This is an automated email from the ASF dual-hosted git repository.
tingchen pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/pinot.git
The following commit(s) were added to refs/heads/master by this push:
new cfd72684709 [BugFix] Fix NOT TEXT_MATCH fence to exclude all docs when
Lucene searcher has zero visible docs (#18006)
cfd72684709 is described below
commit cfd72684709037500cda024bfed035e2fc3ff1a9
Author: heng-kuang-777 <[email protected]>
AuthorDate: Fri Mar 27 15:26:50 2026 -0700
[BugFix] Fix NOT TEXT_MATCH fence to exclude all docs when Lucene searcher
has zero visible docs (#18006)
In `TextMatchFilterOperator.getSearchableDocCount()`, the condition
`searchableDocCount > 0` caused a fallback to `numDocs` when the Lucene
searcher had refreshed to 0 visible documents. This can occur on a newly
created consuming segment before the first Lucene SearcherManager refresh
has completed. This made NOT TEXT_MATCH invert over the full `[0, numDocs)`
range, producing false positives for all tail documents.
Fix by changing the guard to `searchableDocCount >= 0`, so a zero
searchable count is respected and NOT results are correctly empty.
Adds a regression test covering this corner case.
Co-authored-by: heng kuang <[email protected]>
---
.../pinot/core/operator/filter/TextMatchFilterOperator.java | 2 +-
.../pinot/core/operator/filter/FilterOperatorUtilsTest.java | 8 ++++++++
2 files changed, 9 insertions(+), 1 deletion(-)
diff --git
a/pinot-core/src/main/java/org/apache/pinot/core/operator/filter/TextMatchFilterOperator.java
b/pinot-core/src/main/java/org/apache/pinot/core/operator/filter/TextMatchFilterOperator.java
index b7e775d5d80..b03b6147a49 100644
---
a/pinot-core/src/main/java/org/apache/pinot/core/operator/filter/TextMatchFilterOperator.java
+++
b/pinot-core/src/main/java/org/apache/pinot/core/operator/filter/TextMatchFilterOperator.java
@@ -56,7 +56,7 @@ public class TextMatchFilterOperator extends
BaseFilterOperator {
private static int getSearchableDocCount(TextIndexReader reader, int
numDocs) {
int searchableDocCount = reader.getSearchableDocCount();
- return searchableDocCount > 0 && searchableDocCount < numDocs ?
searchableDocCount : numDocs;
+ return searchableDocCount >= 0 && searchableDocCount < numDocs ?
searchableDocCount : numDocs;
}
public TextMatchFilterOperator(TextIndexReader textIndexReader,
TextMatchPredicate predicate, int numDocs) {
diff --git
a/pinot-core/src/test/java/org/apache/pinot/core/operator/filter/FilterOperatorUtilsTest.java
b/pinot-core/src/test/java/org/apache/pinot/core/operator/filter/FilterOperatorUtilsTest.java
index d805c8d2e66..0dd0d3c871b 100644
---
a/pinot-core/src/test/java/org/apache/pinot/core/operator/filter/FilterOperatorUtilsTest.java
+++
b/pinot-core/src/test/java/org/apache/pinot/core/operator/filter/FilterOperatorUtilsTest.java
@@ -247,6 +247,14 @@ public class FilterOperatorUtilsTest {
assertEquals(result.size(), TOTAL_DOCS - LUCENE_MATCH_DOC_IDS.length);
}
+ @Test
+ public void testTextMatchNotFenceZeroSearchableDocsReturnsEmpty() {
+ // searchableDocCount = 0 means Lucene searcher has not yet refreshed; NOT
result must be empty
+ NotFilterOperator notOp = new NotFilterOperator(textMatchOp(TOTAL_DOCS,
0), TOTAL_DOCS, false);
+ List<Integer> result = TestUtils.getDocIds(notOp.getTrues());
+ assertEquals(result.size(), 0, "NOT result must be empty when no docs are
visible to Lucene yet");
+ }
+
@Test
public void testTextMatchNotFenceNoMatchesReturnsAllSearchable() {
TextMatchPredicate predicate =
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
