href=#908 id=908 data-nosnippet>908</a> MissingColBehavior::MightMatch,
+<a href=#909 id=909 data-nosnippet>909</a> )
+<a href=#910 id=910 data-nosnippet>910</a> }
<a href=#911 id=911 data-nosnippet>911</a>
-<a href=#912 id=912 data-nosnippet>912</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
-<a href=#913 id=913 data-nosnippet>913</a> },
-<a href=#914 id=914 data-nosnippet>914</a> MissingColBehavior::MightMatch,
-<a href=#915 id=915 data-nosnippet>915</a> )
-<a href=#916 id=916 data-nosnippet>916</a> }
-<a href=#917 id=917 data-nosnippet>917</a>
-<a href=#918 id=918 data-nosnippet>918</a> <span class="kw">fn </span>r#in(
-<a href=#919 id=919 data-nosnippet>919</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#920 id=920 data-nosnippet>920</a> reference: <span class="kw-2">&</span>BoundReference,
-<a href=#921 id=921 data-nosnippet>921</a> literals: <span class="kw-2">&</span>FnvHashSet<Datum>,
-<a href=#922 id=922 data-nosnippet>922</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#923 id=923 data-nosnippet>923</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#924 id=924 data-nosnippet>924</a> <span class="kw">let </span>field_id = reference.field().id;
-<a href=#925 id=925 data-nosnippet>925</a>
-<a href=#926 id=926 data-nosnippet>926</a> <span class="kw">if </span>literals.len() > IN_PREDICATE_LIMIT {
-<a href=#927 id=927 data-nosnippet>927</a> <span class="comment">// skip evaluating the predicate if the number of values is too big
-<a href=#928 id=928 data-nosnippet>928</a> </span><span class="kw">return </span><span class="self">self</span>.select_all_rows();
-<a href=#929 id=929 data-nosnippet>929</a> }
-<a href=#930 id=930 data-nosnippet>930</a> <span class="self">self</span>.calc_row_selection(
-<a href=#931 id=931 data-nosnippet>931</a> field_id,
-<a href=#932 id=932 data-nosnippet>932</a> |min, max, nulls| {
-<a href=#933 id=933 data-nosnippet>933</a> <span class="kw">if </span><span class="macro">matches!</span>(nulls, PageNullCount::AllNull) {
-<a href=#934 id=934 data-nosnippet>934</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#935 id=935 data-nosnippet>935</a> }
-<a href=#936 id=936 data-nosnippet>936</a>
-<a href=#937 id=937 data-nosnippet>937</a> <span class="kw">match </span>(min, max) {
-<a href=#938 id=938 data-nosnippet>938</a> (<span class="prelude-val">Some</span>(min), <span class="prelude-val">Some</span>(max))
-<a href=#939 id=939 data-nosnippet>939</a> <span class="kw">if </span>literals
-<a href=#940 id=940 data-nosnippet>940</a> .iter()
-<a href=#941 id=941 data-nosnippet>941</a> .all(|datum| datum.lt(<span class="kw-2">&</span>min) || datum.gt(<span class="kw-2">&</span>max)) =>
-<a href=#942 id=942 data-nosnippet>942</a> {
-<a href=#943 id=943 data-nosnippet>943</a> <span class="comment">// if all values are outside the bounds, no rows can match
-<a href=#944 id=944 data-nosnippet>944</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#945 id=945 data-nosnippet>945</a> }
-<a href=#946 id=946 data-nosnippet>946</a> (<span class="prelude-val">Some</span>(min), <span class="kw">_</span>) <span class="kw">if </span>!literals.iter().any(|datum| datum.ge(<span class="kw-2">&</span>min)) => {
-<a href=#947 id=947 data-nosnippet>947</a> <span class="comment">// if no values are within the min bound, no rows can match
-<a href=#948 id=948 data-nosnippet>948</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#949 id=949 data-nosnippet>949</a> }
-<a href=#950 id=950 data-nosnippet>950</a> (<span class="kw">_</span>, <span class="prelude-val">Some</span>(max)) <span class="kw">if </span>!literals.iter().any(|datum| datum.le(<span class="kw-2">&</span>max)) => {
-<a href=#951 id=951 data-nosnippet>951</a> <span class="comment">// if no values are within the max bound, no rows can match
-<a href=#952 id=952 data-nosnippet>952</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
-<a href=#953 id=953 data-nosnippet>953</a> }
-<a href=#954 id=954 data-nosnippet>954</a> <span class="kw">_ </span>=> {}
-<a href=#955 id=955 data-nosnippet>955</a> }
+<a href=#912 id=912 data-nosnippet>912</a> <span class="kw">fn </span>r#in(
+<a href=#913 id=913 data-nosnippet>913</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#914 id=914 data-nosnippet>914</a> reference: <span class="kw-2">&</span>BoundReference,
+<a href=#915 id=915 data-nosnippet>915</a> literals: <span class="kw-2">&</span>FnvHashSet<Datum>,
+<a href=#916 id=916 data-nosnippet>916</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#917 id=917 data-nosnippet>917</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#918 id=918 data-nosnippet>918</a> <span class="kw">let </span>field_id = reference.field().id;
+<a href=#919 id=919 data-nosnippet>919</a>
+<a href=#920 id=920 data-nosnippet>920</a> <span class="kw">if </span>literals.len() > IN_PREDICATE_LIMIT {
+<a href=#921 id=921 data-nosnippet>921</a> <span class="comment">// skip evaluating the predicate if the number of values is too big
+<a href=#922 id=922 data-nosnippet>922</a> </span><span class="kw">return </span><span class="self">self</span>.select_all_rows();
+<a href=#923 id=923 data-nosnippet>923</a> }
+<a href=#924 id=924 data-nosnippet>924</a> <span class="self">self</span>.calc_row_selection(
+<a href=#925 id=925 data-nosnippet>925</a> field_id,
+<a href=#926 id=926 data-nosnippet>926</a> |min, max, nulls| {
+<a href=#927 id=927 data-nosnippet>927</a> <span class="kw">if </span><span class="macro">matches!</span>(nulls, PageNullCount::AllNull) {
+<a href=#928 id=928 data-nosnippet>928</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#929 id=929 data-nosnippet>929</a> }
+<a href=#930 id=930 data-nosnippet>930</a>
+<a href=#931 id=931 data-nosnippet>931</a> <span class="kw">match </span>(min, max) {
+<a href=#932 id=932 data-nosnippet>932</a> (<span class="prelude-val">Some</span>(min), <span class="prelude-val">Some</span>(max))
+<a href=#933 id=933 data-nosnippet>933</a> <span class="kw">if </span>literals
+<a href=#934 id=934 data-nosnippet>934</a> .iter()
+<a href=#935 id=935 data-nosnippet>935</a> .all(|datum| datum.lt(<span class="kw-2">&</span>min) || datum.gt(<span class="kw-2">&</span>max)) =>
+<a href=#936 id=936 data-nosnippet>936</a> {
+<a href=#937 id=937 data-nosnippet>937</a> <span class="comment">// if all values are outside the bounds, no rows can match
+<a href=#938 id=938 data-nosnippet>938</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#939 id=939 data-nosnippet>939</a> }
+<a href=#940 id=940 data-nosnippet>940</a> (<span class="prelude-val">Some</span>(min), <span class="kw">_</span>) <span class="kw">if </span>!literals.iter().any(|datum| datum.ge(<span class="kw-2">&</span>min)) => {
+<a href=#941 id=941 data-nosnippet>941</a> <span class="comment">// if no values are within the min bound, no rows can match
+<a href=#942 id=942 data-nosnippet>942</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#943 id=943 data-nosnippet>943</a> }
+<a href=#944 id=944 data-nosnippet>944</a> (<span class="kw">_</span>, <span class="prelude-val">Some</span>(max)) <span class="kw">if </span>!literals.iter().any(|datum| datum.le(<span class="kw-2">&</span>max)) => {
+<a href=#945 id=945 data-nosnippet>945</a> <span class="comment">// if no values are within the max bound, no rows can match
+<a href=#946 id=946 data-nosnippet>946</a> </span><span class="kw">return </span><span class="prelude-val">Ok</span>(<span class="bool-val">false</span>);
+<a href=#947 id=947 data-nosnippet>947</a> }
+<a href=#948 id=948 data-nosnippet>948</a> <span class="kw">_ </span>=> {}
+<a href=#949 id=949 data-nosnippet>949</a> }
+<a href=#950 id=950 data-nosnippet>950</a>
+<a href=#951 id=951 data-nosnippet>951</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
+<a href=#952 id=952 data-nosnippet>952</a> },
+<a href=#953 id=953 data-nosnippet>953</a> MissingColBehavior::CantMatch,
+<a href=#954 id=954 data-nosnippet>954</a> )
+<a href=#955 id=955 data-nosnippet>955</a> }
<a href=#956 id=956 data-nosnippet>956</a>
-<a href=#957 id=957 data-nosnippet>957</a> <span class="prelude-val">Ok</span>(<span class="bool-val">true</span>)
-<a href=#958 id=958 data-nosnippet>958</a> },
-<a href=#959 id=959 data-nosnippet>959</a> MissingColBehavior::CantMatch,
-<a href=#960 id=960 data-nosnippet>960</a> )
-<a href=#961 id=961 data-nosnippet>961</a> }
-<a href=#962 id=962 data-nosnippet>962</a>
-<a href=#963 id=963 data-nosnippet>963</a> <span class="kw">fn </span>not_in(
-<a href=#964 id=964 data-nosnippet>964</a> <span class="kw-2">&mut </span><span class="self">self</span>,
-<a href=#965 id=965 data-nosnippet>965</a> _reference: <span class="kw-2">&</span>BoundReference,
-<a href=#966 id=966 data-nosnippet>966</a> _literals: <span class="kw-2">&</span>FnvHashSet<Datum>,
-<a href=#967 id=967 data-nosnippet>967</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
-<a href=#968 id=968 data-nosnippet>968</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
-<a href=#969 id=969 data-nosnippet>969</a> <span class="comment">// Because the bounds are not necessarily a min or max value,
-<a href=#970 id=970 data-nosnippet>970</a> // this cannot be answered using them. notIn(col, {X, ...})
-<a href=#971 id=971 data-nosnippet>971</a> // with (X, Y) doesn't guarantee that X is a value in col.
-<a href=#972 id=972 data-nosnippet>972</a> </span><span class="self">self</span>.select_all_rows()
-<a href=#973 id=973 data-nosnippet>973</a> }
-<a href=#974 id=974 data-nosnippet>974</a>}
-<a href=#975 id=975 data-nosnippet>975</a>
-<a href=#976 id=976 data-nosnippet>976</a><span class="attr">#[cfg(test)]
-<a href=#977 id=977 data-nosnippet>977</a></span><span class="kw">mod </span>tests {
-<a href=#978 id=978 data-nosnippet>978</a> <span class="kw">use </span>std::collections::HashMap;
-<a href=#979 id=979 data-nosnippet>979</a> <span class="kw">use </span>std::sync::Arc;
-<a href=#980 id=980 data-nosnippet>980</a>
-<a href=#981 id=981 data-nosnippet>981</a> <span class="kw">use </span>arrow_array::{
-<a href=#982 id=982 data-nosnippet>982</a> ArrayRef, Decimal128Array, FixedSizeBinaryArray, Float32Array, LargeBinaryArray,
-<a href=#983 id=983 data-nosnippet>983</a> RecordBatch, StringArray,
-<a href=#984 id=984 data-nosnippet>984</a> };
-<a href=#985 id=985 data-nosnippet>985</a> <span class="kw">use </span>arrow_schema::{DataType, Field, Schema <span class="kw">as </span>ArrowSchema};
-<a href=#986 id=986 data-nosnippet>986</a> <span class="kw">use </span>parquet::arrow::ArrowWriter;
-<a href=#987 id=987 data-nosnippet>987</a> <span class="kw">use </span>parquet::arrow::arrow_reader::{
-<a href=#988 id=988 data-nosnippet>988</a> ArrowReaderOptions, ParquetRecordBatchReaderBuilder, RowSelector,
-<a href=#989 id=989 data-nosnippet>989</a> };
-<a href=#990 id=990 data-nosnippet>990</a> <span class="kw">use </span>parquet::file::metadata::{PageIndexPolicy, ParquetMetaData};
-<a href=#991 id=991 data-nosnippet>991</a> <span class="kw">use </span>parquet::file::properties::WriterProperties;
-<a href=#992 id=992 data-nosnippet>992</a> <span class="kw">use </span>rand::Rng;
-<a href=#993 id=993 data-nosnippet>993</a> <span class="kw">use </span>tempfile::NamedTempFile;
+<a href=#957 id=957 data-nosnippet>957</a> <span class="kw">fn </span>not_in(
+<a href=#958 id=958 data-nosnippet>958</a> <span class="kw-2">&mut </span><span class="self">self</span>,
+<a href=#959 id=959 data-nosnippet>959</a> _reference: <span class="kw-2">&</span>BoundReference,
+<a href=#960 id=960 data-nosnippet>960</a> _literals: <span class="kw-2">&</span>FnvHashSet<Datum>,
+<a href=#961 id=961 data-nosnippet>961</a> _predicate: <span class="kw-2">&</span>BoundPredicate,
+<a href=#962 id=962 data-nosnippet>962</a> ) -> <span class="prelude-ty">Result</span><RowSelection> {
+<a href=#963 id=963 data-nosnippet>963</a> <span class="comment">// Because the bounds are not necessarily a min or max value,
+<a href=#964 id=964 data-nosnippet>964</a> // this cannot be answered using them. notIn(col, {X, ...})
+<a href=#965 id=965 data-nosnippet>965</a> // with (X, Y) doesn't guarantee that X is a value in col.
+<a href=#966 id=966 data-nosnippet>966</a> </span><span class="self">self</span>.select_all_rows()
+<a href=#967 id=967 data-nosnippet>967</a> }
+<a href=#968 id=968 data-nosnippet>968</a>}
+<a href=#969 id=969 data-nosnippet>969</a>
+<a href=#970 id=970 data-nosnippet>970</a><span class="attr">#[cfg(test)]
+<a href=#971 id=971 data-nosnippet>971</a></span><span class="kw">mod </span>tests {
+<a href=#972 id=972 data-nosnippet>972</a> <span class="kw">use </span>std::collections::HashMap;
+<a href=#973 id=973 data-nosnippet>973</a> <span class="kw">use </span>std::sync::Arc;
+<a href=#974 id=974 data-nosnippet>974</a>
+<a href=#975 id=975 data-nosnippet>975</a> <span class="kw">use </span>arrow_array::{
+<a href=#976 id=976 data-nosnippet>976</a> ArrayRef, Decimal128Array, FixedSizeBinaryArray, Float32Array, LargeBinaryArray,
+<a href=#977 id=977 data-nosnippet>977</a> RecordBatch, StringArray,
+<a href=#978 id=978 data-nosnippet>978</a> };
+<a href=#979 id=979 data-nosnippet>979</a> <span class="kw">use </span>arrow_schema::{DataType, Field, Schema <span class="kw">as </span>ArrowSchema};
+<a href=#980 id=980 data-nosnippet>980</a> <span class="kw">use </span>parquet::arrow::ArrowWriter;
+<a href=#981 id=981 data-nosnippet>981</a> <span class="kw">use </span>parquet::arrow::arrow_reader::{
+<a href=#982 id=982 data-nosnippet>982</a> ArrowReaderOptions, ParquetRecordBatchReaderBuilder, RowSelector,
+<a href=#983 id=983 data-nosnippet>983</a> };
+<a href=#984 id=984 data-nosnippet>984</a> <span class="kw">use </span>parquet::file::metadata::{PageIndexPolicy, ParquetMetaData};
+<a href=#985 id=985 data-nosnippet>985</a> <span class="kw">use </span>parquet::file::properties::WriterProperties;
+<a href=#986 id=986 data-nosnippet>986</a> <span class="kw">use </span>rand::Rng;
+<a href=#987 id=987 data-nosnippet>987</a> <span class="kw">use </span>tempfile::NamedTempFile;
+<a href=#988 id=988 data-nosnippet>988</a>
+<a href=#989 id=989 data-nosnippet>989</a> <span class="kw">use </span><span class="kw">super</span>::PageIndexEvaluator;
+<a href=#990 id=990 data-nosnippet>990</a> <span class="kw">use </span><span class="kw">crate</span>::expr::{Bind, Reference};
+<a href=#991 id=991 data-nosnippet>991</a> <span class="kw">use </span><span class="kw">crate</span>::spec::decimal_utils::decimal_from_i128_with_scale;
+<a href=#992 id=992 data-nosnippet>992</a> <span class="kw">use </span><span class="kw">crate</span>::spec::{Datum, NestedField, PrimitiveType, Schema, Type};
+<a href=#993 id=993 data-nosnippet>993</a> <span class="kw">use </span>crate::{ErrorKind, <span class="prelude-ty">Result</span>};
<a href=#994 id=994 data-nosnippet>994</a>
-<a href=#995 id=995 data-nosnippet>995</a> <span class="kw">use </span><span class="kw">super</span>::PageIndexEvaluator;
-<a href=#996 id=996 data-nosnippet>996</a> <span class="kw">use </span><span class="kw">crate</span>::expr::{Bind, Reference};
-<a href=#997 id=997 data-nosnippet>997</a> <span class="kw">use </span><span class="kw">crate</span>::spec::decimal_utils::decimal_from_i128_with_scale;
-<a href=#998 id=998 data-nosnippet>998</a> <span class="kw">use </span><span class="kw">crate</span>::spec::{Datum, NestedField, PrimitiveType, Schema, Type};
-<a href=#999 id=999 data-nosnippet>999</a> <span class="kw">use </span>crate::{ErrorKind, <span class="prelude-ty">Result</span>};
-<a href=#1000 id=1000 data-nosnippet>1000</a>
-<a href=#1001 id=1001 data-nosnippet>1001</a> <span class="doccomment">/// Helper function to create a test parquet file with page indexes
-<a href=#1002 id=1002 data-nosnippet>1002</a> /// and return the metadata needed for testing
-<a href=#1003 id=1003 data-nosnippet>1003</a> </span><span class="kw">fn </span>create_test_parquet_file() -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
-<a href=#1004 id=1004 data-nosnippet>1004</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#1005 id=1005 data-nosnippet>1005</a> Field::new(<span class="string">"col_float"</span>, DataType::Float32, <span class="bool-val">true</span>),
-<a href=#1006 id=1006 data-nosnippet>1006</a> Field::new(<span class="string">"col_string"</span>, DataType::Utf8, <span class="bool-val">true</span>),
-<a href=#1007 id=1007 data-nosnippet>1007</a> ]));
-<a href=#1008 id=1008 data-nosnippet>1008</a>
-<a href=#1009 id=1009 data-nosnippet>1009</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
-<a href=#1010 id=1010 data-nosnippet>1010</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1011 id=1011 data-nosnippet>1011</a>
-<a href=#1012 id=1012 data-nosnippet>1012</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#1013 id=1013 data-nosnippet>1013</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
-<a href=#1014 id=1014 data-nosnippet>1014</a> .set_write_batch_size(<span class="number">512</span>)
-<a href=#1015 id=1015 data-nosnippet>1015</a> .build();
-<a href=#1016 id=1016 data-nosnippet>1016</a>
-<a href=#1017 id=1017 data-nosnippet>1017</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#1018 id=1018 data-nosnippet>1018</a>
-<a href=#1019 id=1019 data-nosnippet>1019</a> <span class="kw">let </span><span class="kw-2">mut </span>batches = <span class="macro">vec!</span>[];
-<a href=#1020 id=1020 data-nosnippet>1020</a>
-<a href=#1021 id=1021 data-nosnippet>1021</a> <span class="comment">// Batch 1: 1024 rows - strings with AARDVARK, BEAR, BISON
-<a href=#1022 id=1022 data-nosnippet>1022</a> </span><span class="kw">let </span>float_vals: Vec<<span class="prelude-ty">Option</span><f32>> = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; <span class="number">1024</span>];
-<a href=#1023 id=1023 data-nosnippet>1023</a> <span class="kw">let </span><span class="kw-2">mut </span>string_vals = <span class="macro">vec!</span>[];
-<a href=#1024 id=1024 data-nosnippet>1024</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"AARDVARK"</span>.to_string()));
-<a href=#1025 id=1025 data-nosnippet>1025</a> <span class="kw">for _ in </span><span class="number">1</span>..<span class="number">1023 </span>{
-<a href=#1026 id=1026 data-nosnippet>1026</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"BEAR"</span>.to_string()));
-<a href=#1027 id=1027 data-nosnippet>1027</a> }
-<a href=#1028 id=1028 data-nosnippet>1028</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"BISON"</span>.to_string()));
-<a href=#1029 id=1029 data-nosnippet>1029</a>
-<a href=#1030 id=1030 data-nosnippet>1030</a> batches.push(
-<a href=#1031 id=1031 data-nosnippet>1031</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
-<a href=#1032 id=1032 data-nosnippet>1032</a> Arc::new(Float32Array::from(float_vals)),
-<a href=#1033 id=1033 data-nosnippet>1033</a> Arc::new(StringArray::from(string_vals)),
-<a href=#1034 id=1034 data-nosnippet>1034</a> ])
-<a href=#1035 id=1035 data-nosnippet>1035</a> .unwrap(),
-<a href=#1036 id=1036 data-nosnippet>1036</a> );
-<a href=#1037 id=1037 data-nosnippet>1037</a>
-<a href=#1038 id=1038 data-nosnippet>1038</a> <span class="comment">// Batch 2: 1024 rows - all DEER
-<a href=#1039 id=1039 data-nosnippet>1039</a> </span><span class="kw">let </span>float_vals: Vec<<span class="prelude-ty">Option</span><f32>> = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; <span class="number">1024</span>];
-<a href=#1040 id=1040 data-nosnippet>1040</a> <span class="kw">let </span>string_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"DEER"</span>.to_string()); <span class="number">1024</span>];
-<a href=#1041 id=1041 data-nosnippet>1041</a>
-<a href=#1042 id=1042 data-nosnippet>1042</a> batches.push(
-<a href=#1043 id=1043 data-nosnippet>1043</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
-<a href=#1044 id=1044 data-nosnippet>1044</a> Arc::new(Float32Array::from(float_vals)),
-<a href=#1045 id=1045 data-nosnippet>1045</a> Arc::new(StringArray::from(string_vals)),
-<a href=#1046 id=1046 data-nosnippet>1046</a> ])
-<a href=#1047 id=1047 data-nosnippet>1047</a> .unwrap(),
-<a href=#1048 id=1048 data-nosnippet>1048</a> );
-<a href=#1049 id=1049 data-nosnippet>1049</a>
-<a href=#1050 id=1050 data-nosnippet>1050</a> <span class="comment">// Batch 3: 1024 rows - float 0-10
-<a href=#1051 id=1051 data-nosnippet>1051</a> </span><span class="kw">let </span><span class="kw-2">mut </span>float_vals = <span class="macro">vec!</span>[];
-<a href=#1052 id=1052 data-nosnippet>1052</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..<span class="number">1024 </span>{
-<a href=#1053 id=1053 data-nosnippet>1053</a> float_vals.push(<span class="prelude-val">Some</span>(i <span class="kw">as </span>f32 * <span class="number">10.0 </span>/ <span class="number">1024.0</span>));
+<a href=#995 id=995 data-nosnippet>995</a> <span class="doccomment">/// Helper function to create a test parquet file with page indexes
+<a href=#996 id=996 data-nosnippet>996</a> /// and return the metadata needed for testing
+<a href=#997 id=997 data-nosnippet>997</a> </span><span class="kw">fn </span>create_test_parquet_file() -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
+<a href=#998 id=998 data-nosnippet>998</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#999 id=999 data-nosnippet>999</a> Field::new(<span class="string">"col_float"</span>, DataType::Float32, <span class="bool-val">true</span>),
+<a href=#1000 id=1000 data-nosnippet>1000</a> Field::new(<span class="string">"col_string"</span>, DataType::Utf8, <span class="bool-val">true</span>),
+<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>temp_file = NamedTempFile::new().unwrap();
+<a href=#1004 id=1004 data-nosnippet>1004</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1005 id=1005 data-nosnippet>1005</a>
+<a href=#1006 id=1006 data-nosnippet>1006</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#1007 id=1007 data-nosnippet>1007</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
+<a href=#1008 id=1008 data-nosnippet>1008</a> .set_write_batch_size(<span class="number">512</span>)
+<a href=#1009 id=1009 data-nosnippet>1009</a> .build();
+<a href=#1010 id=1010 data-nosnippet>1010</a>
+<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, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#1012 id=1012 data-nosnippet>1012</a>
+<a href=#1013 id=1013 data-nosnippet>1013</a> <span class="kw">let </span><span class="kw-2">mut </span>batches = <span class="macro">vec!</span>[];
+<a href=#1014 id=1014 data-nosnippet>1014</a>
+<a href=#1015 id=1015 data-nosnippet>1015</a> <span class="comment">// Batch 1: 1024 rows - strings with AARDVARK, BEAR, BISON
+<a href=#1016 id=1016 data-nosnippet>1016</a> </span><span class="kw">let </span>float_vals: Vec<<span class="prelude-ty">Option</span><f32>> = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; <span class="number">1024</span>];
+<a href=#1017 id=1017 data-nosnippet>1017</a> <span class="kw">let </span><span class="kw-2">mut </span>string_vals = <span class="macro">vec!</span>[];
+<a href=#1018 id=1018 data-nosnippet>1018</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"AARDVARK"</span>.to_string()));
+<a href=#1019 id=1019 data-nosnippet>1019</a> <span class="kw">for _ in </span><span class="number">1</span>..<span class="number">1023 </span>{
+<a href=#1020 id=1020 data-nosnippet>1020</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"BEAR"</span>.to_string()));
+<a href=#1021 id=1021 data-nosnippet>1021</a> }
+<a href=#1022 id=1022 data-nosnippet>1022</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"BISON"</span>.to_string()));
+<a href=#1023 id=1023 data-nosnippet>1023</a>
+<a href=#1024 id=1024 data-nosnippet>1024</a> batches.push(
+<a href=#1025 id=1025 data-nosnippet>1025</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
+<a href=#1026 id=1026 data-nosnippet>1026</a> Arc::new(Float32Array::from(float_vals)),
+<a href=#1027 id=1027 data-nosnippet>1027</a> Arc::new(StringArray::from(string_vals)),
+<a href=#1028 id=1028 data-nosnippet>1028</a> ])
+<a href=#1029 id=1029 data-nosnippet>1029</a> .unwrap(),
+<a href=#1030 id=1030 data-nosnippet>1030</a> );
+<a href=#1031 id=1031 data-nosnippet>1031</a>
+<a href=#1032 id=1032 data-nosnippet>1032</a> <span class="comment">// Batch 2: 1024 rows - all DEER
+<a href=#1033 id=1033 data-nosnippet>1033</a> </span><span class="kw">let </span>float_vals: Vec<<span class="prelude-ty">Option</span><f32>> = <span class="macro">vec!</span>[<span class="prelude-val">None</span>; <span class="number">1024</span>];
+<a href=#1034 id=1034 data-nosnippet>1034</a> <span class="kw">let </span>string_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"DEER"</span>.to_string()); <span class="number">1024</span>];
+<a href=#1035 id=1035 data-nosnippet>1035</a>
+<a href=#1036 id=1036 data-nosnippet>1036</a> batches.push(
+<a href=#1037 id=1037 data-nosnippet>1037</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
+<a href=#1038 id=1038 data-nosnippet>1038</a> Arc::new(Float32Array::from(float_vals)),
+<a href=#1039 id=1039 data-nosnippet>1039</a> Arc::new(StringArray::from(string_vals)),
+<a href=#1040 id=1040 data-nosnippet>1040</a> ])
+<a href=#1041 id=1041 data-nosnippet>1041</a> .unwrap(),
+<a href=#1042 id=1042 data-nosnippet>1042</a> );
+<a href=#1043 id=1043 data-nosnippet>1043</a>
+<a href=#1044 id=1044 data-nosnippet>1044</a> <span class="comment">// Batch 3: 1024 rows - float 0-10
+<a href=#1045 id=1045 data-nosnippet>1045</a> </span><span class="kw">let </span><span class="kw-2">mut </span>float_vals = <span class="macro">vec!</span>[];
+<a href=#1046 id=1046 data-nosnippet>1046</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..<span class="number">1024 </span>{
+<a href=#1047 id=1047 data-nosnippet>1047</a> float_vals.push(<span class="prelude-val">Some</span>(i <span class="kw">as </span>f32 * <span class="number">10.0 </span>/ <span class="number">1024.0</span>));
+<a href=#1048 id=1048 data-nosnippet>1048</a> }
+<a href=#1049 id=1049 data-nosnippet>1049</a> <span class="kw">let </span><span class="kw-2">mut </span>string_vals = <span class="macro">vec!</span>[];
+<a href=#1050 id=1050 data-nosnippet>1050</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"GIRAFFE"</span>.to_string()));
+<a href=#1051 id=1051 data-nosnippet>1051</a> string_vals.push(<span class="prelude-val">None</span>);
+<a href=#1052 id=1052 data-nosnippet>1052</a> <span class="kw">for _ in </span><span class="number">2</span>..<span class="number">1024 </span>{
+<a href=#1053 id=1053 data-nosnippet>1053</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"HIPPO"</span>.to_string()));
<a href=#1054 id=1054 data-nosnippet>1054</a> }
-<a href=#1055 id=1055 data-nosnippet>1055</a> <span class="kw">let </span><span class="kw-2">mut </span>string_vals = <span class="macro">vec!</span>[];
-<a href=#1056 id=1056 data-nosnippet>1056</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"GIRAFFE"</span>.to_string()));
-<a href=#1057 id=1057 data-nosnippet>1057</a> string_vals.push(<span class="prelude-val">None</span>);
-<a href=#1058 id=1058 data-nosnippet>1058</a> <span class="kw">for _ in </span><span class="number">2</span>..<span class="number">1024 </span>{
-<a href=#1059 id=1059 data-nosnippet>1059</a> string_vals.push(<span class="prelude-val">Some</span>(<span class="string">"HIPPO"</span>.to_string()));
-<a href=#1060 id=1060 data-nosnippet>1060</a> }
-<a href=#1061 id=1061 data-nosnippet>1061</a>
-<a href=#1062 id=1062 data-nosnippet>1062</a> batches.push(
-<a href=#1063 id=1063 data-nosnippet>1063</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
-<a href=#1064 id=1064 data-nosnippet>1064</a> Arc::new(Float32Array::from(float_vals)),
-<a href=#1065 id=1065 data-nosnippet>1065</a> Arc::new(StringArray::from(string_vals)),
-<a href=#1066 id=1066 data-nosnippet>1066</a> ])
-<a href=#1067 id=1067 data-nosnippet>1067</a> .unwrap(),
-<a href=#1068 id=1068 data-nosnippet>1068</a> );
-<a href=#1069 id=1069 data-nosnippet>1069</a>
-<a href=#1070 id=1070 data-nosnippet>1070</a> <span class="comment">// Batch 4: 1024 rows - float 10-20
-<a href=#1071 id=1071 data-nosnippet>1071</a> </span><span class="kw">let </span><span class="kw-2">mut </span>float_vals = <span class="macro">vec!</span>[<span class="prelude-val">None</span>];
-<a href=#1072 id=1072 data-nosnippet>1072</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">1</span>..<span class="number">1024 </span>{
-<a href=#1073 id=1073 data-nosnippet>1073</a> float_vals.push(<span class="prelude-val">Some</span>(<span class="number">10.0 </span>+ i <span class="kw">as </span>f32 * <span class="number">10.0 </span>/ <span class="number">1024.0</span>));
-<a href=#1074 id=1074 data-nosnippet>1074</a> }
-<a href=#1075 id=1075 data-nosnippet>1075</a> <span class="kw">let </span>string_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"HIPPO"</span>.to_string()); <span class="number">1024</span>];
-<a href=#1076 id=1076 data-nosnippet>1076</a>
-<a href=#1077 id=1077 data-nosnippet>1077</a> batches.push(
-<a href=#1078 id=1078 data-nosnippet>1078</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
-<a href=#1079 id=1079 data-nosnippet>1079</a> Arc::new(Float32Array::from(float_vals)),
-<a href=#1080 id=1080 data-nosnippet>1080</a> Arc::new(StringArray::from(string_vals)),
-<a href=#1081 id=1081 data-nosnippet>1081</a> ])
-<a href=#1082 id=1082 data-nosnippet>1082</a> .unwrap(),
-<a href=#1083 id=1083 data-nosnippet>1083</a> );
-<a href=#1084 id=1084 data-nosnippet>1084</a>
-<a href=#1085 id=1085 data-nosnippet>1085</a> <span class="comment">// Write rows one at a time to give the writer a chance to split into pages
-<a href=#1086 id=1086 data-nosnippet>1086</a> </span><span class="kw">for </span>batch <span class="kw">in </span><span class="kw-2">&</span>batches {
-<a href=#1087 id=1087 data-nosnippet>1087</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
-<a href=#1088 id=1088 data-nosnippet>1088</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
-<a href=#1089 id=1089 data-nosnippet>1089</a> }
-<a href=#1090 id=1090 data-nosnippet>1090</a> }
-<a href=#1091 id=1091 data-nosnippet>1091</a>
-<a href=#1092 id=1092 data-nosnippet>1092</a> writer.close().unwrap();
-<a href=#1093 id=1093 data-nosnippet>1093</a>
-<a href=#1094 id=1094 data-nosnippet>1094</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1095 id=1095 data-nosnippet>1095</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
-<a href=#1096 id=1096 data-nosnippet>1096</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
-<a href=#1097 id=1097 data-nosnippet>1097</a> <span class="kw">let </span>metadata = reader.metadata().clone();
-<a href=#1098 id=1098 data-nosnippet>1098</a>
-<a href=#1099 id=1099 data-nosnippet>1099</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
-<a href=#1100 id=1100 data-nosnippet>1100</a> }
-<a href=#1101 id=1101 data-nosnippet>1101</a>
-<a href=#1102 id=1102 data-nosnippet>1102</a> <span class="doccomment">/// Creates a single-column parquet file whose `col_binary` page bounds hold
-<a href=#1103 id=1103 data-nosnippet>1103</a> /// non-UTF-8 bytes, backing the binary path through `BYTE_ARRAY` column
-<a href=#1104 id=1104 data-nosnippet>1104</a> /// indexes. Page 0 is `[0x01]`, page 1 is `[0xff, 0x00]`.
-<a href=#1105 id=1105 data-nosnippet>1105</a> </span><span class="kw">fn </span>create_binary_parquet_file() -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
-<a href=#1106 id=1106 data-nosnippet>1106</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[Field::new(
-<a href=#1107 id=1107 data-nosnippet>1107</a> <span class="string">"col_binary"</span>,
-<a href=#1108 id=1108 data-nosnippet>1108</a> DataType::LargeBinary,
-<a href=#1109 id=1109 data-nosnippet>1109</a> <span class="bool-val">true</span>,
-<a href=#1110 id=1110 data-nosnippet>1110</a> )]));
-<a href=#1111 id=1111 data-nosnippet>1111</a>
-<a href=#1112 id=1112 data-nosnippet>1112</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
-<a href=#1113 id=1113 data-nosnippet>1113</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1114 id=1114 data-nosnippet>1114</a>
-<a href=#1115 id=1115 data-nosnippet>1115</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#1116 id=1116 data-nosnippet>1116</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
-<a href=#1117 id=1117 data-nosnippet>1117</a> .set_write_batch_size(<span class="number">512</span>)
-<a href=#1118 id=1118 data-nosnippet>1118</a> .build();
-<a href=#1119 id=1119 data-nosnippet>1119</a>
-<a href=#1120 id=1120 data-nosnippet>1120</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#1121 id=1121 data-nosnippet>1121</a>
-<a href=#1122 id=1122 data-nosnippet>1122</a> <span class="comment">// Each page holds 1024 rows of a single value, one page per byte pattern.
-<a href=#1123 id=1123 data-nosnippet>1123</a> </span><span class="kw">let </span>pages: [<span class="kw-2">&</span>[u8]; <span class="number">2</span>] = [<span class="kw-2">&</span>[<span class="number">0x01</span>], <span class="kw-2">&</span>[<span class="number">0xff</span>, <span class="number">0x00</span>]];
-<a href=#1124 id=1124 data-nosnippet>1124</a>
-<a href=#1125 id=1125 data-nosnippet>1125</a> <span class="kw">for </span>value <span class="kw">in </span>pages {
-<a href=#1126 id=1126 data-nosnippet>1126</a> <span class="kw">let </span>array = Arc::new(LargeBinaryArray::from_iter_values(std::iter::repeat_n(
-<a href=#1127 id=1127 data-nosnippet>1127</a> value, <span class="number">1024</span>,
-<a href=#1128 id=1128 data-nosnippet>1128</a> ))) <span class="kw">as </span>ArrayRef;
-<a href=#1129 id=1129 data-nosnippet>1129</a> <span class="kw">let </span>batch = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[array]).unwrap();
-<a href=#1130 id=1130 data-nosnippet>1130</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
-<a href=#1131 id=1131 data-nosnippet>1131</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
-<a href=#1132 id=1132 data-nosnippet>1132</a> }
-<a href=#1133 id=1133 data-nosnippet>1133</a> }
-<a href=#1134 id=1134 data-nosnippet>1134</a>
-<a href=#1135 id=1135 data-nosnippet>1135</a> writer.close().unwrap();
-<a href=#1136 id=1136 data-nosnippet>1136</a>
-<a href=#1137 id=1137 data-nosnippet>1137</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1138 id=1138 data-nosnippet>1138</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
-<a href=#1139 id=1139 data-nosnippet>1139</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
-<a href=#1140 id=1140 data-nosnippet>1140</a> <span class="kw">let </span>metadata = reader.metadata().clone();
-<a href=#1141 id=1141 data-nosnippet>1141</a>
-<a href=#1142 id=1142 data-nosnippet>1142</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
-<a href=#1143 id=1143 data-nosnippet>1143</a> }
-<a href=#1144 id=1144 data-nosnippet>1144</a>
-<a href=#1145 id=1145 data-nosnippet>1145</a> <span class="doccomment">/// Creates a single-column parquet file with a `FixedSizeBinary(2)` column,
-<a href=#1146 id=1146 data-nosnippet>1146</a> /// producing `FIXED_LEN_BYTE_ARRAY` page bounds. Page 0 is `[0x01, 0x01]`,
-<a href=#1147 id=1147 data-nosnippet>1147</a> /// page 1 is `[0x02, 0x02]`.
-<a href=#1148 id=1148 data-nosnippet>1148</a> </span><span class="kw">fn </span>create_fixed_len_byte_array_parquet_file() -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
-<a href=#1149 id=1149 data-nosnippet>1149</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[Field::new(
-<a href=#1150 id=1150 data-nosnippet>1150</a> <span class="string">"col_fixed"</span>,
-<a href=#1151 id=1151 data-nosnippet>1151</a> DataType::FixedSizeBinary(<span class="number">2</span>),
-<a href=#1152 id=1152 data-nosnippet>1152</a> <span class="bool-val">true</span>,
-<a href=#1153 id=1153 data-nosnippet>1153</a> )]));
-<a href=#1154 id=1154 data-nosnippet>1154</a>
-<a href=#1155 id=1155 data-nosnippet>1155</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
-<a href=#1156 id=1156 data-nosnippet>1156</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1157 id=1157 data-nosnippet>1157</a>
-<a href=#1158 id=1158 data-nosnippet>1158</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#1159 id=1159 data-nosnippet>1159</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
-<a href=#1160 id=1160 data-nosnippet>1160</a> .set_write_batch_size(<span class="number">512</span>)
-<a href=#1161 id=1161 data-nosnippet>1161</a> .build();
-<a href=#1162 id=1162 data-nosnippet>1162</a>
-<a href=#1163 id=1163 data-nosnippet>1163</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#1164 id=1164 data-nosnippet>1164</a>
-<a href=#1165 id=1165 data-nosnippet>1165</a> <span class="comment">// Each page holds 1024 rows of a single value, one page per byte pattern.
-<a href=#1166 id=1166 data-nosnippet>1166</a> </span><span class="kw">let </span>pages: [<span class="kw-2">&</span>[u8]; <span class="number">2</span>] = [<span class="kw-2">&</span>[<span class="number">0x01</span>, <span class="number">0x01</span>], <span class="kw-2">&</span>[<span class="number">0x02</span>, <span class="number">0x02</span>]];
-<a href=#1167 id=1167 data-nosnippet>1167</a>
-<a href=#1168 id=1168 data-nosnippet>1168</a> <span class="kw">for </span>value <span class="kw">in </span>pages {
-<a href=#1169 id=1169 data-nosnippet>1169</a> <span class="kw">let </span>array = Arc::new(
-<a href=#1170 id=1170 data-nosnippet>1170</a> FixedSizeBinaryArray::try_from_iter(std::iter::repeat_n(value, <span class="number">1024</span>)).unwrap(),
-<a href=#1171 id=1171 data-nosnippet>1171</a> ) <span class="kw">as </span>ArrayRef;
-<a href=#1172 id=1172 data-nosnippet>1172</a> <span class="kw">let </span>batch = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[array]).unwrap();
-<a href=#1173 id=1173 data-nosnippet>1173</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
-<a href=#1174 id=1174 data-nosnippet>1174</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
-<a href=#1175 id=1175 data-nosnippet>1175</a> }
-<a href=#1176 id=1176 data-nosnippet>1176</a> }
-<a href=#1177 id=1177 data-nosnippet>1177</a>
-<a href=#1178 id=1178 data-nosnippet>1178</a> writer.close().unwrap();
-<a href=#1179 id=1179 data-nosnippet>1179</a>
-<a href=#1180 id=1180 data-nosnippet>1180</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1181 id=1181 data-nosnippet>1181</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
-<a href=#1182 id=1182 data-nosnippet>1182</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
-<a href=#1183 id=1183 data-nosnippet>1183</a> <span class="kw">let </span>metadata = reader.metadata().clone();
-<a href=#1184 id=1184 data-nosnippet>1184</a>
-<a href=#1185 id=1185 data-nosnippet>1185</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
-<a href=#1186 id=1186 data-nosnippet>1186</a> }
-<a href=#1187 id=1187 data-nosnippet>1187</a>
-<a href=#1188 id=1188 data-nosnippet>1188</a> <span class="doccomment">/// Creates a single-column `FixedSizeBinary(width)` parquet file whose column
-<a href=#1189 id=1189 data-nosnippet>1189</a> /// index is truncated to `truncate_length` bytes (`< width`), producing
-<a href=#1190 id=1190 data-nosnippet>1190</a> /// `FIXED_LEN_BYTE_ARRAY` page bounds narrower than the declared `type_length`.
-<a href=#1191 id=1191 data-nosnippet>1191</a> /// Writes one 1024-row page per value in `pages`; each value must be `width`
-<a href=#1192 id=1192 data-nosnippet>1192</a> /// bytes.
-<a href=#1193 id=1193 data-nosnippet>1193</a> </span><span class="kw">fn </span>create_truncated_flba_file(
-<a href=#1194 id=1194 data-nosnippet>1194</a> width: i32,
-<a href=#1195 id=1195 data-nosnippet>1195</a> truncate_length: usize,
-<a href=#1196 id=1196 data-nosnippet>1196</a> pages: <span class="kw-2">&</span>[<span class="kw-2">&</span>[u8]],
-<a href=#1197 id=1197 data-nosnippet>1197</a> ) -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
-<a href=#1198 id=1198 data-nosnippet>1198</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[Field::new(
-<a href=#1199 id=1199 data-nosnippet>1199</a> <span class="string">"col_fixed"</span>,
-<a href=#1200 id=1200 data-nosnippet>1200</a> DataType::FixedSizeBinary(width),
-<a href=#1201 id=1201 data-nosnippet>1201</a> <span class="bool-val">true</span>,
-<a href=#1202 id=1202 data-nosnippet>1202</a> )]));
-<a href=#1203 id=1203 data-nosnippet>1203</a>
-<a href=#1204 id=1204 data-nosnippet>1204</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
-<a href=#1205 id=1205 data-nosnippet>1205</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1055 id=1055 data-nosnippet>1055</a>
+<a href=#1056 id=1056 data-nosnippet>1056</a> batches.push(
+<a href=#1057 id=1057 data-nosnippet>1057</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
+<a href=#1058 id=1058 data-nosnippet>1058</a> Arc::new(Float32Array::from(float_vals)),
+<a href=#1059 id=1059 data-nosnippet>1059</a> Arc::new(StringArray::from(string_vals)),
+<a href=#1060 id=1060 data-nosnippet>1060</a> ])
+<a href=#1061 id=1061 data-nosnippet>1061</a> .unwrap(),
+<a href=#1062 id=1062 data-nosnippet>1062</a> );
+<a href=#1063 id=1063 data-nosnippet>1063</a>
+<a href=#1064 id=1064 data-nosnippet>1064</a> <span class="comment">// Batch 4: 1024 rows - float 10-20
+<a href=#1065 id=1065 data-nosnippet>1065</a> </span><span class="kw">let </span><span class="kw-2">mut </span>float_vals = <span class="macro">vec!</span>[<span class="prelude-val">None</span>];
+<a href=#1066 id=1066 data-nosnippet>1066</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">1</span>..<span class="number">1024 </span>{
+<a href=#1067 id=1067 data-nosnippet>1067</a> float_vals.push(<span class="prelude-val">Some</span>(<span class="number">10.0 </span>+ i <span class="kw">as </span>f32 * <span class="number">10.0 </span>/ <span class="number">1024.0</span>));
+<a href=#1068 id=1068 data-nosnippet>1068</a> }
+<a href=#1069 id=1069 data-nosnippet>1069</a> <span class="kw">let </span>string_vals = <span class="macro">vec!</span>[<span class="prelude-val">Some</span>(<span class="string">"HIPPO"</span>.to_string()); <span class="number">1024</span>];
+<a href=#1070 id=1070 data-nosnippet>1070</a>
+<a href=#1071 id=1071 data-nosnippet>1071</a> batches.push(
+<a href=#1072 id=1072 data-nosnippet>1072</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[
+<a href=#1073 id=1073 data-nosnippet>1073</a> Arc::new(Float32Array::from(float_vals)),
+<a href=#1074 id=1074 data-nosnippet>1074</a> Arc::new(StringArray::from(string_vals)),
+<a href=#1075 id=1075 data-nosnippet>1075</a> ])
+<a href=#1076 id=1076 data-nosnippet>1076</a> .unwrap(),
+<a href=#1077 id=1077 data-nosnippet>1077</a> );
+<a href=#1078 id=1078 data-nosnippet>1078</a>
+<a href=#1079 id=1079 data-nosnippet>1079</a> <span class="comment">// Write rows one at a time to give the writer a chance to split into pages
+<a href=#1080 id=1080 data-nosnippet>1080</a> </span><span class="kw">for </span>batch <span class="kw">in </span><span class="kw-2">&</span>batches {
+<a href=#1081 id=1081 data-nosnippet>1081</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
+<a href=#1082 id=1082 data-nosnippet>1082</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
+<a href=#1083 id=1083 data-nosnippet>1083</a> }
+<a href=#1084 id=1084 data-nosnippet>1084</a> }
+<a href=#1085 id=1085 data-nosnippet>1085</a>
+<a href=#1086 id=1086 data-nosnippet>1086</a> writer.close().unwrap();
+<a href=#1087 id=1087 data-nosnippet>1087</a>
+<a href=#1088 id=1088 data-nosnippet>1088</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1089 id=1089 data-nosnippet>1089</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
+<a href=#1090 id=1090 data-nosnippet>1090</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
+<a href=#1091 id=1091 data-nosnippet>1091</a> <span class="kw">let </span>metadata = reader.metadata().clone();
+<a href=#1092 id=1092 data-nosnippet>1092</a>
+<a href=#1093 id=1093 data-nosnippet>1093</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
+<a href=#1094 id=1094 data-nosnippet>1094</a> }
+<a href=#1095 id=1095 data-nosnippet>1095</a>
+<a href=#1096 id=1096 data-nosnippet>1096</a> <span class="doccomment">/// Creates a single-column parquet file whose `col_binary` page bounds hold
+<a href=#1097 id=1097 data-nosnippet>1097</a> /// non-UTF-8 bytes, backing the binary path through `BYTE_ARRAY` column
+<a href=#1098 id=1098 data-nosnippet>1098</a> /// indexes. Page 0 is `[0x01]`, page 1 is `[0xff, 0x00]`.
+<a href=#1099 id=1099 data-nosnippet>1099</a> </span><span class="kw">fn </span>create_binary_parquet_file() -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
+<a href=#1100 id=1100 data-nosnippet>1100</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[Field::new(
+<a href=#1101 id=1101 data-nosnippet>1101</a> <span class="string">"col_binary"</span>,
+<a href=#1102 id=1102 data-nosnippet>1102</a> DataType::LargeBinary,
+<a href=#1103 id=1103 data-nosnippet>1103</a> <span class="bool-val">true</span>,
+<a href=#1104 id=1104 data-nosnippet>1104</a> )]));
+<a href=#1105 id=1105 data-nosnippet>1105</a>
+<a href=#1106 id=1106 data-nosnippet>1106</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
+<a href=#1107 id=1107 data-nosnippet>1107</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1108 id=1108 data-nosnippet>1108</a>
+<a href=#1109 id=1109 data-nosnippet>1109</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#1110 id=1110 data-nosnippet>1110</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
+<a href=#1111 id=1111 data-nosnippet>1111</a> .set_write_batch_size(<span class="number">512</span>)
+<a href=#1112 id=1112 data-nosnippet>1112</a> .build();
+<a href=#1113 id=1113 data-nosnippet>1113</a>
+<a href=#1114 id=1114 data-nosnippet>1114</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#1115 id=1115 data-nosnippet>1115</a>
+<a href=#1116 id=1116 data-nosnippet>1116</a> <span class="comment">// Each page holds 1024 rows of a single value, one page per byte pattern.
+<a href=#1117 id=1117 data-nosnippet>1117</a> </span><span class="kw">let </span>pages: [<span class="kw-2">&</span>[u8]; <span class="number">2</span>] = [<span class="kw-2">&</span>[<span class="number">0x01</span>], <span class="kw-2">&</span>[<span class="number">0xff</span>, <span class="number">0x00</span>]];
+<a href=#1118 id=1118 data-nosnippet>1118</a>
+<a href=#1119 id=1119 data-nosnippet>1119</a> <span class="kw">for </span>value <span class="kw">in </span>pages {
+<a href=#1120 id=1120 data-nosnippet>1120</a> <span class="kw">let </span>array = Arc::new(LargeBinaryArray::from_iter_values(std::iter::repeat_n(
+<a href=#1121 id=1121 data-nosnippet>1121</a> value, <span class="number">1024</span>,
+<a href=#1122 id=1122 data-nosnippet>1122</a> ))) <span class="kw">as </span>ArrayRef;
+<a href=#1123 id=1123 data-nosnippet>1123</a> <span class="kw">let </span>batch = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[array]).unwrap();
+<a href=#1124 id=1124 data-nosnippet>1124</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
+<a href=#1125 id=1125 data-nosnippet>1125</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
+<a href=#1126 id=1126 data-nosnippet>1126</a> }
+<a href=#1127 id=1127 data-nosnippet>1127</a> }
+<a href=#1128 id=1128 data-nosnippet>1128</a>
+<a href=#1129 id=1129 data-nosnippet>1129</a> writer.close().unwrap();
+<a href=#1130 id=1130 data-nosnippet>1130</a>
+<a href=#1131 id=1131 data-nosnippet>1131</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1132 id=1132 data-nosnippet>1132</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
+<a href=#1133 id=1133 data-nosnippet>1133</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
+<a href=#1134 id=1134 data-nosnippet>1134</a> <span class="kw">let </span>metadata = reader.metadata().clone();
+<a href=#1135 id=1135 data-nosnippet>1135</a>
+<a href=#1136 id=1136 data-nosnippet>1136</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
+<a href=#1137 id=1137 data-nosnippet>1137</a> }
+<a href=#1138 id=1138 data-nosnippet>1138</a>
+<a href=#1139 id=1139 data-nosnippet>1139</a> <span class="doccomment">/// Creates a single-column parquet file with a `FixedSizeBinary(2)` column,
+<a href=#1140 id=1140 data-nosnippet>1140</a> /// producing `FIXED_LEN_BYTE_ARRAY` page bounds. Page 0 is `[0x01, 0x01]`,
+<a href=#1141 id=1141 data-nosnippet>1141</a> /// page 1 is `[0x02, 0x02]`.
+<a href=#1142 id=1142 data-nosnippet>1142</a> </span><span class="kw">fn </span>create_fixed_len_byte_array_parquet_file() -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
+<a href=#1143 id=1143 data-nosnippet>1143</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[Field::new(
+<a href=#1144 id=1144 data-nosnippet>1144</a> <span class="string">"col_fixed"</span>,
+<a href=#1145 id=1145 data-nosnippet>1145</a> DataType::FixedSizeBinary(<span class="number">2</span>),
+<a href=#1146 id=1146 data-nosnippet>1146</a> <span class="bool-val">true</span>,
+<a href=#1147 id=1147 data-nosnippet>1147</a> )]));
+<a href=#1148 id=1148 data-nosnippet>1148</a>
+<a href=#1149 id=1149 data-nosnippet>1149</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
+<a href=#1150 id=1150 data-nosnippet>1150</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1151 id=1151 data-nosnippet>1151</a>
+<a href=#1152 id=1152 data-nosnippet>1152</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#1153 id=1153 data-nosnippet>1153</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
+<a href=#1154 id=1154 data-nosnippet>1154</a> .set_write_batch_size(<span class="number">512</span>)
+<a href=#1155 id=1155 data-nosnippet>1155</a> .build();
+<a href=#1156 id=1156 data-nosnippet>1156</a>
+<a href=#1157 id=1157 data-nosnippet>1157</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#1158 id=1158 data-nosnippet>1158</a>
+<a href=#1159 id=1159 data-nosnippet>1159</a> <span class="comment">// Each page holds 1024 rows of a single value, one page per byte pattern.
+<a href=#1160 id=1160 data-nosnippet>1160</a> </span><span class="kw">let </span>pages: [<span class="kw-2">&</span>[u8]; <span class="number">2</span>] = [<span class="kw-2">&</span>[<span class="number">0x01</span>, <span class="number">0x01</span>], <span class="kw-2">&</span>[<span class="number">0x02</span>, <span class="number">0x02</span>]];
+<a href=#1161 id=1161 data-nosnippet>1161</a>
+<a href=#1162 id=1162 data-nosnippet>1162</a> <span class="kw">for </span>value <span class="kw">in </span>pages {
+<a href=#1163 id=1163 data-nosnippet>1163</a> <span class="kw">let </span>array = Arc::new(
+<a href=#1164 id=1164 data-nosnippet>1164</a> FixedSizeBinaryArray::try_from_iter(std::iter::repeat_n(value, <span class="number">1024</span>)).unwrap(),
+<a href=#1165 id=1165 data-nosnippet>1165</a> ) <span class="kw">as </span>ArrayRef;
+<a href=#1166 id=1166 data-nosnippet>1166</a> <span class="kw">let </span>batch = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[array]).unwrap();
+<a href=#1167 id=1167 data-nosnippet>1167</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
+<a href=#1168 id=1168 data-nosnippet>1168</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
+<a href=#1169 id=1169 data-nosnippet>1169</a> }
+<a href=#1170 id=1170 data-nosnippet>1170</a> }
+<a href=#1171 id=1171 data-nosnippet>1171</a>
+<a href=#1172 id=1172 data-nosnippet>1172</a> writer.close().unwrap();
+<a href=#1173 id=1173 data-nosnippet>1173</a>
+<a href=#1174 id=1174 data-nosnippet>1174</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1175 id=1175 data-nosnippet>1175</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
+<a href=#1176 id=1176 data-nosnippet>1176</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
+<a href=#1177 id=1177 data-nosnippet>1177</a> <span class="kw">let </span>metadata = reader.metadata().clone();
+<a href=#1178 id=1178 data-nosnippet>1178</a>
+<a href=#1179 id=1179 data-nosnippet>1179</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
+<a href=#1180 id=1180 data-nosnippet>1180</a> }
+<a href=#1181 id=1181 data-nosnippet>1181</a>
+<a href=#1182 id=1182 data-nosnippet>1182</a> <span class="doccomment">/// Creates a single-column `FixedSizeBinary(width)` parquet file whose column
+<a href=#1183 id=1183 data-nosnippet>1183</a> /// index is truncated to `truncate_length` bytes (`< width`), producing
+<a href=#1184 id=1184 data-nosnippet>1184</a> /// `FIXED_LEN_BYTE_ARRAY` page bounds narrower than the declared `type_length`.
+<a href=#1185 id=1185 data-nosnippet>1185</a> /// Writes one 1024-row page per value in `pages`; each value must be `width`
+<a href=#1186 id=1186 data-nosnippet>1186</a> /// bytes.
+<a href=#1187 id=1187 data-nosnippet>1187</a> </span><span class="kw">fn </span>create_truncated_flba_file(
+<a href=#1188 id=1188 data-nosnippet>1188</a> width: i32,
+<a href=#1189 id=1189 data-nosnippet>1189</a> truncate_length: usize,
+<a href=#1190 id=1190 data-nosnippet>1190</a> pages: <span class="kw-2">&</span>[<span class="kw-2">&</span>[u8]],
+<a href=#1191 id=1191 data-nosnippet>1191</a> ) -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
+<a href=#1192 id=1192 data-nosnippet>1192</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[Field::new(
+<a href=#1193 id=1193 data-nosnippet>1193</a> <span class="string">"col_fixed"</span>,
+<a href=#1194 id=1194 data-nosnippet>1194</a> DataType::FixedSizeBinary(width),
+<a href=#1195 id=1195 data-nosnippet>1195</a> <span class="bool-val">true</span>,
+<a href=#1196 id=1196 data-nosnippet>1196</a> )]));
+<a href=#1197 id=1197 data-nosnippet>1197</a>
+<a href=#1198 id=1198 data-nosnippet>1198</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
+<a href=#1199 id=1199 data-nosnippet>1199</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1200 id=1200 data-nosnippet>1200</a>
+<a href=#1201 id=1201 data-nosnippet>1201</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#1202 id=1202 data-nosnippet>1202</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
+<a href=#1203 id=1203 data-nosnippet>1203</a> .set_write_batch_size(<span class="number">512</span>)
+<a href=#1204 id=1204 data-nosnippet>1204</a> .set_column_index_truncate_length(<span class="prelude-val">Some</span>(truncate_length))
+<a href=#1205 id=1205 data-nosnippet>1205</a> .build();
<a href=#1206 id=1206 data-nosnippet>1206</a>
-<a href=#1207 id=1207 data-nosnippet>1207</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#1208 id=1208 data-nosnippet>1208</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
-<a href=#1209 id=1209 data-nosnippet>1209</a> .set_write_batch_size(<span class="number">512</span>)
-<a href=#1210 id=1210 data-nosnippet>1210</a> .set_column_index_truncate_length(<span class="prelude-val">Some</span>(truncate_length))
-<a href=#1211 id=1211 data-nosnippet>1211</a> .build();
-<a href=#1212 id=1212 data-nosnippet>1212</a>
-<a href=#1213 id=1213 data-nosnippet>1213</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#1214 id=1214 data-nosnippet>1214</a>
-<a href=#1215 id=1215 data-nosnippet>1215</a> <span class="kw">for </span>value <span class="kw">in </span>pages {
-<a href=#1216 id=1216 data-nosnippet>1216</a> <span class="kw">let </span>array = Arc::new(
-<a href=#1217 id=1217 data-nosnippet>1217</a> FixedSizeBinaryArray::try_from_iter(std::iter::repeat_n(<span class="kw-2">*</span>value, <span class="number">1024</span>)).unwrap(),
-<a href=#1218 id=1218 data-nosnippet>1218</a> ) <span class="kw">as </span>ArrayRef;
-<a href=#1219 id=1219 data-nosnippet>1219</a> <span class="kw">let </span>batch = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[array]).unwrap();
-<a href=#1220 id=1220 data-nosnippet>1220</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
-<a href=#1221 id=1221 data-nosnippet>1221</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
-<a href=#1222 id=1222 data-nosnippet>1222</a> }
-<a href=#1223 id=1223 data-nosnippet>1223</a> }
-<a href=#1224 id=1224 data-nosnippet>1224</a>
-<a href=#1225 id=1225 data-nosnippet>1225</a> writer.close().unwrap();
-<a href=#1226 id=1226 data-nosnippet>1226</a>
-<a href=#1227 id=1227 data-nosnippet>1227</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1228 id=1228 data-nosnippet>1228</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
-<a href=#1229 id=1229 data-nosnippet>1229</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
-<a href=#1230 id=1230 data-nosnippet>1230</a> <span class="kw">let </span>metadata = reader.metadata().clone();
-<a href=#1231 id=1231 data-nosnippet>1231</a>
-<a href=#1232 id=1232 data-nosnippet>1232</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
-<a href=#1233 id=1233 data-nosnippet>1233</a> }
-<a href=#1234 id=1234 data-nosnippet>1234</a>
-<a href=#1235 id=1235 data-nosnippet>1235</a> <span class="doccomment">/// Creates a single-column `Decimal128(precision, scale)` parquet file.
-<a href=#1236 id=1236 data-nosnippet>1236</a> /// Parquet encodes precision <= 9 as `INT32`, precision 10..=18 as `INT64`,
-<a href=#1237 id=1237 data-nosnippet>1237</a> /// and precision > 18 as `FIXED_LEN_BYTE_ARRAY`, so `precision` selects which
-<a href=#1238 id=1238 data-nosnippet>1238</a> /// page-index encoding is exercised. Writes one 1024-row page per value in
-<a href=#1239 id=1239 data-nosnippet>1239</a> /// `unscaled`, so page bounds partition the value range.
-<a href=#1240 id=1240 data-nosnippet>1240</a> </span><span class="kw">fn </span>create_decimal_parquet_file(
-<a href=#1241 id=1241 data-nosnippet>1241</a> precision: u8,
-<a href=#1242 id=1242 data-nosnippet>1242</a> scale: i8,
-<a href=#1243 id=1243 data-nosnippet>1243</a> unscaled: <span class="kw-2">&</span>[i128],
-<a href=#1244 id=1244 data-nosnippet>1244</a> ) -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
-<a href=#1245 id=1245 data-nosnippet>1245</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[Field::new(
-<a href=#1246 id=1246 data-nosnippet>1246</a> <span class="string">"col_decimal"</span>,
-<a href=#1247 id=1247 data-nosnippet>1247</a> DataType::Decimal128(precision, scale),
-<a href=#1248 id=1248 data-nosnippet>1248</a> <span class="bool-val">true</span>,
-<a href=#1249 id=1249 data-nosnippet>1249</a> )]));
-<a href=#1250 id=1250 data-nosnippet>1250</a>
-<a href=#1251 id=1251 data-nosnippet>1251</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
-<a href=#1252 id=1252 data-nosnippet>1252</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1253 id=1253 data-nosnippet>1253</a>
-<a href=#1254 id=1254 data-nosnippet>1254</a> <span class="kw">let </span>props = WriterProperties::builder()
-<a href=#1255 id=1255 data-nosnippet>1255</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
-<a href=#1256 id=1256 data-nosnippet>1256</a> .set_write_batch_size(<span class="number">512</span>)
-<a href=#1257 id=1257 data-nosnippet>1257</a> .build();
-<a href=#1258 id=1258 data-nosnippet>1258</a>
-<a href=#1259 id=1259 data-nosnippet>1259</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
-<a href=#1260 id=1260 data-nosnippet>1260</a>
-<a href=#1261 id=1261 data-nosnippet>1261</a> <span class="kw">for </span><span class="kw-2">&</span>unscaled <span class="kw">in </span>unscaled {
-<a href=#1262 id=1262 data-nosnippet>1262</a> <span class="kw">let </span>array = Arc::new(
-<a href=#1263 id=1263 data-nosnippet>1263</a> Decimal128Array::from_iter_values(std::iter::repeat_n(unscaled, <span class="number">1024</span>))
-<a href=#1264 id=1264 data-nosnippet>1264</a> .with_precision_and_scale(precision, scale)
-<a href=#1265 id=1265 data-nosnippet>1265</a> .unwrap(),
-<a href=#1266 id=1266 data-nosnippet>1266</a> ) <span class="kw">as </span>ArrayRef;
-<a href=#1267 id=1267 data-nosnippet>1267</a> <span class="kw">let </span>batch = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[array]).unwrap();
-<a href=#1268 id=1268 data-nosnippet>1268</a> <span class="comment">// Write rows one at a time so the writer splits into per-value pages.
-<a href=#1269 id=1269 data-nosnippet>1269</a> </span><span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
-<a href=#1270 id=1270 data-nosnippet>1270</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
-<a href=#1271 id=1271 data-nosnippet>1271</a> }
-<a href=#1272 id=1272 data-nosnippet>1272</a> }
-<a href=#1273 id=1273 data-nosnippet>1273</a>
-<a href=#1274 id=1274 data-nosnippet>1274</a> writer.close().unwrap();
-<a href=#1275 id=1275 data-nosnippet>1275</a>
-<a href=#1276 id=1276 data-nosnippet>1276</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1277 id=1277 data-nosnippet>1277</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
-<a href=#1278 id=1278 data-nosnippet>1278</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
-<a href=#1279 id=1279 data-nosnippet>1279</a> <span class="kw">let </span>metadata = reader.metadata().clone();
-<a href=#1280 id=1280 data-nosnippet>1280</a>
-<a href=#1281 id=1281 data-nosnippet>1281</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
-<a href=#1282 id=1282 data-nosnippet>1282</a> }
-<a href=#1283 id=1283 data-nosnippet>1283</a>
-<a href=#1284 id=1284 data-nosnippet>1284</a> <span class="doccomment">/// Get the test metadata components for testing
-<a href=#1285 id=1285 data-nosnippet>1285</a> </span><span class="kw">fn </span>get_test_metadata(
-<a href=#1286 id=1286 data-nosnippet>1286</a> metadata: <span class="kw-2">&</span>ParquetMetaData,
-<a href=#1287 id=1287 data-nosnippet>1287</a> ) -> (
-<a href=#1288 id=1288 data-nosnippet>1288</a> Vec<parquet::file::page_index::column_index::ColumnIndexMetaData>,
-<a href=#1289 id=1289 data-nosnippet>1289</a> Vec<parquet::file::page_index::offset_index::OffsetIndexMetaData>,
-<a href=#1290 id=1290 data-nosnippet>1290</a> <span class="kw-2">&</span>parquet::file::metadata::RowGroupMetaData,
-<a href=#1291 id=1291 data-nosnippet>1291</a> ) {
-<a href=#1292 id=1292 data-nosnippet>1292</a> <span class="kw">let </span>row_group_metadata = metadata.row_group(<span class="number">0</span>);
-<a href=#1293 id=1293 data-nosnippet>1293</a> <span class="kw">let </span>column_index = metadata.column_index().unwrap()[<span class="number">0</span>].to_vec();
-<a href=#1294 id=1294 data-nosnippet>1294</a> <span class="kw">let </span>offset_index = metadata.offset_index().unwrap()[<span class="number">0</span>].to_vec();
-<a href=#1295 id=1295 data-nosnippet>1295</a> (column_index, offset_index, row_group_metadata)
-<a href=#1296 id=1296 data-nosnippet>1296</a> }
-<a href=#1297 id=1297 data-nosnippet>1297</a>
-<a href=#1298 id=1298 data-nosnippet>1298</a> <span class="attr">#[test]
-<a href=#1299 id=1299 data-nosnippet>1299</a> </span><span class="kw">fn </span>eval_matches_no_rows_for_empty_row_group() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1300 id=1300 data-nosnippet>1300</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
-<a href=#1301 id=1301 data-nosnippet>1301</a> Field::new(<span class="string">"col_float"</span>, DataType::Float32, <span class="bool-val">true</span>),
-<a href=#1302 id=1302 data-nosnippet>1302</a> Field::new(<span class="string">"col_string"</span>, DataType::Utf8, <span class="bool-val">true</span>),
-<a href=#1303 id=1303 data-nosnippet>1303</a> ]));
-<a href=#1304 id=1304 data-nosnippet>1304</a>
-<a href=#1305 id=1305 data-nosnippet>1305</a> <span class="kw">let </span>empty_float: ArrayRef = Arc::new(Float32Array::from(Vec::<<span class="prelude-ty">Option</span><f32>>::new()));
-<a href=#1306 id=1306 data-nosnippet>1306</a> <span class="kw">let </span>empty_string: ArrayRef = Arc::new(StringArray::from(Vec::<<span class="prelude-ty">Option</span><String>>::new()));
-<a href=#1307 id=1307 data-nosnippet>1307</a> <span class="kw">let </span>empty_batch =
-<a href=#1308 id=1308 data-nosnippet>1308</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[empty_float, empty_string]).unwrap();
-<a href=#1309 id=1309 data-nosnippet>1309</a>
-<a href=#1310 id=1310 data-nosnippet>1310</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
-<a href=#1311 id=1311 data-nosnippet>1311</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1312 id=1312 data-nosnippet>1312</a>
-<a href=#1313 id=1313 data-nosnippet>1313</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema, <span class="prelude-val">None</span>).unwrap();
-<a href=#1314 id=1314 data-nosnippet>1314</a> writer.write(<span class="kw-2">&</span>empty_batch).unwrap();
-<a href=#1315 id=1315 data-nosnippet>1315</a> writer.close().unwrap();
-<a href=#1316 id=1316 data-nosnippet>1316</a>
-<a href=#1317 id=1317 data-nosnippet>1317</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
-<a href=#1318 id=1318 data-nosnippet>1318</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
-<a href=#1319 id=1319 data-nosnippet>1319</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
-<a href=#1320 id=1320 data-nosnippet>1320</a> <span class="kw">let </span>metadata = reader.metadata();
-<a href=#1321 id=1321 data-nosnippet>1321</a>
-<a href=#1322 id=1322 data-nosnippet>1322</a> <span class="kw">if </span>metadata.num_row_groups() == <span class="number">0 </span>|| metadata.row_group(<span class="number">0</span>).num_rows() == <span class="number">0 </span>{
-<a href=#1323 id=1323 data-nosnippet>1323</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(());
-<a href=#1324 id=1324 data-nosnippet>1324</a> }
-<a href=#1325 id=1325 data-nosnippet>1325</a>
-<a href=#1326 id=1326 data-nosnippet>1326</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1207 id=1207 data-nosnippet>1207</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#1208 id=1208 data-nosnippet>1208</a>
+<a href=#1209 id=1209 data-nosnippet>1209</a> <span class="kw">for </span>value <span class="kw">in </span>pages {
+<a href=#1210 id=1210 data-nosnippet>1210</a> <span class="kw">let </span>array = Arc::new(
+<a href=#1211 id=1211 data-nosnippet>1211</a> FixedSizeBinaryArray::try_from_iter(std::iter::repeat_n(<span class="kw-2">*</span>value, <span class="number">1024</span>)).unwrap(),
+<a href=#1212 id=1212 data-nosnippet>1212</a> ) <span class="kw">as </span>ArrayRef;
+<a href=#1213 id=1213 data-nosnippet>1213</a> <span class="kw">let </span>batch = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[array]).unwrap();
+<a href=#1214 id=1214 data-nosnippet>1214</a> <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
+<a href=#1215 id=1215 data-nosnippet>1215</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
+<a href=#1216 id=1216 data-nosnippet>1216</a> }
+<a href=#1217 id=1217 data-nosnippet>1217</a> }
+<a href=#1218 id=1218 data-nosnippet>1218</a>
+<a href=#1219 id=1219 data-nosnippet>1219</a> writer.close().unwrap();
+<a href=#1220 id=1220 data-nosnippet>1220</a>
+<a href=#1221 id=1221 data-nosnippet>1221</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1222 id=1222 data-nosnippet>1222</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
+<a href=#1223 id=1223 data-nosnippet>1223</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
+<a href=#1224 id=1224 data-nosnippet>1224</a> <span class="kw">let </span>metadata = reader.metadata().clone();
+<a href=#1225 id=1225 data-nosnippet>1225</a>
+<a href=#1226 id=1226 data-nosnippet>1226</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
+<a href=#1227 id=1227 data-nosnippet>1227</a> }
+<a href=#1228 id=1228 data-nosnippet>1228</a>
+<a href=#1229 id=1229 data-nosnippet>1229</a> <span class="doccomment">/// Creates a single-column `Decimal128(precision, scale)` parquet file.
+<a href=#1230 id=1230 data-nosnippet>1230</a> /// Parquet encodes precision <= 9 as `INT32`, precision 10..=18 as `INT64`,
+<a href=#1231 id=1231 data-nosnippet>1231</a> /// and precision > 18 as `FIXED_LEN_BYTE_ARRAY`, so `precision` selects which
+<a href=#1232 id=1232 data-nosnippet>1232</a> /// page-index encoding is exercised. Writes one 1024-row page per value in
+<a href=#1233 id=1233 data-nosnippet>1233</a> /// `unscaled`, so page bounds partition the value range.
+<a href=#1234 id=1234 data-nosnippet>1234</a> </span><span class="kw">fn </span>create_decimal_parquet_file(
+<a href=#1235 id=1235 data-nosnippet>1235</a> precision: u8,
+<a href=#1236 id=1236 data-nosnippet>1236</a> scale: i8,
+<a href=#1237 id=1237 data-nosnippet>1237</a> unscaled: <span class="kw-2">&</span>[i128],
+<a href=#1238 id=1238 data-nosnippet>1238</a> ) -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
+<a href=#1239 id=1239 data-nosnippet>1239</a> <span class="kw">let </span>pages: Vec<<span class="prelude-ty">Option</span><i128>> = unscaled.iter().copied().map(<span class="prelude-val">Some</span>).collect();
+<a href=#1240 id=1240 data-nosnippet>1240</a> create_decimal_parquet_file_with_null_pages(precision, scale, <span class="kw-2">&</span>pages)
+<a href=#1241 id=1241 data-nosnippet>1241</a> }
+<a href=#1242 id=1242 data-nosnippet>1242</a>
+<a href=#1243 id=1243 data-nosnippet>1243</a> <span class="doccomment">/// Like [`create_decimal_parquet_file`], but each entry in `pages` is one
+<a href=#1244 id=1244 data-nosnippet>1244</a> /// 1024-row page: `Some(v)` writes that repeated unscaled value, `None`
+<a href=#1245 id=1245 data-nosnippet>1245</a> /// writes an all-null page (whose column index carries no min/max bound).
+<a href=#1246 id=1246 data-nosnippet>1246</a> </span><span class="kw">fn </span>create_decimal_parquet_file_with_null_pages(
+<a href=#1247 id=1247 data-nosnippet>1247</a> precision: u8,
+<a href=#1248 id=1248 data-nosnippet>1248</a> scale: i8,
+<a href=#1249 id=1249 data-nosnippet>1249</a> pages: <span class="kw-2">&</span>[<span class="prelude-ty">Option</span><i128>],
+<a href=#1250 id=1250 data-nosnippet>1250</a> ) -> <span class="prelude-ty">Result</span><(Arc<ParquetMetaData>, NamedTempFile)> {
+<a href=#1251 id=1251 data-nosnippet>1251</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[Field::new(
+<a href=#1252 id=1252 data-nosnippet>1252</a> <span class="string">"col_decimal"</span>,
+<a href=#1253 id=1253 data-nosnippet>1253</a> DataType::Decimal128(precision, scale),
+<a href=#1254 id=1254 data-nosnippet>1254</a> <span class="bool-val">true</span>,
+<a href=#1255 id=1255 data-nosnippet>1255</a> )]));
+<a href=#1256 id=1256 data-nosnippet>1256</a>
+<a href=#1257 id=1257 data-nosnippet>1257</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
+<a href=#1258 id=1258 data-nosnippet>1258</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1259 id=1259 data-nosnippet>1259</a>
+<a href=#1260 id=1260 data-nosnippet>1260</a> <span class="kw">let </span>props = WriterProperties::builder()
+<a href=#1261 id=1261 data-nosnippet>1261</a> .set_data_page_row_count_limit(<span class="number">1024</span>)
+<a href=#1262 id=1262 data-nosnippet>1262</a> .set_write_batch_size(<span class="number">512</span>)
+<a href=#1263 id=1263 data-nosnippet>1263</a> .build();
+<a href=#1264 id=1264 data-nosnippet>1264</a>
+<a href=#1265 id=1265 data-nosnippet>1265</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema.clone(), <span class="prelude-val">Some</span>(props)).unwrap();
+<a href=#1266 id=1266 data-nosnippet>1266</a>
+<a href=#1267 id=1267 data-nosnippet>1267</a> <span class="kw">for </span>page <span class="kw">in </span>pages {
+<a href=#1268 id=1268 data-nosnippet>1268</a> <span class="kw">let </span>array = Arc::new(
+<a href=#1269 id=1269 data-nosnippet>1269</a> <span class="kw">match </span>page {
+<a href=#1270 id=1270 data-nosnippet>1270</a> <span class="prelude-val">Some</span>(unscaled) => {
+<a href=#1271 id=1271 data-nosnippet>1271</a> Decimal128Array::from_iter_values(std::iter::repeat_n(<span class="kw-2">*</span>unscaled, <span class="number">1024</span>))
+<a href=#1272 id=1272 data-nosnippet>1272</a> }
+<a href=#1273 id=1273 data-nosnippet>1273</a> <span class="prelude-val">None </span>=> Decimal128Array::from(<span class="macro">vec!</span>[<span class="prelude-val">None</span>; <span class="number">1024</span>]),
+<a href=#1274 id=1274 data-nosnippet>1274</a> }
+<a href=#1275 id=1275 data-nosnippet>1275</a> .with_precision_and_scale(precision, scale)
+<a href=#1276 id=1276 data-nosnippet>1276</a> .unwrap(),
+<a href=#1277 id=1277 data-nosnippet>1277</a> ) <span class="kw">as </span>ArrayRef;
+<a href=#1278 id=1278 data-nosnippet>1278</a> <span class="kw">let </span>batch = RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[array]).unwrap();
+<a href=#1279 id=1279 data-nosnippet>1279</a> <span class="comment">// Write rows one at a time so the writer splits into per-value pages.
+<a href=#1280 id=1280 data-nosnippet>1280</a> </span><span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..batch.num_rows() {
+<a href=#1281 id=1281 data-nosnippet>1281</a> writer.write(<span class="kw-2">&</span>batch.slice(i, <span class="number">1</span>)).unwrap();
+<a href=#1282 id=1282 data-nosnippet>1282</a> }
+<a href=#1283 id=1283 data-nosnippet>1283</a> }
+<a href=#1284 id=1284 data-nosnippet>1284</a>
+<a href=#1285 id=1285 data-nosnippet>1285</a> writer.close().unwrap();
+<a href=#1286 id=1286 data-nosnippet>1286</a>
+<a href=#1287 id=1287 data-nosnippet>1287</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1288 id=1288 data-nosnippet>1288</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
+<a href=#1289 id=1289 data-nosnippet>1289</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
+<a href=#1290 id=1290 data-nosnippet>1290</a> <span class="kw">let </span>metadata = reader.metadata().clone();
+<a href=#1291 id=1291 data-nosnippet>1291</a>
+<a href=#1292 id=1292 data-nosnippet>1292</a> <span class="prelude-val">Ok</span>((metadata, temp_file))
+<a href=#1293 id=1293 data-nosnippet>1293</a> }
+<a href=#1294 id=1294 data-nosnippet>1294</a>
+<a href=#1295 id=1295 data-nosnippet>1295</a> <span class="doccomment">/// Get the test metadata components for testing
+<a href=#1296 id=1296 data-nosnippet>1296</a> </span><span class="kw">fn </span>get_test_metadata(
+<a href=#1297 id=1297 data-nosnippet>1297</a> metadata: <span class="kw-2">&</span>ParquetMetaData,
+<a href=#1298 id=1298 data-nosnippet>1298</a> ) -> (
+<a href=#1299 id=1299 data-nosnippet>1299</a> Vec<parquet::file::page_index::column_index::ColumnIndexMetaData>,
+<a href=#1300 id=1300 data-nosnippet>1300</a> Vec<parquet::file::page_index::offset_index::OffsetIndexMetaData>,
+<a href=#1301 id=1301 data-nosnippet>1301</a> <span class="kw-2">&</span>parquet::file::metadata::RowGroupMetaData,
+<a href=#1302 id=1302 data-nosnippet>1302</a> ) {
+<a href=#1303 id=1303 data-nosnippet>1303</a> <span class="kw">let </span>row_group_metadata = metadata.row_group(<span class="number">0</span>);
+<a href=#1304 id=1304 data-nosnippet>1304</a> <span class="kw">let </span>column_index = metadata.column_index().unwrap()[<span class="number">0</span>].to_vec();
+<a href=#1305 id=1305 data-nosnippet>1305</a> <span class="kw">let </span>offset_index = metadata.offset_index().unwrap()[<span class="number">0</span>].to_vec();
+<a href=#1306 id=1306 data-nosnippet>1306</a> (column_index, offset_index, row_group_metadata)
+<a href=#1307 id=1307 data-nosnippet>1307</a> }
+<a href=#1308 id=1308 data-nosnippet>1308</a>
+<a href=#1309 id=1309 data-nosnippet>1309</a> <span class="attr">#[test]
+<a href=#1310 id=1310 data-nosnippet>1310</a> </span><span class="kw">fn </span>eval_matches_no_rows_for_empty_row_group() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1311 id=1311 data-nosnippet>1311</a> <span class="kw">let </span>arrow_schema = Arc::new(ArrowSchema::new(<span class="macro">vec!</span>[
+<a href=#1312 id=1312 data-nosnippet>1312</a> Field::new(<span class="string">"col_float"</span>, DataType::Float32, <span class="bool-val">true</span>),
+<a href=#1313 id=1313 data-nosnippet>1313</a> Field::new(<span class="string">"col_string"</span>, DataType::Utf8, <span class="bool-val">true</span>),
+<a href=#1314 id=1314 data-nosnippet>1314</a> ]));
+<a href=#1315 id=1315 data-nosnippet>1315</a>
+<a href=#1316 id=1316 data-nosnippet>1316</a> <span class="kw">let </span>empty_float: ArrayRef = Arc::new(Float32Array::from(Vec::<<span class="prelude-ty">Option</span><f32>>::new()));
+<a href=#1317 id=1317 data-nosnippet>1317</a> <span class="kw">let </span>empty_string: ArrayRef = Arc::new(StringArray::from(Vec::<<span class="prelude-ty">Option</span><String>>::new()));
+<a href=#1318 id=1318 data-nosnippet>1318</a> <span class="kw">let </span>empty_batch =
+<a href=#1319 id=1319 data-nosnippet>1319</a> RecordBatch::try_new(arrow_schema.clone(), <span class="macro">vec!</span>[empty_float, empty_string]).unwrap();
+<a href=#1320 id=1320 data-nosnippet>1320</a>
+<a href=#1321 id=1321 data-nosnippet>1321</a> <span class="kw">let </span>temp_file = NamedTempFile::new().unwrap();
+<a href=#1322 id=1322 data-nosnippet>1322</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1323 id=1323 data-nosnippet>1323</a>
+<a href=#1324 id=1324 data-nosnippet>1324</a> <span class="kw">let </span><span class="kw-2">mut </span>writer = ArrowWriter::try_new(file, arrow_schema, <span class="prelude-val">None</span>).unwrap();
+<a href=#1325 id=1325 data-nosnippet>1325</a> writer.write(<span class="kw-2">&</span>empty_batch).unwrap();
+<a href=#1326 id=1326 data-nosnippet>1326</a> writer.close().unwrap();
<a href=#1327 id=1327 data-nosnippet>1327</a>
-<a href=#1328 id=1328 data-nosnippet>1328</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1329 id=1329 data-nosnippet>1329</a> .greater_than(Datum::float(<span class="number">1.0_f32</span>))
-<a href=#1330 id=1330 data-nosnippet>1330</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1331 id=1331 data-nosnippet>1331</a>
-<a href=#1332 id=1332 data-nosnippet>1332</a> <span class="kw">let </span>row_group_metadata = metadata.row_group(<span class="number">0</span>);
-<a href=#1333 id=1333 data-nosnippet>1333</a> <span class="kw">let </span>column_index = metadata.column_index().unwrap()[<span class="number">0</span>].to_vec();
-<a href=#1334 id=1334 data-nosnippet>1334</a> <span class="kw">let </span>offset_index = metadata.offset_index().unwrap()[<span class="number">0</span>].to_vec();
-<a href=#1335 id=1335 data-nosnippet>1335</a>
-<a href=#1336 id=1336 data-nosnippet>1336</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1337 id=1337 data-nosnippet>1337</a> <span class="kw-2">&</span>filter,
-<a href=#1338 id=1338 data-nosnippet>1338</a> <span class="kw-2">&</span>column_index,
-<a href=#1339 id=1339 data-nosnippet>1339</a> <span class="kw-2">&</span>offset_index,
-<a href=#1340 id=1340 data-nosnippet>1340</a> row_group_metadata,
-<a href=#1341 id=1341 data-nosnippet>1341</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1342 id=1342 data-nosnippet>1342</a> iceberg_schema_ref.as_ref(),
-<a href=#1343 id=1343 data-nosnippet>1343</a> )<span class="question-mark">?</span>;
-<a href=#1344 id=1344 data-nosnippet>1344</a>
-<a href=#1345 id=1345 data-nosnippet>1345</a> <span class="macro">assert_eq!</span>(result.len(), <span class="number">0</span>);
+<a href=#1328 id=1328 data-nosnippet>1328</a> <span class="kw">let </span>file = temp_file.reopen().unwrap();
+<a href=#1329 id=1329 data-nosnippet>1329</a> <span class="kw">let </span>options = ArrowReaderOptions::new().with_page_index_policy(PageIndexPolicy::Required);
+<a href=#1330 id=1330 data-nosnippet>1330</a> <span class="kw">let </span>reader = ParquetRecordBatchReaderBuilder::try_new_with_options(file, options).unwrap();
+<a href=#1331 id=1331 data-nosnippet>1331</a> <span class="kw">let </span>metadata = reader.metadata();
+<a href=#1332 id=1332 data-nosnippet>1332</a>
+<a href=#1333 id=1333 data-nosnippet>1333</a> <span class="kw">if </span>metadata.num_row_groups() == <span class="number">0 </span>|| metadata.row_group(<span class="number">0</span>).num_rows() == <span class="number">0 </span>{
+<a href=#1334 id=1334 data-nosnippet>1334</a> <span class="kw">return </span><span class="prelude-val">Ok</span>(());
+<a href=#1335 id=1335 data-nosnippet>1335</a> }
+<a href=#1336 id=1336 data-nosnippet>1336</a>
+<a href=#1337 id=1337 data-nosnippet>1337</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1338 id=1338 data-nosnippet>1338</a>
+<a href=#1339 id=1339 data-nosnippet>1339</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1340 id=1340 data-nosnippet>1340</a> .greater_than(Datum::float(<span class="number">1.0_f32</span>))
+<a href=#1341 id=1341 data-nosnippet>1341</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1342 id=1342 data-nosnippet>1342</a>
+<a href=#1343 id=1343 data-nosnippet>1343</a> <span class="kw">let </span>row_group_metadata = metadata.row_group(<span class="number">0</span>);
+<a href=#1344 id=1344 data-nosnippet>1344</a> <span class="kw">let </span>column_index = metadata.column_index().unwrap()[<span class="number">0</span>].to_vec();
+<a href=#1345 id=1345 data-nosnippet>1345</a> <span class="kw">let </span>offset_index = metadata.offset_index().unwrap()[<span class="number">0</span>].to_vec();
<a href=#1346 id=1346 data-nosnippet>1346</a>
-<a href=#1347 id=1347 data-nosnippet>1347</a> <span class="prelude-val">Ok</span>(())
-<a href=#1348 id=1348 data-nosnippet>1348</a> }
-<a href=#1349 id=1349 data-nosnippet>1349</a>
-<a href=#1350 id=1350 data-nosnippet>1350</a> <span class="attr">#[test]
-<a href=#1351 id=1351 data-nosnippet>1351</a> </span><span class="kw">fn </span>eval_is_null_select_only_pages_with_nulls() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1352 id=1352 data-nosnippet>1352</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1353 id=1353 data-nosnippet>1353</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1354 id=1354 data-nosnippet>1354</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1347 id=1347 data-nosnippet>1347</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1348 id=1348 data-nosnippet>1348</a> <span class="kw-2">&</span>filter,
+<a href=#1349 id=1349 data-nosnippet>1349</a> <span class="kw-2">&</span>column_index,
+<a href=#1350 id=1350 data-nosnippet>1350</a> <span class="kw-2">&</span>offset_index,
+<a href=#1351 id=1351 data-nosnippet>1351</a> row_group_metadata,
+<a href=#1352 id=1352 data-nosnippet>1352</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1353 id=1353 data-nosnippet>1353</a> iceberg_schema_ref.as_ref(),
+<a href=#1354 id=1354 data-nosnippet>1354</a> )<span class="question-mark">?</span>;
<a href=#1355 id=1355 data-nosnippet>1355</a>
-<a href=#1356 id=1356 data-nosnippet>1356</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1357 id=1357 data-nosnippet>1357</a> .is_null()
-<a href=#1358 id=1358 data-nosnippet>1358</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1359 id=1359 data-nosnippet>1359</a>
-<a href=#1360 id=1360 data-nosnippet>1360</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1361 id=1361 data-nosnippet>1361</a> <span class="kw-2">&</span>filter,
-<a href=#1362 id=1362 data-nosnippet>1362</a> <span class="kw-2">&</span>column_index,
-<a href=#1363 id=1363 data-nosnippet>1363</a> <span class="kw-2">&</span>offset_index,
-<a href=#1364 id=1364 data-nosnippet>1364</a> row_group_metadata,
-<a href=#1365 id=1365 data-nosnippet>1365</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1366 id=1366 data-nosnippet>1366</a> iceberg_schema_ref.as_ref(),
-<a href=#1367 id=1367 data-nosnippet>1367</a> )<span class="question-mark">?</span>;
-<a href=#1368 id=1368 data-nosnippet>1368</a>
-<a href=#1369 id=1369 data-nosnippet>1369</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[
-<a href=#1370 id=1370 data-nosnippet>1370</a> RowSelector::select(<span class="number">2048</span>),
-<a href=#1371 id=1371 data-nosnippet>1371</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#1372 id=1372 data-nosnippet>1372</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#1373 id=1373 data-nosnippet>1373</a> ];
-<a href=#1374 id=1374 data-nosnippet>1374</a>
-<a href=#1375 id=1375 data-nosnippet>1375</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1376 id=1376 data-nosnippet>1376</a>
-<a href=#1377 id=1377 data-nosnippet>1377</a> <span class="prelude-val">Ok</span>(())
-<a href=#1378 id=1378 data-nosnippet>1378</a> }
+<a href=#1356 id=1356 data-nosnippet>1356</a> <span class="macro">assert_eq!</span>(result.len(), <span class="number">0</span>);
+<a href=#1357 id=1357 data-nosnippet>1357</a>
+<a href=#1358 id=1358 data-nosnippet>1358</a> <span class="prelude-val">Ok</span>(())
+<a href=#1359 id=1359 data-nosnippet>1359</a> }
+<a href=#1360 id=1360 data-nosnippet>1360</a>
+<a href=#1361 id=1361 data-nosnippet>1361</a> <span class="attr">#[test]
+<a href=#1362 id=1362 data-nosnippet>1362</a> </span><span class="kw">fn </span>eval_is_null_select_only_pages_with_nulls() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1363 id=1363 data-nosnippet>1363</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1364 id=1364 data-nosnippet>1364</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1365 id=1365 data-nosnippet>1365</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1366 id=1366 data-nosnippet>1366</a>
+<a href=#1367 id=1367 data-nosnippet>1367</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1368 id=1368 data-nosnippet>1368</a> .is_null()
+<a href=#1369 id=1369 data-nosnippet>1369</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1370 id=1370 data-nosnippet>1370</a>
+<a href=#1371 id=1371 data-nosnippet>1371</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1372 id=1372 data-nosnippet>1372</a> <span class="kw-2">&</span>filter,
+<a href=#1373 id=1373 data-nosnippet>1373</a> <span class="kw-2">&</span>column_index,
+<a href=#1374 id=1374 data-nosnippet>1374</a> <span class="kw-2">&</span>offset_index,
+<a href=#1375 id=1375 data-nosnippet>1375</a> row_group_metadata,
+<a href=#1376 id=1376 data-nosnippet>1376</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1377 id=1377 data-nosnippet>1377</a> iceberg_schema_ref.as_ref(),
+<a href=#1378 id=1378 data-nosnippet>1378</a> )<span class="question-mark">?</span>;
<a href=#1379 id=1379 data-nosnippet>1379</a>
-<a href=#1380 id=1380 data-nosnippet>1380</a> <span class="attr">#[test]
-<a href=#1381 id=1381 data-nosnippet>1381</a> </span><span class="kw">fn </span>eval_is_not_null_dont_select_pages_with_all_nulls() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1382 id=1382 data-nosnippet>1382</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1383 id=1383 data-nosnippet>1383</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1384 id=1384 data-nosnippet>1384</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1380 id=1380 data-nosnippet>1380</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[
+<a href=#1381 id=1381 data-nosnippet>1381</a> RowSelector::select(<span class="number">2048</span>),
+<a href=#1382 id=1382 data-nosnippet>1382</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#1383 id=1383 data-nosnippet>1383</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#1384 id=1384 data-nosnippet>1384</a> ];
<a href=#1385 id=1385 data-nosnippet>1385</a>
-<a href=#1386 id=1386 data-nosnippet>1386</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1387 id=1387 data-nosnippet>1387</a> .is_not_null()
-<a href=#1388 id=1388 data-nosnippet>1388</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1389 id=1389 data-nosnippet>1389</a>
-<a href=#1390 id=1390 data-nosnippet>1390</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1391 id=1391 data-nosnippet>1391</a> <span class="kw-2">&</span>filter,
-<a href=#1392 id=1392 data-nosnippet>1392</a> <span class="kw-2">&</span>column_index,
-<a href=#1393 id=1393 data-nosnippet>1393</a> <span class="kw-2">&</span>offset_index,
-<a href=#1394 id=1394 data-nosnippet>1394</a> row_group_metadata,
-<a href=#1395 id=1395 data-nosnippet>1395</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1396 id=1396 data-nosnippet>1396</a> iceberg_schema_ref.as_ref(),
-<a href=#1397 id=1397 data-nosnippet>1397</a> )<span class="question-mark">?</span>;
-<a href=#1398 id=1398 data-nosnippet>1398</a>
-<a href=#1399 id=1399 data-nosnippet>1399</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::skip(<span class="number">2048</span>), RowSelector::select(<span class="number">2048</span>)];
+<a href=#1386 id=1386 data-nosnippet>1386</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1387 id=1387 data-nosnippet>1387</a>
+<a href=#1388 id=1388 data-nosnippet>1388</a> <span class="prelude-val">Ok</span>(())
+<a href=#1389 id=1389 data-nosnippet>1389</a> }
+<a href=#1390 id=1390 data-nosnippet>1390</a>
+<a href=#1391 id=1391 data-nosnippet>1391</a> <span class="attr">#[test]
+<a href=#1392 id=1392 data-nosnippet>1392</a> </span><span class="kw">fn </span>eval_is_not_null_dont_select_pages_with_all_nulls() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1393 id=1393 data-nosnippet>1393</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1394 id=1394 data-nosnippet>1394</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1395 id=1395 data-nosnippet>1395</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1396 id=1396 data-nosnippet>1396</a>
+<a href=#1397 id=1397 data-nosnippet>1397</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1398 id=1398 data-nosnippet>1398</a> .is_not_null()
+<a href=#1399 id=1399 data-nosnippet>1399</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
<a href=#1400 id=1400 data-nosnippet>1400</a>
-<a href=#1401 id=1401 data-nosnippet>1401</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1402 id=1402 data-nosnippet>1402</a>
-<a href=#1403 id=1403 data-nosnippet>1403</a> <span class="prelude-val">Ok</span>(())
-<a href=#1404 id=1404 data-nosnippet>1404</a> }
-<a href=#1405 id=1405 data-nosnippet>1405</a>
-<a href=#1406 id=1406 data-nosnippet>1406</a> <span class="attr">#[test]
-<a href=#1407 id=1407 data-nosnippet>1407</a> </span><span class="kw">fn </span>eval_is_nan_select_all() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1408 id=1408 data-nosnippet>1408</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1409 id=1409 data-nosnippet>1409</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1410 id=1410 data-nosnippet>1410</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1401 id=1401 data-nosnippet>1401</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1402 id=1402 data-nosnippet>1402</a> <span class="kw-2">&</span>filter,
+<a href=#1403 id=1403 data-nosnippet>1403</a> <span class="kw-2">&</span>column_index,
+<a href=#1404 id=1404 data-nosnippet>1404</a> <span class="kw-2">&</span>offset_index,
+<a href=#1405 id=1405 data-nosnippet>1405</a> row_group_metadata,
+<a href=#1406 id=1406 data-nosnippet>1406</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1407 id=1407 data-nosnippet>1407</a> iceberg_schema_ref.as_ref(),
+<a href=#1408 id=1408 data-nosnippet>1408</a> )<span class="question-mark">?</span>;
+<a href=#1409 id=1409 data-nosnippet>1409</a>
+<a href=#1410 id=1410 data-nosnippet>1410</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::skip(<span class="number">2048</span>), RowSelector::select(<span class="number">2048</span>)];
<a href=#1411 id=1411 data-nosnippet>1411</a>
-<a href=#1412 id=1412 data-nosnippet>1412</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1413 id=1413 data-nosnippet>1413</a> .is_nan()
-<a href=#1414 id=1414 data-nosnippet>1414</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1415 id=1415 data-nosnippet>1415</a>
-<a href=#1416 id=1416 data-nosnippet>1416</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1417 id=1417 data-nosnippet>1417</a> <span class="kw-2">&</span>filter,
-<a href=#1418 id=1418 data-nosnippet>1418</a> <span class="kw-2">&</span>column_index,
-<a href=#1419 id=1419 data-nosnippet>1419</a> <span class="kw-2">&</span>offset_index,
-<a href=#1420 id=1420 data-nosnippet>1420</a> row_group_metadata,
-<a href=#1421 id=1421 data-nosnippet>1421</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1422 id=1422 data-nosnippet>1422</a> iceberg_schema_ref.as_ref(),
-<a href=#1423 id=1423 data-nosnippet>1423</a> )<span class="question-mark">?</span>;
-<a href=#1424 id=1424 data-nosnippet>1424</a>
-<a href=#1425 id=1425 data-nosnippet>1425</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">4096</span>)];
+<a href=#1412 id=1412 data-nosnippet>1412</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1413 id=1413 data-nosnippet>1413</a>
+<a href=#1414 id=1414 data-nosnippet>1414</a> <span class="prelude-val">Ok</span>(())
+<a href=#1415 id=1415 data-nosnippet>1415</a> }
+<a href=#1416 id=1416 data-nosnippet>1416</a>
+<a href=#1417 id=1417 data-nosnippet>1417</a> <span class="attr">#[test]
+<a href=#1418 id=1418 data-nosnippet>1418</a> </span><span class="kw">fn </span>eval_is_nan_select_all() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1419 id=1419 data-nosnippet>1419</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1420 id=1420 data-nosnippet>1420</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1421 id=1421 data-nosnippet>1421</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1422 id=1422 data-nosnippet>1422</a>
+<a href=#1423 id=1423 data-nosnippet>1423</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1424 id=1424 data-nosnippet>1424</a> .is_nan()
+<a href=#1425 id=1425 data-nosnippet>1425</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
<a href=#1426 id=1426 data-nosnippet>1426</a>
-<a href=#1427 id=1427 data-nosnippet>1427</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1428 id=1428 data-nosnippet>1428</a>
-<a href=#1429 id=1429 data-nosnippet>1429</a> <span class="prelude-val">Ok</span>(())
-<a href=#1430 id=1430 data-nosnippet>1430</a> }
-<a href=#1431 id=1431 data-nosnippet>1431</a>
-<a href=#1432 id=1432 data-nosnippet>1432</a> <span class="attr">#[test]
-<a href=#1433 id=1433 data-nosnippet>1433</a> </span><span class="kw">fn </span>eval_not_nan_select_all() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1434 id=1434 data-nosnippet>1434</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1435 id=1435 data-nosnippet>1435</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1436 id=1436 data-nosnippet>1436</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1427 id=1427 data-nosnippet>1427</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1428 id=1428 data-nosnippet>1428</a> <span class="kw-2">&</span>filter,
+<a href=#1429 id=1429 data-nosnippet>1429</a> <span class="kw-2">&</span>column_index,
+<a href=#1430 id=1430 data-nosnippet>1430</a> <span class="kw-2">&</span>offset_index,
+<a href=#1431 id=1431 data-nosnippet>1431</a> row_group_metadata,
+<a href=#1432 id=1432 data-nosnippet>1432</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1433 id=1433 data-nosnippet>1433</a> iceberg_schema_ref.as_ref(),
+<a href=#1434 id=1434 data-nosnippet>1434</a> )<span class="question-mark">?</span>;
+<a href=#1435 id=1435 data-nosnippet>1435</a>
+<a href=#1436 id=1436 data-nosnippet>1436</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">4096</span>)];
<a href=#1437 id=1437 data-nosnippet>1437</a>
-<a href=#1438 id=1438 data-nosnippet>1438</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1439 id=1439 data-nosnippet>1439</a> .is_not_nan()
-<a href=#1440 id=1440 data-nosnippet>1440</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1441 id=1441 data-nosnippet>1441</a>
-<a href=#1442 id=1442 data-nosnippet>1442</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1443 id=1443 data-nosnippet>1443</a> <span class="kw-2">&</span>filter,
-<a href=#1444 id=1444 data-nosnippet>1444</a> <span class="kw-2">&</span>column_index,
-<a href=#1445 id=1445 data-nosnippet>1445</a> <span class="kw-2">&</span>offset_index,
-<a href=#1446 id=1446 data-nosnippet>1446</a> row_group_metadata,
-<a href=#1447 id=1447 data-nosnippet>1447</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1448 id=1448 data-nosnippet>1448</a> iceberg_schema_ref.as_ref(),
-<a href=#1449 id=1449 data-nosnippet>1449</a> )<span class="question-mark">?</span>;
-<a href=#1450 id=1450 data-nosnippet>1450</a>
-<a href=#1451 id=1451 data-nosnippet>1451</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">4096</span>)];
+<a href=#1438 id=1438 data-nosnippet>1438</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1439 id=1439 data-nosnippet>1439</a>
+<a href=#1440 id=1440 data-nosnippet>1440</a> <span class="prelude-val">Ok</span>(())
+<a href=#1441 id=1441 data-nosnippet>1441</a> }
+<a href=#1442 id=1442 data-nosnippet>1442</a>
+<a href=#1443 id=1443 data-nosnippet>1443</a> <span class="attr">#[test]
+<a href=#1444 id=1444 data-nosnippet>1444</a> </span><span class="kw">fn </span>eval_not_nan_select_all() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1445 id=1445 data-nosnippet>1445</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1446 id=1446 data-nosnippet>1446</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1447 id=1447 data-nosnippet>1447</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1448 id=1448 data-nosnippet>1448</a>
+<a href=#1449 id=1449 data-nosnippet>1449</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1450 id=1450 data-nosnippet>1450</a> .is_not_nan()
+<a href=#1451 id=1451 data-nosnippet>1451</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
<a href=#1452 id=1452 data-nosnippet>1452</a>
-<a href=#1453 id=1453 data-nosnippet>1453</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1454 id=1454 data-nosnippet>1454</a>
-<a href=#1455 id=1455 data-nosnippet>1455</a> <span class="prelude-val">Ok</span>(())
-<a href=#1456 id=1456 data-nosnippet>1456</a> }
-<a href=#1457 id=1457 data-nosnippet>1457</a>
-<a href=#1458 id=1458 data-nosnippet>1458</a> <span class="attr">#[test]
-<a href=#1459 id=1459 data-nosnippet>1459</a> </span><span class="kw">fn </span>eval_inequality_nan_datum_all_rows_except_all_null_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1460 id=1460 data-nosnippet>1460</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1461 id=1461 data-nosnippet>1461</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1462 id=1462 data-nosnippet>1462</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1453 id=1453 data-nosnippet>1453</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1454 id=1454 data-nosnippet>1454</a> <span class="kw-2">&</span>filter,
+<a href=#1455 id=1455 data-nosnippet>1455</a> <span class="kw-2">&</span>column_index,
+<a href=#1456 id=1456 data-nosnippet>1456</a> <span class="kw-2">&</span>offset_index,
+<a href=#1457 id=1457 data-nosnippet>1457</a> row_group_metadata,
+<a href=#1458 id=1458 data-nosnippet>1458</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1459 id=1459 data-nosnippet>1459</a> iceberg_schema_ref.as_ref(),
+<a href=#1460 id=1460 data-nosnippet>1460</a> )<span class="question-mark">?</span>;
+<a href=#1461 id=1461 data-nosnippet>1461</a>
+<a href=#1462 id=1462 data-nosnippet>1462</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">4096</span>)];
<a href=#1463 id=1463 data-nosnippet>1463</a>
-<a href=#1464 id=1464 data-nosnippet>1464</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1465 id=1465 data-nosnippet>1465</a> .less_than(Datum::float(f32::NAN))
-<a href=#1466 id=1466 data-nosnippet>1466</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1467 id=1467 data-nosnippet>1467</a>
-<a href=#1468 id=1468 data-nosnippet>1468</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1469 id=1469 data-nosnippet>1469</a> <span class="kw-2">&</span>filter,
-<a href=#1470 id=1470 data-nosnippet>1470</a> <span class="kw-2">&</span>column_index,
-<a href=#1471 id=1471 data-nosnippet>1471</a> <span class="kw-2">&</span>offset_index,
-<a href=#1472 id=1472 data-nosnippet>1472</a> row_group_metadata,
-<a href=#1473 id=1473 data-nosnippet>1473</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1474 id=1474 data-nosnippet>1474</a> iceberg_schema_ref.as_ref(),
-<a href=#1475 id=1475 data-nosnippet>1475</a> )<span class="question-mark">?</span>;
-<a href=#1476 id=1476 data-nosnippet>1476</a>
-<a href=#1477 id=1477 data-nosnippet>1477</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::skip(<span class="number">2048</span>), RowSelector::select(<span class="number">2048</span>)];
+<a href=#1464 id=1464 data-nosnippet>1464</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1465 id=1465 data-nosnippet>1465</a>
+<a href=#1466 id=1466 data-nosnippet>1466</a> <span class="prelude-val">Ok</span>(())
+<a href=#1467 id=1467 data-nosnippet>1467</a> }
+<a href=#1468 id=1468 data-nosnippet>1468</a>
+<a href=#1469 id=1469 data-nosnippet>1469</a> <span class="attr">#[test]
+<a href=#1470 id=1470 data-nosnippet>1470</a> </span><span class="kw">fn </span>eval_inequality_nan_datum_all_rows_except_all_null_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1471 id=1471 data-nosnippet>1471</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1472 id=1472 data-nosnippet>1472</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1473 id=1473 data-nosnippet>1473</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1474 id=1474 data-nosnippet>1474</a>
+<a href=#1475 id=1475 data-nosnippet>1475</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1476 id=1476 data-nosnippet>1476</a> .less_than(Datum::float(f32::NAN))
+<a href=#1477 id=1477 data-nosnippet>1477</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
<a href=#1478 id=1478 data-nosnippet>1478</a>
-<a href=#1479 id=1479 data-nosnippet>1479</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1480 id=1480 data-nosnippet>1480</a>
-<a href=#1481 id=1481 data-nosnippet>1481</a> <span class="prelude-val">Ok</span>(())
-<a href=#1482 id=1482 data-nosnippet>1482</a> }
-<a href=#1483 id=1483 data-nosnippet>1483</a>
-<a href=#1484 id=1484 data-nosnippet>1484</a> <span class="attr">#[test]
-<a href=#1485 id=1485 data-nosnippet>1485</a> </span><span class="kw">fn </span>eval_inequality_pages_containing_value_except_all_null_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1486 id=1486 data-nosnippet>1486</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1487 id=1487 data-nosnippet>1487</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1488 id=1488 data-nosnippet>1488</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1479 id=1479 data-nosnippet>1479</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1480 id=1480 data-nosnippet>1480</a> <span class="kw-2">&</span>filter,
+<a href=#1481 id=1481 data-nosnippet>1481</a> <span class="kw-2">&</span>column_index,
+<a href=#1482 id=1482 data-nosnippet>1482</a> <span class="kw-2">&</span>offset_index,
+<a href=#1483 id=1483 data-nosnippet>1483</a> row_group_metadata,
+<a href=#1484 id=1484 data-nosnippet>1484</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1485 id=1485 data-nosnippet>1485</a> iceberg_schema_ref.as_ref(),
+<a href=#1486 id=1486 data-nosnippet>1486</a> )<span class="question-mark">?</span>;
+<a href=#1487 id=1487 data-nosnippet>1487</a>
+<a href=#1488 id=1488 data-nosnippet>1488</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::skip(<span class="number">2048</span>), RowSelector::select(<span class="number">2048</span>)];
<a href=#1489 id=1489 data-nosnippet>1489</a>
-<a href=#1490 id=1490 data-nosnippet>1490</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1491 id=1491 data-nosnippet>1491</a> .less_than(Datum::float(<span class="number">5.0_f32</span>))
-<a href=#1492 id=1492 data-nosnippet>1492</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1493 id=1493 data-nosnippet>1493</a>
-<a href=#1494 id=1494 data-nosnippet>1494</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1495 id=1495 data-nosnippet>1495</a> <span class="kw-2">&</span>filter,
-<a href=#1496 id=1496 data-nosnippet>1496</a> <span class="kw-2">&</span>column_index,
-<a href=#1497 id=1497 data-nosnippet>1497</a> <span class="kw-2">&</span>offset_index,
-<a href=#1498 id=1498 data-nosnippet>1498</a> row_group_metadata,
-<a href=#1499 id=1499 data-nosnippet>1499</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1500 id=1500 data-nosnippet>1500</a> iceberg_schema_ref.as_ref(),
-<a href=#1501 id=1501 data-nosnippet>1501</a> )<span class="question-mark">?</span>;
-<a href=#1502 id=1502 data-nosnippet>1502</a>
-<a href=#1503 id=1503 data-nosnippet>1503</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[
-<a href=#1504 id=1504 data-nosnippet>1504</a> RowSelector::skip(<span class="number">2048</span>),
-<a href=#1505 id=1505 data-nosnippet>1505</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#1506 id=1506 data-nosnippet>1506</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#1507 id=1507 data-nosnippet>1507</a> ];
-<a href=#1508 id=1508 data-nosnippet>1508</a>
-<a href=#1509 id=1509 data-nosnippet>1509</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1510 id=1510 data-nosnippet>1510</a>
-<a href=#1511 id=1511 data-nosnippet>1511</a> <span class="prelude-val">Ok</span>(())
-<a href=#1512 id=1512 data-nosnippet>1512</a> }
+<a href=#1490 id=1490 data-nosnippet>1490</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1491 id=1491 data-nosnippet>1491</a>
+<a href=#1492 id=1492 data-nosnippet>1492</a> <span class="prelude-val">Ok</span>(())
+<a href=#1493 id=1493 data-nosnippet>1493</a> }
+<a href=#1494 id=1494 data-nosnippet>1494</a>
+<a href=#1495 id=1495 data-nosnippet>1495</a> <span class="attr">#[test]
+<a href=#1496 id=1496 data-nosnippet>1496</a> </span><span class="kw">fn </span>eval_inequality_pages_containing_value_except_all_null_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1497 id=1497 data-nosnippet>1497</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1498 id=1498 data-nosnippet>1498</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1499 id=1499 data-nosnippet>1499</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1500 id=1500 data-nosnippet>1500</a>
+<a href=#1501 id=1501 data-nosnippet>1501</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1502 id=1502 data-nosnippet>1502</a> .less_than(Datum::float(<span class="number">5.0_f32</span>))
+<a href=#1503 id=1503 data-nosnippet>1503</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1504 id=1504 data-nosnippet>1504</a>
+<a href=#1505 id=1505 data-nosnippet>1505</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1506 id=1506 data-nosnippet>1506</a> <span class="kw-2">&</span>filter,
+<a href=#1507 id=1507 data-nosnippet>1507</a> <span class="kw-2">&</span>column_index,
+<a href=#1508 id=1508 data-nosnippet>1508</a> <span class="kw-2">&</span>offset_index,
+<a href=#1509 id=1509 data-nosnippet>1509</a> row_group_metadata,
+<a href=#1510 id=1510 data-nosnippet>1510</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1511 id=1511 data-nosnippet>1511</a> iceberg_schema_ref.as_ref(),
+<a href=#1512 id=1512 data-nosnippet>1512</a> )<span class="question-mark">?</span>;
<a href=#1513 id=1513 data-nosnippet>1513</a>
-<a href=#1514 id=1514 data-nosnippet>1514</a> <span class="attr">#[test]
-<a href=#1515 id=1515 data-nosnippet>1515</a> </span><span class="kw">fn </span>eval_eq_pages_containing_value_except_all_null_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1516 id=1516 data-nosnippet>1516</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1517 id=1517 data-nosnippet>1517</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1518 id=1518 data-nosnippet>1518</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1514 id=1514 data-nosnippet>1514</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[
+<a href=#1515 id=1515 data-nosnippet>1515</a> RowSelector::skip(<span class="number">2048</span>),
+<a href=#1516 id=1516 data-nosnippet>1516</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#1517 id=1517 data-nosnippet>1517</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#1518 id=1518 data-nosnippet>1518</a> ];
<a href=#1519 id=1519 data-nosnippet>1519</a>
-<a href=#1520 id=1520 data-nosnippet>1520</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1521 id=1521 data-nosnippet>1521</a> .equal_to(Datum::float(<span class="number">5.0_f32</span>))
-<a href=#1522 id=1522 data-nosnippet>1522</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1523 id=1523 data-nosnippet>1523</a>
-<a href=#1524 id=1524 data-nosnippet>1524</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1525 id=1525 data-nosnippet>1525</a> <span class="kw-2">&</span>filter,
-<a href=#1526 id=1526 data-nosnippet>1526</a> <span class="kw-2">&</span>column_index,
-<a href=#1527 id=1527 data-nosnippet>1527</a> <span class="kw-2">&</span>offset_index,
-<a href=#1528 id=1528 data-nosnippet>1528</a> row_group_metadata,
-<a href=#1529 id=1529 data-nosnippet>1529</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1530 id=1530 data-nosnippet>1530</a> iceberg_schema_ref.as_ref(),
-<a href=#1531 id=1531 data-nosnippet>1531</a> )<span class="question-mark">?</span>;
-<a href=#1532 id=1532 data-nosnippet>1532</a>
-<a href=#1533 id=1533 data-nosnippet>1533</a> <span class="comment">// Pages 0-1: all null (skip)
-<a href=#1534 id=1534 data-nosnippet>1534</a> // Page 2: 0-10 (select, might contain 5.0)
-<a href=#1535 id=1535 data-nosnippet>1535</a> // Page 3: 10-20 (skip, min > 5.0)
-<a href=#1536 id=1536 data-nosnippet>1536</a> </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[
-<a href=#1537 id=1537 data-nosnippet>1537</a> RowSelector::skip(<span class="number">2048</span>),
-<a href=#1538 id=1538 data-nosnippet>1538</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#1539 id=1539 data-nosnippet>1539</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#1540 id=1540 data-nosnippet>1540</a> ];
-<a href=#1541 id=1541 data-nosnippet>1541</a>
-<a href=#1542 id=1542 data-nosnippet>1542</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1520 id=1520 data-nosnippet>1520</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1521 id=1521 data-nosnippet>1521</a>
+<a href=#1522 id=1522 data-nosnippet>1522</a> <span class="prelude-val">Ok</span>(())
+<a href=#1523 id=1523 data-nosnippet>1523</a> }
+<a href=#1524 id=1524 data-nosnippet>1524</a>
+<a href=#1525 id=1525 data-nosnippet>1525</a> <span class="attr">#[test]
+<a href=#1526 id=1526 data-nosnippet>1526</a> </span><span class="kw">fn </span>eval_eq_pages_containing_value_except_all_null_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1527 id=1527 data-nosnippet>1527</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1528 id=1528 data-nosnippet>1528</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1529 id=1529 data-nosnippet>1529</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1530 id=1530 data-nosnippet>1530</a>
+<a href=#1531 id=1531 data-nosnippet>1531</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1532 id=1532 data-nosnippet>1532</a> .equal_to(Datum::float(<span class="number">5.0_f32</span>))
+<a href=#1533 id=1533 data-nosnippet>1533</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1534 id=1534 data-nosnippet>1534</a>
+<a href=#1535 id=1535 data-nosnippet>1535</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1536 id=1536 data-nosnippet>1536</a> <span class="kw-2">&</span>filter,
+<a href=#1537 id=1537 data-nosnippet>1537</a> <span class="kw-2">&</span>column_index,
+<a href=#1538 id=1538 data-nosnippet>1538</a> <span class="kw-2">&</span>offset_index,
+<a href=#1539 id=1539 data-nosnippet>1539</a> row_group_metadata,
+<a href=#1540 id=1540 data-nosnippet>1540</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1541 id=1541 data-nosnippet>1541</a> iceberg_schema_ref.as_ref(),
+<a href=#1542 id=1542 data-nosnippet>1542</a> )<span class="question-mark">?</span>;
<a href=#1543 id=1543 data-nosnippet>1543</a>
-<a href=#1544 id=1544 data-nosnippet>1544</a> <span class="prelude-val">Ok</span>(())
-<a href=#1545 id=1545 data-nosnippet>1545</a> }
-<a href=#1546 id=1546 data-nosnippet>1546</a>
-<a href=#1547 id=1547 data-nosnippet>1547</a> <span class="attr">#[test]
-<a href=#1548 id=1548 data-nosnippet>1548</a> </span><span class="kw">fn </span>eval_not_eq_all_rows() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1549 id=1549 data-nosnippet>1549</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1550 id=1550 data-nosnippet>1550</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1551 id=1551 data-nosnippet>1551</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1544 id=1544 data-nosnippet>1544</a> <span class="comment">// Pages 0-1: all null (skip)
+<a href=#1545 id=1545 data-nosnippet>1545</a> // Page 2: 0-10 (select, might contain 5.0)
+<a href=#1546 id=1546 data-nosnippet>1546</a> // Page 3: 10-20 (skip, min > 5.0)
+<a href=#1547 id=1547 data-nosnippet>1547</a> </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[
+<a href=#1548 id=1548 data-nosnippet>1548</a> RowSelector::skip(<span class="number">2048</span>),
+<a href=#1549 id=1549 data-nosnippet>1549</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#1550 id=1550 data-nosnippet>1550</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#1551 id=1551 data-nosnippet>1551</a> ];
<a href=#1552 id=1552 data-nosnippet>1552</a>
-<a href=#1553 id=1553 data-nosnippet>1553</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1554 id=1554 data-nosnippet>1554</a> .not_equal_to(Datum::float(<span class="number">5.0_f32</span>))
-<a href=#1555 id=1555 data-nosnippet>1555</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1556 id=1556 data-nosnippet>1556</a>
-<a href=#1557 id=1557 data-nosnippet>1557</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1558 id=1558 data-nosnippet>1558</a> <span class="kw-2">&</span>filter,
-<a href=#1559 id=1559 data-nosnippet>1559</a> <span class="kw-2">&</span>column_index,
-<a href=#1560 id=1560 data-nosnippet>1560</a> <span class="kw-2">&</span>offset_index,
-<a href=#1561 id=1561 data-nosnippet>1561</a> row_group_metadata,
-<a href=#1562 id=1562 data-nosnippet>1562</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1563 id=1563 data-nosnippet>1563</a> iceberg_schema_ref.as_ref(),
-<a href=#1564 id=1564 data-nosnippet>1564</a> )<span class="question-mark">?</span>;
-<a href=#1565 id=1565 data-nosnippet>1565</a>
-<a href=#1566 id=1566 data-nosnippet>1566</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">4096</span>)];
+<a href=#1553 id=1553 data-nosnippet>1553</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1554 id=1554 data-nosnippet>1554</a>
+<a href=#1555 id=1555 data-nosnippet>1555</a> <span class="prelude-val">Ok</span>(())
+<a href=#1556 id=1556 data-nosnippet>1556</a> }
+<a href=#1557 id=1557 data-nosnippet>1557</a>
+<a href=#1558 id=1558 data-nosnippet>1558</a> <span class="attr">#[test]
+<a href=#1559 id=1559 data-nosnippet>1559</a> </span><span class="kw">fn </span>eval_not_eq_all_rows() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1560 id=1560 data-nosnippet>1560</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1561 id=1561 data-nosnippet>1561</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1562 id=1562 data-nosnippet>1562</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1563 id=1563 data-nosnippet>1563</a>
+<a href=#1564 id=1564 data-nosnippet>1564</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1565 id=1565 data-nosnippet>1565</a> .not_equal_to(Datum::float(<span class="number">5.0_f32</span>))
+<a href=#1566 id=1566 data-nosnippet>1566</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
<a href=#1567 id=1567 data-nosnippet>1567</a>
-<a href=#1568 id=1568 data-nosnippet>1568</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1569 id=1569 data-nosnippet>1569</a>
-<a href=#1570 id=1570 data-nosnippet>1570</a> <span class="prelude-val">Ok</span>(())
-<a href=#1571 id=1571 data-nosnippet>1571</a> }
-<a href=#1572 id=1572 data-nosnippet>1572</a>
-<a href=#1573 id=1573 data-nosnippet>1573</a> <span class="attr">#[test]
-<a href=#1574 id=1574 data-nosnippet>1574</a> </span><span class="kw">fn </span>eval_inequality_prunes_binary_pages_with_non_utf8_bounds() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1575 id=1575 data-nosnippet>1575</a> <span class="kw">let </span>(metadata, _temp_file) = create_binary_parquet_file()<span class="question-mark">?</span>;
-<a href=#1576 id=1576 data-nosnippet>1576</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1577 id=1577 data-nosnippet>1577</a>
-<a href=#1578 id=1578 data-nosnippet>1578</a> <span class="kw">let </span>iceberg_schema = Arc::new(
-<a href=#1579 id=1579 data-nosnippet>1579</a> Schema::builder()
-<a href=#1580 id=1580 data-nosnippet>1580</a> .with_fields([Arc::new(NestedField::new(
-<a href=#1581 id=1581 data-nosnippet>1581</a> <span class="number">1</span>,
-<a href=#1582 id=1582 data-nosnippet>1582</a> <span class="string">"col_binary"</span>,
-<a href=#1583 id=1583 data-nosnippet>1583</a> Type::Primitive(PrimitiveType::Binary),
-<a href=#1584 id=1584 data-nosnippet>1584</a> <span class="bool-val">true</span>,
-<a href=#1585 id=1585 data-nosnippet>1585</a> ))])
-<a href=#1586 id=1586 data-nosnippet>1586</a> .build()<span class="question-mark">?</span>,
-<a href=#1587 id=1587 data-nosnippet>1587</a> );
-<a href=#1588 id=1588 data-nosnippet>1588</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
-<a href=#1589 id=1589 data-nosnippet>1589</a>
-<a href=#1590 id=1590 data-nosnippet>1590</a> <span class="comment">// Page 0 bounds are [0x01], page 1 bounds are [0xff, 0x00]; only page 1
-<a href=#1591 id=1591 data-nosnippet>1591</a> // exceeds [0x80]. Decoding [0xff, 0x00] as UTF-8 would panic.
-<a href=#1592 id=1592 data-nosnippet>1592</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_binary"</span>)
-<a href=#1593 id=1593 data-nosnippet>1593</a> .greater_than(Datum::binary(<span class="macro">vec!</span>[<span class="number">0x80u8</span>]))
-<a href=#1594 id=1594 data-nosnippet>1594</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1595 id=1595 data-nosnippet>1595</a>
-<a href=#1596 id=1596 data-nosnippet>1596</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1597 id=1597 data-nosnippet>1597</a> <span class="kw-2">&</span>filter,
-<a href=#1598 id=1598 data-nosnippet>1598</a> <span class="kw-2">&</span>column_index,
-<a href=#1599 id=1599 data-nosnippet>1599</a> <span class="kw-2">&</span>offset_index,
-<a href=#1600 id=1600 data-nosnippet>1600</a> row_group_metadata,
-<a href=#1601 id=1601 data-nosnippet>1601</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1602 id=1602 data-nosnippet>1602</a> iceberg_schema.as_ref(),
-<a href=#1603 id=1603 data-nosnippet>1603</a> )<span class="question-mark">?</span>;
-<a href=#1604 id=1604 data-nosnippet>1604</a>
-<a href=#1605 id=1605 data-nosnippet>1605</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::skip(<span class="number">1024</span>), RowSelector::select(<span class="number">1024</span>)];
+<a href=#1568 id=1568 data-nosnippet>1568</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1569 id=1569 data-nosnippet>1569</a> <span class="kw-2">&</span>filter,
+<a href=#1570 id=1570 data-nosnippet>1570</a> <span class="kw-2">&</span>column_index,
+<a href=#1571 id=1571 data-nosnippet>1571</a> <span class="kw-2">&</span>offset_index,
+<a href=#1572 id=1572 data-nosnippet>1572</a> row_group_metadata,
+<a href=#1573 id=1573 data-nosnippet>1573</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1574 id=1574 data-nosnippet>1574</a> iceberg_schema_ref.as_ref(),
+<a href=#1575 id=1575 data-nosnippet>1575</a> )<span class="question-mark">?</span>;
+<a href=#1576 id=1576 data-nosnippet>1576</a>
+<a href=#1577 id=1577 data-nosnippet>1577</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">4096</span>)];
+<a href=#1578 id=1578 data-nosnippet>1578</a>
+<a href=#1579 id=1579 data-nosnippet>1579</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1580 id=1580 data-nosnippet>1580</a>
+<a href=#1581 id=1581 data-nosnippet>1581</a> <span class="prelude-val">Ok</span>(())
+<a href=#1582 id=1582 data-nosnippet>1582</a> }
+<a href=#1583 id=1583 data-nosnippet>1583</a>
+<a href=#1584 id=1584 data-nosnippet>1584</a> <span class="attr">#[test]
+<a href=#1585 id=1585 data-nosnippet>1585</a> </span><span class="kw">fn </span>eval_inequality_prunes_binary_pages_with_non_utf8_bounds() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1586 id=1586 data-nosnippet>1586</a> <span class="kw">let </span>(metadata, _temp_file) = create_binary_parquet_file()<span class="question-mark">?</span>;
+<a href=#1587 id=1587 data-nosnippet>1587</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1588 id=1588 data-nosnippet>1588</a>
+<a href=#1589 id=1589 data-nosnippet>1589</a> <span class="kw">let </span>iceberg_schema = Arc::new(
+<a href=#1590 id=1590 data-nosnippet>1590</a> Schema::builder()
+<a href=#1591 id=1591 data-nosnippet>1591</a> .with_fields([Arc::new(NestedField::new(
+<a href=#1592 id=1592 data-nosnippet>1592</a> <span class="number">1</span>,
+<a href=#1593 id=1593 data-nosnippet>1593</a> <span class="string">"col_binary"</span>,
+<a href=#1594 id=1594 data-nosnippet>1594</a> Type::Primitive(PrimitiveType::Binary),
+<a href=#1595 id=1595 data-nosnippet>1595</a> <span class="bool-val">true</span>,
+<a href=#1596 id=1596 data-nosnippet>1596</a> ))])
+<a href=#1597 id=1597 data-nosnippet>1597</a> .build()<span class="question-mark">?</span>,
+<a href=#1598 id=1598 data-nosnippet>1598</a> );
+<a href=#1599 id=1599 data-nosnippet>1599</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
+<a href=#1600 id=1600 data-nosnippet>1600</a>
+<a href=#1601 id=1601 data-nosnippet>1601</a> <span class="comment">// Page 0 bounds are [0x01], page 1 bounds are [0xff, 0x00]; only page 1
+<a href=#1602 id=1602 data-nosnippet>1602</a> // exceeds [0x80]. Decoding [0xff, 0x00] as UTF-8 would panic.
+<a href=#1603 id=1603 data-nosnippet>1603</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_binary"</span>)
+<a href=#1604 id=1604 data-nosnippet>1604</a> .greater_than(Datum::binary(<span class="macro">vec!</span>[<span class="number">0x80u8</span>]))
+<a href=#1605 id=1605 data-nosnippet>1605</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
<a href=#1606 id=1606 data-nosnippet>1606</a>
-<a href=#1607 id=1607 data-nosnippet>1607</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1608 id=1608 data-nosnippet>1608</a>
-<a href=#1609 id=1609 data-nosnippet>1609</a> <span class="prelude-val">Ok</span>(())
-<a href=#1610 id=1610 data-nosnippet>1610</a> }
-<a href=#1611 id=1611 data-nosnippet>1611</a>
-<a href=#1612 id=1612 data-nosnippet>1612</a> <span class="attr">#[test]
-<a href=#1613 id=1613 data-nosnippet>1613</a> </span><span class="kw">fn </span>eval_skips_pruning_for_byte_array_decimal_bounds() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1614 id=1614 data-nosnippet>1614</a> <span class="comment">// A non-spec writer can store a decimal column as BYTE_ARRAY. Those
-<a href=#1615 id=1615 data-nosnippet>1615</a> // bounds can't be decoded safely (Parquet may truncate them, and
-<a href=#1616 id=1616 data-nosnippet>1616</a> // truncation only preserves lexicographic byte order, not decimal
-<a href=#1617 id=1617 data-nosnippet>1617</a> // order), so the evaluator skips page pruning rather than prune pages
-<a href=#1618 id=1618 data-nosnippet>1618</a> // that might match.
-<a href=#1619 id=1619 data-nosnippet>1619</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_binary_parquet_file()<span class="question-mark">?</span>;
-<a href=#1620 id=1620 data-nosnippet>1620</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1621 id=1621 data-nosnippet>1621</a>
-<a href=#1622 id=1622 data-nosnippet>1622</a> <span class="kw">let </span>iceberg_schema = Arc::new(
-<a href=#1623 id=1623 data-nosnippet>1623</a> Schema::builder()
-<a href=#1624 id=1624 data-nosnippet>1624</a> .with_fields([Arc::new(NestedField::new(
-<a href=#1625 id=1625 data-nosnippet>1625</a> <span class="number">1</span>,
-<a href=#1626 id=1626 data-nosnippet>1626</a> <span class="string">"col_decimal"</span>,
-<a href=#1627 id=1627 data-nosnippet>1627</a> Type::Primitive(PrimitiveType::Decimal {
-<a href=#1628 id=1628 data-nosnippet>1628</a> precision: <span class="number">10</span>,
-<a href=#1629 id=1629 data-nosnippet>1629</a> scale: <span class="number">2</span>,
-<a href=#1630 id=1630 data-nosnippet>1630</a> }),
-<a href=#1631 id=1631 data-nosnippet>1631</a> <span class="bool-val">true</span>,
-<a href=#1632 id=1632 data-nosnippet>1632</a> ))])
-<a href=#1633 id=1633 data-nosnippet>1633</a> .build()<span class="question-mark">?</span>,
-<a href=#1634 id=1634 data-nosnippet>1634</a> );
-<a href=#1635 id=1635 data-nosnippet>1635</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
-<a href=#1636 id=1636 data-nosnippet>1636</a>
-<a href=#1637 id=1637 data-nosnippet>1637</a> <span class="comment">// A predicate that would prune every page if the bounds were decoded.
-<a href=#1638 id=1638 data-nosnippet>1638</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#1639 id=1639 data-nosnippet>1639</a> .greater_than(Datum::decimal_with_precision(
-<a href=#1640 id=1640 data-nosnippet>1640</a> decimal_from_i128_with_scale(<span class="number">99999</span>, <span class="number">2</span>),
-<a href=#1641 id=1641 data-nosnippet>1641</a> <span class="number">10</span>,
-<a href=#1642 id=1642 data-nosnippet>1642</a> )<span class="question-mark">?</span>)
-<a href=#1643 id=1643 data-nosnippet>1643</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1644 id=1644 data-nosnippet>1644</a>
-<a href=#1645 id=1645 data-nosnippet>1645</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1646 id=1646 data-nosnippet>1646</a> <span class="kw-2">&</span>filter,
-<a href=#1647 id=1647 data-nosnippet>1647</a> <span class="kw-2">&</span>column_index,
-<a href=#1648 id=1648 data-nosnippet>1648</a> <span class="kw-2">&</span>offset_index,
-<a href=#1649 id=1649 data-nosnippet>1649</a> row_group_metadata,
-<a href=#1650 id=1650 data-nosnippet>1650</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1651 id=1651 data-nosnippet>1651</a> iceberg_schema.as_ref(),
-<a href=#1652 id=1652 data-nosnippet>1652</a> )<span class="question-mark">?</span>;
-<a href=#1653 id=1653 data-nosnippet>1653</a>
-<a href=#1654 id=1654 data-nosnippet>1654</a> <span class="comment">// Both 1024-row pages survive: no page pruning is applied.
-<a href=#1655 id=1655 data-nosnippet>1655</a> </span><span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[RowSelector::select(<span class="number">2048</span>)]);
-<a href=#1656 id=1656 data-nosnippet>1656</a>
-<a href=#1657 id=1657 data-nosnippet>1657</a> <span class="prelude-val">Ok</span>(())
-<a href=#1658 id=1658 data-nosnippet>1658</a> }
-<a href=#1659 id=1659 data-nosnippet>1659</a>
-<a href=#1660 id=1660 data-nosnippet>1660</a> <span class="attr">#[test]
-<a href=#1661 id=1661 data-nosnippet>1661</a> </span><span class="kw">fn </span>eval_skips_pruning_for_fixed_len_byte_array() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1662 id=1662 data-nosnippet>1662</a> <span class="comment">// FIXED_LEN_BYTE_ARRAY page indexes back spec-conforming fixed, uuid,
-<a href=#1663 id=1663 data-nosnippet>1663</a> // and decimal(P > 18) columns. Only decimals are decoded; fixed and uuid
-<a href=#1664 id=1664 data-nosnippet>1664</a> // bounds aren't, so the evaluator skips page pruning for them rather than
-<a href=#1665 id=1665 data-nosnippet>1665</a> // abort the scan.
-<a href=#1666 id=1666 data-nosnippet>1666</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_fixed_len_byte_array_parquet_file()<span class="question-mark">?</span>;
-<a href=#1667 id=1667 data-nosnippet>1667</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1668 id=1668 data-nosnippet>1668</a>
-<a href=#1669 id=1669 data-nosnippet>1669</a> <span class="kw">let </span>iceberg_schema = Arc::new(
-<a href=#1670 id=1670 data-nosnippet>1670</a> Schema::builder()
-<a href=#1671 id=1671 data-nosnippet>1671</a> .with_fields([Arc::new(NestedField::new(
-<a href=#1672 id=1672 data-nosnippet>1672</a> <span class="number">1</span>,
-<a href=#1673 id=1673 data-nosnippet>1673</a> <span class="string">"col_fixed"</span>,
-<a href=#1674 id=1674 data-nosnippet>1674</a> Type::Primitive(PrimitiveType::Fixed(<span class="number">2</span>)),
-<a href=#1675 id=1675 data-nosnippet>1675</a> <span class="bool-val">true</span>,
-<a href=#1676 id=1676 data-nosnippet>1676</a> ))])
-<a href=#1677 id=1677 data-nosnippet>1677</a> .build()<span class="question-mark">?</span>,
-<a href=#1678 id=1678 data-nosnippet>1678</a> );
-<a href=#1679 id=1679 data-nosnippet>1679</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
-<a href=#1680 id=1680 data-nosnippet>1680</a>
-<a href=#1681 id=1681 data-nosnippet>1681</a> <span class="comment">// A predicate that would prune every page if the bounds were decoded.
-<a href=#1682 id=1682 data-nosnippet>1682</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_fixed"</span>)
-<a href=#1683 id=1683 data-nosnippet>1683</a> .greater_than(Datum::fixed(<span class="macro">vec!</span>[<span class="number">0xffu8</span>, <span class="number">0xff</span>]))
-<a href=#1684 id=1684 data-nosnippet>1684</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1685 id=1685 data-nosnippet>1685</a>
-<a href=#1686 id=1686 data-nosnippet>1686</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1687 id=1687 data-nosnippet>1687</a> <span class="kw-2">&</span>filter,
-<a href=#1688 id=1688 data-nosnippet>1688</a> <span class="kw-2">&</span>column_index,
-<a href=#1689 id=1689 data-nosnippet>1689</a> <span class="kw-2">&</span>offset_index,
-<a href=#1690 id=1690 data-nosnippet>1690</a> row_group_metadata,
-<a href=#1691 id=1691 data-nosnippet>1691</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1692 id=1692 data-nosnippet>1692</a> iceberg_schema.as_ref(),
-<a href=#1693 id=1693 data-nosnippet>1693</a> )<span class="question-mark">?</span>;
-<a href=#1694 id=1694 data-nosnippet>1694</a>
-<a href=#1695 id=1695 data-nosnippet>1695</a> <span class="comment">// All rows survive: no page pruning is applied for FIXED_LEN_BYTE_ARRAY.
-<a href=#1696 id=1696 data-nosnippet>1696</a> </span><span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[RowSelector::select(<span class="number">2048</span>)]);
-<a href=#1697 id=1697 data-nosnippet>1697</a>
-<a href=#1698 id=1698 data-nosnippet>1698</a> <span class="prelude-val">Ok</span>(())
-<a href=#1699 id=1699 data-nosnippet>1699</a> }
-<a href=#1700 id=1700 data-nosnippet>1700</a>
-<a href=#1701 id=1701 data-nosnippet>1701</a> <span class="attr">#[test]
-<a href=#1702 id=1702 data-nosnippet>1702</a> </span><span class="kw">fn </span>eval_skips_pruning_for_non_utf8_string_bound() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1703 id=1703 data-nosnippet>1703</a> <span class="comment">// A writer can truncate a string min/max stat mid-UTF-8-sequence,
-<a href=#1704 id=1704 data-nosnippet>1704</a> // leaving a BYTE_ARRAY page bound that isn't valid UTF-8. Decoding it
-<a href=#1705 id=1705 data-nosnippet>1705</a> // as a string fails, so the evaluator skips page pruning for the column
-<a href=#1706 id=1706 data-nosnippet>1706</a> // rather than aborting the whole file read.
-<a href=#1707 id=1707 data-nosnippet>1707</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_binary_parquet_file()<span class="question-mark">?</span>;
-<a href=#1708 id=1708 data-nosnippet>1708</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1709 id=1709 data-nosnippet>1709</a>
-<a href=#1710 id=1710 data-nosnippet>1710</a> <span class="kw">let </span>iceberg_schema = Arc::new(
-<a href=#1711 id=1711 data-nosnippet>1711</a> Schema::builder()
-<a href=#1712 id=1712 data-nosnippet>1712</a> .with_fields([Arc::new(NestedField::new(
-<a href=#1713 id=1713 data-nosnippet>1713</a> <span class="number">1</span>,
-<a href=#1714 id=1714 data-nosnippet>1714</a> <span class="string">"col_string"</span>,
-<a href=#1715 id=1715 data-nosnippet>1715</a> Type::Primitive(PrimitiveType::String),
-<a href=#1716 id=1716 data-nosnippet>1716</a> <span class="bool-val">true</span>,
-<a href=#1717 id=1717 data-nosnippet>1717</a> ))])
-<a href=#1718 id=1718 data-nosnippet>1718</a> .build()<span class="question-mark">?</span>,
-<a href=#1719 id=1719 data-nosnippet>1719</a> );
-<a href=#1720 id=1720 data-nosnippet>1720</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
-<a href=#1721 id=1721 data-nosnippet>1721</a>
-<a href=#1722 id=1722 data-nosnippet>1722</a> <span class="comment">// Page 1's bound [0xff, 0x00] isn't valid UTF-8. A predicate that would
-<a href=#1723 id=1723 data-nosnippet>1723</a> // otherwise prune every page must not: the undecodable bound forces a
-<a href=#1724 id=1724 data-nosnippet>1724</a> // skip rather than an error.
-<a href=#1725 id=1725 data-nosnippet>1725</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_string"</span>)
-<a href=#1726 id=1726 data-nosnippet>1726</a> .greater_than(Datum::string(<span class="string">"zzz"</span>))
-<a href=#1727 id=1727 data-nosnippet>1727</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1728 id=1728 data-nosnippet>1728</a>
-<a href=#1729 id=1729 data-nosnippet>1729</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1730 id=1730 data-nosnippet>1730</a> <span class="kw-2">&</span>filter,
-<a href=#1731 id=1731 data-nosnippet>1731</a> <span class="kw-2">&</span>column_index,
-<a href=#1732 id=1732 data-nosnippet>1732</a> <span class="kw-2">&</span>offset_index,
-<a href=#1733 id=1733 data-nosnippet>1733</a> row_group_metadata,
-<a href=#1734 id=1734 data-nosnippet>1734</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1735 id=1735 data-nosnippet>1735</a> iceberg_schema.as_ref(),
-<a href=#1736 id=1736 data-nosnippet>1736</a> )<span class="question-mark">?</span>;
-<a href=#1737 id=1737 data-nosnippet>1737</a>
-<a href=#1738 id=1738 data-nosnippet>1738</a> <span class="comment">// Both pages survive: no page pruning is applied when a bound won't decode.
-<a href=#1739 id=1739 data-nosnippet>1739</a> </span><span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[RowSelector::select(<span class="number">2048</span>)]);
-<a href=#1740 id=1740 data-nosnippet>1740</a>
-<a href=#1741 id=1741 data-nosnippet>1741</a> <span class="prelude-val">Ok</span>(())
-<a href=#1742 id=1742 data-nosnippet>1742</a> }
-<a href=#1743 id=1743 data-nosnippet>1743</a>
-<a href=#1744 id=1744 data-nosnippet>1744</a> <span class="attr">#[test]
-<a href=#1745 id=1745 data-nosnippet>1745</a> </span><span class="kw">fn </span>eval_starts_with_error_float_col() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1746 id=1746 data-nosnippet>1746</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1747 id=1747 data-nosnippet>1747</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1748 id=1748 data-nosnippet>1748</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
-<a href=#1749 id=1749 data-nosnippet>1749</a>
-<a href=#1750 id=1750 data-nosnippet>1750</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1751 id=1751 data-nosnippet>1751</a> .starts_with(Datum::float(<span class="number">5.0_f32</span>))
-<a href=#1752 id=1752 data-nosnippet>1752</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1753 id=1753 data-nosnippet>1753</a>
-<a href=#1754 id=1754 data-nosnippet>1754</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1755 id=1755 data-nosnippet>1755</a> <span class="kw-2">&</span>filter,
-<a href=#1756 id=1756 data-nosnippet>1756</a> <span class="kw-2">&</span>column_index,
-<a href=#1757 id=1757 data-nosnippet>1757</a> <span class="kw-2">&</span>offset_index,
-<a href=#1758 id=1758 data-nosnippet>1758</a> row_group_metadata,
-<a href=#1759 id=1759 data-nosnippet>1759</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1760 id=1760 data-nosnippet>1760</a> iceberg_schema_ref.as_ref(),
-<a href=#1761 id=1761 data-nosnippet>1761</a> );
-<a href=#1762 id=1762 data-nosnippet>1762</a>
-<a href=#1763 id=1763 data-nosnippet>1763</a> <span class="macro">assert_eq!</span>(result.unwrap_err().kind(), ErrorKind::Unexpected);
+<a href=#1607 id=1607 data-nosnippet>1607</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1608 id=1608 data-nosnippet>1608</a> <span class="kw-2">&</span>filter,
+<a href=#1609 id=1609 data-nosnippet>1609</a> <span class="kw-2">&</span>column_index,
+<a href=#1610 id=1610 data-nosnippet>1610</a> <span class="kw-2">&</span>offset_index,
+<a href=#1611 id=1611 data-nosnippet>1611</a> row_group_metadata,
+<a href=#1612 id=1612 data-nosnippet>1612</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1613 id=1613 data-nosnippet>1613</a> iceberg_schema.as_ref(),
+<a href=#1614 id=1614 data-nosnippet>1614</a> )<span class="question-mark">?</span>;
+<a href=#1615 id=1615 data-nosnippet>1615</a>
+<a href=#1616 id=1616 data-nosnippet>1616</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::skip(<span class="number">1024</span>), RowSelector::select(<span class="number">1024</span>)];
+<a href=#1617 id=1617 data-nosnippet>1617</a>
+<a href=#1618 id=1618 data-nosnippet>1618</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1619 id=1619 data-nosnippet>1619</a>
+<a href=#1620 id=1620 data-nosnippet>1620</a> <span class="prelude-val">Ok</span>(())
+<a href=#1621 id=1621 data-nosnippet>1621</a> }
+<a href=#1622 id=1622 data-nosnippet>1622</a>
+<a href=#1623 id=1623 data-nosnippet>1623</a> <span class="attr">#[test]
+<a href=#1624 id=1624 data-nosnippet>1624</a> </span><span class="kw">fn </span>eval_skips_pruning_for_byte_array_decimal_bounds() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1625 id=1625 data-nosnippet>1625</a> <span class="comment">// A non-spec writer can store a decimal column as BYTE_ARRAY. Those
+<a href=#1626 id=1626 data-nosnippet>1626</a> // bounds can't be decoded safely (Parquet may truncate them, and
+<a href=#1627 id=1627 data-nosnippet>1627</a> // truncation only preserves lexicographic byte order, not decimal
+<a href=#1628 id=1628 data-nosnippet>1628</a> // order), so the evaluator skips page pruning rather than prune pages
+<a href=#1629 id=1629 data-nosnippet>1629</a> // that might match.
+<a href=#1630 id=1630 data-nosnippet>1630</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_binary_parquet_file()<span class="question-mark">?</span>;
+<a href=#1631 id=1631 data-nosnippet>1631</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1632 id=1632 data-nosnippet>1632</a>
+<a href=#1633 id=1633 data-nosnippet>1633</a> <span class="kw">let </span>iceberg_schema = Arc::new(
+<a href=#1634 id=1634 data-nosnippet>1634</a> Schema::builder()
+<a href=#1635 id=1635 data-nosnippet>1635</a> .with_fields([Arc::new(NestedField::new(
+<a href=#1636 id=1636 data-nosnippet>1636</a> <span class="number">1</span>,
+<a href=#1637 id=1637 data-nosnippet>1637</a> <span class="string">"col_decimal"</span>,
+<a href=#1638 id=1638 data-nosnippet>1638</a> Type::Primitive(PrimitiveType::Decimal {
+<a href=#1639 id=1639 data-nosnippet>1639</a> precision: <span class="number">10</span>,
+<a href=#1640 id=1640 data-nosnippet>1640</a> scale: <span class="number">2</span>,
+<a href=#1641 id=1641 data-nosnippet>1641</a> }),
+<a href=#1642 id=1642 data-nosnippet>1642</a> <span class="bool-val">true</span>,
+<a href=#1643 id=1643 data-nosnippet>1643</a> ))])
+<a href=#1644 id=1644 data-nosnippet>1644</a> .build()<span class="question-mark">?</span>,
+<a href=#1645 id=1645 data-nosnippet>1645</a> );
+<a href=#1646 id=1646 data-nosnippet>1646</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
+<a href=#1647 id=1647 data-nosnippet>1647</a>
+<a href=#1648 id=1648 data-nosnippet>1648</a> <span class="comment">// A predicate that would prune every page if the bounds were decoded.
+<a href=#1649 id=1649 data-nosnippet>1649</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#1650 id=1650 data-nosnippet>1650</a> .greater_than(Datum::decimal_with_precision(
+<a href=#1651 id=1651 data-nosnippet>1651</a> decimal_from_i128_with_scale(<span class="number">99999</span>, <span class="number">2</span>),
+<a href=#1652 id=1652 data-nosnippet>1652</a> <span class="number">10</span>,
+<a href=#1653 id=1653 data-nosnippet>1653</a> )<span class="question-mark">?</span>)
+<a href=#1654 id=1654 data-nosnippet>1654</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1655 id=1655 data-nosnippet>1655</a>
+<a href=#1656 id=1656 data-nosnippet>1656</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1657 id=1657 data-nosnippet>1657</a> <span class="kw-2">&</span>filter,
+<a href=#1658 id=1658 data-nosnippet>1658</a> <span class="kw-2">&</span>column_index,
+<a href=#1659 id=1659 data-nosnippet>1659</a> <span class="kw-2">&</span>offset_index,
+<a href=#1660 id=1660 data-nosnippet>1660</a> row_group_metadata,
+<a href=#1661 id=1661 data-nosnippet>1661</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1662 id=1662 data-nosnippet>1662</a> iceberg_schema.as_ref(),
+<a href=#1663 id=1663 data-nosnippet>1663</a> )<span class="question-mark">?</span>;
+<a href=#1664 id=1664 data-nosnippet>1664</a>
+<a href=#1665 id=1665 data-nosnippet>1665</a> <span class="comment">// Both 1024-row pages survive: no page pruning is applied.
+<a href=#1666 id=1666 data-nosnippet>1666</a> </span><span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[RowSelector::select(<span class="number">2048</span>)]);
+<a href=#1667 id=1667 data-nosnippet>1667</a>
+<a href=#1668 id=1668 data-nosnippet>1668</a> <span class="prelude-val">Ok</span>(())
+<a href=#1669 id=1669 data-nosnippet>1669</a> }
+<a href=#1670 id=1670 data-nosnippet>1670</a>
+<a href=#1671 id=1671 data-nosnippet>1671</a> <span class="attr">#[test]
+<a href=#1672 id=1672 data-nosnippet>1672</a> </span><span class="kw">fn </span>eval_skips_pruning_for_fixed_len_byte_array() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1673 id=1673 data-nosnippet>1673</a> <span class="comment">// FIXED_LEN_BYTE_ARRAY page indexes back spec-conforming fixed, uuid,
+<a href=#1674 id=1674 data-nosnippet>1674</a> // and decimal(P > 18) columns. Only decimals are decoded; fixed and uuid
+<a href=#1675 id=1675 data-nosnippet>1675</a> // bounds aren't, so the evaluator skips page pruning for them rather than
+<a href=#1676 id=1676 data-nosnippet>1676</a> // abort the scan.
+<a href=#1677 id=1677 data-nosnippet>1677</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_fixed_len_byte_array_parquet_file()<span class="question-mark">?</span>;
+<a href=#1678 id=1678 data-nosnippet>1678</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1679 id=1679 data-nosnippet>1679</a>
+<a href=#1680 id=1680 data-nosnippet>1680</a> <span class="kw">let </span>iceberg_schema = Arc::new(
+<a href=#1681 id=1681 data-nosnippet>1681</a> Schema::builder()
+<a href=#1682 id=1682 data-nosnippet>1682</a> .with_fields([Arc::new(NestedField::new(
+<a href=#1683 id=1683 data-nosnippet>1683</a> <span class="number">1</span>,
+<a href=#1684 id=1684 data-nosnippet>1684</a> <span class="string">"col_fixed"</span>,
+<a href=#1685 id=1685 data-nosnippet>1685</a> Type::Primitive(PrimitiveType::Fixed(<span class="number">2</span>)),
+<a href=#1686 id=1686 data-nosnippet>1686</a> <span class="bool-val">true</span>,
+<a href=#1687 id=1687 data-nosnippet>1687</a> ))])
+<a href=#1688 id=1688 data-nosnippet>1688</a> .build()<span class="question-mark">?</span>,
+<a href=#1689 id=1689 data-nosnippet>1689</a> );
+<a href=#1690 id=1690 data-nosnippet>1690</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
+<a href=#1691 id=1691 data-nosnippet>1691</a>
+<a href=#1692 id=1692 data-nosnippet>1692</a> <span class="comment">// A predicate that would prune every page if the bounds were decoded.
+<a href=#1693 id=1693 data-nosnippet>1693</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_fixed"</span>)
+<a href=#1694 id=1694 data-nosnippet>1694</a> .greater_than(Datum::fixed(<span class="macro">vec!</span>[<span class="number">0xffu8</span>, <span class="number">0xff</span>]))
+<a href=#1695 id=1695 data-nosnippet>1695</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1696 id=1696 data-nosnippet>1696</a>
+<a href=#1697 id=1697 data-nosnippet>1697</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1698 id=1698 data-nosnippet>1698</a> <span class="kw-2">&</span>filter,
+<a href=#1699 id=1699 data-nosnippet>1699</a> <span class="kw-2">&</span>column_index,
+<a href=#1700 id=1700 data-nosnippet>1700</a> <span class="kw-2">&</span>offset_index,
+<a href=#1701 id=1701 data-nosnippet>1701</a> row_group_metadata,
+<a href=#1702 id=1702 data-nosnippet>1702</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1703 id=1703 data-nosnippet>1703</a> iceberg_schema.as_ref(),
+<a href=#1704 id=1704 data-nosnippet>1704</a> )<span class="question-mark">?</span>;
+<a href=#1705 id=1705 data-nosnippet>1705</a>
+<a href=#1706 id=1706 data-nosnippet>1706</a> <span class="comment">// All rows survive: no page pruning is applied for FIXED_LEN_BYTE_ARRAY.
+<a href=#1707 id=1707 data-nosnippet>1707</a> </span><span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[RowSelector::select(<span class="number">2048</span>)]);
+<a href=#1708 id=1708 data-nosnippet>1708</a>
+<a href=#1709 id=1709 data-nosnippet>1709</a> <span class="prelude-val">Ok</span>(())
+<a href=#1710 id=1710 data-nosnippet>1710</a> }
+<a href=#1711 id=1711 data-nosnippet>1711</a>
+<a href=#1712 id=1712 data-nosnippet>1712</a> <span class="attr">#[test]
+<a href=#1713 id=1713 data-nosnippet>1713</a> </span><span class="kw">fn </span>eval_skips_pruning_for_non_utf8_string_bound() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1714 id=1714 data-nosnippet>1714</a> <span class="comment">// A writer can truncate a string min/max stat mid-UTF-8-sequence,
+<a href=#1715 id=1715 data-nosnippet>1715</a> // leaving a BYTE_ARRAY page bound that isn't valid UTF-8. Decoding it
+<a href=#1716 id=1716 data-nosnippet>1716</a> // as a string fails, so the evaluator skips page pruning for the column
+<a href=#1717 id=1717 data-nosnippet>1717</a> // rather than aborting the whole file read.
+<a href=#1718 id=1718 data-nosnippet>1718</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_binary_parquet_file()<span class="question-mark">?</span>;
+<a href=#1719 id=1719 data-nosnippet>1719</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1720 id=1720 data-nosnippet>1720</a>
+<a href=#1721 id=1721 data-nosnippet>1721</a> <span class="kw">let </span>iceberg_schema = Arc::new(
+<a href=#1722 id=1722 data-nosnippet>1722</a> Schema::builder()
+<a href=#1723 id=1723 data-nosnippet>1723</a> .with_fields([Arc::new(NestedField::new(
+<a href=#1724 id=1724 data-nosnippet>1724</a> <span class="number">1</span>,
+<a href=#1725 id=1725 data-nosnippet>1725</a> <span class="string">"col_string"</span>,
+<a href=#1726 id=1726 data-nosnippet>1726</a> Type::Primitive(PrimitiveType::String),
+<a href=#1727 id=1727 data-nosnippet>1727</a> <span class="bool-val">true</span>,
+<a href=#1728 id=1728 data-nosnippet>1728</a> ))])
+<a href=#1729 id=1729 data-nosnippet>1729</a> .build()<span class="question-mark">?</span>,
+<a href=#1730 id=1730 data-nosnippet>1730</a> );
+<a href=#1731 id=1731 data-nosnippet>1731</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
+<a href=#1732 id=1732 data-nosnippet>1732</a>
+<a href=#1733 id=1733 data-nosnippet>1733</a> <span class="comment">// Page 1's bound [0xff, 0x00] isn't valid UTF-8. A predicate that would
+<a href=#1734 id=1734 data-nosnippet>1734</a> // otherwise prune every page must not: the undecodable bound forces a
+<a href=#1735 id=1735 data-nosnippet>1735</a> // skip rather than an error.
+<a href=#1736 id=1736 data-nosnippet>1736</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_string"</span>)
+<a href=#1737 id=1737 data-nosnippet>1737</a> .greater_than(Datum::string(<span class="string">"zzz"</span>))
+<a href=#1738 id=1738 data-nosnippet>1738</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1739 id=1739 data-nosnippet>1739</a>
+<a href=#1740 id=1740 data-nosnippet>1740</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1741 id=1741 data-nosnippet>1741</a> <span class="kw-2">&</span>filter,
+<a href=#1742 id=1742 data-nosnippet>1742</a> <span class="kw-2">&</span>column_index,
+<a href=#1743 id=1743 data-nosnippet>1743</a> <span class="kw-2">&</span>offset_index,
+<a href=#1744 id=1744 data-nosnippet>1744</a> row_group_metadata,
+<a href=#1745 id=1745 data-nosnippet>1745</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1746 id=1746 data-nosnippet>1746</a> iceberg_schema.as_ref(),
+<a href=#1747 id=1747 data-nosnippet>1747</a> )<span class="question-mark">?</span>;
+<a href=#1748 id=1748 data-nosnippet>1748</a>
+<a href=#1749 id=1749 data-nosnippet>1749</a> <span class="comment">// Both pages survive: no page pruning is applied when a bound won't decode.
+<a href=#1750 id=1750 data-nosnippet>1750</a> </span><span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[RowSelector::select(<span class="number">2048</span>)]);
+<a href=#1751 id=1751 data-nosnippet>1751</a>
+<a href=#1752 id=1752 data-nosnippet>1752</a> <span class="prelude-val">Ok</span>(())
+<a href=#1753 id=1753 data-nosnippet>1753</a> }
+<a href=#1754 id=1754 data-nosnippet>1754</a>
+<a href=#1755 id=1755 data-nosnippet>1755</a> <span class="attr">#[test]
+<a href=#1756 id=1756 data-nosnippet>1756</a> </span><span class="kw">fn </span>eval_starts_with_error_float_col() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1757 id=1757 data-nosnippet>1757</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1758 id=1758 data-nosnippet>1758</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1759 id=1759 data-nosnippet>1759</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1760 id=1760 data-nosnippet>1760</a>
+<a href=#1761 id=1761 data-nosnippet>1761</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1762 id=1762 data-nosnippet>1762</a> .starts_with(Datum::float(<span class="number">5.0_f32</span>))
+<a href=#1763 id=1763 data-nosnippet>1763</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
<a href=#1764 id=1764 data-nosnippet>1764</a>
-<a href=#1765 id=1765 data-nosnippet>1765</a> <span class="prelude-val">Ok</span>(())
-<a href=#1766 id=1766 data-nosnippet>1766</a> }
-<a href=#1767 id=1767 data-nosnippet>1767</a>
-<a href=#1768 id=1768 data-nosnippet>1768</a> <span class="attr">#[test]
-<a href=#1769 id=1769 data-nosnippet>1769</a> </span><span class="kw">fn </span>eval_starts_with_pages_containing_value_except_all_null_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1770 id=1770 data-nosnippet>1770</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1771 id=1771 data-nosnippet>1771</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1772 id=1772 data-nosnippet>1772</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1765 id=1765 data-nosnippet>1765</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1766 id=1766 data-nosnippet>1766</a> <span class="kw-2">&</span>filter,
+<a href=#1767 id=1767 data-nosnippet>1767</a> <span class="kw-2">&</span>column_index,
+<a href=#1768 id=1768 data-nosnippet>1768</a> <span class="kw-2">&</span>offset_index,
+<a href=#1769 id=1769 data-nosnippet>1769</a> row_group_metadata,
+<a href=#1770 id=1770 data-nosnippet>1770</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1771 id=1771 data-nosnippet>1771</a> iceberg_schema_ref.as_ref(),
+<a href=#1772 id=1772 data-nosnippet>1772</a> );
<a href=#1773 id=1773 data-nosnippet>1773</a>
-<a href=#1774 id=1774 data-nosnippet>1774</a> <span class="comment">// Test starts_with on string column where only some pages match
-<a href=#1775 id=1775 data-nosnippet>1775</a> // Our file has 4 pages: ["AARDVARK".."BISON"], ["DEER"], ["GIRAFFE".."HIPPO"], ["HIPPO"]
-<a href=#1776 id=1776 data-nosnippet>1776</a> // Testing starts_with("B") should select only page 0
-<a href=#1777 id=1777 data-nosnippet>1777</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_string"</span>)
-<a href=#1778 id=1778 data-nosnippet>1778</a> .starts_with(Datum::string(<span class="string">"B"</span>))
-<a href=#1779 id=1779 data-nosnippet>1779</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1780 id=1780 data-nosnippet>1780</a>
-<a href=#1781 id=1781 data-nosnippet>1781</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1782 id=1782 data-nosnippet>1782</a> <span class="kw-2">&</span>filter,
-<a href=#1783 id=1783 data-nosnippet>1783</a> <span class="kw-2">&</span>column_index,
-<a href=#1784 id=1784 data-nosnippet>1784</a> <span class="kw-2">&</span>offset_index,
-<a href=#1785 id=1785 data-nosnippet>1785</a> row_group_metadata,
-<a href=#1786 id=1786 data-nosnippet>1786</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1787 id=1787 data-nosnippet>1787</a> iceberg_schema_ref.as_ref(),
-<a href=#1788 id=1788 data-nosnippet>1788</a> )<span class="question-mark">?</span>;
-<a href=#1789 id=1789 data-nosnippet>1789</a>
-<a href=#1790 id=1790 data-nosnippet>1790</a> <span class="comment">// Page 0 has "BEAR" and "BISON" (starts with B), rest don't
-<a href=#1791 id=1791 data-nosnippet>1791</a> </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">1024</span>), RowSelector::skip(<span class="number">3072</span>)];
-<a href=#1792 id=1792 data-nosnippet>1792</a>
-<a href=#1793 id=1793 data-nosnippet>1793</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1794 id=1794 data-nosnippet>1794</a>
-<a href=#1795 id=1795 data-nosnippet>1795</a> <span class="prelude-val">Ok</span>(())
-<a href=#1796 id=1796 data-nosnippet>1796</a> }
-<a href=#1797 id=1797 data-nosnippet>1797</a>
-<a href=#1798 id=1798 data-nosnippet>1798</a> <span class="attr">#[test]
-<a href=#1799 id=1799 data-nosnippet>1799</a> </span><span class="kw">fn </span>eval_not_starts_with_pages_containing_value_except_pages_with_min_and_max_equal_to_prefix_and_all_null_pages()
-<a href=#1800 id=1800 data-nosnippet>1800</a> -> <span class="prelude-ty">Result</span><()> {
-<a href=#1801 id=1801 data-nosnippet>1801</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1802 id=1802 data-nosnippet>1802</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1803 id=1803 data-nosnippet>1803</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
-<a href=#1804 id=1804 data-nosnippet>1804</a>
-<a href=#1805 id=1805 data-nosnippet>1805</a> <span class="comment">// Test not_starts_with where one page has ALL values starting with prefix
-<a href=#1806 id=1806 data-nosnippet>1806</a> // Our file has page 1 with all "DEER" (min="DEER", max="DEER")
-<a href=#1807 id=1807 data-nosnippet>1807</a> // Testing not_starts_with("DE") should skip page 1 where all values start with "DE"
-<a href=#1808 id=1808 data-nosnippet>1808</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_string"</span>)
-<a href=#1809 id=1809 data-nosnippet>1809</a> .not_starts_with(Datum::string(<span class="string">"DE"</span>))
-<a href=#1810 id=1810 data-nosnippet>1810</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1811 id=1811 data-nosnippet>1811</a>
-<a href=#1812 id=1812 data-nosnippet>1812</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1813 id=1813 data-nosnippet>1813</a> <span class="kw-2">&</span>filter,
-<a href=#1814 id=1814 data-nosnippet>1814</a> <span class="kw-2">&</span>column_index,
-<a href=#1815 id=1815 data-nosnippet>1815</a> <span class="kw-2">&</span>offset_index,
-<a href=#1816 id=1816 data-nosnippet>1816</a> row_group_metadata,
-<a href=#1817 id=1817 data-nosnippet>1817</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1818 id=1818 data-nosnippet>1818</a> iceberg_schema_ref.as_ref(),
-<a href=#1819 id=1819 data-nosnippet>1819</a> )<span class="question-mark">?</span>;
-<a href=#1820 id=1820 data-nosnippet>1820</a>
-<a href=#1821 id=1821 data-nosnippet>1821</a> <span class="comment">// Page 0: mixed values (select)
-<a href=#1822 id=1822 data-nosnippet>1822</a> // Page 1: all "DEER" starting with "DE" (skip)
-<a href=#1823 id=1823 data-nosnippet>1823</a> // Pages 2-3: other values not all starting with "DE" (select)
-<a href=#1824 id=1824 data-nosnippet>1824</a> </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[
-<a href=#1825 id=1825 data-nosnippet>1825</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#1826 id=1826 data-nosnippet>1826</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#1827 id=1827 data-nosnippet>1827</a> RowSelector::select(<span class="number">2048</span>),
-<a href=#1828 id=1828 data-nosnippet>1828</a> ];
-<a href=#1829 id=1829 data-nosnippet>1829</a>
-<a href=#1830 id=1830 data-nosnippet>1830</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1774 id=1774 data-nosnippet>1774</a> <span class="macro">assert_eq!</span>(result.unwrap_err().kind(), ErrorKind::Unexpected);
+<a href=#1775 id=1775 data-nosnippet>1775</a>
+<a href=#1776 id=1776 data-nosnippet>1776</a> <span class="prelude-val">Ok</span>(())
+<a href=#1777 id=1777 data-nosnippet>1777</a> }
+<a href=#1778 id=1778 data-nosnippet>1778</a>
+<a href=#1779 id=1779 data-nosnippet>1779</a> <span class="attr">#[test]
+<a href=#1780 id=1780 data-nosnippet>1780</a> </span><span class="kw">fn </span>eval_starts_with_pages_containing_value_except_all_null_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1781 id=1781 data-nosnippet>1781</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1782 id=1782 data-nosnippet>1782</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1783 id=1783 data-nosnippet>1783</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1784 id=1784 data-nosnippet>1784</a>
+<a href=#1785 id=1785 data-nosnippet>1785</a> <span class="comment">// Test starts_with on string column where only some pages match
+<a href=#1786 id=1786 data-nosnippet>1786</a> // Our file has 4 pages: ["AARDVARK".."BISON"], ["DEER"], ["GIRAFFE".."HIPPO"], ["HIPPO"]
+<a href=#1787 id=1787 data-nosnippet>1787</a> // Testing starts_with("B") should select only page 0
+<a href=#1788 id=1788 data-nosnippet>1788</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_string"</span>)
+<a href=#1789 id=1789 data-nosnippet>1789</a> .starts_with(Datum::string(<span class="string">"B"</span>))
+<a href=#1790 id=1790 data-nosnippet>1790</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1791 id=1791 data-nosnippet>1791</a>
+<a href=#1792 id=1792 data-nosnippet>1792</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1793 id=1793 data-nosnippet>1793</a> <span class="kw-2">&</span>filter,
+<a href=#1794 id=1794 data-nosnippet>1794</a> <span class="kw-2">&</span>column_index,
+<a href=#1795 id=1795 data-nosnippet>1795</a> <span class="kw-2">&</span>offset_index,
+<a href=#1796 id=1796 data-nosnippet>1796</a> row_group_metadata,
+<a href=#1797 id=1797 data-nosnippet>1797</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1798 id=1798 data-nosnippet>1798</a> iceberg_schema_ref.as_ref(),
+<a href=#1799 id=1799 data-nosnippet>1799</a> )<span class="question-mark">?</span>;
+<a href=#1800 id=1800 data-nosnippet>1800</a>
+<a href=#1801 id=1801 data-nosnippet>1801</a> <span class="comment">// Page 0 has "BEAR" and "BISON" (starts with B), rest don't
+<a href=#1802 id=1802 data-nosnippet>1802</a> </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">1024</span>), RowSelector::skip(<span class="number">3072</span>)];
+<a href=#1803 id=1803 data-nosnippet>1803</a>
+<a href=#1804 id=1804 data-nosnippet>1804</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1805 id=1805 data-nosnippet>1805</a>
+<a href=#1806 id=1806 data-nosnippet>1806</a> <span class="prelude-val">Ok</span>(())
+<a href=#1807 id=1807 data-nosnippet>1807</a> }
+<a href=#1808 id=1808 data-nosnippet>1808</a>
+<a href=#1809 id=1809 data-nosnippet>1809</a> <span class="attr">#[test]
+<a href=#1810 id=1810 data-nosnippet>1810</a> </span><span class="kw">fn </span>eval_not_starts_with_pages_containing_value_except_pages_with_min_and_max_equal_to_prefix_and_all_null_pages()
+<a href=#1811 id=1811 data-nosnippet>1811</a> -> <span class="prelude-ty">Result</span><()> {
+<a href=#1812 id=1812 data-nosnippet>1812</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1813 id=1813 data-nosnippet>1813</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1814 id=1814 data-nosnippet>1814</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1815 id=1815 data-nosnippet>1815</a>
+<a href=#1816 id=1816 data-nosnippet>1816</a> <span class="comment">// Test not_starts_with where one page has ALL values starting with prefix
+<a href=#1817 id=1817 data-nosnippet>1817</a> // Our file has page 1 with all "DEER" (min="DEER", max="DEER")
+<a href=#1818 id=1818 data-nosnippet>1818</a> // Testing not_starts_with("DE") should skip page 1 where all values start with "DE"
+<a href=#1819 id=1819 data-nosnippet>1819</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_string"</span>)
+<a href=#1820 id=1820 data-nosnippet>1820</a> .not_starts_with(Datum::string(<span class="string">"DE"</span>))
+<a href=#1821 id=1821 data-nosnippet>1821</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1822 id=1822 data-nosnippet>1822</a>
+<a href=#1823 id=1823 data-nosnippet>1823</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1824 id=1824 data-nosnippet>1824</a> <span class="kw-2">&</span>filter,
+<a href=#1825 id=1825 data-nosnippet>1825</a> <span class="kw-2">&</span>column_index,
+<a href=#1826 id=1826 data-nosnippet>1826</a> <span class="kw-2">&</span>offset_index,
+<a href=#1827 id=1827 data-nosnippet>1827</a> row_group_metadata,
+<a href=#1828 id=1828 data-nosnippet>1828</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1829 id=1829 data-nosnippet>1829</a> iceberg_schema_ref.as_ref(),
+<a href=#1830 id=1830 data-nosnippet>1830</a> )<span class="question-mark">?</span>;
<a href=#1831 id=1831 data-nosnippet>1831</a>
-<a href=#1832 id=1832 data-nosnippet>1832</a> <span class="prelude-val">Ok</span>(())
-<a href=#1833 id=1833 data-nosnippet>1833</a> }
-<a href=#1834 id=1834 data-nosnippet>1834</a>
-<a href=#1835 id=1835 data-nosnippet>1835</a> <span class="attr">#[test]
-<a href=#1836 id=1836 data-nosnippet>1836</a> </span><span class="kw">fn </span>eval_in_length_of_set_above_limit_all_rows() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1837 id=1837 data-nosnippet>1837</a> <span class="kw">let </span><span class="kw-2">mut </span>rng = rand::rng();
-<a href=#1838 id=1838 data-nosnippet>1838</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1839 id=1839 data-nosnippet>1839</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1840 id=1840 data-nosnippet>1840</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
-<a href=#1841 id=1841 data-nosnippet>1841</a>
-<a href=#1842 id=1842 data-nosnippet>1842</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
-<a href=#1843 id=1843 data-nosnippet>1843</a> .is_in(
-<a href=#1844 id=1844 data-nosnippet>1844</a> std::iter::repeat_with(|| Datum::float(rng.random_range(<span class="number">0.0_f32</span>..<span class="number">10.0_f32</span>)))
-<a href=#1845 id=1845 data-nosnippet>1845</a> .take(<span class="number">1000</span>),
-<a href=#1846 id=1846 data-nosnippet>1846</a> )
-<a href=#1847 id=1847 data-nosnippet>1847</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1848 id=1848 data-nosnippet>1848</a>
-<a href=#1849 id=1849 data-nosnippet>1849</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1850 id=1850 data-nosnippet>1850</a> <span class="kw-2">&</span>filter,
-<a href=#1851 id=1851 data-nosnippet>1851</a> <span class="kw-2">&</span>column_index,
-<a href=#1852 id=1852 data-nosnippet>1852</a> <span class="kw-2">&</span>offset_index,
-<a href=#1853 id=1853 data-nosnippet>1853</a> row_group_metadata,
-<a href=#1854 id=1854 data-nosnippet>1854</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1855 id=1855 data-nosnippet>1855</a> iceberg_schema_ref.as_ref(),
-<a href=#1856 id=1856 data-nosnippet>1856</a> )<span class="question-mark">?</span>;
-<a href=#1857 id=1857 data-nosnippet>1857</a>
-<a href=#1858 id=1858 data-nosnippet>1858</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">4096</span>)];
+<a href=#1832 id=1832 data-nosnippet>1832</a> <span class="comment">// Page 0: mixed values (select)
+<a href=#1833 id=1833 data-nosnippet>1833</a> // Page 1: all "DEER" starting with "DE" (skip)
+<a href=#1834 id=1834 data-nosnippet>1834</a> // Pages 2-3: other values not all starting with "DE" (select)
+<a href=#1835 id=1835 data-nosnippet>1835</a> </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[
+<a href=#1836 id=1836 data-nosnippet>1836</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#1837 id=1837 data-nosnippet>1837</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#1838 id=1838 data-nosnippet>1838</a> RowSelector::select(<span class="number">2048</span>),
+<a href=#1839 id=1839 data-nosnippet>1839</a> ];
+<a href=#1840 id=1840 data-nosnippet>1840</a>
+<a href=#1841 id=1841 data-nosnippet>1841</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1842 id=1842 data-nosnippet>1842</a>
+<a href=#1843 id=1843 data-nosnippet>1843</a> <span class="prelude-val">Ok</span>(())
+<a href=#1844 id=1844 data-nosnippet>1844</a> }
+<a href=#1845 id=1845 data-nosnippet>1845</a>
+<a href=#1846 id=1846 data-nosnippet>1846</a> <span class="attr">#[test]
+<a href=#1847 id=1847 data-nosnippet>1847</a> </span><span class="kw">fn </span>eval_in_length_of_set_above_limit_all_rows() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1848 id=1848 data-nosnippet>1848</a> <span class="kw">let </span><span class="kw-2">mut </span>rng = rand::rng();
+<a href=#1849 id=1849 data-nosnippet>1849</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1850 id=1850 data-nosnippet>1850</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1851 id=1851 data-nosnippet>1851</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1852 id=1852 data-nosnippet>1852</a>
+<a href=#1853 id=1853 data-nosnippet>1853</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_float"</span>)
+<a href=#1854 id=1854 data-nosnippet>1854</a> .is_in(
+<a href=#1855 id=1855 data-nosnippet>1855</a> std::iter::repeat_with(|| Datum::float(rng.random_range(<span class="number">0.0_f32</span>..<span class="number">10.0_f32</span>)))
+<a href=#1856 id=1856 data-nosnippet>1856</a> .take(<span class="number">1000</span>),
+<a href=#1857 id=1857 data-nosnippet>1857</a> )
+<a href=#1858 id=1858 data-nosnippet>1858</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
<a href=#1859 id=1859 data-nosnippet>1859</a>
-<a href=#1860 id=1860 data-nosnippet>1860</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1861 id=1861 data-nosnippet>1861</a>
-<a href=#1862 id=1862 data-nosnippet>1862</a> <span class="prelude-val">Ok</span>(())
-<a href=#1863 id=1863 data-nosnippet>1863</a> }
-<a href=#1864 id=1864 data-nosnippet>1864</a>
-<a href=#1865 id=1865 data-nosnippet>1865</a> <span class="attr">#[test]
-<a href=#1866 id=1866 data-nosnippet>1866</a> </span><span class="kw">fn </span>eval_in_valid_set_size_some_rows() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1867 id=1867 data-nosnippet>1867</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
-<a href=#1868 id=1868 data-nosnippet>1868</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1869 id=1869 data-nosnippet>1869</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1860 id=1860 data-nosnippet>1860</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1861 id=1861 data-nosnippet>1861</a> <span class="kw-2">&</span>filter,
+<a href=#1862 id=1862 data-nosnippet>1862</a> <span class="kw-2">&</span>column_index,
+<a href=#1863 id=1863 data-nosnippet>1863</a> <span class="kw-2">&</span>offset_index,
+<a href=#1864 id=1864 data-nosnippet>1864</a> row_group_metadata,
+<a href=#1865 id=1865 data-nosnippet>1865</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1866 id=1866 data-nosnippet>1866</a> iceberg_schema_ref.as_ref(),
+<a href=#1867 id=1867 data-nosnippet>1867</a> )<span class="question-mark">?</span>;
+<a href=#1868 id=1868 data-nosnippet>1868</a>
+<a href=#1869 id=1869 data-nosnippet>1869</a> <span class="kw">let </span>expected = <span class="macro">vec!</span>[RowSelector::select(<span class="number">4096</span>)];
<a href=#1870 id=1870 data-nosnippet>1870</a>
-<a href=#1871 id=1871 data-nosnippet>1871</a> <span class="comment">// Test is_in with multiple values using min/max bounds
-<a href=#1872 id=1872 data-nosnippet>1872</a> // Our file has 4 pages: ["AARDVARK".."BISON"], ["DEER"], ["GIRAFFE".."HIPPO"], ["HIPPO"]
-<a href=#1873 id=1873 data-nosnippet>1873</a> // Testing is_in(["AARDVARK", "GIRAFFE"]) - both are in different pages
-<a href=#1874 id=1874 data-nosnippet>1874</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_string"</span>)
-<a href=#1875 id=1875 data-nosnippet>1875</a> .is_in([Datum::string(<span class="string">"AARDVARK"</span>), Datum::string(<span class="string">"GIRAFFE"</span>)])
-<a href=#1876 id=1876 data-nosnippet>1876</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1877 id=1877 data-nosnippet>1877</a>
-<a href=#1878 id=1878 data-nosnippet>1878</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1879 id=1879 data-nosnippet>1879</a> <span class="kw-2">&</span>filter,
-<a href=#1880 id=1880 data-nosnippet>1880</a> <span class="kw-2">&</span>column_index,
-<a href=#1881 id=1881 data-nosnippet>1881</a> <span class="kw-2">&</span>offset_index,
-<a href=#1882 id=1882 data-nosnippet>1882</a> row_group_metadata,
-<a href=#1883 id=1883 data-nosnippet>1883</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1884 id=1884 data-nosnippet>1884</a> iceberg_schema_ref.as_ref(),
-<a href=#1885 id=1885 data-nosnippet>1885</a> )<span class="question-mark">?</span>;
-<a href=#1886 id=1886 data-nosnippet>1886</a>
-<a href=#1887 id=1887 data-nosnippet>1887</a> <span class="comment">// Page 0 contains "AARDVARK", page 1 doesn't contain either, page 2 contains "GIRAFFE", page 3 doesn't
-<a href=#1888 id=1888 data-nosnippet>1888</a> </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[
-<a href=#1889 id=1889 data-nosnippet>1889</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#1890 id=1890 data-nosnippet>1890</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#1891 id=1891 data-nosnippet>1891</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#1892 id=1892 data-nosnippet>1892</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#1893 id=1893 data-nosnippet>1893</a> ];
-<a href=#1894 id=1894 data-nosnippet>1894</a>
-<a href=#1895 id=1895 data-nosnippet>1895</a> <span class="macro">assert_eq!</span>(result, expected);
-<a href=#1896 id=1896 data-nosnippet>1896</a>
-<a href=#1897 id=1897 data-nosnippet>1897</a> <span class="prelude-val">Ok</span>(())
-<a href=#1898 id=1898 data-nosnippet>1898</a> }
-<a href=#1899 id=1899 data-nosnippet>1899</a>
-<a href=#1900 id=1900 data-nosnippet>1900</a> <span class="attr">#[test]
-<a href=#1901 id=1901 data-nosnippet>1901</a> </span><span class="kw">fn </span>eval_inequality_prunes_int32_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1902 id=1902 data-nosnippet>1902</a> <span class="comment">// precision 9 -> Parquet INT32 page bounds.
-<a href=#1903 id=1903 data-nosnippet>1903</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">9</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
-<a href=#1904 id=1904 data-nosnippet>1904</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1905 id=1905 data-nosnippet>1905</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">9</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#1906 id=1906 data-nosnippet>1906</a>
-<a href=#1907 id=1907 data-nosnippet>1907</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. `> 2.50` keeps the pages whose
-<a href=#1908 id=1908 data-nosnippet>1908</a> // upper bound exceeds 2.50 (3.00 and 4.00).
-<a href=#1909 id=1909 data-nosnippet>1909</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#1910 id=1910 data-nosnippet>1910</a> .greater_than(decimal_datum(<span class="number">250</span>, <span class="number">2</span>, <span class="number">9</span>)<span class="question-mark">?</span>)
-<a href=#1911 id=1911 data-nosnippet>1911</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1912 id=1912 data-nosnippet>1912</a>
-<a href=#1913 id=1913 data-nosnippet>1913</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1914 id=1914 data-nosnippet>1914</a> <span class="kw-2">&</span>filter,
-<a href=#1915 id=1915 data-nosnippet>1915</a> <span class="kw-2">&</span>column_index,
-<a href=#1916 id=1916 data-nosnippet>1916</a> <span class="kw-2">&</span>offset_index,
-<a href=#1917 id=1917 data-nosnippet>1917</a> row_group_metadata,
-<a href=#1918 id=1918 data-nosnippet>1918</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1919 id=1919 data-nosnippet>1919</a> iceberg_schema.as_ref(),
-<a href=#1920 id=1920 data-nosnippet>1920</a> )<span class="question-mark">?</span>;
-<a href=#1921 id=1921 data-nosnippet>1921</a>
-<a href=#1922 id=1922 data-nosnippet>1922</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
-<a href=#1923 id=1923 data-nosnippet>1923</a> RowSelector::skip(<span class="number">2048</span>),
-<a href=#1924 id=1924 data-nosnippet>1924</a> RowSelector::select(<span class="number">2048</span>)
-<a href=#1925 id=1925 data-nosnippet>1925</a> ]);
-<a href=#1926 id=1926 data-nosnippet>1926</a>
-<a href=#1927 id=1927 data-nosnippet>1927</a> <span class="prelude-val">Ok</span>(())
-<a href=#1928 id=1928 data-nosnippet>1928</a> }
-<a href=#1929 id=1929 data-nosnippet>1929</a>
-<a href=#1930 id=1930 data-nosnippet>1930</a> <span class="attr">#[test]
-<a href=#1931 id=1931 data-nosnippet>1931</a> </span><span class="kw">fn </span>eval_in_prunes_int32_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1932 id=1932 data-nosnippet>1932</a> <span class="comment">// precision 9 -> Parquet INT32 page bounds.
-<a href=#1933 id=1933 data-nosnippet>1933</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">9</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
-<a href=#1934 id=1934 data-nosnippet>1934</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1935 id=1935 data-nosnippet>1935</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">9</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#1936 id=1936 data-nosnippet>1936</a>
-<a href=#1937 id=1937 data-nosnippet>1937</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. IN (2.00, 4.00) keeps only the
-<a href=#1938 id=1938 data-nosnippet>1938</a> // pages whose single value is one of the literals.
-<a href=#1939 id=1939 data-nosnippet>1939</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#1940 id=1940 data-nosnippet>1940</a> .is_in([decimal_datum(<span class="number">200</span>, <span class="number">2</span>, <span class="number">9</span>)<span class="question-mark">?</span>, decimal_datum(<span class="number">400</span>, <span class="number">2</span>, <span class="number">9</span>)<span class="question-mark">?</span>])
-<a href=#1941 id=1941 data-nosnippet>1941</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1942 id=1942 data-nosnippet>1942</a>
-<a href=#1943 id=1943 data-nosnippet>1943</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1944 id=1944 data-nosnippet>1944</a> <span class="kw-2">&</span>filter,
-<a href=#1945 id=1945 data-nosnippet>1945</a> <span class="kw-2">&</span>column_index,
-<a href=#1946 id=1946 data-nosnippet>1946</a> <span class="kw-2">&</span>offset_index,
-<a href=#1947 id=1947 data-nosnippet>1947</a> row_group_metadata,
-<a href=#1948 id=1948 data-nosnippet>1948</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1949 id=1949 data-nosnippet>1949</a> iceberg_schema.as_ref(),
-<a href=#1950 id=1950 data-nosnippet>1950</a> )<span class="question-mark">?</span>;
-<a href=#1951 id=1951 data-nosnippet>1951</a>
-<a href=#1952 id=1952 data-nosnippet>1952</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
-<a href=#1953 id=1953 data-nosnippet>1953</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#1954 id=1954 data-nosnippet>1954</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#1955 id=1955 data-nosnippet>1955</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#1956 id=1956 data-nosnippet>1956</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#1957 id=1957 data-nosnippet>1957</a> ]);
-<a href=#1958 id=1958 data-nosnippet>1958</a>
-<a href=#1959 id=1959 data-nosnippet>1959</a> <span class="prelude-val">Ok</span>(())
-<a href=#1960 id=1960 data-nosnippet>1960</a> }
-<a href=#1961 id=1961 data-nosnippet>1961</a>
-<a href=#1962 id=1962 data-nosnippet>1962</a> <span class="attr">#[test]
-<a href=#1963 id=1963 data-nosnippet>1963</a> </span><span class="kw">fn </span>eval_inequality_prunes_int64_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1964 id=1964 data-nosnippet>1964</a> <span class="comment">// precision 18 -> Parquet INT64 page bounds.
-<a href=#1965 id=1965 data-nosnippet>1965</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">18</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
-<a href=#1966 id=1966 data-nosnippet>1966</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1967 id=1967 data-nosnippet>1967</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">18</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#1968 id=1968 data-nosnippet>1968</a>
-<a href=#1969 id=1969 data-nosnippet>1969</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. `> 2.50` keeps the pages whose
-<a href=#1970 id=1970 data-nosnippet>1970</a> // upper bound exceeds 2.50 (3.00 and 4.00).
-<a href=#1971 id=1971 data-nosnippet>1971</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#1972 id=1972 data-nosnippet>1972</a> .greater_than(decimal_datum(<span class="number">250</span>, <span class="number">2</span>, <span class="number">18</span>)<span class="question-mark">?</span>)
-<a href=#1973 id=1973 data-nosnippet>1973</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#1974 id=1974 data-nosnippet>1974</a>
-<a href=#1975 id=1975 data-nosnippet>1975</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#1976 id=1976 data-nosnippet>1976</a> <span class="kw-2">&</span>filter,
-<a href=#1977 id=1977 data-nosnippet>1977</a> <span class="kw-2">&</span>column_index,
-<a href=#1978 id=1978 data-nosnippet>1978</a> <span class="kw-2">&</span>offset_index,
-<a href=#1979 id=1979 data-nosnippet>1979</a> row_group_metadata,
-<a href=#1980 id=1980 data-nosnippet>1980</a> <span class="kw-2">&</span>field_id_map,
-<a href=#1981 id=1981 data-nosnippet>1981</a> iceberg_schema.as_ref(),
-<a href=#1982 id=1982 data-nosnippet>1982</a> )<span class="question-mark">?</span>;
-<a href=#1983 id=1983 data-nosnippet>1983</a>
-<a href=#1984 id=1984 data-nosnippet>1984</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
-<a href=#1985 id=1985 data-nosnippet>1985</a> RowSelector::skip(<span class="number">2048</span>),
-<a href=#1986 id=1986 data-nosnippet>1986</a> RowSelector::select(<span class="number">2048</span>)
-<a href=#1987 id=1987 data-nosnippet>1987</a> ]);
-<a href=#1988 id=1988 data-nosnippet>1988</a>
-<a href=#1989 id=1989 data-nosnippet>1989</a> <span class="prelude-val">Ok</span>(())
-<a href=#1990 id=1990 data-nosnippet>1990</a> }
-<a href=#1991 id=1991 data-nosnippet>1991</a>
-<a href=#1992 id=1992 data-nosnippet>1992</a> <span class="attr">#[test]
-<a href=#1993 id=1993 data-nosnippet>1993</a> </span><span class="kw">fn </span>eval_in_prunes_int64_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#1994 id=1994 data-nosnippet>1994</a> <span class="comment">// precision 18 -> Parquet INT64 page bounds.
-<a href=#1995 id=1995 data-nosnippet>1995</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">18</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
-<a href=#1996 id=1996 data-nosnippet>1996</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#1997 id=1997 data-nosnippet>1997</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">18</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#1998 id=1998 data-nosnippet>1998</a>
-<a href=#1999 id=1999 data-nosnippet>1999</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. IN (2.00, 4.00) keeps only the
-<a href=#2000 id=2000 data-nosnippet>2000</a> // pages whose single value is one of the literals.
-<a href=#2001 id=2001 data-nosnippet>2001</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#2002 id=2002 data-nosnippet>2002</a> .is_in([decimal_datum(<span class="number">200</span>, <span class="number">2</span>, <span class="number">18</span>)<span class="question-mark">?</span>, decimal_datum(<span class="number">400</span>, <span class="number">2</span>, <span class="number">18</span>)<span class="question-mark">?</span>])
-<a href=#2003 id=2003 data-nosnippet>2003</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#2004 id=2004 data-nosnippet>2004</a>
-<a href=#2005 id=2005 data-nosnippet>2005</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#2006 id=2006 data-nosnippet>2006</a> <span class="kw-2">&</span>filter,
-<a href=#2007 id=2007 data-nosnippet>2007</a> <span class="kw-2">&</span>column_index,
-<a href=#2008 id=2008 data-nosnippet>2008</a> <span class="kw-2">&</span>offset_index,
-<a href=#2009 id=2009 data-nosnippet>2009</a> row_group_metadata,
-<a href=#2010 id=2010 data-nosnippet>2010</a> <span class="kw-2">&</span>field_id_map,
-<a href=#2011 id=2011 data-nosnippet>2011</a> iceberg_schema.as_ref(),
-<a href=#2012 id=2012 data-nosnippet>2012</a> )<span class="question-mark">?</span>;
-<a href=#2013 id=2013 data-nosnippet>2013</a>
-<a href=#2014 id=2014 data-nosnippet>2014</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
-<a href=#2015 id=2015 data-nosnippet>2015</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#2016 id=2016 data-nosnippet>2016</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#2017 id=2017 data-nosnippet>2017</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#2018 id=2018 data-nosnippet>2018</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#2019 id=2019 data-nosnippet>2019</a> ]);
-<a href=#2020 id=2020 data-nosnippet>2020</a>
-<a href=#2021 id=2021 data-nosnippet>2021</a> <span class="prelude-val">Ok</span>(())
-<a href=#2022 id=2022 data-nosnippet>2022</a> }
-<a href=#2023 id=2023 data-nosnippet>2023</a>
-<a href=#2024 id=2024 data-nosnippet>2024</a> <span class="attr">#[test]
-<a href=#2025 id=2025 data-nosnippet>2025</a> </span><span class="kw">fn </span>eval_inequality_prunes_negative_decimal_pages_at_boundary() -> <span class="prelude-ty">Result</span><()> {
-<a href=#2026 id=2026 data-nosnippet>2026</a> <span class="comment">// precision 9 -> Parquet INT32 page bounds, spanning negative values.
-<a href=#2027 id=2027 data-nosnippet>2027</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">9</span>, <span class="number">2</span>, <span class="kw-2">&</span>[-<span class="number">400</span>, -<span class="number">200</span>, <span class="number">100</span>, <span class="number">300</span>])<span class="question-mark">?</span>;
-<a href=#2028 id=2028 data-nosnippet>2028</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#2029 id=2029 data-nosnippet>2029</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">9</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#2030 id=2030 data-nosnippet>2030</a>
-<a href=#2031 id=2031 data-nosnippet>2031</a> <span class="comment">// Pages hold -4.00, -2.00, 1.00, 3.00. `>= -2.00` keeps the pages whose
-<a href=#2032 id=2032 data-nosnippet>2032</a> // upper bound is at least -2.00, including page 1 whose bound equals it.
-<a href=#2033 id=2033 data-nosnippet>2033</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#2034 id=2034 data-nosnippet>2034</a> .greater_than_or_equal_to(decimal_datum(-<span class="number">200</span>, <span class="number">2</span>, <span class="number">9</span>)<span class="question-mark">?</span>)
-<a href=#2035 id=2035 data-nosnippet>2035</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#2036 id=2036 data-nosnippet>2036</a>
-<a href=#2037 id=2037 data-nosnippet>2037</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#2038 id=2038 data-nosnippet>2038</a> <span class="kw-2">&</span>filter,
-<a href=#2039 id=2039 data-nosnippet>2039</a> <span class="kw-2">&</span>column_index,
-<a href=#2040 id=2040 data-nosnippet>2040</a> <span class="kw-2">&</span>offset_index,
-<a href=#2041 id=2041 data-nosnippet>2041</a> row_group_metadata,
-<a href=#2042 id=2042 data-nosnippet>2042</a> <span class="kw-2">&</span>field_id_map,
-<a href=#2043 id=2043 data-nosnippet>2043</a> iceberg_schema.as_ref(),
-<a href=#2044 id=2044 data-nosnippet>2044</a> )<span class="question-mark">?</span>;
-<a href=#2045 id=2045 data-nosnippet>2045</a>
-<a href=#2046 id=2046 data-nosnippet>2046</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
-<a href=#2047 id=2047 data-nosnippet>2047</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#2048 id=2048 data-nosnippet>2048</a> RowSelector::select(<span class="number">3072</span>)
-<a href=#2049 id=2049 data-nosnippet>2049</a> ]);
-<a href=#2050 id=2050 data-nosnippet>2050</a>
-<a href=#2051 id=2051 data-nosnippet>2051</a> <span class="prelude-val">Ok</span>(())
-<a href=#2052 id=2052 data-nosnippet>2052</a> }
-<a href=#2053 id=2053 data-nosnippet>2053</a>
-<a href=#2054 id=2054 data-nosnippet>2054</a> <span class="attr">#[test]
-<a href=#2055 id=2055 data-nosnippet>2055</a> </span><span class="kw">fn </span>eval_inequality_prunes_fixed_len_byte_array_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#2056 id=2056 data-nosnippet>2056</a> <span class="comment">// precision 30 -> Parquet FIXED_LEN_BYTE_ARRAY page bounds.
-<a href=#2057 id=2057 data-nosnippet>2057</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">30</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
-<a href=#2058 id=2058 data-nosnippet>2058</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#2059 id=2059 data-nosnippet>2059</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#2060 id=2060 data-nosnippet>2060</a>
-<a href=#2061 id=2061 data-nosnippet>2061</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. `> 2.50` keeps the pages whose
-<a href=#2062 id=2062 data-nosnippet>2062</a> // upper bound exceeds 2.50 (3.00 and 4.00).
-<a href=#2063 id=2063 data-nosnippet>2063</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#2064 id=2064 data-nosnippet>2064</a> .greater_than(decimal_datum(<span class="number">250</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>)
-<a href=#2065 id=2065 data-nosnippet>2065</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#2066 id=2066 data-nosnippet>2066</a>
-<a href=#2067 id=2067 data-nosnippet>2067</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#2068 id=2068 data-nosnippet>2068</a> <span class="kw-2">&</span>filter,
-<a href=#2069 id=2069 data-nosnippet>2069</a> <span class="kw-2">&</span>column_index,
-<a href=#2070 id=2070 data-nosnippet>2070</a> <span class="kw-2">&</span>offset_index,
-<a href=#2071 id=2071 data-nosnippet>2071</a> row_group_metadata,
-<a href=#2072 id=2072 data-nosnippet>2072</a> <span class="kw-2">&</span>field_id_map,
-<a href=#2073 id=2073 data-nosnippet>2073</a> iceberg_schema.as_ref(),
-<a href=#2074 id=2074 data-nosnippet>2074</a> )<span class="question-mark">?</span>;
-<a href=#2075 id=2075 data-nosnippet>2075</a>
-<a href=#2076 id=2076 data-nosnippet>2076</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
-<a href=#2077 id=2077 data-nosnippet>2077</a> RowSelector::skip(<span class="number">2048</span>),
-<a href=#2078 id=2078 data-nosnippet>2078</a> RowSelector::select(<span class="number">2048</span>)
-<a href=#2079 id=2079 data-nosnippet>2079</a> ]);
-<a href=#2080 id=2080 data-nosnippet>2080</a>
-<a href=#2081 id=2081 data-nosnippet>2081</a> <span class="prelude-val">Ok</span>(())
-<a href=#2082 id=2082 data-nosnippet>2082</a> }
-<a href=#2083 id=2083 data-nosnippet>2083</a>
-<a href=#2084 id=2084 data-nosnippet>2084</a> <span class="attr">#[test]
-<a href=#2085 id=2085 data-nosnippet>2085</a> </span><span class="kw">fn </span>eval_in_prunes_fixed_len_byte_array_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#2086 id=2086 data-nosnippet>2086</a> <span class="comment">// precision 30 -> Parquet FIXED_LEN_BYTE_ARRAY page bounds.
-<a href=#2087 id=2087 data-nosnippet>2087</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">30</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
-<a href=#2088 id=2088 data-nosnippet>2088</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#2089 id=2089 data-nosnippet>2089</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#2090 id=2090 data-nosnippet>2090</a>
-<a href=#2091 id=2091 data-nosnippet>2091</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. IN (2.00, 4.00) keeps only the
-<a href=#2092 id=2092 data-nosnippet>2092</a> // pages whose single value is one of the literals.
-<a href=#2093 id=2093 data-nosnippet>2093</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#2094 id=2094 data-nosnippet>2094</a> .is_in([decimal_datum(<span class="number">200</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>, decimal_datum(<span class="number">400</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>])
-<a href=#2095 id=2095 data-nosnippet>2095</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#2096 id=2096 data-nosnippet>2096</a>
-<a href=#2097 id=2097 data-nosnippet>2097</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#2098 id=2098 data-nosnippet>2098</a> <span class="kw-2">&</span>filter,
-<a href=#2099 id=2099 data-nosnippet>2099</a> <span class="kw-2">&</span>column_index,
-<a href=#2100 id=2100 data-nosnippet>2100</a> <span class="kw-2">&</span>offset_index,
-<a href=#2101 id=2101 data-nosnippet>2101</a> row_group_metadata,
-<a href=#2102 id=2102 data-nosnippet>2102</a> <span class="kw-2">&</span>field_id_map,
-<a href=#2103 id=2103 data-nosnippet>2103</a> iceberg_schema.as_ref(),
-<a href=#2104 id=2104 data-nosnippet>2104</a> )<span class="question-mark">?</span>;
-<a href=#2105 id=2105 data-nosnippet>2105</a>
-<a href=#2106 id=2106 data-nosnippet>2106</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
-<a href=#2107 id=2107 data-nosnippet>2107</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#2108 id=2108 data-nosnippet>2108</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#2109 id=2109 data-nosnippet>2109</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#2110 id=2110 data-nosnippet>2110</a> RowSelector::select(<span class="number">1024</span>),
-<a href=#2111 id=2111 data-nosnippet>2111</a> ]);
-<a href=#2112 id=2112 data-nosnippet>2112</a>
-<a href=#2113 id=2113 data-nosnippet>2113</a> <span class="prelude-val">Ok</span>(())
-<a href=#2114 id=2114 data-nosnippet>2114</a> }
-<a href=#2115 id=2115 data-nosnippet>2115</a>
-<a href=#2116 id=2116 data-nosnippet>2116</a> <span class="attr">#[test]
-<a href=#2117 id=2117 data-nosnippet>2117</a> </span><span class="kw">fn </span>eval_inequality_prunes_negative_fixed_len_byte_array_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
-<a href=#2118 id=2118 data-nosnippet>2118</a> <span class="comment">// precision 30 -> Parquet FIXED_LEN_BYTE_ARRAY page bounds, spanning
-<a href=#2119 id=2119 data-nosnippet>2119</a> // negative values. Negative decimals have a 0xff high byte in two's
-<a href=#2120 id=2120 data-nosnippet>2120</a> // complement, so this only prunes correctly if the bounds are compared
-<a href=#2121 id=2121 data-nosnippet>2121</a> // numerically rather than by raw byte order.
-<a href=#2122 id=2122 data-nosnippet>2122</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">30</span>, <span class="number">2</span>, <span class="kw-2">&</span>[-<span class="number">400</span>, -<span class="number">200</span>, <span class="number">100</span>, <span class="number">300</span>])<span class="question-mark">?</span>;
-<a href=#2123 id=2123 data-nosnippet>2123</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#2124 id=2124 data-nosnippet>2124</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#2125 id=2125 data-nosnippet>2125</a>
-<a href=#2126 id=2126 data-nosnippet>2126</a> <span class="comment">// Pages hold -4.00, -2.00, 1.00, 3.00. `>= -2.00` keeps the pages whose
-<a href=#2127 id=2127 data-nosnippet>2127</a> // upper bound is at least -2.00, including page 1 whose bound equals it.
-<a href=#2128 id=2128 data-nosnippet>2128</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#2129 id=2129 data-nosnippet>2129</a> .greater_than_or_equal_to(decimal_datum(-<span class="number">200</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>)
-<a href=#2130 id=2130 data-nosnippet>2130</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#2131 id=2131 data-nosnippet>2131</a>
-<a href=#2132 id=2132 data-nosnippet>2132</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#2133 id=2133 data-nosnippet>2133</a> <span class="kw-2">&</span>filter,
-<a href=#2134 id=2134 data-nosnippet>2134</a> <span class="kw-2">&</span>column_index,
-<a href=#2135 id=2135 data-nosnippet>2135</a> <span class="kw-2">&</span>offset_index,
-<a href=#2136 id=2136 data-nosnippet>2136</a> row_group_metadata,
-<a href=#2137 id=2137 data-nosnippet>2137</a> <span class="kw-2">&</span>field_id_map,
-<a href=#2138 id=2138 data-nosnippet>2138</a> iceberg_schema.as_ref(),
-<a href=#2139 id=2139 data-nosnippet>2139</a> )<span class="question-mark">?</span>;
-<a href=#2140 id=2140 data-nosnippet>2140</a>
-<a href=#2141 id=2141 data-nosnippet>2141</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
-<a href=#2142 id=2142 data-nosnippet>2142</a> RowSelector::skip(<span class="number">1024</span>),
-<a href=#2143 id=2143 data-nosnippet>2143</a> RowSelector::select(<span class="number">3072</span>)
-<a href=#2144 id=2144 data-nosnippet>2144</a> ]);
-<a href=#2145 id=2145 data-nosnippet>2145</a>
-<a href=#2146 id=2146 data-nosnippet>2146</a> <span class="prelude-val">Ok</span>(())
-<a href=#2147 id=2147 data-nosnippet>2147</a> }
-<a href=#2148 id=2148 data-nosnippet>2148</a>
-<a href=#2149 id=2149 data-nosnippet>2149</a> <span class="attr">#[test]
-<a href=#2150 id=2150 data-nosnippet>2150</a> </span><span class="kw">fn </span>eval_skips_pruning_for_truncated_fixed_len_byte_array_decimal_bound() -> <span class="prelude-ty">Result</span><()> {
-<a href=#2151 id=2151 data-nosnippet>2151</a> <span class="comment">// A non-Iceberg writer can store a decimal as a plain FIXED_LEN_BYTE_ARRAY
-<a href=#2152 id=2152 data-nosnippet>2152</a> // and truncate its column index (parquet-rs truncates FLBA bounds but not
-<a href=#2153 id=2153 data-nosnippet>2153</a> // decimal/float16 ones). A bound narrower than the column's declared
-<a href=#2154 id=2154 data-nosnippet>2154</a> // type_length can't be decoded as an i128 without changing its value, so
-<a href=#2155 id=2155 data-nosnippet>2155</a> // the evaluator skips page pruning for the column rather than aborting the
-<a href=#2156 id=2156 data-nosnippet>2156</a> // file read.
-<a href=#2157 id=2157 data-nosnippet>2157</a> //
-<a href=#2158 id=2158 data-nosnippet>2158</a> // Page 0 is 0.00, page 1 is 2.55; with full-width bounds `> 1.00` would
-<a href=#2159 id=2159 data-nosnippet>2159</a> // prune page 0. The 13-byte values are truncated to 4-byte column-index
-<a href=#2160 id=2160 data-nosnippet>2160</a> // bounds (!= type_length 13), so neither page is pruned instead.
-<a href=#2161 id=2161 data-nosnippet>2161</a> </span><span class="kw">let </span>page0 = [<span class="number">0u8</span>; <span class="number">13</span>]; <span class="comment">// unscaled 0 -> 0.00
-<a href=#2162 id=2162 data-nosnippet>2162</a> </span><span class="kw">let </span><span class="kw-2">mut </span>page1 = [<span class="number">0u8</span>; <span class="number">13</span>];
-<a href=#2163 id=2163 data-nosnippet>2163</a> page1[<span class="number">12</span>] = <span class="number">0xff</span>; <span class="comment">// unscaled 255 -> 2.55
-<a href=#2164 id=2164 data-nosnippet>2164</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_truncated_flba_file(<span class="number">13</span>, <span class="number">4</span>, <span class="kw-2">&</span>[<span class="kw-2">&</span>page0, <span class="kw-2">&</span>page1])<span class="question-mark">?</span>;
-<a href=#2165 id=2165 data-nosnippet>2165</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
-<a href=#2166 id=2166 data-nosnippet>2166</a>
-<a href=#2167 id=2167 data-nosnippet>2167</a> <span class="comment">// The Iceberg schema calls the 13-byte column a Decimal(30,2).
-<a href=#2168 id=2168 data-nosnippet>2168</a> </span><span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
-<a href=#2169 id=2169 data-nosnippet>2169</a>
-<a href=#2170 id=2170 data-nosnippet>2170</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
-<a href=#2171 id=2171 data-nosnippet>2171</a> .greater_than(decimal_datum(<span class="number">100</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>) <span class="comment">// > 1.00
-<a href=#2172 id=2172 data-nosnippet>2172</a> </span>.bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
-<a href=#2173 id=2173 data-nosnippet>2173</a>
-<a href=#2174 id=2174 data-nosnippet>2174</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
-<a href=#2175 id=2175 data-nosnippet>2175</a> <span class="kw-2">&</span>filter,
-<a href=#2176 id=2176 data-nosnippet>2176</a> <span class="kw-2">&</span>column_index,
-<a href=#2177 id=2177 data-nosnippet>2177</a> <span class="kw-2">&</span>offset_index,
-<a href=#2178 id=2178 data-nosnippet>2178</a> row_group_metadata,
-<a href=#2179 id=2179 data-nosnippet>2179</a> <span class="kw-2">&</span>field_id_map,
-<a href=#2180 id=2180 data-nosnippet>2180</a> iceberg_schema.as_ref(),
-<a href=#2181 id=2181 data-nosnippet>2181</a> )<span class="question-mark">?</span>;
-<a href=#2182 id=2182 data-nosnippet>2182</a>
-<a href=#2183 id=2183 data-nosnippet>2183</a> <span class="comment">// Truncated bounds -> no page pruning: every row survives.
-<a href=#2184 id=2184 data-nosnippet>2184</a> </span><span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[RowSelector::select(<span class="number">2048</span>)]);
-<a href=#2185 id=2185 data-nosnippet>2185</a>
-<a href=#2186 id=2186 data-nosnippet>2186</a> <span class="prelude-val">Ok</span>(())
-<a href=#2187 id=2187 data-nosnippet>2187</a> }
-<a href=#2188 id=2188 data-nosnippet>2188</a>
-<a href=#2189 id=2189 data-nosnippet>2189</a> <span class="doccomment">/// Builds a decimal query [`Datum`] from an `unscaled` mantissa, matching
-<a href=#2190 id=2190 data-nosnippet>2190</a> /// the field's `scale` and `precision`.
-<a href=#2191 id=2191 data-nosnippet>2191</a> </span><span class="kw">fn </span>decimal_datum(unscaled: i128, scale: u32, precision: u32) -> <span class="prelude-ty">Result</span><Datum> {
-<a href=#2192 id=2192 data-nosnippet>2192</a> Datum::decimal_with_precision(decimal_from_i128_with_scale(unscaled, scale), precision)
-<a href=#2193 id=2193 data-nosnippet>2193</a> }
-<a href=#2194 id=2194 data-nosnippet>2194</a>
-<a href=#2195 id=2195 data-nosnippet>2195</a> <span class="kw">fn </span>build_decimal_schema_and_field_map(
-<a href=#2196 id=2196 data-nosnippet>2196</a> precision: u32,
-<a href=#2197 id=2197 data-nosnippet>2197</a> scale: u32,
-<a href=#2198 id=2198 data-nosnippet>2198</a> ) -> <span class="prelude-ty">Result</span><(Arc<Schema>, HashMap<i32, usize>)> {
-<a href=#2199 id=2199 data-nosnippet>2199</a> <span class="kw">let </span>iceberg_schema = Arc::new(
-<a href=#2200 id=2200 data-nosnippet>2200</a> Schema::builder()
-<a href=#2201 id=2201 data-nosnippet>2201</a> .with_fields([Arc::new(NestedField::new(
-<a href=#2202 id=2202 data-nosnippet>2202</a> <span class="number">1</span>,
-<a href=#2203 id=2203 data-nosnippet>2203</a> <span class="string">"col_decimal"</span>,
-<a href=#2204 id=2204 data-nosnippet>2204</a> Type::Primitive(PrimitiveType::Decimal { precision, scale }),
-<a href=#2205 id=2205 data-nosnippet>2205</a> <span class="bool-val">true</span>,
-<a href=#2206 id=2206 data-nosnippet>2206</a> ))])
-<a href=#2207 id=2207 data-nosnippet>2207</a> .build()<span class="question-mark">?</span>,
-<a href=#2208 id=2208 data-nosnippet>2208</a> );
-<a href=#2209 id=2209 data-nosnippet>2209</a>
-<a href=#2210 id=2210 data-nosnippet>2210</a> <span class="prelude-val">Ok</span>((iceberg_schema, HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)])))
-<a href=#2211 id=2211 data-nosnippet>2211</a> }
-<a href=#2212 id=2212 data-nosnippet>2212</a>
-<a href=#2213 id=2213 data-nosnippet>2213</a> <span class="kw">fn </span>build_iceberg_schema_and_field_map() -> <span class="prelude-ty">Result</span><(Arc<Schema>, HashMap<i32, usize>)> {
-<a href=#2214 id=2214 data-nosnippet>2214</a> <span class="kw">let </span>iceberg_schema = Schema::builder()
-<a href=#2215 id=2215 data-nosnippet>2215</a> .with_fields([
-<a href=#2216 id=2216 data-nosnippet>2216</a> Arc::new(NestedField::new(
-<a href=#2217 id=2217 data-nosnippet>2217</a> <span class="number">1</span>,
-<a href=#2218 id=2218 data-nosnippet>2218</a> <span class="string">"col_float"</span>,
-<a href=#2219 id=2219 data-nosnippet>2219</a> Type::Primitive(PrimitiveType::Float),
-<a href=#2220 id=2220 data-nosnippet>2220</a> <span class="bool-val">false</span>,
-<a href=#2221 id=2221 data-nosnippet>2221</a> )),
-<a href=#2222 id=2222 data-nosnippet>2222</a> Arc::new(NestedField::new(
-<a href=#2223 id=2223 data-nosnippet>2223</a> <span class="number">2</span>,
-<a href=#2224 id=2224 data-nosnippet>2224</a> <span class="string">"col_string"</span>,
-<a href=#2225 id=2225 data-nosnippet>2225</a> Type::Primitive(PrimitiveType::String),
-<a href=#2226 id=2226 data-nosnippet>2226</a> <span class="bool-val">false</span>,
-<a href=#2227 id=2227 data-nosnippet>2227</a> )),
-<a href=#2228 id=2228 data-nosnippet>2228</a> ])
-<a href=#2229 id=2229 data-nosnippet>2229</a> .build()<span class="question-mark">?</span>;
-<a href=#2230 id=2230 data-nosnippet>2230</a> <span class="kw">let </span>iceberg_schema_ref = Arc::new(iceberg_schema);
-<a href=#2231 id=2231 data-nosnippet>2231</a>
-<a href=#2232 id=2232 data-nosnippet>2232</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>), (<span class="number">2</span>, <span class="number">1</span>)]);
-<a href=#2233 id=2233 data-nosnippet>2233</a>
-<a href=#2234 id=2234 data-nosnippet>2234</a> <span class="prelude-val">Ok</span>((iceberg_schema_ref, field_id_map))
-<a href=#2235 id=2235 data-nosnippet>2235</a> }
-<a href=#2236 id=2236 data-nosnippet>2236</a>}
+<a href=#1871 id=1871 data-nosnippet>1871</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1872 id=1872 data-nosnippet>1872</a>
+<a href=#1873 id=1873 data-nosnippet>1873</a> <span class="prelude-val">Ok</span>(())
+<a href=#1874 id=1874 data-nosnippet>1874</a> }
+<a href=#1875 id=1875 data-nosnippet>1875</a>
+<a href=#1876 id=1876 data-nosnippet>1876</a> <span class="attr">#[test]
+<a href=#1877 id=1877 data-nosnippet>1877</a> </span><span class="kw">fn </span>eval_in_valid_set_size_some_rows() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1878 id=1878 data-nosnippet>1878</a> <span class="kw">let </span>(metadata, _temp_file) = create_test_parquet_file()<span class="question-mark">?</span>;
+<a href=#1879 id=1879 data-nosnippet>1879</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1880 id=1880 data-nosnippet>1880</a> <span class="kw">let </span>(iceberg_schema_ref, field_id_map) = build_iceberg_schema_and_field_map()<span class="question-mark">?</span>;
+<a href=#1881 id=1881 data-nosnippet>1881</a>
+<a href=#1882 id=1882 data-nosnippet>1882</a> <span class="comment">// Test is_in with multiple values using min/max bounds
+<a href=#1883 id=1883 data-nosnippet>1883</a> // Our file has 4 pages: ["AARDVARK".."BISON"], ["DEER"], ["GIRAFFE".."HIPPO"], ["HIPPO"]
+<a href=#1884 id=1884 data-nosnippet>1884</a> // Testing is_in(["AARDVARK", "GIRAFFE"]) - both are in different pages
+<a href=#1885 id=1885 data-nosnippet>1885</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_string"</span>)
+<a href=#1886 id=1886 data-nosnippet>1886</a> .is_in([Datum::string(<span class="string">"AARDVARK"</span>), Datum::string(<span class="string">"GIRAFFE"</span>)])
+<a href=#1887 id=1887 data-nosnippet>1887</a> .bind(iceberg_schema_ref.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1888 id=1888 data-nosnippet>1888</a>
+<a href=#1889 id=1889 data-nosnippet>1889</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1890 id=1890 data-nosnippet>1890</a> <span class="kw-2">&</span>filter,
+<a href=#1891 id=1891 data-nosnippet>1891</a> <span class="kw-2">&</span>column_index,
+<a href=#1892 id=1892 data-nosnippet>1892</a> <span class="kw-2">&</span>offset_index,
+<a href=#1893 id=1893 data-nosnippet>1893</a> row_group_metadata,
+<a href=#1894 id=1894 data-nosnippet>1894</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1895 id=1895 data-nosnippet>1895</a> iceberg_schema_ref.as_ref(),
+<a href=#1896 id=1896 data-nosnippet>1896</a> )<span class="question-mark">?</span>;
+<a href=#1897 id=1897 data-nosnippet>1897</a>
+<a href=#1898 id=1898 data-nosnippet>1898</a> <span class="comment">// Page 0 contains "AARDVARK", page 1 doesn't contain either, page 2 contains "GIRAFFE", page 3 doesn't
+<a href=#1899 id=1899 data-nosnippet>1899</a> </span><span class="kw">let </span>expected = <span class="macro">vec!</span>[
+<a href=#1900 id=1900 data-nosnippet>1900</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#1901 id=1901 data-nosnippet>1901</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#1902 id=1902 data-nosnippet>1902</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#1903 id=1903 data-nosnippet>1903</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#1904 id=1904 data-nosnippet>1904</a> ];
+<a href=#1905 id=1905 data-nosnippet>1905</a>
+<a href=#1906 id=1906 data-nosnippet>1906</a> <span class="macro">assert_eq!</span>(result, expected);
+<a href=#1907 id=1907 data-nosnippet>1907</a>
+<a href=#1908 id=1908 data-nosnippet>1908</a> <span class="prelude-val">Ok</span>(())
+<a href=#1909 id=1909 data-nosnippet>1909</a> }
+<a href=#1910 id=1910 data-nosnippet>1910</a>
+<a href=#1911 id=1911 data-nosnippet>1911</a> <span class="attr">#[test]
+<a href=#1912 id=1912 data-nosnippet>1912</a> </span><span class="kw">fn </span>eval_inequality_prunes_int32_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1913 id=1913 data-nosnippet>1913</a> <span class="comment">// precision 9 -> Parquet INT32 page bounds.
+<a href=#1914 id=1914 data-nosnippet>1914</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">9</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
+<a href=#1915 id=1915 data-nosnippet>1915</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1916 id=1916 data-nosnippet>1916</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">9</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#1917 id=1917 data-nosnippet>1917</a>
+<a href=#1918 id=1918 data-nosnippet>1918</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. `> 2.50` keeps the pages whose
+<a href=#1919 id=1919 data-nosnippet>1919</a> // upper bound exceeds 2.50 (3.00 and 4.00).
+<a href=#1920 id=1920 data-nosnippet>1920</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#1921 id=1921 data-nosnippet>1921</a> .greater_than(decimal_datum(<span class="number">250</span>, <span class="number">2</span>, <span class="number">9</span>)<span class="question-mark">?</span>)
+<a href=#1922 id=1922 data-nosnippet>1922</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1923 id=1923 data-nosnippet>1923</a>
+<a href=#1924 id=1924 data-nosnippet>1924</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1925 id=1925 data-nosnippet>1925</a> <span class="kw-2">&</span>filter,
+<a href=#1926 id=1926 data-nosnippet>1926</a> <span class="kw-2">&</span>column_index,
+<a href=#1927 id=1927 data-nosnippet>1927</a> <span class="kw-2">&</span>offset_index,
+<a href=#1928 id=1928 data-nosnippet>1928</a> row_group_metadata,
+<a href=#1929 id=1929 data-nosnippet>1929</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1930 id=1930 data-nosnippet>1930</a> iceberg_schema.as_ref(),
+<a href=#1931 id=1931 data-nosnippet>1931</a> )<span class="question-mark">?</span>;
+<a href=#1932 id=1932 data-nosnippet>1932</a>
+<a href=#1933 id=1933 data-nosnippet>1933</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#1934 id=1934 data-nosnippet>1934</a> RowSelector::skip(<span class="number">2048</span>),
+<a href=#1935 id=1935 data-nosnippet>1935</a> RowSelector::select(<span class="number">2048</span>)
+<a href=#1936 id=1936 data-nosnippet>1936</a> ]);
+<a href=#1937 id=1937 data-nosnippet>1937</a>
+<a href=#1938 id=1938 data-nosnippet>1938</a> <span class="prelude-val">Ok</span>(())
+<a href=#1939 id=1939 data-nosnippet>1939</a> }
+<a href=#1940 id=1940 data-nosnippet>1940</a>
+<a href=#1941 id=1941 data-nosnippet>1941</a> <span class="attr">#[test]
+<a href=#1942 id=1942 data-nosnippet>1942</a> </span><span class="kw">fn </span>eval_in_prunes_int32_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1943 id=1943 data-nosnippet>1943</a> <span class="comment">// precision 9 -> Parquet INT32 page bounds.
+<a href=#1944 id=1944 data-nosnippet>1944</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">9</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
+<a href=#1945 id=1945 data-nosnippet>1945</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1946 id=1946 data-nosnippet>1946</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">9</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#1947 id=1947 data-nosnippet>1947</a>
+<a href=#1948 id=1948 data-nosnippet>1948</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. IN (2.00, 4.00) keeps only the
+<a href=#1949 id=1949 data-nosnippet>1949</a> // pages whose single value is one of the literals.
+<a href=#1950 id=1950 data-nosnippet>1950</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#1951 id=1951 data-nosnippet>1951</a> .is_in([decimal_datum(<span class="number">200</span>, <span class="number">2</span>, <span class="number">9</span>)<span class="question-mark">?</span>, decimal_datum(<span class="number">400</span>, <span class="number">2</span>, <span class="number">9</span>)<span class="question-mark">?</span>])
+<a href=#1952 id=1952 data-nosnippet>1952</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1953 id=1953 data-nosnippet>1953</a>
+<a href=#1954 id=1954 data-nosnippet>1954</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1955 id=1955 data-nosnippet>1955</a> <span class="kw-2">&</span>filter,
+<a href=#1956 id=1956 data-nosnippet>1956</a> <span class="kw-2">&</span>column_index,
+<a href=#1957 id=1957 data-nosnippet>1957</a> <span class="kw-2">&</span>offset_index,
+<a href=#1958 id=1958 data-nosnippet>1958</a> row_group_metadata,
+<a href=#1959 id=1959 data-nosnippet>1959</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1960 id=1960 data-nosnippet>1960</a> iceberg_schema.as_ref(),
+<a href=#1961 id=1961 data-nosnippet>1961</a> )<span class="question-mark">?</span>;
+<a href=#1962 id=1962 data-nosnippet>1962</a>
+<a href=#1963 id=1963 data-nosnippet>1963</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#1964 id=1964 data-nosnippet>1964</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#1965 id=1965 data-nosnippet>1965</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#1966 id=1966 data-nosnippet>1966</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#1967 id=1967 data-nosnippet>1967</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#1968 id=1968 data-nosnippet>1968</a> ]);
+<a href=#1969 id=1969 data-nosnippet>1969</a>
+<a href=#1970 id=1970 data-nosnippet>1970</a> <span class="prelude-val">Ok</span>(())
+<a href=#1971 id=1971 data-nosnippet>1971</a> }
+<a href=#1972 id=1972 data-nosnippet>1972</a>
+<a href=#1973 id=1973 data-nosnippet>1973</a> <span class="attr">#[test]
+<a href=#1974 id=1974 data-nosnippet>1974</a> </span><span class="kw">fn </span>eval_inequality_prunes_int64_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#1975 id=1975 data-nosnippet>1975</a> <span class="comment">// precision 18 -> Parquet INT64 page bounds.
+<a href=#1976 id=1976 data-nosnippet>1976</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">18</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
+<a href=#1977 id=1977 data-nosnippet>1977</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#1978 id=1978 data-nosnippet>1978</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">18</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#1979 id=1979 data-nosnippet>1979</a>
+<a href=#1980 id=1980 data-nosnippet>1980</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. `> 2.50` keeps the pages whose
+<a href=#1981 id=1981 data-nosnippet>1981</a> // upper bound exceeds 2.50 (3.00 and 4.00).
+<a href=#1982 id=1982 data-nosnippet>1982</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#1983 id=1983 data-nosnippet>1983</a> .greater_than(decimal_datum(<span class="number">250</span>, <span class="number">2</span>, <span class="number">18</span>)<span class="question-mark">?</span>)
+<a href=#1984 id=1984 data-nosnippet>1984</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#1985 id=1985 data-nosnippet>1985</a>
+<a href=#1986 id=1986 data-nosnippet>1986</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#1987 id=1987 data-nosnippet>1987</a> <span class="kw-2">&</span>filter,
+<a href=#1988 id=1988 data-nosnippet>1988</a> <span class="kw-2">&</span>column_index,
+<a href=#1989 id=1989 data-nosnippet>1989</a> <span class="kw-2">&</span>offset_index,
+<a href=#1990 id=1990 data-nosnippet>1990</a> row_group_metadata,
+<a href=#1991 id=1991 data-nosnippet>1991</a> <span class="kw-2">&</span>field_id_map,
+<a href=#1992 id=1992 data-nosnippet>1992</a> iceberg_schema.as_ref(),
+<a href=#1993 id=1993 data-nosnippet>1993</a> )<span class="question-mark">?</span>;
+<a href=#1994 id=1994 data-nosnippet>1994</a>
+<a href=#1995 id=1995 data-nosnippet>1995</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#1996 id=1996 data-nosnippet>1996</a> RowSelector::skip(<span class="number">2048</span>),
+<a href=#1997 id=1997 data-nosnippet>1997</a> RowSelector::select(<span class="number">2048</span>)
+<a href=#1998 id=1998 data-nosnippet>1998</a> ]);
+<a href=#1999 id=1999 data-nosnippet>1999</a>
+<a href=#2000 id=2000 data-nosnippet>2000</a> <span class="prelude-val">Ok</span>(())
+<a href=#2001 id=2001 data-nosnippet>2001</a> }
+<a href=#2002 id=2002 data-nosnippet>2002</a>
+<a href=#2003 id=2003 data-nosnippet>2003</a> <span class="attr">#[test]
+<a href=#2004 id=2004 data-nosnippet>2004</a> </span><span class="kw">fn </span>eval_in_prunes_int64_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2005 id=2005 data-nosnippet>2005</a> <span class="comment">// precision 18 -> Parquet INT64 page bounds.
+<a href=#2006 id=2006 data-nosnippet>2006</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">18</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
+<a href=#2007 id=2007 data-nosnippet>2007</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2008 id=2008 data-nosnippet>2008</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">18</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#2009 id=2009 data-nosnippet>2009</a>
+<a href=#2010 id=2010 data-nosnippet>2010</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. IN (2.00, 4.00) keeps only the
+<a href=#2011 id=2011 data-nosnippet>2011</a> // pages whose single value is one of the literals.
+<a href=#2012 id=2012 data-nosnippet>2012</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2013 id=2013 data-nosnippet>2013</a> .is_in([decimal_datum(<span class="number">200</span>, <span class="number">2</span>, <span class="number">18</span>)<span class="question-mark">?</span>, decimal_datum(<span class="number">400</span>, <span class="number">2</span>, <span class="number">18</span>)<span class="question-mark">?</span>])
+<a href=#2014 id=2014 data-nosnippet>2014</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2015 id=2015 data-nosnippet>2015</a>
+<a href=#2016 id=2016 data-nosnippet>2016</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2017 id=2017 data-nosnippet>2017</a> <span class="kw-2">&</span>filter,
+<a href=#2018 id=2018 data-nosnippet>2018</a> <span class="kw-2">&</span>column_index,
+<a href=#2019 id=2019 data-nosnippet>2019</a> <span class="kw-2">&</span>offset_index,
+<a href=#2020 id=2020 data-nosnippet>2020</a> row_group_metadata,
+<a href=#2021 id=2021 data-nosnippet>2021</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2022 id=2022 data-nosnippet>2022</a> iceberg_schema.as_ref(),
+<a href=#2023 id=2023 data-nosnippet>2023</a> )<span class="question-mark">?</span>;
+<a href=#2024 id=2024 data-nosnippet>2024</a>
+<a href=#2025 id=2025 data-nosnippet>2025</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#2026 id=2026 data-nosnippet>2026</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#2027 id=2027 data-nosnippet>2027</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#2028 id=2028 data-nosnippet>2028</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#2029 id=2029 data-nosnippet>2029</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#2030 id=2030 data-nosnippet>2030</a> ]);
+<a href=#2031 id=2031 data-nosnippet>2031</a>
+<a href=#2032 id=2032 data-nosnippet>2032</a> <span class="prelude-val">Ok</span>(())
+<a href=#2033 id=2033 data-nosnippet>2033</a> }
+<a href=#2034 id=2034 data-nosnippet>2034</a>
+<a href=#2035 id=2035 data-nosnippet>2035</a> <span class="attr">#[test]
+<a href=#2036 id=2036 data-nosnippet>2036</a> </span><span class="kw">fn </span>eval_inequality_prunes_negative_decimal_pages_at_boundary() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2037 id=2037 data-nosnippet>2037</a> <span class="comment">// precision 9 -> Parquet INT32 page bounds, spanning negative values.
+<a href=#2038 id=2038 data-nosnippet>2038</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">9</span>, <span class="number">2</span>, <span class="kw-2">&</span>[-<span class="number">400</span>, -<span class="number">200</span>, <span class="number">100</span>, <span class="number">300</span>])<span class="question-mark">?</span>;
+<a href=#2039 id=2039 data-nosnippet>2039</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2040 id=2040 data-nosnippet>2040</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">9</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#2041 id=2041 data-nosnippet>2041</a>
+<a href=#2042 id=2042 data-nosnippet>2042</a> <span class="comment">// Pages hold -4.00, -2.00, 1.00, 3.00. `>= -2.00` keeps the pages whose
+<a href=#2043 id=2043 data-nosnippet>2043</a> // upper bound is at least -2.00, including page 1 whose bound equals it.
+<a href=#2044 id=2044 data-nosnippet>2044</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2045 id=2045 data-nosnippet>2045</a> .greater_than_or_equal_to(decimal_datum(-<span class="number">200</span>, <span class="number">2</span>, <span class="number">9</span>)<span class="question-mark">?</span>)
+<a href=#2046 id=2046 data-nosnippet>2046</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2047 id=2047 data-nosnippet>2047</a>
+<a href=#2048 id=2048 data-nosnippet>2048</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2049 id=2049 data-nosnippet>2049</a> <span class="kw-2">&</span>filter,
+<a href=#2050 id=2050 data-nosnippet>2050</a> <span class="kw-2">&</span>column_index,
+<a href=#2051 id=2051 data-nosnippet>2051</a> <span class="kw-2">&</span>offset_index,
+<a href=#2052 id=2052 data-nosnippet>2052</a> row_group_metadata,
+<a href=#2053 id=2053 data-nosnippet>2053</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2054 id=2054 data-nosnippet>2054</a> iceberg_schema.as_ref(),
+<a href=#2055 id=2055 data-nosnippet>2055</a> )<span class="question-mark">?</span>;
+<a href=#2056 id=2056 data-nosnippet>2056</a>
+<a href=#2057 id=2057 data-nosnippet>2057</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#2058 id=2058 data-nosnippet>2058</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#2059 id=2059 data-nosnippet>2059</a> RowSelector::select(<span class="number">3072</span>)
+<a href=#2060 id=2060 data-nosnippet>2060</a> ]);
+<a href=#2061 id=2061 data-nosnippet>2061</a>
+<a href=#2062 id=2062 data-nosnippet>2062</a> <span class="prelude-val">Ok</span>(())
+<a href=#2063 id=2063 data-nosnippet>2063</a> }
+<a href=#2064 id=2064 data-nosnippet>2064</a>
+<a href=#2065 id=2065 data-nosnippet>2065</a> <span class="attr">#[test]
+<a href=#2066 id=2066 data-nosnippet>2066</a> </span><span class="kw">fn </span>eval_inequality_prunes_fixed_len_byte_array_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2067 id=2067 data-nosnippet>2067</a> <span class="comment">// precision 30 -> Parquet FIXED_LEN_BYTE_ARRAY page bounds.
+<a href=#2068 id=2068 data-nosnippet>2068</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">30</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
+<a href=#2069 id=2069 data-nosnippet>2069</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2070 id=2070 data-nosnippet>2070</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#2071 id=2071 data-nosnippet>2071</a>
+<a href=#2072 id=2072 data-nosnippet>2072</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. `> 2.50` keeps the pages whose
+<a href=#2073 id=2073 data-nosnippet>2073</a> // upper bound exceeds 2.50 (3.00 and 4.00).
+<a href=#2074 id=2074 data-nosnippet>2074</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2075 id=2075 data-nosnippet>2075</a> .greater_than(decimal_datum(<span class="number">250</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>)
+<a href=#2076 id=2076 data-nosnippet>2076</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2077 id=2077 data-nosnippet>2077</a>
+<a href=#2078 id=2078 data-nosnippet>2078</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2079 id=2079 data-nosnippet>2079</a> <span class="kw-2">&</span>filter,
+<a href=#2080 id=2080 data-nosnippet>2080</a> <span class="kw-2">&</span>column_index,
+<a href=#2081 id=2081 data-nosnippet>2081</a> <span class="kw-2">&</span>offset_index,
+<a href=#2082 id=2082 data-nosnippet>2082</a> row_group_metadata,
+<a href=#2083 id=2083 data-nosnippet>2083</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2084 id=2084 data-nosnippet>2084</a> iceberg_schema.as_ref(),
+<a href=#2085 id=2085 data-nosnippet>2085</a> )<span class="question-mark">?</span>;
+<a href=#2086 id=2086 data-nosnippet>2086</a>
+<a href=#2087 id=2087 data-nosnippet>2087</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#2088 id=2088 data-nosnippet>2088</a> RowSelector::skip(<span class="number">2048</span>),
+<a href=#2089 id=2089 data-nosnippet>2089</a> RowSelector::select(<span class="number">2048</span>)
+<a href=#2090 id=2090 data-nosnippet>2090</a> ]);
+<a href=#2091 id=2091 data-nosnippet>2091</a>
+<a href=#2092 id=2092 data-nosnippet>2092</a> <span class="prelude-val">Ok</span>(())
+<a href=#2093 id=2093 data-nosnippet>2093</a> }
+<a href=#2094 id=2094 data-nosnippet>2094</a>
+<a href=#2095 id=2095 data-nosnippet>2095</a> <span class="attr">#[test]
+<a href=#2096 id=2096 data-nosnippet>2096</a> </span><span class="kw">fn </span>eval_in_prunes_fixed_len_byte_array_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2097 id=2097 data-nosnippet>2097</a> <span class="comment">// precision 30 -> Parquet FIXED_LEN_BYTE_ARRAY page bounds.
+<a href=#2098 id=2098 data-nosnippet>2098</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">30</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="number">100</span>, <span class="number">200</span>, <span class="number">300</span>, <span class="number">400</span>])<span class="question-mark">?</span>;
+<a href=#2099 id=2099 data-nosnippet>2099</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2100 id=2100 data-nosnippet>2100</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#2101 id=2101 data-nosnippet>2101</a>
+<a href=#2102 id=2102 data-nosnippet>2102</a> <span class="comment">// Pages hold 1.00, 2.00, 3.00, 4.00. IN (2.00, 4.00) keeps only the
+<a href=#2103 id=2103 data-nosnippet>2103</a> // pages whose single value is one of the literals.
+<a href=#2104 id=2104 data-nosnippet>2104</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2105 id=2105 data-nosnippet>2105</a> .is_in([decimal_datum(<span class="number">200</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>, decimal_datum(<span class="number">400</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>])
+<a href=#2106 id=2106 data-nosnippet>2106</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2107 id=2107 data-nosnippet>2107</a>
+<a href=#2108 id=2108 data-nosnippet>2108</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2109 id=2109 data-nosnippet>2109</a> <span class="kw-2">&</span>filter,
+<a href=#2110 id=2110 data-nosnippet>2110</a> <span class="kw-2">&</span>column_index,
+<a href=#2111 id=2111 data-nosnippet>2111</a> <span class="kw-2">&</span>offset_index,
+<a href=#2112 id=2112 data-nosnippet>2112</a> row_group_metadata,
+<a href=#2113 id=2113 data-nosnippet>2113</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2114 id=2114 data-nosnippet>2114</a> iceberg_schema.as_ref(),
+<a href=#2115 id=2115 data-nosnippet>2115</a> )<span class="question-mark">?</span>;
+<a href=#2116 id=2116 data-nosnippet>2116</a>
+<a href=#2117 id=2117 data-nosnippet>2117</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#2118 id=2118 data-nosnippet>2118</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#2119 id=2119 data-nosnippet>2119</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#2120 id=2120 data-nosnippet>2120</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#2121 id=2121 data-nosnippet>2121</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#2122 id=2122 data-nosnippet>2122</a> ]);
+<a href=#2123 id=2123 data-nosnippet>2123</a>
+<a href=#2124 id=2124 data-nosnippet>2124</a> <span class="prelude-val">Ok</span>(())
+<a href=#2125 id=2125 data-nosnippet>2125</a> }
+<a href=#2126 id=2126 data-nosnippet>2126</a>
+<a href=#2127 id=2127 data-nosnippet>2127</a> <span class="attr">#[test]
+<a href=#2128 id=2128 data-nosnippet>2128</a> </span><span class="kw">fn </span>eval_inequality_prunes_negative_fixed_len_byte_array_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2129 id=2129 data-nosnippet>2129</a> <span class="comment">// precision 30 -> Parquet FIXED_LEN_BYTE_ARRAY page bounds, spanning
+<a href=#2130 id=2130 data-nosnippet>2130</a> // negative values. Negative decimals have a 0xff high byte in two's
+<a href=#2131 id=2131 data-nosnippet>2131</a> // complement, so this only prunes correctly if the bounds are compared
+<a href=#2132 id=2132 data-nosnippet>2132</a> // numerically rather than by raw byte order.
+<a href=#2133 id=2133 data-nosnippet>2133</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">30</span>, <span class="number">2</span>, <span class="kw-2">&</span>[-<span class="number">400</span>, -<span class="number">200</span>, <span class="number">100</span>, <span class="number">300</span>])<span class="question-mark">?</span>;
+<a href=#2134 id=2134 data-nosnippet>2134</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2135 id=2135 data-nosnippet>2135</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#2136 id=2136 data-nosnippet>2136</a>
+<a href=#2137 id=2137 data-nosnippet>2137</a> <span class="comment">// Pages hold -4.00, -2.00, 1.00, 3.00. `>= -2.00` keeps the pages whose
+<a href=#2138 id=2138 data-nosnippet>2138</a> // upper bound is at least -2.00, including page 1 whose bound equals it.
+<a href=#2139 id=2139 data-nosnippet>2139</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2140 id=2140 data-nosnippet>2140</a> .greater_than_or_equal_to(decimal_datum(-<span class="number">200</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>)
+<a href=#2141 id=2141 data-nosnippet>2141</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2142 id=2142 data-nosnippet>2142</a>
+<a href=#2143 id=2143 data-nosnippet>2143</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2144 id=2144 data-nosnippet>2144</a> <span class="kw-2">&</span>filter,
+<a href=#2145 id=2145 data-nosnippet>2145</a> <span class="kw-2">&</span>column_index,
+<a href=#2146 id=2146 data-nosnippet>2146</a> <span class="kw-2">&</span>offset_index,
+<a href=#2147 id=2147 data-nosnippet>2147</a> row_group_metadata,
+<a href=#2148 id=2148 data-nosnippet>2148</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2149 id=2149 data-nosnippet>2149</a> iceberg_schema.as_ref(),
+<a href=#2150 id=2150 data-nosnippet>2150</a> )<span class="question-mark">?</span>;
+<a href=#2151 id=2151 data-nosnippet>2151</a>
+<a href=#2152 id=2152 data-nosnippet>2152</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#2153 id=2153 data-nosnippet>2153</a> RowSelector::skip(<span class="number">1024</span>),
+<a href=#2154 id=2154 data-nosnippet>2154</a> RowSelector::select(<span class="number">3072</span>)
+<a href=#2155 id=2155 data-nosnippet>2155</a> ]);
+<a href=#2156 id=2156 data-nosnippet>2156</a>
+<a href=#2157 id=2157 data-nosnippet>2157</a> <span class="prelude-val">Ok</span>(())
+<a href=#2158 id=2158 data-nosnippet>2158</a> }
+<a href=#2159 id=2159 data-nosnippet>2159</a>
+<a href=#2160 id=2160 data-nosnippet>2160</a> <span class="attr">#[test]
+<a href=#2161 id=2161 data-nosnippet>2161</a> </span><span class="kw">fn </span>eval_skips_pruning_for_truncated_fixed_len_byte_array_decimal_bound() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2162 id=2162 data-nosnippet>2162</a> <span class="comment">// A non-Iceberg writer can store a decimal as a plain FIXED_LEN_BYTE_ARRAY
+<a href=#2163 id=2163 data-nosnippet>2163</a> // and truncate its column index (parquet-rs truncates FLBA bounds but not
+<a href=#2164 id=2164 data-nosnippet>2164</a> // decimal/float16 ones). A bound narrower than the column's declared
+<a href=#2165 id=2165 data-nosnippet>2165</a> // type_length can't be decoded as an i128 without changing its value, so
+<a href=#2166 id=2166 data-nosnippet>2166</a> // the evaluator skips page pruning for the column rather than aborting the
+<a href=#2167 id=2167 data-nosnippet>2167</a> // file read.
+<a href=#2168 id=2168 data-nosnippet>2168</a> //
+<a href=#2169 id=2169 data-nosnippet>2169</a> // Page 0 is 0.00, page 1 is 2.55; with full-width bounds `> 1.00` would
+<a href=#2170 id=2170 data-nosnippet>2170</a> // prune page 0. The 13-byte values are truncated to 4-byte column-index
+<a href=#2171 id=2171 data-nosnippet>2171</a> // bounds (!= type_length 13), so neither page is pruned instead.
+<a href=#2172 id=2172 data-nosnippet>2172</a> </span><span class="kw">let </span>page0 = [<span class="number">0u8</span>; <span class="number">13</span>]; <span class="comment">// unscaled 0 -> 0.00
+<a href=#2173 id=2173 data-nosnippet>2173</a> </span><span class="kw">let </span><span class="kw-2">mut </span>page1 = [<span class="number">0u8</span>; <span class="number">13</span>];
+<a href=#2174 id=2174 data-nosnippet>2174</a> page1[<span class="number">12</span>] = <span class="number">0xff</span>; <span class="comment">// unscaled 255 -> 2.55
+<a href=#2175 id=2175 data-nosnippet>2175</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_truncated_flba_file(<span class="number">13</span>, <span class="number">4</span>, <span class="kw-2">&</span>[<span class="kw-2">&</span>page0, <span class="kw-2">&</span>page1])<span class="question-mark">?</span>;
+<a href=#2176 id=2176 data-nosnippet>2176</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2177 id=2177 data-nosnippet>2177</a>
+<a href=#2178 id=2178 data-nosnippet>2178</a> <span class="comment">// The Iceberg schema calls the 13-byte column a Decimal(30,2).
+<a href=#2179 id=2179 data-nosnippet>2179</a> </span><span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#2180 id=2180 data-nosnippet>2180</a>
+<a href=#2181 id=2181 data-nosnippet>2181</a> <span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2182 id=2182 data-nosnippet>2182</a> .greater_than(decimal_datum(<span class="number">100</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>) <span class="comment">// > 1.00
+<a href=#2183 id=2183 data-nosnippet>2183</a> </span>.bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2184 id=2184 data-nosnippet>2184</a>
+<a href=#2185 id=2185 data-nosnippet>2185</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2186 id=2186 data-nosnippet>2186</a> <span class="kw-2">&</span>filter,
+<a href=#2187 id=2187 data-nosnippet>2187</a> <span class="kw-2">&</span>column_index,
+<a href=#2188 id=2188 data-nosnippet>2188</a> <span class="kw-2">&</span>offset_index,
+<a href=#2189 id=2189 data-nosnippet>2189</a> row_group_metadata,
+<a href=#2190 id=2190 data-nosnippet>2190</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2191 id=2191 data-nosnippet>2191</a> iceberg_schema.as_ref(),
+<a href=#2192 id=2192 data-nosnippet>2192</a> )<span class="question-mark">?</span>;
+<a href=#2193 id=2193 data-nosnippet>2193</a>
+<a href=#2194 id=2194 data-nosnippet>2194</a> <span class="comment">// Truncated bounds -> no page pruning: every row survives.
+<a href=#2195 id=2195 data-nosnippet>2195</a> </span><span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[RowSelector::select(<span class="number">2048</span>)]);
+<a href=#2196 id=2196 data-nosnippet>2196</a>
+<a href=#2197 id=2197 data-nosnippet>2197</a> <span class="prelude-val">Ok</span>(())
+<a href=#2198 id=2198 data-nosnippet>2198</a> }
+<a href=#2199 id=2199 data-nosnippet>2199</a>
+<a href=#2200 id=2200 data-nosnippet>2200</a> <span class="attr">#[test]
+<a href=#2201 id=2201 data-nosnippet>2201</a> </span><span class="kw">fn </span>eval_inequality_prunes_precision_38_fixed_len_byte_array_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2202 id=2202 data-nosnippet>2202</a> <span class="comment">// precision 38 is the max Iceberg decimal precision -> 16-byte
+<a href=#2203 id=2203 data-nosnippet>2203</a> // FIXED_LEN_BYTE_ARRAY bounds, the width extreme where decoding fills
+<a href=#2204 id=2204 data-nosnippet>2204</a> // the full i128 with no sign-extension padding. Values near +/-(10^38-1)
+<a href=#2205 id=2205 data-nosnippet>2205</a> // carry significant bits in the high-order bytes, and span both signs.
+<a href=#2206 id=2206 data-nosnippet>2206</a> </span><span class="kw">let </span>max = <span class="number">10_i128</span>.pow(<span class="number">38</span>) - <span class="number">1</span>;
+<a href=#2207 id=2207 data-nosnippet>2207</a> <span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">38</span>, <span class="number">2</span>, <span class="kw-2">&</span>[-max, <span class="number">0</span>, max])<span class="question-mark">?</span>;
+<a href=#2208 id=2208 data-nosnippet>2208</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2209 id=2209 data-nosnippet>2209</a> <span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">38</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#2210 id=2210 data-nosnippet>2210</a>
+<a href=#2211 id=2211 data-nosnippet>2211</a> <span class="comment">// `> 0` keeps only the +max page; the -max and 0 pages are pruned.
+<a href=#2212 id=2212 data-nosnippet>2212</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2213 id=2213 data-nosnippet>2213</a> .greater_than(decimal_datum(<span class="number">0</span>, <span class="number">2</span>, <span class="number">38</span>)<span class="question-mark">?</span>)
+<a href=#2214 id=2214 data-nosnippet>2214</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2215 id=2215 data-nosnippet>2215</a>
+<a href=#2216 id=2216 data-nosnippet>2216</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2217 id=2217 data-nosnippet>2217</a> <span class="kw-2">&</span>filter,
+<a href=#2218 id=2218 data-nosnippet>2218</a> <span class="kw-2">&</span>column_index,
+<a href=#2219 id=2219 data-nosnippet>2219</a> <span class="kw-2">&</span>offset_index,
+<a href=#2220 id=2220 data-nosnippet>2220</a> row_group_metadata,
+<a href=#2221 id=2221 data-nosnippet>2221</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2222 id=2222 data-nosnippet>2222</a> iceberg_schema.as_ref(),
+<a href=#2223 id=2223 data-nosnippet>2223</a> )<span class="question-mark">?</span>;
+<a href=#2224 id=2224 data-nosnippet>2224</a>
+<a href=#2225 id=2225 data-nosnippet>2225</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#2226 id=2226 data-nosnippet>2226</a> RowSelector::skip(<span class="number">2048</span>),
+<a href=#2227 id=2227 data-nosnippet>2227</a> RowSelector::select(<span class="number">1024</span>)
+<a href=#2228 id=2228 data-nosnippet>2228</a> ]);
+<a href=#2229 id=2229 data-nosnippet>2229</a>
+<a href=#2230 id=2230 data-nosnippet>2230</a> <span class="prelude-val">Ok</span>(())
+<a href=#2231 id=2231 data-nosnippet>2231</a> }
+<a href=#2232 id=2232 data-nosnippet>2232</a>
+<a href=#2233 id=2233 data-nosnippet>2233</a> <span class="attr">#[test]
+<a href=#2234 id=2234 data-nosnippet>2234</a> </span><span class="kw">fn </span>eval_inequality_prunes_precision_widened_fixed_len_byte_array_decimal_pages() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2235 id=2235 data-nosnippet>2235</a> <span class="comment">// Precision widening keeps a file's original FIXED_LEN_BYTE_ARRAY width,
+<a href=#2236 id=2236 data-nosnippet>2236</a> // so a decimal(20,2) file's 9-byte bounds are read here as decimal(30,2).
+<a href=#2237 id=2237 data-nosnippet>2237</a> // The guard keys off the file's type_length (9), not the field
+<a href=#2238 id=2238 data-nosnippet>2238</a> // precision's canonical width (13).
+<a href=#2239 id=2239 data-nosnippet>2239</a> </span><span class="kw">let </span>(metadata, _temp_file) = create_decimal_parquet_file(<span class="number">20</span>, <span class="number">2</span>, <span class="kw-2">&</span>[-<span class="number">400</span>, <span class="number">100</span>, <span class="number">300</span>])<span class="question-mark">?</span>;
+<a href=#2240 id=2240 data-nosnippet>2240</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2241 id=2241 data-nosnippet>2241</a> <span class="comment">// Read the 9-byte column as decimal(30,2).
+<a href=#2242 id=2242 data-nosnippet>2242</a> </span><span class="kw">let </span>(iceberg_schema, field_id_map) = build_decimal_schema_and_field_map(<span class="number">30</span>, <span class="number">2</span>)<span class="question-mark">?</span>;
+<a href=#2243 id=2243 data-nosnippet>2243</a>
+<a href=#2244 id=2244 data-nosnippet>2244</a> <span class="comment">// Pages hold -4.00, 1.00, 3.00. `> 1.50` keeps only the 3.00 page.
+<a href=#2245 id=2245 data-nosnippet>2245</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2246 id=2246 data-nosnippet>2246</a> .greater_than(decimal_datum(<span class="number">150</span>, <span class="number">2</span>, <span class="number">30</span>)<span class="question-mark">?</span>)
+<a href=#2247 id=2247 data-nosnippet>2247</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2248 id=2248 data-nosnippet>2248</a>
+<a href=#2249 id=2249 data-nosnippet>2249</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2250 id=2250 data-nosnippet>2250</a> <span class="kw-2">&</span>filter,
+<a href=#2251 id=2251 data-nosnippet>2251</a> <span class="kw-2">&</span>column_index,
+<a href=#2252 id=2252 data-nosnippet>2252</a> <span class="kw-2">&</span>offset_index,
+<a href=#2253 id=2253 data-nosnippet>2253</a> row_group_metadata,
+<a href=#2254 id=2254 data-nosnippet>2254</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2255 id=2255 data-nosnippet>2255</a> iceberg_schema.as_ref(),
+<a href=#2256 id=2256 data-nosnippet>2256</a> )<span class="question-mark">?</span>;
+<a href=#2257 id=2257 data-nosnippet>2257</a>
+<a href=#2258 id=2258 data-nosnippet>2258</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#2259 id=2259 data-nosnippet>2259</a> RowSelector::skip(<span class="number">2048</span>),
+<a href=#2260 id=2260 data-nosnippet>2260</a> RowSelector::select(<span class="number">1024</span>)
+<a href=#2261 id=2261 data-nosnippet>2261</a> ]);
+<a href=#2262 id=2262 data-nosnippet>2262</a>
+<a href=#2263 id=2263 data-nosnippet>2263</a> <span class="prelude-val">Ok</span>(())
+<a href=#2264 id=2264 data-nosnippet>2264</a> }
+<a href=#2265 id=2265 data-nosnippet>2265</a>
+<a href=#2266 id=2266 data-nosnippet>2266</a> <span class="attr">#[test]
+<a href=#2267 id=2267 data-nosnippet>2267</a> </span><span class="kw">fn </span>eval_is_null_selects_all_null_fixed_len_byte_array_decimal_page() -> <span class="prelude-ty">Result</span><()> {
+<a href=#2268 id=2268 data-nosnippet>2268</a> <span class="comment">// precision 30 -> FIXED_LEN_BYTE_ARRAY page bounds. Page 0 is all-null,
+<a href=#2269 id=2269 data-nosnippet>2269</a> // so its column index carries no min/max; the arm must pass those absent
+<a href=#2270 id=2270 data-nosnippet>2270</a> // bounds straight to the predicate rather than try to decode them. Pages
+<a href=#2271 id=2271 data-nosnippet>2271</a> // 1-2 hold 2.00 and 3.00.
+<a href=#2272 id=2272 data-nosnippet>2272</a> </span><span class="kw">let </span>(metadata, _temp_file) =
+<a href=#2273 id=2273 data-nosnippet>2273</a> create_decimal_parquet_file_with_null_pages(<span class="number">30</span>, <span class="number">2</span>, <span class="kw-2">&</span>[<span class="prelude-val">None</span>, <span class="prelude-val">Some</span>(<span class="number">200</span>), <span class="prelude-val">Some</span>(<span class="number">300</span>)])<span class="question-mark">?</span>;
+<a href=#2274 id=2274 data-nosnippet>2274</a> <span class="kw">let </span>(column_index, offset_index, row_group_metadata) = get_test_metadata(<span class="kw-2">&</span>metadata);
+<a href=#2275 id=2275 data-nosnippet>2275</a>
+<a href=#2276 id=2276 data-nosnippet>2276</a> <span class="comment">// The column is optional so IS NULL binds to a real predicate (on a
+<a href=#2277 id=2277 data-nosnippet>2277</a> // required field it would simplify to always-false).
+<a href=#2278 id=2278 data-nosnippet>2278</a> </span><span class="kw">let </span>iceberg_schema = Arc::new(
+<a href=#2279 id=2279 data-nosnippet>2279</a> Schema::builder()
+<a href=#2280 id=2280 data-nosnippet>2280</a> .with_fields([Arc::new(NestedField::new(
+<a href=#2281 id=2281 data-nosnippet>2281</a> <span class="number">1</span>,
+<a href=#2282 id=2282 data-nosnippet>2282</a> <span class="string">"col_decimal"</span>,
+<a href=#2283 id=2283 data-nosnippet>2283</a> Type::Primitive(PrimitiveType::Decimal {
+<a href=#2284 id=2284 data-nosnippet>2284</a> precision: <span class="number">30</span>,
+<a href=#2285 id=2285 data-nosnippet>2285</a> scale: <span class="number">2</span>,
+<a href=#2286 id=2286 data-nosnippet>2286</a> }),
+<a href=#2287 id=2287 data-nosnippet>2287</a> <span class="bool-val">false</span>,
+<a href=#2288 id=2288 data-nosnippet>2288</a> ))])
+<a href=#2289 id=2289 data-nosnippet>2289</a> .build()<span class="question-mark">?</span>,
+<a href=#2290 id=2290 data-nosnippet>2290</a> );
+<a href=#2291 id=2291 data-nosnippet>2291</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)]);
+<a href=#2292 id=2292 data-nosnippet>2292</a>
+<a href=#2293 id=2293 data-nosnippet>2293</a> <span class="comment">// IS NULL keeps the all-null page and prunes the two value pages.
+<a href=#2294 id=2294 data-nosnippet>2294</a> </span><span class="kw">let </span>filter = Reference::new(<span class="string">"col_decimal"</span>)
+<a href=#2295 id=2295 data-nosnippet>2295</a> .is_null()
+<a href=#2296 id=2296 data-nosnippet>2296</a> .bind(iceberg_schema.clone(), <span class="bool-val">false</span>)<span class="question-mark">?</span>;
+<a href=#2297 id=2297 data-nosnippet>2297</a>
+<a href=#2298 id=2298 data-nosnippet>2298</a> <span class="kw">let </span>result = PageIndexEvaluator::eval(
+<a href=#2299 id=2299 data-nosnippet>2299</a> <span class="kw-2">&</span>filter,
+<a href=#2300 id=2300 data-nosnippet>2300</a> <span class="kw-2">&</span>column_index,
+<a href=#2301 id=2301 data-nosnippet>2301</a> <span class="kw-2">&</span>offset_index,
+<a href=#2302 id=2302 data-nosnippet>2302</a> row_group_metadata,
+<a href=#2303 id=2303 data-nosnippet>2303</a> <span class="kw-2">&</span>field_id_map,
+<a href=#2304 id=2304 data-nosnippet>2304</a> iceberg_schema.as_ref(),
+<a href=#2305 id=2305 data-nosnippet>2305</a> )<span class="question-mark">?</span>;
+<a href=#2306 id=2306 data-nosnippet>2306</a>
+<a href=#2307 id=2307 data-nosnippet>2307</a> <span class="macro">assert_eq!</span>(result, <span class="macro">vec!</span>[
+<a href=#2308 id=2308 data-nosnippet>2308</a> RowSelector::select(<span class="number">1024</span>),
+<a href=#2309 id=2309 data-nosnippet>2309</a> RowSelector::skip(<span class="number">2048</span>)
+<a href=#2310 id=2310 data-nosnippet>2310</a> ]);
+<a href=#2311 id=2311 data-nosnippet>2311</a>
+<a href=#2312 id=2312 data-nosnippet>2312</a> <span class="prelude-val">Ok</span>(())
+<a href=#2313 id=2313 data-nosnippet>2313</a> }
+<a href=#2314 id=2314 data-nosnippet>2314</a>
+<a href=#2315 id=2315 data-nosnippet>2315</a> <span class="doccomment">/// Builds a decimal query [`Datum`] from an `unscaled` mantissa, matching
+<a href=#2316 id=2316 data-nosnippet>2316</a> /// the field's `scale` and `precision`.
+<a href=#2317 id=2317 data-nosnippet>2317</a> </span><span class="kw">fn </span>decimal_datum(unscaled: i128, scale: u32, precision: u32) -> <span class="prelude-ty">Result</span><Datum> {
+<a href=#2318 id=2318 data-nosnippet>2318</a> Datum::decimal_with_precision(decimal_from_i128_with_scale(unscaled, scale), precision)
+<a href=#2319 id=2319 data-nosnippet>2319</a> }
+<a href=#2320 id=2320 data-nosnippet>2320</a>
+<a href=#2321 id=2321 data-nosnippet>2321</a> <span class="kw">fn </span>build_decimal_schema_and_field_map(
+<a href=#2322 id=2322 data-nosnippet>2322</a> precision: u32,
+<a href=#2323 id=2323 data-nosnippet>2323</a> scale: u32,
+<a href=#2324 id=2324 data-nosnippet>2324</a> ) -> <span class="prelude-ty">Result</span><(Arc<Schema>, HashMap<i32, usize>)> {
+<a href=#2325 id=2325 data-nosnippet>2325</a> <span class="kw">let </span>iceberg_schema = Arc::new(
+<a href=#2326 id=2326 data-nosnippet>2326</a> Schema::builder()
+<a href=#2327 id=2327 data-nosnippet>2327</a> .with_fields([Arc::new(NestedField::new(
+<a href=#2328 id=2328 data-nosnippet>2328</a> <span class="number">1</span>,
+<a href=#2329 id=2329 data-nosnippet>2329</a> <span class="string">"col_decimal"</span>,
+<a href=#2330 id=2330 data-nosnippet>2330</a> Type::Primitive(PrimitiveType::Decimal { precision, scale }),
+<a href=#2331 id=2331 data-nosnippet>2331</a> <span class="bool-val">true</span>,
+<a href=#2332 id=2332 data-nosnippet>2332</a> ))])
+<a href=#2333 id=2333 data-nosnippet>2333</a> .build()<span class="question-mark">?</span>,
+<a href=#2334 id=2334 data-nosnippet>2334</a> );
+<a href=#2335 id=2335 data-nosnippet>2335</a>
+<a href=#2336 id=2336 data-nosnippet>2336</a> <span class="prelude-val">Ok</span>((iceberg_schema, HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>)])))
+<a href=#2337 id=2337 data-nosnippet>2337</a> }
+<a href=#2338 id=2338 data-nosnippet>2338</a>
+<a href=#2339 id=2339 data-nosnippet>2339</a> <span class="kw">fn </span>build_iceberg_schema_and_field_map() -> <span class="prelude-ty">Result</span><(Arc<Schema>, HashMap<i32, usize>)> {
+<a href=#2340 id=2340 data-nosnippet>2340</a> <span class="kw">let </span>iceberg_schema = Schema::builder()
+<a href=#2341 id=2341 data-nosnippet>2341</a> .with_fields([
+<a href=#2342 id=2342 data-nosnippet>2342</a> Arc::new(NestedField::new(
+<a href=#2343 id=2343 data-nosnippet>2343</a> <span class="number">1</span>,
+<a href=#2344 id=2344 data-nosnippet>2344</a> <span class="string">"col_float"</span>,
+<a href=#2345 id=2345 data-nosnippet>2345</a> Type::Primitive(PrimitiveType::Float),
+<a href=#2346 id=2346 data-nosnippet>2346</a> <span class="bool-val">false</span>,
+<a href=#2347 id=2347 data-nosnippet>2347</a> )),
+<a href=#2348 id=2348 data-nosnippet>2348</a> Arc::new(NestedField::new(
+<a href=#2349 id=2349 data-nosnippet>2349</a> <span class="number">2</span>,
+<a href=#2350 id=2350 data-nosnippet>2350</a> <span class="string">"col_string"</span>,
+<a href=#2351 id=2351 data-nosnippet>2351</a> Type::Primitive(PrimitiveType::String),
+<a href=#2352 id=2352 data-nosnippet>2352</a> <span class="bool-val">false</span>,
+<a href=#2353 id=2353 data-nosnippet>2353</a> )),
+<a href=#2354 id=2354 data-nosnippet>2354</a> ])
+<a href=#2355 id=2355 data-nosnippet>2355</a> .build()<span class="question-mark">?</span>;
+<a href=#2356 id=2356 data-nosnippet>2356</a> <span class="kw">let </span>iceberg_schema_ref = Arc::new(iceberg_schema);
+<a href=#2357 id=2357 data-nosnippet>2357</a>
+<a href=#2358 id=2358 data-nosnippet>2358</a> <span class="kw">let </span>field_id_map = HashMap::from_iter([(<span class="number">1</span>, <span class="number">0</span>), (<span class="number">2</span>, <span class="number">1</span>)]);
+<a href=#2359 id=2359 data-nosnippet>2359</a>
+<a href=#2360 id=2360 data-nosnippet>2360</a> <span class="prelude-val">Ok</span>((iceberg_schema_ref, field_id_map))
+<a href=#2361 id=2361 data-nosnippet>2361</a> }
+<a href=#2362 id=2362 data-nosnippet>2362</a>}
</code></pre></div></section></main></body></html>
\ No newline at end of file