Examples with FixedBitSet org.apache.lucene.util.FixedBitSet used on opensource projects

Example 71 with FixedBitSet

use of org.apache.lucene.util.FixedBitSet in project lucene-solr by apache.

the class TestConjunctionDISI method testCollapseSubConjunctions.

public void testCollapseSubConjunctions(boolean wrapWithScorer) throws IOException {
    final int iters = atLeast(100);
    for (int iter = 0; iter < iters; ++iter) {
        final int maxDoc = TestUtil.nextInt(random(), 100, 10000);
        final int numIterators = TestUtil.nextInt(random(), 5, 10);
        final FixedBitSet[] sets = new FixedBitSet[numIterators];
        final List<Scorer> scorers = new LinkedList<>();
        for (int i = 0; i < numIterators; ++i) {
            final FixedBitSet set = randomSet(maxDoc);
            if (random().nextBoolean()) {
                // simple iterator
                sets[i] = set;
                scorers.add(new ConstantScoreScorer(null, 0f, new BitDocIdSet(set).iterator()));
            } else {
                // scorer with approximation
                final FixedBitSet confirmed = clearRandomBits(set);
                sets[i] = confirmed;
                final TwoPhaseIterator approximation = approximation(new BitDocIdSet(set).iterator(), confirmed);
                scorers.add(scorer(approximation));
            }
        }
        // make some sub sequences into sub conjunctions
        final int subIters = atLeast(3);
        for (int subIter = 0; subIter < subIters && scorers.size() > 3; ++subIter) {
            final int subSeqStart = TestUtil.nextInt(random(), 0, scorers.size() - 2);
            final int subSeqEnd = TestUtil.nextInt(random(), subSeqStart + 2, scorers.size());
            List<Scorer> subIterators = scorers.subList(subSeqStart, subSeqEnd);
            Scorer subConjunction;
            if (wrapWithScorer) {
                subConjunction = new ConjunctionScorer(null, subIterators, Collections.emptyList());
            } else {
                subConjunction = new ConstantScoreScorer(null, 0f, ConjunctionDISI.intersectScorers(subIterators));
            }
            scorers.set(subSeqStart, subConjunction);
            int toRemove = subSeqEnd - subSeqStart - 1;
            while (toRemove-- > 0) {
                scorers.remove(subSeqStart + 1);
            }
        }
        if (scorers.size() == 1) {
            // ConjunctionDISI needs two iterators
            scorers.add(new ConstantScoreScorer(null, 0f, DocIdSetIterator.all(maxDoc)));
        }
        final DocIdSetIterator conjunction = ConjunctionDISI.intersectScorers(scorers);
        assertEquals(intersect(sets), toBitSet(maxDoc, conjunction));
    }
}

Example 72 with FixedBitSet

use of org.apache.lucene.util.FixedBitSet in project lucene-solr by apache.

the class TestConjunctionDISI method intersect.

private static FixedBitSet intersect(FixedBitSet[] bitSets) {
    final FixedBitSet intersection = new FixedBitSet(bitSets[0].length());
    intersection.or(bitSets[0]);
    for (int i = 1; i < bitSets.length; ++i) {
        intersection.and(bitSets[i]);
    }
    return intersection;
}

Example 73 with FixedBitSet

use of org.apache.lucene.util.FixedBitSet in project Krill by KorAP.

the class Match method getSnippetHTML.

@JsonProperty("snippet")
public String getSnippetHTML() {
    if (!this._processHighlight())
        return null;
    if (this.processed && this.snippetHTML != null)
        return this.snippetHTML;
    if (DEBUG)
        log.trace("Create HTML Snippet");
    StringBuilder sb = new StringBuilder();
    StringBuilder rightContext = new StringBuilder();
    // Remember ids already defined to
    // have joined elements
    HashSet<String> joins = new HashSet<>(100);
    // Snippet stack sizes
    short start = (short) 0;
    short end = this.snippetArray.size();
    // Create context
    sb.append("<span class=\"context-left\">");
    if (this.startMore)
        sb.append("<span class=\"more\"></span>");
    // Set levels for highlights
    FixedBitSet level = new FixedBitSet(255);
    level.set(0, 255);
    byte[] levelCache = new byte[255];
    HighlightCombinatorElement elem;
    end--;
    if (end > 0) {
        // First element of sorted array
        elem = this.snippetArray.getFirst();
        // First element is textual
        if (elem.type == 0) {
            sb.append(elem.toHTML(this, level, levelCache, joins));
            // Move start position
            start++;
        }
        ;
        sb.append("</span>");
        // Last element of sorted array
        elem = this.snippetArray.getLast();
        // Create right context, if there is any
        rightContext.append("<span class=\"context-right\">");
        // Last element is textual
        if (elem != null && elem.type == 0) {
            rightContext.append(elem.toHTML(this, level, levelCache, joins));
            // decrement end
            end--;
        }
        ;
    }
    ;
    if (this.endMore)
        rightContext.append("<span class=\"more\"></span>");
    rightContext.append("</span>");
    // Iterate through all remaining elements
    sb.append("<span class=\"match\">");
    for (short i = start; i <= end; i++) {
        elem = this.snippetArray.get(i);
        // UNTESTED
        if (elem != null) {
            String elemString = elem.toHTML(this, level, levelCache, joins);
            if (DEBUG) {
                log.trace("Add node {}", elemString);
            }
            ;
            sb.append(elemString);
        }
    }
    ;
    sb.append("</span>");
    sb.append(rightContext);
    return (this.snippetHTML = sb.toString());
}

Example 74 with FixedBitSet

use of org.apache.lucene.util.FixedBitSet in project Krill by KorAP.

the class KrillCollection method bits.

/**
 * Create a bit vector representing the live documents of the
 * virtual collection to be used in searches.
 * This will respect deleted documents.
 *
 * @param The
 *            {@link LeafReaderContext} to search in.
 * @return A bit vector representing the live documents of the
 *         virtual collection.
 * @throws IOException
 */
public FixedBitSet bits(LeafReaderContext atomic) throws IOException {
    LeafReader r = atomic.reader();
    FixedBitSet bitset = new FixedBitSet(r.maxDoc());
    DocIdSet docids = this.getDocIdSet(atomic, (Bits) r.getLiveDocs());
    if (docids == null) {
        if (this.cbi != null) {
            bitset.clear(0, bitset.length());
        } else {
            bitset.set(0, bitset.length());
        }
        ;
    } else
        bitset.or(docids.iterator());
    return bitset;
}

Example 75 with FixedBitSet

use of org.apache.lucene.util.FixedBitSet in project SearchServices by Alfresco.

the class BitsFilter method and.

public void and(BitsFilter bitsFilter) {
    List<FixedBitSet> andSets = bitsFilter.bitSets;
    for (int i = 0; i < bitSets.size(); i++) {
        FixedBitSet a = bitSets.get(i);
        FixedBitSet b = andSets.get(i);
        a.and(b);
    }
}