Index: lucene/join/src/java/org/apache/lucene/search/join/ToParentBlockJoinQuery.java IDEA additional info: Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP <+>UTF-8 =================================================================== --- lucene/join/src/java/org/apache/lucene/search/join/ToParentBlockJoinQuery.java (revision 1694685) +++ lucene/join/src/java/org/apache/lucene/search/join/ToParentBlockJoinQuery.java (revision ) @@ -270,12 +270,10 @@ // Parent & child docs are supposed to be // orthogonal: - if (nextChildDoc == parentDoc) { - throw new IllegalStateException("child query must only match non-parent docs, but parent docID=" + nextChildDoc + " matched childScorer=" + childScorer.getClass()); - } + checkParentChildOrthogonality(nextChildDoc, parentDoc); //System.out.println(" parentDoc=" + parentDoc); - assert parentDoc != DocIdSetIterator.NO_MORE_DOCS; + //assert parentDoc != DocIdSetIterator.NO_MORE_DOCS; //System.out.println(" nextChildDoc=" + nextChildDoc); if (acceptDocs != null && !acceptDocs.get(parentDoc)) { @@ -287,9 +285,7 @@ // Parent & child docs are supposed to be // orthogonal: - if (nextChildDoc == parentDoc) { - throw new IllegalStateException("child query must only match non-parent docs, but parent docID=" + nextChildDoc + " matched childScorer=" + childScorer.getClass()); - } + checkParentChildOrthogonality(nextChildDoc, parentDoc); continue; } @@ -330,9 +326,7 @@ // Parent & child docs are supposed to be // orthogonal: - if (nextChildDoc == parentDoc) { - throw new IllegalStateException("child query must only match non-parent docs, but parent docID=" + nextChildDoc + " matched childScorer=" + childScorer.getClass()); - } + checkParentChildOrthogonality(nextChildDoc, parentDoc); switch(scoreMode) { case Avg: @@ -401,13 +395,26 @@ } // Parent & child docs are supposed to be orthogonal: - if (nextChildDoc == prevParentDoc) { - throw new IllegalStateException("child query must only match non-parent docs, but parent docID=" + nextChildDoc + " matched childScorer=" + childScorer.getClass()); - } + checkParentChildOrthogonality(nextChildDoc, prevParentDoc); final int nd = nextDoc(); //System.out.println(" return nextParentDoc=" + nd); return nd; + } + + private void checkParentChildOrthogonality(int childDoc, int parentDoc) { + if (childDoc == parentDoc) { + if (childDoc == NO_MORE_DOCS) { + throw new IllegalStateException("Next child document is not found in current segment." + + " This happens when child documents of the same parent are split into different segments." + + " Please ensure that parent documents are indexed together with their children in a single block."); + } else { + ToParentBlockJoinQuery query = (ToParentBlockJoinQuery) this.weight.getQuery(); + throw new IllegalStateException("Child query must only match non-parent docs, but parent docID=" + childDoc + + ", produced by filter=\"" + query.parentsFilter + "\"," + + " is matched by child query=\"" + query.origChildQuery + "\""); + } + } } public Explanation explain(int docBase) throws IOException { Index: lucene/join/src/test/org/apache/lucene/search/join/TestBlockJoinValidation.java IDEA additional info: Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP <+>UTF-8 =================================================================== --- lucene/join/src/test/org/apache/lucene/search/join/TestBlockJoinValidation.java (revision 1694685) +++ lucene/join/src/test/org/apache/lucene/search/join/TestBlockJoinValidation.java (revision ) @@ -17,6 +17,7 @@ * limitations under the License. */ +import java.io.IOException; import java.util.ArrayList; import java.util.List; @@ -53,20 +54,48 @@ @Override public void setUp() throws Exception { super.setUp(); + init(true); + } + + private void init(boolean indexDocsCorrectly) throws IOException { directory = newDirectory(); final IndexWriterConfig config = new IndexWriterConfig(new MockAnalyzer(random())); - final IndexWriter indexWriter = new IndexWriter(directory, config); - for (int i = 0; i < AMOUNT_OF_SEGMENTS; i++) { - List segmentDocs = createDocsForSegment(i); - indexWriter.addDocuments(segmentDocs); - indexWriter.commit(); - } + IndexWriter indexWriter = new IndexWriter(directory, config); + indexDocs(indexWriter, indexDocsCorrectly); indexReader = DirectoryReader.open(indexWriter, random().nextBoolean()); indexWriter.close(); indexSearcher = new IndexSearcher(indexReader); parentsFilter = new BitDocIdSetCachingWrapperFilter(new QueryWrapperFilter(new WildcardQuery(new Term("parent", "*")))); } + private void indexDocs(IndexWriter indexWriter, boolean indexDocsCorrectly) throws IOException { + if (indexDocsCorrectly) { + indexDocsForSegments(indexWriter, 0, AMOUNT_OF_SEGMENTS); + } else { + int poisonedSegmentNumber = random().nextInt(AMOUNT_OF_SEGMENTS-2); + indexDocsForSegments(indexWriter, 0, poisonedSegmentNumber); + int poisonedParent = random().nextInt(AMOUNT_OF_PARENT_DOCS); + // split next block onto 2 segments + List segmentDocs = createDocsForSegment(poisonedSegmentNumber); + int numberOfSplitChildDocs = random().nextInt(AMOUNT_OF_CHILD_DOCS) + 1; + int numberOfDocsInFirstHalf = poisonedParent * (AMOUNT_OF_CHILD_DOCS + 1) + numberOfSplitChildDocs; + indexWriter.addDocuments(segmentDocs.subList(0, numberOfDocsInFirstHalf)); + indexWriter.commit(); + indexWriter.addDocuments(segmentDocs.subList(numberOfDocsInFirstHalf, segmentDocs.size())); + indexWriter.commit(); + // index rest of segments + indexDocsForSegments(indexWriter, poisonedSegmentNumber + 1, AMOUNT_OF_SEGMENTS); + } + } + + void indexDocsForSegments(IndexWriter indexWriter, int fromSegment, int toSegment) throws IOException { + for (int i = fromSegment; i < toSegment; i++) { + List segmentDocs = createDocsForSegment(i); + indexWriter.addDocuments(segmentDocs); + indexWriter.commit(); + } + } + @Override public void tearDown() throws Exception { indexReader.close(); @@ -81,7 +110,7 @@ indexSearcher.search(blockJoinQuery, 1); fail("didn't get expected exception"); } catch (IllegalStateException expected) { - assertTrue(expected.getMessage() != null && expected.getMessage().contains("child query must only match non-parent docs")); + assertTrue(expected.getMessage() != null && expected.getMessage().contains("Child query must only match non-parent docs")); } } @@ -102,7 +131,7 @@ indexSearcher.search(conjunctionQuery, 1); fail("didn't get expected exception"); } catch (IllegalStateException expected) { - assertTrue(expected.getMessage() != null && expected.getMessage().contains("child query must only match non-parent docs")); + assertTrue(expected.getMessage() != null && expected.getMessage().contains("Child query must only match non-parent docs")); } } @@ -140,7 +169,20 @@ } } - private static List createDocsForSegment(int segmentNumber) { + public void testWrongIndexingValidation() throws Exception { + tearDown(); + init(false); + Query childQuery = new WildcardQuery(new Term("child", "*")); + ToParentBlockJoinQuery blockJoinQuery = new ToParentBlockJoinQuery(childQuery, parentsFilter, ScoreMode.None); + try { + indexSearcher.search(blockJoinQuery, 1); + fail("didn't get expected exception"); + } catch (IllegalStateException expected) { + assertTrue(expected.getMessage() != null && expected.getMessage().contains("Next child document is not found in current segment." )); + } + } + + static List createDocsForSegment(int segmentNumber) { List> blocks = new ArrayList<>(AMOUNT_OF_PARENT_DOCS); for (int i = 0; i < AMOUNT_OF_PARENT_DOCS; i++) { blocks.add(createParentDocWithChildren(segmentNumber, i)); @@ -203,10 +245,6 @@ childQueryWithRandomParent.add(new BooleanClause(parentsQuery, BooleanClause.Occur.SHOULD)); childQueryWithRandomParent.add(new BooleanClause(randomChildQuery(randomChildNumber), BooleanClause.Occur.SHOULD)); return childQueryWithRandomParent; - } - - private static Query createParentQuery() { - return new TermQuery(new Term("id", createFieldValue(getRandomParentId()))); } private static int getRandomParentId() {