honeymoose
/
OpenSearch
mirror of https://github.com/honeymoose/OpenSearch.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Lazily initialize the delegate in BloomFilteredPostingsFormat to prevent unnecessary loading if bloomfilter terminates early

This commit is contained in:
Simon Willnauer 2013-03-20 12:05:36 +01:00
parent 747ce36915
commit 4705eb2959
3 changed files with 183 additions and 175 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
src/main/java/org/elasticsearch/common/lucene/uid/UidField.java
View File

@ -62,13 +62,7 @@ public class UidField extends Field {
if (terms == null) {
return null;
}
// hack to break early if we have a bloom filter...
if (terms instanceof BloomFilterPostingsFormat.BloomFilteredFieldsProducer.BloomFilteredTerms) {
if (!((BloomFilterPostingsFormat.BloomFilteredFieldsProducer.BloomFilteredTerms) terms).getFilter().mightContain(term.bytes())) {
return null;
}
}
TermsEnum termsEnum = terms.iterator(null);
final TermsEnum termsEnum = terms.iterator(null);
if (termsEnum == null) {
return null;
}
@ -110,13 +104,7 @@ public class UidField extends Field {
if (terms == null) {
return -1;
}
// hack to break early if we have a bloom filter...
if (terms instanceof BloomFilterPostingsFormat.BloomFilteredFieldsProducer.BloomFilteredTerms) {
if (!((BloomFilterPostingsFormat.BloomFilteredFieldsProducer.BloomFilteredTerms) terms).getFilter().mightContain(term.bytes())) {
return -1;
}
}
TermsEnum termsEnum = terms.iterator(null);
final TermsEnum termsEnum = terms.iterator(null);
if (termsEnum == null) {
return -1;
}

334
src/main/java/org/elasticsearch/index/codec/postingsformat/BloomFilterPostingsFormat.java
View File

@ -180,167 +180,190 @@ public final class BloomFilterPostingsFormat extends PostingsFormat {
return delegateFieldsProducer.getUniqueTermCount();
}
public class BloomFilteredTerms extends Terms {
private Terms delegateTerms;
private BloomFilter filter;
}
public static final class BloomFilteredTerms extends Terms {
private Terms delegateTerms;
private BloomFilter filter;
public BloomFilteredTerms(Terms terms, BloomFilter filter) {
this.delegateTerms = terms;
this.filter = filter;
}
public BloomFilter getFilter() {
return filter;
}
@Override
public TermsEnum intersect(CompiledAutomaton compiled,
final BytesRef startTerm) throws IOException {
return delegateTerms.intersect(compiled, startTerm);
}
@Override
public TermsEnum iterator(TermsEnum reuse) throws IOException {
TermsEnum result;
if ((reuse != null) && (reuse instanceof BloomFilteredTermsEnum)) {
// recycle the existing BloomFilteredTermsEnum by asking the delegate
// to recycle its contained TermsEnum
BloomFilteredTermsEnum bfte = (BloomFilteredTermsEnum) reuse;
if (bfte.filter == filter) {
bfte.delegateTermsEnum = delegateTerms.iterator(bfte.delegateTermsEnum);
return bfte;
}
}
// We have been handed something we cannot reuse (either null, wrong
// class or wrong filter) so allocate a new object
result = new BloomFilteredTermsEnum(delegateTerms.iterator(reuse), filter);
return result;
}
@Override
public Comparator<BytesRef> getComparator() {
return delegateTerms.getComparator();
}
@Override
public long size() throws IOException {
return delegateTerms.size();
}
@Override
public long getSumTotalTermFreq() throws IOException {
return delegateTerms.getSumTotalTermFreq();
}
@Override
public long getSumDocFreq() throws IOException {
return delegateTerms.getSumDocFreq();
}
@Override
public int getDocCount() throws IOException {
return delegateTerms.getDocCount();
}
@Override
public boolean hasOffsets() {
return delegateTerms.hasOffsets();
}
@Override
public boolean hasPositions() {
return delegateTerms.hasPositions();
}
@Override
public boolean hasPayloads() {
return delegateTerms.hasPayloads();
}
public BloomFilteredTerms(Terms terms, BloomFilter filter) {
this.delegateTerms = terms;
this.filter = filter;
}
class BloomFilteredTermsEnum extends TermsEnum {
TermsEnum delegateTermsEnum;
private BloomFilter filter;
public BloomFilteredTermsEnum(TermsEnum iterator, BloomFilter filter) {
this.delegateTermsEnum = iterator;
this.filter = filter;
}
@Override
public final BytesRef next() throws IOException {
return delegateTermsEnum.next();
}
@Override
public final Comparator<BytesRef> getComparator() {
return delegateTermsEnum.getComparator();
}
@Override
public final boolean seekExact(BytesRef text, boolean useCache)
throws IOException {
// The magical fail-fast speed up that is the entire point of all of
// this code - save a disk seek if there is a match on an in-memory
// structure
// that may occasionally give a false positive but guaranteed no false
// negatives
if (!filter.mightContain(text)) {
return false;
}
return delegateTermsEnum.seekExact(text, useCache);
}
@Override
public final SeekStatus seekCeil(BytesRef text, boolean useCache)
throws IOException {
return delegateTermsEnum.seekCeil(text, useCache);
}
@Override
public final void seekExact(long ord) throws IOException {
delegateTermsEnum.seekExact(ord);
}
@Override
public final BytesRef term() throws IOException {
return delegateTermsEnum.term();
}
@Override
public final long ord() throws IOException {
return delegateTermsEnum.ord();
}
@Override
public final int docFreq() throws IOException {
return delegateTermsEnum.docFreq();
}
@Override
public final long totalTermFreq() throws IOException {
return delegateTermsEnum.totalTermFreq();
}
@Override
public DocsAndPositionsEnum docsAndPositions(Bits liveDocs,
DocsAndPositionsEnum reuse, int flags) throws IOException {
return delegateTermsEnum.docsAndPositions(liveDocs, reuse, flags);
}
@Override
public DocsEnum docs(Bits liveDocs, DocsEnum reuse, int flags)
throws IOException {
return delegateTermsEnum.docs(liveDocs, reuse, flags);
}
public BloomFilter getFilter() {
return filter;
}
@Override
public TermsEnum intersect(CompiledAutomaton compiled,
final BytesRef startTerm) throws IOException {
return delegateTerms.intersect(compiled, startTerm);
}
@Override
public TermsEnum iterator(TermsEnum reuse) throws IOException {
TermsEnum result;
if ((reuse != null) && (reuse instanceof BloomFilteredTermsEnum)) {
// recycle the existing BloomFilteredTermsEnum by asking the delegate
// to recycle its contained TermsEnum
BloomFilteredTermsEnum bfte = (BloomFilteredTermsEnum) reuse;
if (bfte.filter == filter) {
bfte.reset(delegateTerms);
return bfte;
}
reuse = bfte.reuse;
}
// We have been handed something we cannot reuse (either null, wrong
// class or wrong filter) so allocate a new object
result = new BloomFilteredTermsEnum(delegateTerms, reuse, filter);
return result;
}
@Override
public Comparator<BytesRef> getComparator() {
return delegateTerms.getComparator();
}
@Override
public long size() throws IOException {
return delegateTerms.size();
}
@Override
public long getSumTotalTermFreq() throws IOException {
return delegateTerms.getSumTotalTermFreq();
}
@Override
public long getSumDocFreq() throws IOException {
return delegateTerms.getSumDocFreq();
}
@Override
public int getDocCount() throws IOException {
return delegateTerms.getDocCount();
}
@Override
public boolean hasOffsets() {
return delegateTerms.hasOffsets();
}
@Override
public boolean hasPositions() {
return delegateTerms.hasPositions();
}
@Override
public boolean hasPayloads() {
return delegateTerms.hasPayloads();
}
}
static final class BloomFilteredTermsEnum extends TermsEnum {
private Terms delegateTerms;
private TermsEnum delegateTermsEnum;
private TermsEnum reuse;
private BloomFilter filter;
public BloomFilteredTermsEnum(Terms other, TermsEnum reuse, BloomFilter filter) {
this.delegateTerms = other;
this.reuse = reuse;
this.filter = filter;
}
void reset(Terms others) {
reuse = this.delegateTermsEnum;
this.delegateTermsEnum = null;
this.delegateTerms = others;
}
private TermsEnum getDelegate() throws IOException {
if (delegateTermsEnum == null) {
/* pull the iterator only if we really need it -
* this can be a relatively heavy operation depending on the
* delegate postings format and they underlying directory
* (clone IndexInput) */
delegateTermsEnum = delegateTerms.iterator(reuse);
}
return delegateTermsEnum;
}
@Override
public final BytesRef next() throws IOException {
return getDelegate().next();
}
@Override
public final Comparator<BytesRef> getComparator() {
return delegateTerms.getComparator();
}
@Override
public final boolean seekExact(BytesRef text, boolean useCache)
throws IOException {
// The magical fail-fast speed up that is the entire point of all of
// this code - save a disk seek if there is a match on an in-memory
// structure
// that may occasionally give a false positive but guaranteed no false
// negatives
if (!filter.mightContain(text)) {
return false;
}
return getDelegate().seekExact(text, useCache);
}
@Override
public final SeekStatus seekCeil(BytesRef text, boolean useCache)
throws IOException {
return getDelegate().seekCeil(text, useCache);
}
@Override
public final void seekExact(long ord) throws IOException {
getDelegate().seekExact(ord);
}
@Override
public final BytesRef term() throws IOException {
return getDelegate().term();
}
@Override
public final long ord() throws IOException {
return getDelegate().ord();
}
@Override
public final int docFreq() throws IOException {
return getDelegate().docFreq();
}
@Override
public final long totalTermFreq() throws IOException {
return getDelegate().totalTermFreq();
}
@Override
public DocsAndPositionsEnum docsAndPositions(Bits liveDocs,
DocsAndPositionsEnum reuse, int flags) throws IOException {
return getDelegate().docsAndPositions(liveDocs, reuse, flags);
}
@Override
public DocsEnum docs(Bits liveDocs, DocsEnum reuse, int flags)
throws IOException {
return getDelegate().docs(liveDocs, reuse, flags);
}
}
final class BloomFilteredFieldsConsumer extends FieldsConsumer {
private FieldsConsumer delegateFieldsConsumer;
private Map<FieldInfo, BloomFilter> bloomFilters = new HashMap<FieldInfo, BloomFilter>();
@ -381,10 +404,7 @@ public final class BloomFilterPostingsFormat extends PostingsFormat {
List<Entry<FieldInfo, BloomFilter>> nonSaturatedBlooms = new ArrayList<Map.Entry<FieldInfo, BloomFilter>>();
for (Entry<FieldInfo, BloomFilter> entry : bloomFilters.entrySet()) {
BloomFilter bloomFilter = entry.getValue();
//if (!bloomFilterFactory.isSaturated(bloomFilter, entry.getKey())) {
nonSaturatedBlooms.add(entry);
//}
}
String bloomFileName = IndexFileNames.segmentFileName(
state.segmentInfo.name, state.segmentSuffix, BLOOM_EXTENSION);

8
src/test/java/org/elasticsearch/test/unit/index/codec/postingformat/DefaultPostingsFormatTests.java
View File

@ -82,8 +82,8 @@ public class DefaultPostingsFormatTests {
Terms uidTerms = ar.terms(UidFieldMapper.NAME);
assertThat(terms.size(), equalTo(1l));
assertThat(terms, not(instanceOf(BloomFilterPostingsFormat.BloomFilteredFieldsProducer.BloomFilteredTerms.class)));
assertThat(uidTerms, instanceOf(BloomFilterPostingsFormat.BloomFilteredFieldsProducer.BloomFilteredTerms.class));
assertThat(terms, not(instanceOf(BloomFilterPostingsFormat.BloomFilteredTerms.class)));
assertThat(uidTerms, instanceOf(BloomFilterPostingsFormat.BloomFilteredTerms.class));
reader.close();
writer.close();
@ -112,8 +112,8 @@ public class DefaultPostingsFormatTests {
Terms some_other_field = ar.terms("some_other_field");
assertThat(terms.size(), equalTo(2l));
assertThat(terms, not(instanceOf(BloomFilterPostingsFormat.BloomFilteredFieldsProducer.BloomFilteredTerms.class)));
assertThat(some_other_field, not(instanceOf(BloomFilterPostingsFormat.BloomFilteredFieldsProducer.BloomFilteredTerms.class)));
assertThat(terms, not(instanceOf(BloomFilterPostingsFormat.BloomFilteredTerms.class)));
assertThat(some_other_field, not(instanceOf(BloomFilterPostingsFormat.BloomFilteredTerms.class)));
TermsEnum iterator = terms.iterator(null);
Set<String> expected = new HashSet<String>();
expected.add("foo");