-
Notifications
You must be signed in to change notification settings - Fork 6
Commit
A frequency caching terms enum, that also allows to be configured with an optional filter. To be used by both significant terms and phrase suggester. This change extracts the frequency caching into the same code, and allow in the future to add a filter to control/customize the background frequencies
- Loading branch information
There are no files selected for viewing
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,261 @@ | ||
/* | ||
* 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.index; | ||
|
||
import com.google.common.collect.Lists; | ||
import org.apache.lucene.index.*; | ||
import org.apache.lucene.search.DocIdSet; | ||
import org.apache.lucene.search.DocIdSetIterator; | ||
import org.apache.lucene.search.Filter; | ||
import org.apache.lucene.util.Bits; | ||
import org.apache.lucene.util.BytesRef; | ||
import org.elasticsearch.ElasticsearchException; | ||
import org.elasticsearch.ElasticsearchIllegalArgumentException; | ||
import org.elasticsearch.common.Nullable; | ||
import org.elasticsearch.common.lease.Releasable; | ||
import org.elasticsearch.common.lease.Releasables; | ||
import org.elasticsearch.common.lucene.docset.DocIdSets; | ||
import org.elasticsearch.common.lucene.search.ApplyAcceptedDocsFilter; | ||
import org.elasticsearch.common.lucene.search.Queries; | ||
import org.elasticsearch.common.util.BigArrays; | ||
import org.elasticsearch.common.util.BytesRefHash; | ||
import org.elasticsearch.common.util.IntArray; | ||
import org.elasticsearch.common.util.LongArray; | ||
|
||
import java.io.IOException; | ||
import java.util.Comparator; | ||
import java.util.List; | ||
|
||
/** | ||
* A frequency terms enum that maintains a cache of docFreq, totalTermFreq, or both for repeated term lookup. It also | ||
* allows to provide a filter to explicitly compute frequencies only for docs that match the filter (heavier!). | ||
*/ | ||
public class FreqTermsEnum extends TermsEnum implements Releasable { | ||
|
||
static class Holder { | ||
final TermsEnum termsEnum; | ||
@Nullable | ||
DocsEnum docsEnum; | ||
@Nullable | ||
final Bits bits; | ||
|
||
Holder(TermsEnum termsEnum, Bits bits) { | ||
this.termsEnum = termsEnum; | ||
this.bits = bits; | ||
} | ||
} | ||
|
||
static final int INITIAL_NUM_TERM_FREQS_CACHED = 512; | ||
|
||
private final boolean docFreq; | ||
private final boolean totalTermFreq; | ||
private final Holder[] enums; | ||
|
||
private final BigArrays bigArrays; | ||
private IntArray termDocFreqs; | ||
private LongArray termsTotalFreqs; | ||
private BytesRefHash cachedTermOrds; | ||
|
||
private int currentDocFreq = 0; | ||
private long currentTotalTermFreq = 0; | ||
|
||
private BytesRef current; | ||
|
||
public FreqTermsEnum(IndexReader reader, String field, boolean docFreq, boolean totalTermFreq, @Nullable Filter filter, BigArrays bigArrays) throws IOException { | ||
this.docFreq = docFreq; | ||
this.totalTermFreq = totalTermFreq; | ||
if (!docFreq && !totalTermFreq) { | ||
throw new ElasticsearchIllegalArgumentException("either docFreq or totalTermFreq must be true"); | ||
} | ||
List<AtomicReaderContext> leaves = reader.leaves(); | ||
List<Holder> enums = Lists.newArrayListWithExpectedSize(leaves.size()); | ||
for (AtomicReaderContext context : leaves) { | ||
Terms terms = context.reader().terms(field); | ||
if (terms == null) { | ||
continue; | ||
} | ||
TermsEnum termsEnum = terms.iterator(null); | ||
if (termsEnum == null) { | ||
continue; | ||
} | ||
Bits bits = null; | ||
if (filter != null) { | ||
if (filter == Queries.MATCH_ALL_FILTER) { | ||
bits = context.reader().getLiveDocs(); | ||
} else { | ||
// we want to force apply deleted docs | ||
filter = new ApplyAcceptedDocsFilter(filter); | ||
DocIdSet docIdSet = filter.getDocIdSet(context, context.reader().getLiveDocs()); | ||
if (DocIdSets.isEmpty(docIdSet)) { | ||
// fully filtered, none matching, no need to iterate on this | ||
continue; | ||
} | ||
bits = DocIdSets.toSafeBits(context.reader(), docIdSet); | ||
} | ||
} | ||
enums.add(new Holder(termsEnum, bits)); | ||
} | ||
this.bigArrays = bigArrays; | ||
|
||
this.enums = enums.toArray(new Holder[enums.size()]); | ||
|
||
if (docFreq) { | ||
termDocFreqs = bigArrays.newIntArray(INITIAL_NUM_TERM_FREQS_CACHED, false); | ||
} else { | ||
termDocFreqs = null; | ||
} | ||
if (totalTermFreq) { | ||
termsTotalFreqs = bigArrays.newLongArray(INITIAL_NUM_TERM_FREQS_CACHED, false); | ||
} else { | ||
termsTotalFreqs = null; | ||
} | ||
cachedTermOrds = new BytesRefHash(INITIAL_NUM_TERM_FREQS_CACHED, bigArrays); | ||
} | ||
|
||
@Override | ||
public BytesRef term() throws IOException { | ||
return current; | ||
} | ||
|
||
@Override | ||
public boolean seekExact(BytesRef text) throws IOException { | ||
long currentTermOrd = cachedTermOrds.add(text); | ||
if (currentTermOrd < 0) { // already seen, initialize instance data with the cached frequencies | ||
currentTermOrd = -1 - currentTermOrd; | ||
boolean found = true; | ||
if (docFreq) { | ||
currentDocFreq = termDocFreqs.get(currentTermOrd); | ||
if (currentDocFreq == -2) { | ||
found = false; | ||
} | ||
} | ||
if (totalTermFreq) { | ||
currentTotalTermFreq = termsTotalFreqs.get(currentTermOrd); | ||
if (currentTotalTermFreq == -2) { | ||
found = false; | ||
} | ||
} | ||
current = found ? text : null; | ||
return found; | ||
} | ||
|
||
boolean found = false; | ||
int docFreq = 0; | ||
long totalTermFreq = 0; | ||
for (Holder anEnum : enums) { | ||
if (!anEnum.termsEnum.seekExact(text)) { | ||
continue; | ||
} | ||
found = true; | ||
if (anEnum.bits == null) { | ||
docFreq += anEnum.termsEnum.docFreq(); | ||
totalTermFreq += anEnum.termsEnum.totalTermFreq(); | ||
This comment has been minimized.
Sorry, something went wrong.
This comment has been minimized.
Sorry, something went wrong.
kimchy
Author
Owner
|
||
} else { | ||
DocsEnum docsEnum = anEnum.docsEnum = anEnum.termsEnum.docs(anEnum.bits, anEnum.docsEnum, this.totalTermFreq ? DocsEnum.FLAG_FREQS : DocsEnum.FLAG_NONE); | ||
for (int docId = docsEnum.nextDoc(); docId != DocIdSetIterator.NO_MORE_DOCS; docId = docsEnum.nextDoc()) { | ||
docFreq++; | ||
if (this.totalTermFreq) { | ||
totalTermFreq += docsEnum.freq(); | ||
} | ||
} | ||
} | ||
} | ||
|
||
current = found ? text : null; | ||
if (this.docFreq) { | ||
if (!found) { | ||
docFreq = -2; // -2 is used to indicate not found | ||
This comment has been minimized.
Sorry, something went wrong.
This comment has been minimized.
Sorry, something went wrong. |
||
} | ||
currentDocFreq = docFreq; | ||
termDocFreqs = bigArrays.grow(termDocFreqs, currentTermOrd + 1); | ||
termDocFreqs.set(currentTermOrd, docFreq); | ||
} | ||
if (this.totalTermFreq) { | ||
if (!found) { | ||
totalTermFreq = -2; // -2 is used to indicate not found | ||
} else if (totalTermFreq < 0) { | ||
// no freqs really..., blast | ||
totalTermFreq = -1; | ||
This comment has been minimized.
Sorry, something went wrong.
markharwood
|
||
} | ||
currentTotalTermFreq = totalTermFreq; | ||
termsTotalFreqs = bigArrays.grow(termsTotalFreqs, currentTermOrd + 1); | ||
termsTotalFreqs.set(currentTermOrd, totalTermFreq); | ||
} | ||
|
||
return found; | ||
} | ||
|
||
@Override | ||
public int docFreq() throws IOException { | ||
return currentDocFreq; | ||
} | ||
|
||
@Override | ||
public long totalTermFreq() throws IOException { | ||
return currentTotalTermFreq; | ||
} | ||
|
||
@Override | ||
public boolean release() throws ElasticsearchException { | ||
try { | ||
Releasables.release(cachedTermOrds, termDocFreqs, termsTotalFreqs); | ||
} finally { | ||
cachedTermOrds = null; | ||
termDocFreqs = null; | ||
termsTotalFreqs = null; | ||
} | ||
return true; | ||
} | ||
|
||
@Override | ||
public void seekExact(long ord) throws IOException { | ||
throw new UnsupportedOperationException("freq terms enum"); | ||
} | ||
|
||
@Override | ||
public SeekStatus seekCeil(BytesRef text) throws IOException { | ||
throw new UnsupportedOperationException("freq terms enum"); | ||
} | ||
|
||
@Override | ||
public long ord() throws IOException { | ||
throw new UnsupportedOperationException("freq terms enum"); | ||
} | ||
|
||
@Override | ||
public DocsEnum docs(Bits liveDocs, DocsEnum reuse, int flags) throws IOException { | ||
throw new UnsupportedOperationException("freq terms enum"); | ||
} | ||
|
||
@Override | ||
public DocsAndPositionsEnum docsAndPositions(Bits liveDocs, DocsAndPositionsEnum reuse, int flags) throws IOException { | ||
throw new UnsupportedOperationException("freq terms enum"); | ||
} | ||
|
||
@Override | ||
public BytesRef next() throws IOException { | ||
throw new UnsupportedOperationException("freq terms enum"); | ||
} | ||
|
||
@Override | ||
public Comparator<BytesRef> getComparator() { | ||
throw new UnsupportedOperationException("freq terms enum"); | ||
} | ||
} |
1 comment
on commit a45c616
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good but for the comments above.
The alternative strategy I considered for providing "background frequencies" was not disk-based but using a parent agg and FieldData to have a RAM-based cache of stats for any child aggs. Both that and the implementation here are probably desirable with each choosing a trade-off between execution speed and RAM usage.
Avoid the overhead of the enum.totalTermFreq call if the constructor passed false for caching these?