Examples with SortedSetDocValues org.apache.lucene.index.SortedSetDocValues used on opensource projects

Example 41 with SortedSetDocValues

use of org.apache.lucene.index.SortedSetDocValues in project lucene-solr by apache.

the class TestUninvertingReader method testSortedSetInteger.

public void testSortedSetInteger() throws IOException {
    Directory dir = newDirectory();
    IndexWriter iw = new IndexWriter(dir, newIndexWriterConfig(null));
    Document doc = new Document();
    doc.add(new LegacyIntField("foo", 5, Field.Store.NO));
    iw.addDocument(doc);
    doc = new Document();
    doc.add(new LegacyIntField("foo", 5, Field.Store.NO));
    doc.add(new LegacyIntField("foo", -3, Field.Store.NO));
    iw.addDocument(doc);
    iw.forceMerge(1);
    iw.close();
    DirectoryReader ir = UninvertingReader.wrap(DirectoryReader.open(dir), Collections.singletonMap("foo", Type.SORTED_SET_INTEGER));
    LeafReader ar = ir.leaves().get(0).reader();
    SortedSetDocValues v = ar.getSortedSetDocValues("foo");
    assertEquals(2, v.getValueCount());
    assertEquals(0, v.nextDoc());
    assertEquals(1, v.nextOrd());
    assertEquals(SortedSetDocValues.NO_MORE_ORDS, v.nextOrd());
    assertEquals(1, v.nextDoc());
    assertEquals(0, v.nextOrd());
    assertEquals(1, v.nextOrd());
    assertEquals(SortedSetDocValues.NO_MORE_ORDS, v.nextOrd());
    BytesRef value = v.lookupOrd(0);
    assertEquals(-3, LegacyNumericUtils.prefixCodedToInt(value));
    value = v.lookupOrd(1);
    assertEquals(5, LegacyNumericUtils.prefixCodedToInt(value));
    TestUtil.checkReader(ir);
    ir.close();
    dir.close();
}

Example 42 with SortedSetDocValues

use of org.apache.lucene.index.SortedSetDocValues in project lucene-solr by apache.

the class TestUninvertingReader method testSortedSetFloat.

public void testSortedSetFloat() throws IOException {
    Directory dir = newDirectory();
    IndexWriter iw = new IndexWriter(dir, newIndexWriterConfig(null));
    Document doc = new Document();
    doc.add(new LegacyIntField("foo", Float.floatToRawIntBits(5f), Field.Store.NO));
    iw.addDocument(doc);
    doc = new Document();
    doc.add(new LegacyIntField("foo", Float.floatToRawIntBits(5f), Field.Store.NO));
    doc.add(new LegacyIntField("foo", Float.floatToRawIntBits(-3f), Field.Store.NO));
    iw.addDocument(doc);
    iw.forceMerge(1);
    iw.close();
    DirectoryReader ir = UninvertingReader.wrap(DirectoryReader.open(dir), Collections.singletonMap("foo", Type.SORTED_SET_FLOAT));
    LeafReader ar = ir.leaves().get(0).reader();
    SortedSetDocValues v = ar.getSortedSetDocValues("foo");
    assertEquals(2, v.getValueCount());
    assertEquals(0, v.nextDoc());
    assertEquals(1, v.nextOrd());
    assertEquals(SortedSetDocValues.NO_MORE_ORDS, v.nextOrd());
    assertEquals(1, v.nextDoc());
    assertEquals(0, v.nextOrd());
    assertEquals(1, v.nextOrd());
    assertEquals(SortedSetDocValues.NO_MORE_ORDS, v.nextOrd());
    BytesRef value = v.lookupOrd(0);
    assertEquals(Float.floatToRawIntBits(-3f), LegacyNumericUtils.prefixCodedToInt(value));
    value = v.lookupOrd(1);
    assertEquals(Float.floatToRawIntBits(5f), LegacyNumericUtils.prefixCodedToInt(value));
    TestUtil.checkReader(ir);
    ir.close();
    dir.close();
}

Example 43 with SortedSetDocValues

use of org.apache.lucene.index.SortedSetDocValues in project lucene-solr by apache.

the class TestUninvertingReader method testSortedSetDouble.

public void testSortedSetDouble() throws IOException {
    Directory dir = newDirectory();
    IndexWriter iw = new IndexWriter(dir, newIndexWriterConfig(null));
    Document doc = new Document();
    doc.add(new LegacyLongField("foo", Double.doubleToRawLongBits(5d), Field.Store.NO));
    iw.addDocument(doc);
    doc = new Document();
    doc.add(new LegacyLongField("foo", Double.doubleToRawLongBits(5d), Field.Store.NO));
    doc.add(new LegacyLongField("foo", Double.doubleToRawLongBits(-3d), Field.Store.NO));
    iw.addDocument(doc);
    iw.forceMerge(1);
    iw.close();
    DirectoryReader ir = UninvertingReader.wrap(DirectoryReader.open(dir), Collections.singletonMap("foo", Type.SORTED_SET_DOUBLE));
    LeafReader ar = ir.leaves().get(0).reader();
    SortedSetDocValues v = ar.getSortedSetDocValues("foo");
    assertEquals(2, v.getValueCount());
    assertEquals(0, v.nextDoc());
    assertEquals(1, v.nextOrd());
    assertEquals(SortedSetDocValues.NO_MORE_ORDS, v.nextOrd());
    assertEquals(1, v.nextDoc());
    assertEquals(0, v.nextOrd());
    assertEquals(1, v.nextOrd());
    assertEquals(SortedSetDocValues.NO_MORE_ORDS, v.nextOrd());
    BytesRef value = v.lookupOrd(0);
    assertEquals(Double.doubleToRawLongBits(-3d), LegacyNumericUtils.prefixCodedToLong(value));
    value = v.lookupOrd(1);
    assertEquals(Double.doubleToRawLongBits(5d), LegacyNumericUtils.prefixCodedToLong(value));
    TestUtil.checkReader(ir);
    ir.close();
    dir.close();
}

Example 44 with SortedSetDocValues

use of org.apache.lucene.index.SortedSetDocValues in project lucene-solr by apache.

the class DefaultSortedSetDocValuesReaderState method getDocValues.

/** Return top-level doc values. */
@Override
public SortedSetDocValues getDocValues() throws IOException {
    // TODO: this is dup'd from slow composite reader wrapper ... can we factor it out to share?
    OrdinalMap map = null;
    // why are we using a map?
    synchronized (cachedOrdMaps) {
        map = cachedOrdMaps.get(field);
        if (map == null) {
            // uncached, or not a multi dv
            SortedSetDocValues dv = MultiDocValues.getSortedSetValues(reader, field);
            if (dv instanceof MultiDocValues.MultiSortedSetDocValues) {
                map = ((MultiDocValues.MultiSortedSetDocValues) dv).mapping;
                IndexReader.CacheHelper cacheHelper = reader.getReaderCacheHelper();
                if (cacheHelper != null && map.owner == cacheHelper.getKey()) {
                    cachedOrdMaps.put(field, map);
                }
            }
            return dv;
        }
    }
    assert map != null;
    int size = reader.leaves().size();
    final SortedSetDocValues[] values = new SortedSetDocValues[size];
    final int[] starts = new int[size + 1];
    long cost = 0;
    for (int i = 0; i < size; i++) {
        LeafReaderContext context = reader.leaves().get(i);
        final LeafReader reader = context.reader();
        final FieldInfo fieldInfo = reader.getFieldInfos().fieldInfo(field);
        if (fieldInfo != null && fieldInfo.getDocValuesType() != DocValuesType.SORTED_SET) {
            return null;
        }
        SortedSetDocValues v = reader.getSortedSetDocValues(field);
        if (v == null) {
            v = DocValues.emptySortedSet();
        }
        values[i] = v;
        starts[i] = context.docBase;
        cost += v.cost();
    }
    starts[size] = reader.maxDoc();
    return new MultiSortedSetDocValues(values, starts, map, cost);
}

Example 45 with SortedSetDocValues

use of org.apache.lucene.index.SortedSetDocValues in project lucene-solr by apache.

the class SortedSetDocValuesFacetCounts method countOneSegment.

private void countOneSegment(MultiDocValues.OrdinalMap ordinalMap, LeafReader reader, int segOrd, MatchingDocs hits) throws IOException {
    SortedSetDocValues segValues = reader.getSortedSetDocValues(field);
    if (segValues == null) {
        // nothing to count
        return;
    }
    DocIdSetIterator it;
    if (hits == null) {
        it = segValues;
    } else {
        it = ConjunctionDISI.intersectIterators(Arrays.asList(hits.bits.iterator(), segValues));
    }
    // segs)
    if (ordinalMap != null) {
        final LongValues ordMap = ordinalMap.getGlobalOrds(segOrd);
        int numSegOrds = (int) segValues.getValueCount();
        if (hits != null && hits.totalHits < numSegOrds / 10) {
            // Remap every ord to global ord as we iterate:
            for (int doc = it.nextDoc(); doc != DocIdSetIterator.NO_MORE_DOCS; doc = it.nextDoc()) {
                int term = (int) segValues.nextOrd();
                while (term != SortedSetDocValues.NO_MORE_ORDS) {
                    //System.out.println("      segOrd=" + segOrd + " ord=" + term + " globalOrd=" + ordinalMap.getGlobalOrd(segOrd, term));
                    counts[(int) ordMap.get(term)]++;
                    term = (int) segValues.nextOrd();
                }
            }
        } else {
            //System.out.println("    count in seg ord first");
            // First count in seg-ord space:
            final int[] segCounts = new int[numSegOrds];
            for (int doc = it.nextDoc(); doc != DocIdSetIterator.NO_MORE_DOCS; doc = it.nextDoc()) {
                int term = (int) segValues.nextOrd();
                while (term != SortedSetDocValues.NO_MORE_ORDS) {
                    //System.out.println("      ord=" + term);
                    segCounts[term]++;
                    term = (int) segValues.nextOrd();
                }
            }
            // Then, migrate to global ords:
            for (int ord = 0; ord < numSegOrds; ord++) {
                int count = segCounts[ord];
                if (count != 0) {
                    //System.out.println("    migrate segOrd=" + segOrd + " ord=" + ord + " globalOrd=" + ordinalMap.getGlobalOrd(segOrd, ord));
                    counts[(int) ordMap.get(ord)] += count;
                }
            }
        }
    } else {
        // just aggregate directly into counts:
        for (int doc = it.nextDoc(); doc != DocIdSetIterator.NO_MORE_DOCS; doc = it.nextDoc()) {
            int term = (int) segValues.nextOrd();
            while (term != SortedSetDocValues.NO_MORE_ORDS) {
                counts[term]++;
                term = (int) segValues.nextOrd();
            }
        }
    }
}