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

Example 71 with LeafReader

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

the class RangeFieldQuery method createWeight.

@Override
public final Weight createWeight(IndexSearcher searcher, boolean needsScores, float boost) throws IOException {
    return new ConstantScoreWeight(this, boost) {

        final RangeFieldComparator target = new RangeFieldComparator();

        private DocIdSet buildMatchingDocIdSet(LeafReader reader, PointValues values) throws IOException {
            DocIdSetBuilder result = new DocIdSetBuilder(reader.maxDoc(), values, field);
            values.intersect(new IntersectVisitor() {

                DocIdSetBuilder.BulkAdder adder;

                @Override
                public void grow(int count) {
                    adder = result.grow(count);
                }

                @Override
                public void visit(int docID) throws IOException {
                    adder.add(docID);
                }

                @Override
                public void visit(int docID, byte[] leaf) throws IOException {
                    if (target.matches(leaf)) {
                        adder.add(docID);
                    }
                }

                @Override
                public Relation compare(byte[] minPackedValue, byte[] maxPackedValue) {
                    return compareRange(minPackedValue, maxPackedValue);
                }
            });
            return result.build();
        }

        private Relation compareRange(byte[] minPackedValue, byte[] maxPackedValue) {
            byte[] node = getInternalRange(minPackedValue, maxPackedValue);
            // compute range relation for BKD traversal
            if (target.intersects(node) == false) {
                return Relation.CELL_OUTSIDE_QUERY;
            } else if (target.within(node)) {
                // target within cell; continue traversing:
                return Relation.CELL_CROSSES_QUERY;
            } else if (target.contains(node)) {
                // target contains cell; add iff queryType is not a CONTAINS or CROSSES query:
                return (queryType == QueryType.CONTAINS || queryType == QueryType.CROSSES) ? Relation.CELL_OUTSIDE_QUERY : Relation.CELL_INSIDE_QUERY;
            }
            // target intersects cell; continue traversing:
            return Relation.CELL_CROSSES_QUERY;
        }

        @Override
        public Scorer scorer(LeafReaderContext context) throws IOException {
            LeafReader reader = context.reader();
            PointValues values = reader.getPointValues(field);
            if (values == null) {
                // no docs in this segment indexed any ranges
                return null;
            }
            FieldInfo fieldInfo = reader.getFieldInfos().fieldInfo(field);
            if (fieldInfo == null) {
                // no docs in this segment indexed this field
                return null;
            }
            checkFieldInfo(fieldInfo);
            boolean allDocsMatch = false;
            if (values.getDocCount() == reader.maxDoc() && compareRange(values.getMinPackedValue(), values.getMaxPackedValue()) == Relation.CELL_INSIDE_QUERY) {
                allDocsMatch = true;
            }
            DocIdSetIterator iterator = allDocsMatch == true ? DocIdSetIterator.all(reader.maxDoc()) : buildMatchingDocIdSet(reader, values).iterator();
            return new ConstantScoreScorer(this, score(), iterator);
        }

        /** get an encoded byte representation of the internal node; this is
       *  the lower half of the min array and the upper half of the max array */
        private byte[] getInternalRange(byte[] min, byte[] max) {
            byte[] range = new byte[min.length];
            final int dimSize = numDims * bytesPerDim;
            System.arraycopy(min, 0, range, 0, dimSize);
            System.arraycopy(max, dimSize, range, dimSize, dimSize);
            return range;
        }
    };
}

Example 72 with LeafReader

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

the class TestLucene70DocValuesFormat method testSortedSetAroundBlockSize.

@Slow
public void testSortedSetAroundBlockSize() throws IOException {
    final int frontier = 1 << Lucene70DocValuesFormat.DIRECT_MONOTONIC_BLOCK_SHIFT;
    for (int maxDoc = frontier - 1; maxDoc <= frontier + 1; ++maxDoc) {
        final Directory dir = newDirectory();
        IndexWriter w = new IndexWriter(dir, newIndexWriterConfig().setMergePolicy(newLogMergePolicy()));
        RAMFile buffer = new RAMFile();
        RAMOutputStream out = new RAMOutputStream(buffer, false);
        Document doc = new Document();
        SortedSetDocValuesField field1 = new SortedSetDocValuesField("sset", new BytesRef());
        doc.add(field1);
        SortedSetDocValuesField field2 = new SortedSetDocValuesField("sset", new BytesRef());
        doc.add(field2);
        for (int i = 0; i < maxDoc; ++i) {
            BytesRef s1 = new BytesRef(TestUtil.randomSimpleString(random(), 2));
            BytesRef s2 = new BytesRef(TestUtil.randomSimpleString(random(), 2));
            field1.setBytesValue(s1);
            field2.setBytesValue(s2);
            w.addDocument(doc);
            Set<BytesRef> set = new TreeSet<>(Arrays.asList(s1, s2));
            out.writeVInt(set.size());
            for (BytesRef ref : set) {
                out.writeVInt(ref.length);
                out.writeBytes(ref.bytes, ref.offset, ref.length);
            }
        }
        out.close();
        w.forceMerge(1);
        DirectoryReader r = DirectoryReader.open(w);
        w.close();
        LeafReader sr = getOnlyLeafReader(r);
        assertEquals(maxDoc, sr.maxDoc());
        SortedSetDocValues values = sr.getSortedSetDocValues("sset");
        assertNotNull(values);
        RAMInputStream in = new RAMInputStream("", buffer);
        BytesRefBuilder b = new BytesRefBuilder();
        for (int i = 0; i < maxDoc; ++i) {
            assertEquals(i, values.nextDoc());
            final int numValues = in.readVInt();
            for (int j = 0; j < numValues; ++j) {
                b.setLength(in.readVInt());
                b.grow(b.length());
                in.readBytes(b.bytes(), 0, b.length());
                assertEquals(b.get(), values.lookupOrd(values.nextOrd()));
            }
            assertEquals(SortedSetDocValues.NO_MORE_ORDS, values.nextOrd());
        }
        r.close();
        dir.close();
    }
}

Example 73 with LeafReader

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

the class TestLucene70DocValuesFormat method doTestSparseDocValuesVsStoredFields.

private void doTestSparseDocValuesVsStoredFields() throws Exception {
    final long[] values = new long[TestUtil.nextInt(random(), 1, 500)];
    for (int i = 0; i < values.length; ++i) {
        values[i] = random().nextLong();
    }
    Directory dir = newFSDirectory(createTempDir());
    IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random()));
    conf.setMergeScheduler(new SerialMergeScheduler());
    RandomIndexWriter writer = new RandomIndexWriter(random(), dir, conf);
    // sparse compression is only enabled if less than 1% of docs have a value
    final int avgGap = 100;
    final int numDocs = atLeast(200);
    for (int i = random().nextInt(avgGap * 2); i >= 0; --i) {
        writer.addDocument(new Document());
    }
    final int maxNumValuesPerDoc = random().nextBoolean() ? 1 : TestUtil.nextInt(random(), 2, 5);
    for (int i = 0; i < numDocs; ++i) {
        Document doc = new Document();
        // single-valued
        long docValue = values[random().nextInt(values.length)];
        doc.add(new NumericDocValuesField("numeric", docValue));
        doc.add(new SortedDocValuesField("sorted", new BytesRef(Long.toString(docValue))));
        doc.add(new BinaryDocValuesField("binary", new BytesRef(Long.toString(docValue))));
        doc.add(new StoredField("value", docValue));
        // multi-valued
        final int numValues = TestUtil.nextInt(random(), 1, maxNumValuesPerDoc);
        for (int j = 0; j < numValues; ++j) {
            docValue = values[random().nextInt(values.length)];
            doc.add(new SortedNumericDocValuesField("sorted_numeric", docValue));
            doc.add(new SortedSetDocValuesField("sorted_set", new BytesRef(Long.toString(docValue))));
            doc.add(new StoredField("values", docValue));
        }
        writer.addDocument(doc);
        // add a gap
        for (int j = TestUtil.nextInt(random(), 0, avgGap * 2); j >= 0; --j) {
            writer.addDocument(new Document());
        }
    }
    if (random().nextBoolean()) {
        writer.forceMerge(1);
    }
    final IndexReader indexReader = writer.getReader();
    writer.close();
    for (LeafReaderContext context : indexReader.leaves()) {
        final LeafReader reader = context.reader();
        final NumericDocValues numeric = DocValues.getNumeric(reader, "numeric");
        final SortedDocValues sorted = DocValues.getSorted(reader, "sorted");
        final BinaryDocValues binary = DocValues.getBinary(reader, "binary");
        final SortedNumericDocValues sortedNumeric = DocValues.getSortedNumeric(reader, "sorted_numeric");
        final SortedSetDocValues sortedSet = DocValues.getSortedSet(reader, "sorted_set");
        for (int i = 0; i < reader.maxDoc(); ++i) {
            final Document doc = reader.document(i);
            final IndexableField valueField = doc.getField("value");
            final Long value = valueField == null ? null : valueField.numericValue().longValue();
            if (value == null) {
                assertTrue(numeric.docID() + " vs " + i, numeric.docID() < i);
            } else {
                assertEquals(i, numeric.nextDoc());
                assertEquals(i, binary.nextDoc());
                assertEquals(i, sorted.nextDoc());
                assertEquals(value.longValue(), numeric.longValue());
                assertTrue(sorted.ordValue() >= 0);
                assertEquals(new BytesRef(Long.toString(value)), sorted.lookupOrd(sorted.ordValue()));
                assertEquals(new BytesRef(Long.toString(value)), binary.binaryValue());
            }
            final IndexableField[] valuesFields = doc.getFields("values");
            if (valuesFields.length == 0) {
                assertTrue(sortedNumeric.docID() + " vs " + i, sortedNumeric.docID() < i);
            } else {
                final Set<Long> valueSet = new HashSet<>();
                for (IndexableField sf : valuesFields) {
                    valueSet.add(sf.numericValue().longValue());
                }
                assertEquals(i, sortedNumeric.nextDoc());
                assertEquals(valuesFields.length, sortedNumeric.docValueCount());
                for (int j = 0; j < sortedNumeric.docValueCount(); ++j) {
                    assertTrue(valueSet.contains(sortedNumeric.nextValue()));
                }
                assertEquals(i, sortedSet.nextDoc());
                int sortedSetCount = 0;
                while (true) {
                    long ord = sortedSet.nextOrd();
                    if (ord == SortedSetDocValues.NO_MORE_ORDS) {
                        break;
                    }
                    assertTrue(valueSet.contains(Long.parseLong(sortedSet.lookupOrd(ord).utf8ToString())));
                    sortedSetCount++;
                }
                assertEquals(valueSet.size(), sortedSetCount);
            }
        }
    }
    indexReader.close();
    dir.close();
}

Example 74 with LeafReader

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

the class TestLucene70DocValuesFormat method doTestTermsEnumRandom.

// TODO: try to refactor this and some termsenum tests into the base class.
// to do this we need to fix the test class to get a DVF not a Codec so we can setup
// the postings format correctly.
private void doTestTermsEnumRandom(int numDocs, Supplier<String> valuesProducer) throws Exception {
    Directory dir = newFSDirectory(createTempDir());
    IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random()));
    conf.setMergeScheduler(new SerialMergeScheduler());
    // set to duel against a codec which has ordinals:
    final PostingsFormat pf = TestUtil.getPostingsFormatWithOrds(random());
    final DocValuesFormat dv = new Lucene70DocValuesFormat();
    conf.setCodec(new AssertingCodec() {

        @Override
        public PostingsFormat getPostingsFormatForField(String field) {
            return pf;
        }

        @Override
        public DocValuesFormat getDocValuesFormatForField(String field) {
            return dv;
        }
    });
    RandomIndexWriter writer = new RandomIndexWriter(random(), dir, conf);
    // index some docs
    for (int i = 0; i < numDocs; i++) {
        Document doc = new Document();
        Field idField = new StringField("id", Integer.toString(i), Field.Store.NO);
        doc.add(idField);
        int numValues = random().nextInt(17);
        // create a random list of strings
        List<String> values = new ArrayList<>();
        for (int v = 0; v < numValues; v++) {
            values.add(valuesProducer.get());
        }
        // add in any order to the indexed field
        ArrayList<String> unordered = new ArrayList<>(values);
        Collections.shuffle(unordered, random());
        for (String v : values) {
            doc.add(newStringField("indexed", v, Field.Store.NO));
        }
        // add in any order to the dv field
        ArrayList<String> unordered2 = new ArrayList<>(values);
        Collections.shuffle(unordered2, random());
        for (String v : unordered2) {
            doc.add(new SortedSetDocValuesField("dv", new BytesRef(v)));
        }
        writer.addDocument(doc);
        if (random().nextInt(31) == 0) {
            writer.commit();
        }
    }
    // delete some docs
    int numDeletions = random().nextInt(numDocs / 10);
    for (int i = 0; i < numDeletions; i++) {
        int id = random().nextInt(numDocs);
        writer.deleteDocuments(new Term("id", Integer.toString(id)));
    }
    // compare per-segment
    DirectoryReader ir = writer.getReader();
    for (LeafReaderContext context : ir.leaves()) {
        LeafReader r = context.reader();
        Terms terms = r.terms("indexed");
        if (terms != null) {
            SortedSetDocValues ssdv = r.getSortedSetDocValues("dv");
            assertEquals(terms.size(), ssdv.getValueCount());
            TermsEnum expected = terms.iterator();
            TermsEnum actual = r.getSortedSetDocValues("dv").termsEnum();
            assertEquals(terms.size(), expected, actual);
            doTestSortedSetEnumAdvanceIndependently(ssdv);
        }
    }
    ir.close();
    writer.forceMerge(1);
    // now compare again after the merge
    ir = writer.getReader();
    LeafReader ar = getOnlyLeafReader(ir);
    Terms terms = ar.terms("indexed");
    if (terms != null) {
        assertEquals(terms.size(), ar.getSortedSetDocValues("dv").getValueCount());
        TermsEnum expected = terms.iterator();
        TermsEnum actual = ar.getSortedSetDocValues("dv").termsEnum();
        assertEquals(terms.size(), expected, actual);
    }
    ir.close();
    writer.close();
    dir.close();
}

Example 75 with LeafReader

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

the class TestLucene60PointsFormat method testEstimatePointCount.

public void testEstimatePointCount() throws IOException {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = newIndexWriterConfig();
    // number of points per leaf hard to predict
    while (iwc.getMergePolicy() instanceof MockRandomMergePolicy) {
        iwc.setMergePolicy(newMergePolicy());
    }
    IndexWriter w = new IndexWriter(dir, iwc);
    byte[] pointValue = new byte[3];
    byte[] uniquePointValue = new byte[3];
    random().nextBytes(uniquePointValue);
    // make sure we have several leaves
    final int numDocs = atLeast(10000);
    for (int i = 0; i < numDocs; ++i) {
        Document doc = new Document();
        if (i == numDocs / 2) {
            doc.add(new BinaryPoint("f", uniquePointValue));
        } else {
            do {
                random().nextBytes(pointValue);
            } while (Arrays.equals(pointValue, uniquePointValue));
            doc.add(new BinaryPoint("f", pointValue));
        }
        w.addDocument(doc);
    }
    w.forceMerge(1);
    final IndexReader r = DirectoryReader.open(w);
    w.close();
    final LeafReader lr = getOnlyLeafReader(r);
    PointValues points = lr.getPointValues("f");
    // If all points match, then the point count is numLeaves * maxPointsInLeafNode
    final int numLeaves = (int) Math.ceil((double) numDocs / maxPointsInLeafNode);
    assertEquals(numLeaves * maxPointsInLeafNode, points.estimatePointCount(new IntersectVisitor() {

        @Override
        public void visit(int docID, byte[] packedValue) throws IOException {
        }

        @Override
        public void visit(int docID) throws IOException {
        }

        @Override
        public Relation compare(byte[] minPackedValue, byte[] maxPackedValue) {
            return Relation.CELL_INSIDE_QUERY;
        }
    }));
    // Return 0 if no points match
    assertEquals(0, points.estimatePointCount(new IntersectVisitor() {

        @Override
        public void visit(int docID, byte[] packedValue) throws IOException {
        }

        @Override
        public void visit(int docID) throws IOException {
        }

        @Override
        public Relation compare(byte[] minPackedValue, byte[] maxPackedValue) {
            return Relation.CELL_OUTSIDE_QUERY;
        }
    }));
    // If only one point matches, then the point count is (maxPointsInLeafNode + 1) / 2
    // in general, or maybe 2x that if the point is a split value
    final long pointCount = points.estimatePointCount(new IntersectVisitor() {

        @Override
        public void visit(int docID, byte[] packedValue) throws IOException {
        }

        @Override
        public void visit(int docID) throws IOException {
        }

        @Override
        public Relation compare(byte[] minPackedValue, byte[] maxPackedValue) {
            if (StringHelper.compare(3, uniquePointValue, 0, maxPackedValue, 0) > 0 || StringHelper.compare(3, uniquePointValue, 0, minPackedValue, 0) < 0) {
                return Relation.CELL_OUTSIDE_QUERY;
            }
            return Relation.CELL_CROSSES_QUERY;
        }
    });
    assertTrue("" + pointCount, // common case
    pointCount == (maxPointsInLeafNode + 1) / 2 || // if the point is a split value
    pointCount == 2 * ((maxPointsInLeafNode + 1) / 2));
    r.close();
    dir.close();
}