FunctionScoreQuery should implement two-phase iteration.
FunctionScoreQuery should do two things that it doesn't do today: - propagate the two-phase iterator from the wrapped scorer so that things are still executed efficiently eg. if a phrase or geo-distance query is wrapped - filter out docs that don't have a high enough score using two-phase iteration: this way the score is only checked when everything else matches While doing these changes, I noticed that minScore was ignored when scores were not needed and that explain did not take it into account, so I fixed these issues as well.
This commit is contained in:
Adrien Grand
parent
56d2dd701e
commit
07658f58a8
|
|
@ -1,152 +0,0 @@
|
|||
/*
|
||||
* Licensed to Elasticsearch under one or more contributor
|
||||
* license agreements. See the NOTICE file distributed with
|
||||
* this work for additional information regarding copyright
|
||||
* ownership. Elasticsearch 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.elasticsearch.common.lucene.search.function;
|
||||
|
||||
import org.apache.lucene.search.DocIdSetIterator;
|
||||
import org.apache.lucene.search.Scorer;
|
||||
import org.apache.lucene.search.Weight;
|
||||
|
||||
import java.io.IOException;
|
||||
|
||||
abstract class CustomBoostFactorScorer extends Scorer {
|
||||
|
||||
final Scorer scorer;
|
||||
final DocIdSetIterator iterator;
|
||||
final float maxBoost;
|
||||
final CombineFunction scoreCombiner;
|
||||
|
||||
Float minScore;
|
||||
NextDoc nextDoc;
|
||||
|
||||
CustomBoostFactorScorer(Weight w, Scorer scorer, float maxBoost, CombineFunction scoreCombiner, Float minScore)
|
||||
throws IOException {
|
||||
super(w);
|
||||
if (minScore == null) {
|
||||
nextDoc = new AnyNextDoc();
|
||||
} else {
|
||||
nextDoc = new MinScoreNextDoc();
|
||||
}
|
||||
this.scorer = scorer;
|
||||
this.iterator = scorer.iterator();
|
||||
this.maxBoost = maxBoost;
|
||||
this.scoreCombiner = scoreCombiner;
|
||||
this.minScore = minScore;
|
||||
}
|
||||
|
||||
@Override
|
||||
public int docID() {
|
||||
return scorer.docID();
|
||||
}
|
||||
|
||||
@Override
|
||||
public DocIdSetIterator iterator() {
|
||||
return new DocIdSetIterator() {
|
||||
@Override
|
||||
public int nextDoc() throws IOException {
|
||||
return nextDoc.nextDoc();
|
||||
}
|
||||
@Override
|
||||
public int advance(int target) throws IOException {
|
||||
return nextDoc.advance(target);
|
||||
}
|
||||
@Override
|
||||
public long cost() {
|
||||
return iterator.cost();
|
||||
}
|
||||
@Override
|
||||
public int docID() {
|
||||
return iterator.docID();
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
public abstract float innerScore() throws IOException;
|
||||
|
||||
@Override
|
||||
public float score() throws IOException {
|
||||
return nextDoc.score();
|
||||
}
|
||||
|
||||
@Override
|
||||
public int freq() throws IOException {
|
||||
return scorer.freq();
|
||||
}
|
||||
|
||||
public interface NextDoc {
|
||||
public int advance(int target) throws IOException;
|
||||
|
||||
public int nextDoc() throws IOException;
|
||||
|
||||
public float score() throws IOException;
|
||||
}
|
||||
|
||||
public class MinScoreNextDoc implements NextDoc {
|
||||
float currentScore = Float.MAX_VALUE * -1.0f;
|
||||
|
||||
@Override
|
||||
public int nextDoc() throws IOException {
|
||||
int doc;
|
||||
do {
|
||||
doc = iterator.nextDoc();
|
||||
if (doc == DocIdSetIterator.NO_MORE_DOCS) {
|
||||
return doc;
|
||||
}
|
||||
currentScore = innerScore();
|
||||
} while (currentScore < minScore);
|
||||
return doc;
|
||||
}
|
||||
|
||||
@Override
|
||||
public float score() throws IOException {
|
||||
return currentScore;
|
||||
}
|
||||
|
||||
@Override
|
||||
public int advance(int target) throws IOException {
|
||||
int doc = iterator.advance(target);
|
||||
if (doc == DocIdSetIterator.NO_MORE_DOCS) {
|
||||
return doc;
|
||||
}
|
||||
currentScore = innerScore();
|
||||
if (currentScore < minScore) {
|
||||
return iterator.nextDoc();
|
||||
}
|
||||
return doc;
|
||||
}
|
||||
}
|
||||
|
||||
public class AnyNextDoc implements NextDoc {
|
||||
|
||||
@Override
|
||||
public int nextDoc() throws IOException {
|
||||
return iterator.nextDoc();
|
||||
}
|
||||
|
||||
@Override
|
||||
public float score() throws IOException {
|
||||
return innerScore();
|
||||
}
|
||||
|
||||
@Override
|
||||
public int advance(int target) throws IOException {
|
||||
return iterator.advance(target);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -23,6 +23,7 @@ import org.apache.lucene.index.IndexReader;
|
|||
import org.apache.lucene.index.LeafReaderContext;
|
||||
import org.apache.lucene.index.Term;
|
||||
import org.apache.lucene.search.Explanation;
|
||||
import org.apache.lucene.search.FilterScorer;
|
||||
import org.apache.lucene.search.IndexSearcher;
|
||||
import org.apache.lucene.search.Query;
|
||||
import org.apache.lucene.search.Scorer;
|
||||
|
|
@ -142,7 +143,7 @@ public class FiltersFunctionScoreQuery extends Query {
|
|||
|
||||
@Override
|
||||
public Weight createWeight(IndexSearcher searcher, boolean needsScores) throws IOException {
|
||||
if (needsScores == false) {
|
||||
if (needsScores == false && minScore == null) {
|
||||
return subQuery.createWeight(searcher, needsScores);
|
||||
}
|
||||
|
||||
|
|
@ -184,11 +185,7 @@ public class FiltersFunctionScoreQuery extends Query {
|
|||
subQueryWeight.normalize(norm, boost);
|
||||
}
|
||||
|
||||
@Override
|
||||
public Scorer scorer(LeafReaderContext context) throws IOException {
|
||||
// we ignore scoreDocsInOrder parameter, because we need to score in
|
||||
// order if documents are scored with a script. The
|
||||
// ShardLookup depends on in order scoring.
|
||||
private FiltersFunctionFactorScorer functionScorer(LeafReaderContext context) throws IOException {
|
||||
Scorer subQueryScorer = subQueryWeight.scorer(context);
|
||||
if (subQueryScorer == null) {
|
||||
return null;
|
||||
|
|
@ -201,15 +198,24 @@ public class FiltersFunctionScoreQuery extends Query {
|
|||
Scorer filterScorer = filterWeights[i].scorer(context);
|
||||
docSets[i] = Lucene.asSequentialAccessBits(context.reader().maxDoc(), filterScorer);
|
||||
}
|
||||
return new FiltersFunctionFactorScorer(this, subQueryScorer, scoreMode, filterFunctions, maxBoost, functions, docSets, combineFunction, minScore, needsScores);
|
||||
return new FiltersFunctionFactorScorer(this, subQueryScorer, scoreMode, filterFunctions, maxBoost, functions, docSets, combineFunction, needsScores);
|
||||
}
|
||||
|
||||
@Override
|
||||
public Scorer scorer(LeafReaderContext context) throws IOException {
|
||||
Scorer scorer = functionScorer(context);
|
||||
if (scorer != null && minScore != null) {
|
||||
scorer = new MinScoreScorer(this, scorer, minScore);
|
||||
}
|
||||
return scorer;
|
||||
}
|
||||
|
||||
@Override
|
||||
public Explanation explain(LeafReaderContext context, int doc) throws IOException {
|
||||
|
||||
Explanation subQueryExpl = subQueryWeight.explain(context, doc);
|
||||
if (!subQueryExpl.isMatch()) {
|
||||
return subQueryExpl;
|
||||
Explanation expl = subQueryWeight.explain(context, doc);
|
||||
if (!expl.isMatch()) {
|
||||
return expl;
|
||||
}
|
||||
// First: Gather explanations for all filters
|
||||
List<Explanation> filterExplanations = new ArrayList<>();
|
||||
|
|
@ -218,7 +224,7 @@ public class FiltersFunctionScoreQuery extends Query {
|
|||
filterWeights[i].scorer(context));
|
||||
if (docSet.get(doc)) {
|
||||
FilterFunction filterFunction = filterFunctions[i];
|
||||
Explanation functionExplanation = filterFunction.function.getLeafScoreFunction(context).explainScore(doc, subQueryExpl);
|
||||
Explanation functionExplanation = filterFunction.function.getLeafScoreFunction(context).explainScore(doc, expl);
|
||||
double factor = functionExplanation.getValue();
|
||||
float sc = CombineFunction.toFloat(factor);
|
||||
Explanation filterExplanation = Explanation.match(sc, "function score, product of:",
|
||||
|
|
@ -226,46 +232,52 @@ public class FiltersFunctionScoreQuery extends Query {
|
|||
filterExplanations.add(filterExplanation);
|
||||
}
|
||||
}
|
||||
if (filterExplanations.size() == 0) {
|
||||
return subQueryExpl;
|
||||
if (filterExplanations.size() > 0) {
|
||||
FiltersFunctionFactorScorer scorer = functionScorer(context);
|
||||
int actualDoc = scorer.iterator().advance(doc);
|
||||
assert (actualDoc == doc);
|
||||
double score = scorer.computeScore(doc, expl.getValue());
|
||||
Explanation factorExplanation = Explanation.match(
|
||||
CombineFunction.toFloat(score),
|
||||
"function score, score mode [" + scoreMode.toString().toLowerCase(Locale.ROOT) + "]",
|
||||
filterExplanations);
|
||||
expl = combineFunction.explain(expl, factorExplanation, maxBoost);
|
||||
}
|
||||
|
||||
FiltersFunctionFactorScorer scorer = (FiltersFunctionFactorScorer)scorer(context);
|
||||
int actualDoc = scorer.iterator.advance(doc);
|
||||
assert (actualDoc == doc);
|
||||
double score = scorer.computeScore(doc, subQueryExpl.getValue());
|
||||
Explanation factorExplanation = Explanation.match(
|
||||
CombineFunction.toFloat(score),
|
||||
"function score, score mode [" + scoreMode.toString().toLowerCase(Locale.ROOT) + "]",
|
||||
filterExplanations);
|
||||
return combineFunction.explain(subQueryExpl, factorExplanation, maxBoost);
|
||||
if (minScore != null && minScore > expl.getValue()) {
|
||||
expl = Explanation.noMatch("Score value is too low, expected at least " + minScore + " but got " + expl.getValue(), expl);
|
||||
}
|
||||
return expl;
|
||||
}
|
||||
}
|
||||
|
||||
static class FiltersFunctionFactorScorer extends CustomBoostFactorScorer {
|
||||
static class FiltersFunctionFactorScorer extends FilterScorer {
|
||||
private final FilterFunction[] filterFunctions;
|
||||
private final ScoreMode scoreMode;
|
||||
private final LeafScoreFunction[] functions;
|
||||
private final Bits[] docSets;
|
||||
private final CombineFunction scoreCombiner;
|
||||
private final float maxBoost;
|
||||
private final boolean needsScores;
|
||||
|
||||
private FiltersFunctionFactorScorer(CustomBoostFactorWeight w, Scorer scorer, ScoreMode scoreMode, FilterFunction[] filterFunctions,
|
||||
float maxBoost, LeafScoreFunction[] functions, Bits[] docSets, CombineFunction scoreCombiner, Float minScore, boolean needsScores) throws IOException {
|
||||
super(w, scorer, maxBoost, scoreCombiner, minScore);
|
||||
float maxBoost, LeafScoreFunction[] functions, Bits[] docSets, CombineFunction scoreCombiner, boolean needsScores) throws IOException {
|
||||
super(scorer, w);
|
||||
this.scoreMode = scoreMode;
|
||||
this.filterFunctions = filterFunctions;
|
||||
this.functions = functions;
|
||||
this.docSets = docSets;
|
||||
this.scoreCombiner = scoreCombiner;
|
||||
this.maxBoost = maxBoost;
|
||||
this.needsScores = needsScores;
|
||||
}
|
||||
|
||||
@Override
|
||||
public float innerScore() throws IOException {
|
||||
int docId = scorer.docID();
|
||||
public float score() throws IOException {
|
||||
int docId = docID();
|
||||
// Even if the weight is created with needsScores=false, it might
|
||||
// be costly to call score(), so we explicitly check if scores
|
||||
// are needed
|
||||
float subQueryScore = needsScores ? scorer.score() : 0f;
|
||||
float subQueryScore = needsScores ? super.score() : 0f;
|
||||
double factor = computeScore(docId, subQueryScore);
|
||||
return scoreCombiner.combine(subQueryScore, factor, maxBoost);
|
||||
}
|
||||
|
|
|
|||
|
|
@ -23,6 +23,7 @@ import org.apache.lucene.index.IndexReader;
|
|||
import org.apache.lucene.index.LeafReaderContext;
|
||||
import org.apache.lucene.index.Term;
|
||||
import org.apache.lucene.search.Explanation;
|
||||
import org.apache.lucene.search.FilterScorer;
|
||||
import org.apache.lucene.search.IndexSearcher;
|
||||
import org.apache.lucene.search.Query;
|
||||
import org.apache.lucene.search.Scorer;
|
||||
|
|
@ -90,7 +91,7 @@ public class FunctionScoreQuery extends Query {
|
|||
|
||||
@Override
|
||||
public Weight createWeight(IndexSearcher searcher, boolean needsScores) throws IOException {
|
||||
if (needsScores == false) {
|
||||
if (needsScores == false && minScore == null) {
|
||||
return subQuery.createWeight(searcher, needsScores);
|
||||
}
|
||||
|
||||
|
|
@ -128,8 +129,7 @@ public class FunctionScoreQuery extends Query {
|
|||
subQueryWeight.normalize(norm, boost);
|
||||
}
|
||||
|
||||
@Override
|
||||
public Scorer scorer(LeafReaderContext context) throws IOException {
|
||||
private FunctionFactorScorer functionScorer(LeafReaderContext context) throws IOException {
|
||||
Scorer subQueryScorer = subQueryWeight.scorer(context);
|
||||
if (subQueryScorer == null) {
|
||||
return null;
|
||||
|
|
@ -138,7 +138,16 @@ public class FunctionScoreQuery extends Query {
|
|||
if (function != null) {
|
||||
leafFunction = function.getLeafScoreFunction(context);
|
||||
}
|
||||
return new FunctionFactorScorer(this, subQueryScorer, leafFunction, maxBoost, combineFunction, minScore, needsScores);
|
||||
return new FunctionFactorScorer(this, subQueryScorer, leafFunction, maxBoost, combineFunction, needsScores);
|
||||
}
|
||||
|
||||
@Override
|
||||
public Scorer scorer(LeafReaderContext context) throws IOException {
|
||||
Scorer scorer = functionScorer(context);
|
||||
if (scorer != null && minScore != null) {
|
||||
scorer = new MinScoreScorer(this, scorer, minScore);
|
||||
}
|
||||
return scorer;
|
||||
}
|
||||
|
||||
@Override
|
||||
|
|
@ -147,38 +156,47 @@ public class FunctionScoreQuery extends Query {
|
|||
if (!subQueryExpl.isMatch()) {
|
||||
return subQueryExpl;
|
||||
}
|
||||
Explanation expl;
|
||||
if (function != null) {
|
||||
Explanation functionExplanation = function.getLeafScoreFunction(context).explainScore(doc, subQueryExpl);
|
||||
return combineFunction.explain(subQueryExpl, functionExplanation, maxBoost);
|
||||
expl = combineFunction.explain(subQueryExpl, functionExplanation, maxBoost);
|
||||
} else {
|
||||
return subQueryExpl;
|
||||
expl = subQueryExpl;
|
||||
}
|
||||
if (minScore != null && minScore > expl.getValue()) {
|
||||
expl = Explanation.noMatch("Score value is too low, expected at least " + minScore + " but got " + expl.getValue(), expl);
|
||||
}
|
||||
return expl;
|
||||
}
|
||||
}
|
||||
|
||||
static class FunctionFactorScorer extends CustomBoostFactorScorer {
|
||||
static class FunctionFactorScorer extends FilterScorer {
|
||||
|
||||
private final LeafScoreFunction function;
|
||||
private final boolean needsScores;
|
||||
private final CombineFunction scoreCombiner;
|
||||
private final float maxBoost;
|
||||
|
||||
private FunctionFactorScorer(CustomBoostFactorWeight w, Scorer scorer, LeafScoreFunction function, float maxBoost, CombineFunction scoreCombiner, Float minScore, boolean needsScores)
|
||||
private FunctionFactorScorer(CustomBoostFactorWeight w, Scorer scorer, LeafScoreFunction function, float maxBoost, CombineFunction scoreCombiner, boolean needsScores)
|
||||
throws IOException {
|
||||
super(w, scorer, maxBoost, scoreCombiner, minScore);
|
||||
super(scorer, w);
|
||||
this.function = function;
|
||||
this.scoreCombiner = scoreCombiner;
|
||||
this.maxBoost = maxBoost;
|
||||
this.needsScores = needsScores;
|
||||
}
|
||||
|
||||
@Override
|
||||
public float innerScore() throws IOException {
|
||||
public float score() throws IOException {
|
||||
// Even if the weight is created with needsScores=false, it might
|
||||
// be costly to call score(), so we explicitly check if scores
|
||||
// are needed
|
||||
float score = needsScores ? scorer.score() : 0f;
|
||||
float score = needsScores ? super.score() : 0f;
|
||||
if (function == null) {
|
||||
return score;
|
||||
} else {
|
||||
return scoreCombiner.combine(score,
|
||||
function.score(scorer.docID(), score), maxBoost);
|
||||
function.score(docID(), score), maxBoost);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -0,0 +1,95 @@
|
|||
/*
|
||||
* Licensed to Elasticsearch under one or more contributor
|
||||
* license agreements. See the NOTICE file distributed with
|
||||
* this work for additional information regarding copyright
|
||||
* ownership. Elasticsearch 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.elasticsearch.common.lucene.search.function;
|
||||
|
||||
import java.io.IOException;
|
||||
|
||||
import org.apache.lucene.search.DocIdSetIterator;
|
||||
import org.apache.lucene.search.ScoreCachingWrappingScorer;
|
||||
import org.apache.lucene.search.Scorer;
|
||||
import org.apache.lucene.search.TwoPhaseIterator;
|
||||
import org.apache.lucene.search.Weight;
|
||||
|
||||
/** A {@link Scorer} that filters out documents that have a score that is
|
||||
* lower than a configured constant. */
|
||||
final class MinScoreScorer extends Scorer {
|
||||
|
||||
private final Scorer in;
|
||||
private final float minScore;
|
||||
|
||||
MinScoreScorer(Weight weight, Scorer scorer, float minScore) {
|
||||
super(weight);
|
||||
if (scorer instanceof ScoreCachingWrappingScorer == false) {
|
||||
// when minScore is set, scores might be requested twice: once
|
||||
// to verify the match, and once by the collector
|
||||
scorer = new ScoreCachingWrappingScorer(scorer);
|
||||
}
|
||||
this.in = scorer;
|
||||
this.minScore = minScore;
|
||||
}
|
||||
|
||||
public Scorer getScorer() {
|
||||
return in;
|
||||
}
|
||||
|
||||
@Override
|
||||
public int docID() {
|
||||
return in.docID();
|
||||
}
|
||||
|
||||
@Override
|
||||
public float score() throws IOException {
|
||||
return in.score();
|
||||
}
|
||||
|
||||
@Override
|
||||
public int freq() throws IOException {
|
||||
return in.freq();
|
||||
}
|
||||
|
||||
@Override
|
||||
public DocIdSetIterator iterator() {
|
||||
return TwoPhaseIterator.asDocIdSetIterator(twoPhaseIterator());
|
||||
}
|
||||
|
||||
@Override
|
||||
public TwoPhaseIterator twoPhaseIterator() {
|
||||
final TwoPhaseIterator inTwoPhase = this.in.twoPhaseIterator();
|
||||
final DocIdSetIterator approximation = inTwoPhase == null ? in.iterator() : inTwoPhase.approximation();
|
||||
return new TwoPhaseIterator(approximation) {
|
||||
|
||||
@Override
|
||||
public boolean matches() throws IOException {
|
||||
// we need to check the two-phase iterator first
|
||||
// otherwise calling score() is illegal
|
||||
if (inTwoPhase != null && inTwoPhase.matches() == false) {
|
||||
return false;
|
||||
}
|
||||
return in.score() >= minScore;
|
||||
}
|
||||
|
||||
@Override
|
||||
public float matchCost() {
|
||||
return 1000f // random constant for the score computation
|
||||
+ (inTwoPhase == null ? 0 : inTwoPhase.matchCost());
|
||||
}
|
||||
};
|
||||
}
|
||||
}
|
||||
|
|
@ -0,0 +1,173 @@
|
|||
/*
|
||||
* Licensed to Elasticsearch under one or more contributor
|
||||
* license agreements. See the NOTICE file distributed with
|
||||
* this work for additional information regarding copyright
|
||||
* ownership. Elasticsearch 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.elasticsearch.common.lucene.search.function;
|
||||
|
||||
import org.apache.lucene.search.DocIdSetIterator;
|
||||
import org.apache.lucene.search.Scorer;
|
||||
import org.apache.lucene.search.TwoPhaseIterator;
|
||||
import org.apache.lucene.util.LuceneTestCase;
|
||||
import org.apache.lucene.util.TestUtil;
|
||||
|
||||
import java.io.IOException;
|
||||
import java.util.Arrays;
|
||||
import java.util.HashSet;
|
||||
import java.util.Set;
|
||||
|
||||
public class MinScoreScorerTests extends LuceneTestCase {
|
||||
|
||||
private static DocIdSetIterator iterator(final int... docs) {
|
||||
return new DocIdSetIterator() {
|
||||
|
||||
int i = -1;
|
||||
|
||||
@Override
|
||||
public int nextDoc() throws IOException {
|
||||
if (i + 1 == docs.length) {
|
||||
return NO_MORE_DOCS;
|
||||
} else {
|
||||
return docs[++i];
|
||||
}
|
||||
}
|
||||
|
||||
@Override
|
||||
public int docID() {
|
||||
return i < 0 ? -1 : i == docs.length ? NO_MORE_DOCS : docs[i];
|
||||
}
|
||||
|
||||
@Override
|
||||
public long cost() {
|
||||
return docs.length;
|
||||
}
|
||||
|
||||
@Override
|
||||
public int advance(int target) throws IOException {
|
||||
return slowAdvance(target);
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
private static Scorer scorer(int maxDoc, final int[] docs, final float[] scores, final boolean twoPhase) {
|
||||
final DocIdSetIterator iterator = twoPhase ? DocIdSetIterator.all(maxDoc) : iterator(docs);
|
||||
return new Scorer(null) {
|
||||
public DocIdSetIterator iterator() {
|
||||
if (twoPhase) {
|
||||
return TwoPhaseIterator.asDocIdSetIterator(twoPhaseIterator());
|
||||
} else {
|
||||
return iterator;
|
||||
}
|
||||
}
|
||||
|
||||
public TwoPhaseIterator twoPhaseIterator() {
|
||||
if (twoPhase) {
|
||||
return new TwoPhaseIterator(iterator) {
|
||||
|
||||
@Override
|
||||
public boolean matches() throws IOException {
|
||||
return Arrays.binarySearch(docs, iterator.docID()) >= 0;
|
||||
}
|
||||
|
||||
@Override
|
||||
public float matchCost() {
|
||||
return 10;
|
||||
}
|
||||
};
|
||||
} else {
|
||||
return null;
|
||||
}
|
||||
}
|
||||
|
||||
@Override
|
||||
public int docID() {
|
||||
return iterator.docID();
|
||||
}
|
||||
|
||||
@Override
|
||||
public float score() throws IOException {
|
||||
final int idx = Arrays.binarySearch(docs, docID());
|
||||
return scores[idx];
|
||||
}
|
||||
|
||||
@Override
|
||||
public int freq() throws IOException {
|
||||
return 1;
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
public void doTestRandom(boolean twoPhase) throws IOException {
|
||||
final int maxDoc = TestUtil.nextInt(random(), 10, 10000);
|
||||
final int numDocs = TestUtil.nextInt(random(), 1, maxDoc / 2);
|
||||
final Set<Integer> uniqueDocs = new HashSet<>();
|
||||
while (uniqueDocs.size() < numDocs) {
|
||||
uniqueDocs.add(random().nextInt(maxDoc));
|
||||
}
|
||||
final int[] docs = new int[numDocs];
|
||||
int i = 0;
|
||||
for (int doc : uniqueDocs) {
|
||||
docs[i++] = doc;
|
||||
}
|
||||
Arrays.sort(docs);
|
||||
final float[] scores = new float[numDocs];
|
||||
for (i = 0; i < numDocs; ++i) {
|
||||
scores[i] = random().nextFloat();
|
||||
}
|
||||
Scorer scorer = scorer(maxDoc, docs, scores, twoPhase);
|
||||
final float minScore = random().nextFloat();
|
||||
Scorer minScoreScorer = new MinScoreScorer(null, scorer, minScore);
|
||||
int doc = -1;
|
||||
while (doc != DocIdSetIterator.NO_MORE_DOCS) {
|
||||
final int target;
|
||||
if (random().nextBoolean()) {
|
||||
target = doc + 1;
|
||||
doc = minScoreScorer.iterator().nextDoc();
|
||||
} else {
|
||||
target = doc + TestUtil.nextInt(random(), 1, 10);
|
||||
doc = minScoreScorer.iterator().advance(target);
|
||||
}
|
||||
int idx = Arrays.binarySearch(docs, target);
|
||||
if (idx < 0) {
|
||||
idx = -1 - idx;
|
||||
}
|
||||
while (idx < docs.length && scores[idx] < minScore) {
|
||||
idx += 1;
|
||||
}
|
||||
if (idx == docs.length) {
|
||||
assertEquals(DocIdSetIterator.NO_MORE_DOCS, doc);
|
||||
} else {
|
||||
assertEquals(docs[idx], doc);
|
||||
assertEquals(scores[idx], scorer.score(), 0f);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
public void testRegularIterator() throws IOException {
|
||||
final int iters = atLeast(5);
|
||||
for (int iter = 0; iter < iters; ++iter) {
|
||||
doTestRandom(false);
|
||||
}
|
||||
}
|
||||
|
||||
public void testTwoPhaseIterator() throws IOException {
|
||||
final int iters = atLeast(5);
|
||||
for (int iter = 0; iter < iters; ++iter) {
|
||||
doTestRandom(true);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -0,0 +1,71 @@
|
|||
/*
|
||||
* Licensed to Elasticsearch under one or more contributor
|
||||
* license agreements. See the NOTICE file distributed with
|
||||
* this work for additional information regarding copyright
|
||||
* ownership. Elasticsearch 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.elasticsearch.index.query.functionscore;
|
||||
|
||||
import org.apache.lucene.index.Term;
|
||||
import org.apache.lucene.search.MatchNoDocsQuery;
|
||||
import org.apache.lucene.search.Query;
|
||||
import org.apache.lucene.search.RandomApproximationQuery;
|
||||
import org.apache.lucene.search.SearchEquivalenceTestBase;
|
||||
import org.apache.lucene.search.TermQuery;
|
||||
import org.elasticsearch.common.lucene.search.function.CombineFunction;
|
||||
import org.elasticsearch.common.lucene.search.function.FiltersFunctionScoreQuery;
|
||||
import org.elasticsearch.common.lucene.search.function.FiltersFunctionScoreQuery.FilterFunction;
|
||||
import org.elasticsearch.common.lucene.search.function.FiltersFunctionScoreQuery.ScoreMode;
|
||||
import org.elasticsearch.common.lucene.search.function.FunctionScoreQuery;
|
||||
|
||||
public class FunctionScoreEquivalenceTests extends SearchEquivalenceTestBase {
|
||||
|
||||
public void testMinScoreAllIncluded() throws Exception {
|
||||
Term term = randomTerm();
|
||||
Query query = new TermQuery(term);
|
||||
|
||||
FunctionScoreQuery fsq = new FunctionScoreQuery(query, null, 0f, null, Float.POSITIVE_INFINITY);
|
||||
assertSameScores(query, fsq);
|
||||
|
||||
FiltersFunctionScoreQuery ffsq = new FiltersFunctionScoreQuery(query, ScoreMode.SUM, new FilterFunction[0], Float.POSITIVE_INFINITY, 0f, CombineFunction.MULTIPLY);
|
||||
assertSameScores(query, ffsq);
|
||||
}
|
||||
|
||||
public void testMinScoreAllExcluded() throws Exception {
|
||||
Term term = randomTerm();
|
||||
Query query = new TermQuery(term);
|
||||
|
||||
FunctionScoreQuery fsq = new FunctionScoreQuery(query, null, Float.POSITIVE_INFINITY, null, Float.POSITIVE_INFINITY);
|
||||
assertSameScores(new MatchNoDocsQuery(), fsq);
|
||||
|
||||
FiltersFunctionScoreQuery ffsq = new FiltersFunctionScoreQuery(query, ScoreMode.SUM, new FilterFunction[0], Float.POSITIVE_INFINITY, Float.POSITIVE_INFINITY, CombineFunction.MULTIPLY);
|
||||
assertSameScores(new MatchNoDocsQuery(), ffsq);
|
||||
}
|
||||
|
||||
public void testTwoPhaseMinScore() throws Exception {
|
||||
Term term = randomTerm();
|
||||
Query query = new TermQuery(term);
|
||||
Float minScore = random().nextFloat();
|
||||
|
||||
FunctionScoreQuery fsq1 = new FunctionScoreQuery(query, null, minScore, null, Float.POSITIVE_INFINITY);
|
||||
FunctionScoreQuery fsq2 = new FunctionScoreQuery(new RandomApproximationQuery(query, random()), null, minScore, null, Float.POSITIVE_INFINITY);
|
||||
assertSameScores(fsq1, fsq2);
|
||||
|
||||
FiltersFunctionScoreQuery ffsq1 = new FiltersFunctionScoreQuery(query, ScoreMode.SUM, new FilterFunction[0], Float.POSITIVE_INFINITY, minScore, CombineFunction.MULTIPLY);
|
||||
FiltersFunctionScoreQuery ffsq2 = new FiltersFunctionScoreQuery(query, ScoreMode.SUM, new FilterFunction[0], Float.POSITIVE_INFINITY, minScore, CombineFunction.MULTIPLY);
|
||||
assertSameScores(ffsq1, ffsq2);
|
||||
}
|
||||
|
||||
}
|
||||
|
|
@ -31,6 +31,9 @@ import org.apache.lucene.index.Term;
|
|||
import org.apache.lucene.search.Explanation;
|
||||
import org.apache.lucene.search.IndexSearcher;
|
||||
import org.apache.lucene.search.MatchAllDocsQuery;
|
||||
import org.apache.lucene.search.Query;
|
||||
import org.apache.lucene.search.RandomApproximationQuery;
|
||||
import org.apache.lucene.search.Scorer;
|
||||
import org.apache.lucene.search.TermQuery;
|
||||
import org.apache.lucene.search.TopDocs;
|
||||
import org.apache.lucene.search.Weight;
|
||||
|
|
@ -46,6 +49,8 @@ import org.elasticsearch.common.lucene.search.function.LeafScoreFunction;
|
|||
import org.elasticsearch.common.lucene.search.function.RandomScoreFunction;
|
||||
import org.elasticsearch.common.lucene.search.function.ScoreFunction;
|
||||
import org.elasticsearch.common.lucene.search.function.WeightFactorFunction;
|
||||
import org.elasticsearch.common.lucene.search.function.FiltersFunctionScoreQuery.FilterFunction;
|
||||
import org.elasticsearch.common.lucene.search.function.FiltersFunctionScoreQuery.ScoreMode;
|
||||
import org.elasticsearch.index.Index;
|
||||
import org.elasticsearch.index.fielddata.AtomicFieldData;
|
||||
import org.elasticsearch.index.fielddata.AtomicNumericFieldData;
|
||||
|
|
@ -559,4 +564,51 @@ public class FunctionScoreTests extends ESTestCase {
|
|||
float score = topDocsWithWeights.scoreDocs[0].score;
|
||||
assertThat(score, equalTo(2.0f));
|
||||
}
|
||||
|
||||
public void testMinScoreExplain() throws IOException {
|
||||
Query query = new MatchAllDocsQuery();
|
||||
Explanation queryExpl = searcher.explain(query, 0);
|
||||
|
||||
FunctionScoreQuery fsq = new FunctionScoreQuery(query, null, 0f, null, Float.POSITIVE_INFINITY);
|
||||
Explanation fsqExpl = searcher.explain(fsq, 0);
|
||||
assertTrue(fsqExpl.isMatch());
|
||||
assertEquals(queryExpl.getValue(), fsqExpl.getValue(), 0f);
|
||||
assertEquals(queryExpl.getDescription(), fsqExpl.getDescription());
|
||||
|
||||
fsq = new FunctionScoreQuery(query, null, 10f, null, Float.POSITIVE_INFINITY);
|
||||
fsqExpl = searcher.explain(fsq, 0);
|
||||
assertFalse(fsqExpl.isMatch());
|
||||
assertEquals("Score value is too low, expected at least 10.0 but got 1.0", fsqExpl.getDescription());
|
||||
|
||||
FiltersFunctionScoreQuery ffsq = new FiltersFunctionScoreQuery(query, ScoreMode.SUM, new FilterFunction[0], Float.POSITIVE_INFINITY, 0f, CombineFunction.MULTIPLY);
|
||||
Explanation ffsqExpl = searcher.explain(ffsq, 0);
|
||||
assertTrue(ffsqExpl.isMatch());
|
||||
assertEquals(queryExpl.getValue(), ffsqExpl.getValue(), 0f);
|
||||
assertEquals(queryExpl.getDescription(), ffsqExpl.getDescription());
|
||||
|
||||
ffsq = new FiltersFunctionScoreQuery(query, ScoreMode.SUM, new FilterFunction[0], Float.POSITIVE_INFINITY, 10f, CombineFunction.MULTIPLY);
|
||||
ffsqExpl = searcher.explain(ffsq, 0);
|
||||
assertFalse(ffsqExpl.isMatch());
|
||||
assertEquals("Score value is too low, expected at least 10.0 but got 1.0", ffsqExpl.getDescription());
|
||||
}
|
||||
|
||||
public void testPropagatesApproximations() throws IOException {
|
||||
Query query = new RandomApproximationQuery(new MatchAllDocsQuery(), random());
|
||||
IndexSearcher searcher = newSearcher(reader);
|
||||
searcher.setQueryCache(null); // otherwise we could get a cached entry that does not have approximations
|
||||
|
||||
FunctionScoreQuery fsq = new FunctionScoreQuery(query, null, null, null, Float.POSITIVE_INFINITY);
|
||||
for (boolean needsScores : new boolean[] {true, false}) {
|
||||
Weight weight = searcher.createWeight(fsq, needsScores);
|
||||
Scorer scorer = weight.scorer(reader.leaves().get(0));
|
||||
assertNotNull(scorer.twoPhaseIterator());
|
||||
}
|
||||
|
||||
FiltersFunctionScoreQuery ffsq = new FiltersFunctionScoreQuery(query, ScoreMode.SUM, new FilterFunction[0], Float.POSITIVE_INFINITY, null, CombineFunction.MULTIPLY);
|
||||
for (boolean needsScores : new boolean[] {true, false}) {
|
||||
Weight weight = searcher.createWeight(ffsq, needsScores);
|
||||
Scorer scorer = weight.scorer(reader.leaves().get(0));
|
||||
assertNotNull(scorer.twoPhaseIterator());
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in New Issue