iprithv commented on code in PR #15991: URL: https://github.com/apache/lucene/pull/15991#discussion_r3218594119
########## lucene/core/src/java/org/apache/lucene/search/PrimarySortAlignable.java: ########## @@ -0,0 +1,49 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package org.apache.lucene.search; + +import java.io.IOException; +import org.apache.lucene.index.LeafReaderContext; + +/** + * Contract for filter {@link Query} types that may expose a contiguous doc id interval on a leaf + * when the index's primary sort order aligns with the filter's field. Used to narrow bulk scoring + * for boolean queries with a single such {@link BooleanClause.Occur#FILTER} clause. + * + * <p>Each implementation must return intervals that are exact for matching documents on the leaf; + * otherwise boolean results can be wrong. When in doubt, return {@code null} from {@link + * #denseDocIdRangeOrNull} so execution falls back to the unoptimized boolean {@link Weight}. + * + * <p>Implementations include {@link IndexSortSortedNumericDocValuesRangeQuery}, {@link TermQuery}, + * {@link PointRangeQuery} (1D int/long ranges), and package-private sorted doc-value range queries + * in {@code org.apache.lucene.document}. + * + * @lucene.experimental + */ +public interface PrimarySortAlignable { + + /** Field constrained by this query. */ + String getField(); + + /** Whether this filter may participate in the optimization on the given index. */ + boolean canOptimize(IndexSearcher searcher) throws IOException; Review Comment: @iverase The density and single value checks are actually already there, but they live in denseDocIdRangeOrNull rather than canOptimize. The reason is that canOptimize is called once at the index level (across all segments) while density is a per leaf property, a segment that went through a merge without forceMerge might be dense, another might not. So canOptimize is intentionally a query eligibility check, and denseDocIdRangeOrNull is where each leaf either confirms a tight dense range or returns null and falls back. Also, TermQuery.denseDocIdRangeOrNull guards with docFreq == range.maxDoc() - range.minDoc(), and IndexSortSortedNumericDocValuesRangeQuery checks pointValues.getDocCount() == reader.maxDoc() before returning a dense range (otherwise it falls back to sparse). So the safety net is there, it's just not visible from canOptimize. Also on deletions, IndexSortSortedNumericDocValuesRangeQuery.getDocIdSetIteratorOrNull already bails early with hasDeletions() == false, so deleted segments don't participate in the optimization. It's not obvious which checks belong in canOptimize vs denseDocIdRangeOrNull. I'll add a note to the interface javadoc if you prefer, to make that boundary explicit. Thanks! -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
