Apache Lucene open-source search software
Optimize MultiTermQueryConstantScoreWrapper for case when a term matches all docs in a segment. (#11738)
Greg Miller734841d6c0f8
diff --git a/lucene/CHANGES.txt b/lucene/CHANGES.txtindex ea22d302723..da470df6e54 100644--- a/lucene/CHANGES.txt+++ b/lucene/CHANGES.txt@@ -115,6 +115,11 @@ Bug Fixestrying to apply a dictionary whose size is greater than the maximum supportedwindow size for LZ4. (Adrien Grand)+Optimizations+---------------------+* GITHUB#11738: Optimize MultiTermQueryConstantScoreWrapper when a term is present that matches all+ docs in a segment. (Greg Miller)+* GITHUB#11735: KeywordRepeatFilter + OpenNLPLemmatizer always drops last token of a stream.(Luke Kot-Zaniewski)diff --git a/lucene/core/src/java/org/apache/lucene/search/MultiTermQueryConstantScoreWrapper.java b/lucene/core/src/java/org/apache/lucene/search/MultiTermQueryConstantScoreWrapper.javaindex f729ed6421c..1271b3c5571 100644--- a/lucene/core/src/java/org/apache/lucene/search/MultiTermQueryConstantScoreWrapper.java+++ b/lucene/core/src/java/org/apache/lucene/search/MultiTermQueryConstantScoreWrapper.java@@ -125,11 +125,11 @@ final class MultiTermQueryConstantScoreWrapper<Q extends MultiTermQuery> extendsreturn new ConstantScoreWeight(this, boost) {/**- * Try to collect terms from the given terms enum and return true iff all terms could be- * collected. If {@code false} is returned, the enum is left positioned on the next term.+ * Try to collect terms from the given terms enum and return true if all terms could be+ * collected or if one of the iterated terms contains all docs for the field. If {@code false}+ * is returned, the enum is left positioned on the next term.*/- private boolean collectTerms(- LeafReaderContext context, TermsEnum termsEnum, List<TermAndState> terms)+ private boolean collectTerms(int fieldDocCount, TermsEnum termsEnum, List<TermAndState> terms)throws IOException {final int threshold =Math.min(BOOLEAN_REWRITE_TERM_COUNT_THRESHOLD, IndexSearcher.getMaxClauseCount());@@ -139,12 +139,18 @@ final class MultiTermQueryConstantScoreWrapper<Q extends MultiTermQuery> extendsreturn true;}TermState state = termsEnum.termState();- terms.add(+ int docFreq = termsEnum.docFreq();+ TermAndState termAndState =new TermAndState(- BytesRef.deepCopyOf(term),- state,- termsEnum.docFreq(),- termsEnum.totalTermFreq()));+ BytesRef.deepCopyOf(term), state, docFreq, termsEnum.totalTermFreq());+ if (fieldDocCount == docFreq) {+ // If the term contains every document with a value for the field, we can ignore all+ // other terms:+ terms.clear();+ terms.add(termAndState);+ return true;+ }+ terms.add(termAndState);}return termsEnum.next() == null;}@@ -160,13 +166,14 @@ final class MultiTermQueryConstantScoreWrapper<Q extends MultiTermQuery> extendsreturn new WeightOrDocIdSet((DocIdSet) null);}+ final int fieldDocCount = terms.getDocCount();final TermsEnum termsEnum = query.getTermsEnum(terms);assert termsEnum != null;PostingsEnum docs = null;final List<TermAndState> collectedTerms = new ArrayList<>();- if (collectTerms(context, termsEnum, collectedTerms)) {+ if (collectTerms(fieldDocCount, termsEnum, collectedTerms)) {// build a boolean queryBooleanQuery.Builder bq = new BooleanQuery.Builder();for (TermAndState t : collectedTerms) {@@ -193,6 +200,19 @@ final class MultiTermQueryConstantScoreWrapper<Q extends MultiTermQuery> extends// Then keep filling the bit set with remaining termsdo {docs = termsEnum.postings(docs, PostingsEnum.NONE);+ // If a term contains all docs with a value for the specified field, we can discard the+ // other terms and just use the dense term's postings:+ int docFreq = termsEnum.docFreq();+ if (fieldDocCount == docFreq) {+ TermStates termStates = new TermStates(searcher.getTopReaderContext());+ termStates.register(+ termsEnum.termState(), context.ord, docFreq, termsEnum.totalTermFreq());+ Query q =+ new ConstantScoreQuery(+ new TermQuery(new Term(query.field, termsEnum.term()), termStates));+ Weight weight = searcher.rewrite(q).createWeight(searcher, scoreMode, score());+ return new WeightOrDocIdSet(weight);+ }builder.add(docs);} while (termsEnum.next() != null);