org.apache.lucene.sandbox.search.QueryProfilerWeight Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of lucene-sandbox Show documentation
Show all versions of lucene-sandbox Show documentation
Apache Lucene (module: sandbox)
/*
* 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.sandbox.search;
import java.io.IOException;
import org.apache.lucene.index.LeafReaderContext;
import org.apache.lucene.search.BulkScorer;
import org.apache.lucene.search.Explanation;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.Scorer;
import org.apache.lucene.search.ScorerSupplier;
import org.apache.lucene.search.Weight;
/**
* Weight wrapper that will compute how much time it takes to build the {@link Scorer} and then
* return a {@link Scorer} that is wrapped in order to compute timings as well.
*/
class QueryProfilerWeight extends Weight {
private final Weight subQueryWeight;
private final QueryProfilerBreakdown profile;
public QueryProfilerWeight(Query query, Weight subQueryWeight, QueryProfilerBreakdown profile) {
super(query);
this.subQueryWeight = subQueryWeight;
this.profile = profile;
}
@Override
public int count(LeafReaderContext context) throws IOException {
QueryProfilerTimer timer = profile.getTimer(QueryProfilerTimingType.COUNT);
timer.start();
try {
return subQueryWeight.count(context);
} finally {
timer.stop();
}
}
@Override
public Scorer scorer(LeafReaderContext context) throws IOException {
ScorerSupplier supplier = scorerSupplier(context);
if (supplier == null) {
return null;
}
return supplier.get(Long.MAX_VALUE);
}
@Override
public ScorerSupplier scorerSupplier(LeafReaderContext context) throws IOException {
QueryProfilerTimer timer = profile.getTimer(QueryProfilerTimingType.BUILD_SCORER);
timer.start();
final ScorerSupplier subQueryScorerSupplier;
try {
subQueryScorerSupplier = subQueryWeight.scorerSupplier(context);
} finally {
timer.stop();
}
if (subQueryScorerSupplier == null) {
return null;
}
final QueryProfilerWeight weight = this;
return new ScorerSupplier() {
@Override
public Scorer get(long loadCost) throws IOException {
timer.start();
try {
return new QueryProfilerScorer(weight, subQueryScorerSupplier.get(loadCost), profile);
} finally {
timer.stop();
}
}
@Override
public long cost() {
timer.start();
try {
return subQueryScorerSupplier.cost();
} finally {
timer.stop();
}
}
};
}
@Override
public BulkScorer bulkScorer(LeafReaderContext context) throws IOException {
// We use the default bulk scorer instead of the specialized one. The reason
// is that Lucene's BulkScorers do everything at once: finding matches,
// scoring them and calling the collector, so they make it impossible to
// see where time is spent, which is the purpose of query profiling.
// The default bulk scorer will pull a scorer and iterate over matches,
// this might be a significantly different execution path for some queries
// like disjunctions, but in general this is what is done anyway
return super.bulkScorer(context);
}
@Override
public Explanation explain(LeafReaderContext context, int doc) throws IOException {
return subQueryWeight.explain(context, doc);
}
@Override
public boolean isCacheable(LeafReaderContext ctx) {
return false;
}
}