Examples with Intersection com.greplin.lucene.util.Intersection used on opensource projects

Example 1 with Intersection

use of com.greplin.lucene.util.Intersection in project greplin-lucene-utils by Cue.

the class PhraseFilter method getDocIdSet.

@Override
public DocIdSet getDocIdSet(final IndexReader reader) throws IOException {
    List<IndexReader> subReaders = IndexReaders.gatherSubReaders(reader);
    PhraseFilterMatchList[] results = new PhraseFilterMatchList[subReaders.size()];
    int matchCount = 0;
    int readerNumber = 0;
    for (IndexReader subReader : subReaders) {
        SortedSet<TermWithFrequency> termsOrderedByFrequency = Sets.newTreeSet();
        for (int i = 0; i < this.terms.length; i++) {
            Term t = this.terms[i];
            termsOrderedByFrequency.add(new TermWithFrequency(t, subReader.docFreq(t), i));
        }
        PhraseFilterMatchList matches = null;
        TermPositions termPositions = subReader.termPositions();
        try {
            for (TermWithFrequency term : termsOrderedByFrequency) {
                if (term.docFreq == 0) {
                    break;
                }
                termPositions.seek(term.term);
                if (matches == null) {
                    // If this is the first term, collect all matches that intersect
                    // with the provided initial document set.
                    Intersection intersection = this.intersectionProvider.get(reader);
                    matches = new PhraseFilterMatchList(term.docFreq);
                    while (intersection.advanceToNextIntersection(termPositions)) {
                        int freq = termPositions.freq();
                        PhraseFilterIntList list = new PhraseFilterIntList(freq);
                        for (int i = 0; i < freq; i++) {
                            list.add(termPositions.nextPosition() - term.offset);
                        }
                        matches.add(termPositions.doc(), list);
                    }
                } else {
                    // Otherwise, intersect with the existing matches.
                    matches.intersect(termPositions, term.offset);
                }
                if (matches.getCount() == 0) {
                    break;
                }
            }
        } finally {
            termPositions.close();
        }
        if (matches != null) {
            results[readerNumber] = matches;
            matchCount += matches.getCount();
        }
        readerNumber++;
    }
    // 2^5 = 32
    final int bitsPerIntPowerLogTwo = 5;
    if (matchCount > reader.maxDoc() >> bitsPerIntPowerLogTwo) {
        FixedBitSet result = new FixedBitSet(reader.maxDoc());
        int readerOffset = 0;
        for (int readerIndex = 0; readerIndex < results.length; readerIndex++) {
            PhraseFilterMatchList matches = results[readerIndex];
            if (matches != null) {
                int count = matches.getCount();
                int[] docIds = matches.getDocIds();
                for (int i = 0; i < count; i++) {
                    result.set(docIds[i] + readerOffset);
                }
            }
            readerOffset += subReaders.get(readerIndex).maxDoc();
        }
        return result;
    } else if (matchCount == 0) {
        return DocIdSets.EMPTY;
    } else {
        int[] result = new int[matchCount];
        int base = 0;
        int readerOffset = 0;
        for (int readerIndex = 0; readerIndex < results.length; readerIndex++) {
            PhraseFilterMatchList matches = results[readerIndex];
            if (matches != null) {
                int count = matches.getCount();
                int[] docIds = matches.getDocIds();
                for (int i = 0; i < count; i++) {
                    result[base + i] = docIds[i] + readerOffset;
                }
                base += count;
            }
            readerOffset += subReaders.get(readerIndex).maxDoc();
        }
        return new SortedIntArrayDocIdSet(result);
    }
}