private void saveAppropriatelySizedBloomFilter(
IndexOutput bloomOutput, FuzzySet bloomFilter, FieldInfo fieldInfo) throws IOException {
FuzzySet rightSizedSet = bloomFilterFactory.downsize(fieldInfo, bloomFilter);
if (rightSizedSet == null) {
rightSizedSet = bloomFilter;
}
rightSizedSet.serialize(bloomOutput);
}
public BloomFilteredFieldsProducer(SegmentReadState state) throws IOException {
String bloomFileName =
IndexFileNames.segmentFileName(
state.segmentInfo.name, state.segmentSuffix, BLOOM_EXTENSION);
IndexInput bloomIn = null;
boolean success = false;
try {
bloomIn = state.directory.openInput(bloomFileName, state.context);
CodecUtil.checkHeader(bloomIn, BLOOM_CODEC_NAME, BLOOM_CODEC_VERSION, BLOOM_CODEC_VERSION);
// // Load the hash function used in the BloomFilter
// hashFunction = HashFunction.forName(bloomIn.readString());
// Load the delegate postings format
PostingsFormat delegatePostingsFormat = PostingsFormat.forName(bloomIn.readString());
this.delegateFieldsProducer = delegatePostingsFormat.fieldsProducer(state);
int numBlooms = bloomIn.readInt();
for (int i = 0; i < numBlooms; i++) {
int fieldNum = bloomIn.readInt();
FuzzySet bloom = FuzzySet.deserialize(bloomIn);
FieldInfo fieldInfo = state.fieldInfos.fieldInfo(fieldNum);
bloomsByFieldName.put(fieldInfo.name, bloom);
}
IOUtils.close(bloomIn);
success = true;
} finally {
if (!success) {
IOUtils.closeWhileHandlingException(bloomIn, delegateFieldsProducer);
}
}
}
@Override
public void finishTerm(BytesRef text, TermStats stats) throws IOException {
// Record this term in our BloomFilter
if (stats.docFreq > 0) {
bloomFilter.addValue(text);
}
delegateTermsConsumer.finishTerm(text, stats);
}
@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.contains(text) == ContainsResult.NO) {
return false;
}
return delegate().seekExact(text, useCache);
}