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

Example 16 with PointValues

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

the class TestPointValues method testSparsePoints.

public void testSparsePoints() throws Exception {
    Directory dir = newDirectory();
    int numDocs = atLeast(1000);
    int numFields = TestUtil.nextInt(random(), 1, 10);
    RandomIndexWriter w = new RandomIndexWriter(random(), dir);
    int[] fieldDocCounts = new int[numFields];
    int[] fieldSizes = new int[numFields];
    for (int i = 0; i < numDocs; i++) {
        Document doc = new Document();
        for (int field = 0; field < numFields; field++) {
            String fieldName = "int" + field;
            if (random().nextInt(100) == 17) {
                doc.add(new IntPoint(fieldName, random().nextInt()));
                fieldDocCounts[field]++;
                fieldSizes[field]++;
                if (random().nextInt(10) == 5) {
                    // add same field again!
                    doc.add(new IntPoint(fieldName, random().nextInt()));
                    fieldSizes[field]++;
                }
            }
        }
        w.addDocument(doc);
    }
    IndexReader r = w.getReader();
    for (int field = 0; field < numFields; field++) {
        int docCount = 0;
        int size = 0;
        String fieldName = "int" + field;
        for (LeafReaderContext ctx : r.leaves()) {
            PointValues points = ctx.reader().getPointValues(fieldName);
            if (points != null) {
                docCount += points.getDocCount();
                size += points.size();
            }
        }
        assertEquals(fieldDocCounts[field], docCount);
        assertEquals(fieldSizes[field], size);
    }
    r.close();
    w.close();
    dir.close();
}

Example 17 with PointValues

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

the class LatLonPointDistanceQuery method createWeight.

@Override
public Weight createWeight(IndexSearcher searcher, boolean needsScores, float boost) throws IOException {
    Rectangle box = Rectangle.fromPointDistance(latitude, longitude, radiusMeters);
    // create bounding box(es) for the distance range
    // these are pre-encoded with LatLonPoint's encoding
    final byte[] minLat = new byte[Integer.BYTES];
    final byte[] maxLat = new byte[Integer.BYTES];
    final byte[] minLon = new byte[Integer.BYTES];
    final byte[] maxLon = new byte[Integer.BYTES];
    // second set of longitude ranges to check (for cross-dateline case)
    final byte[] minLon2 = new byte[Integer.BYTES];
    NumericUtils.intToSortableBytes(encodeLatitude(box.minLat), minLat, 0);
    NumericUtils.intToSortableBytes(encodeLatitude(box.maxLat), maxLat, 0);
    // crosses dateline: split
    if (box.crossesDateline()) {
        // box1
        NumericUtils.intToSortableBytes(Integer.MIN_VALUE, minLon, 0);
        NumericUtils.intToSortableBytes(encodeLongitude(box.maxLon), maxLon, 0);
        // box2
        NumericUtils.intToSortableBytes(encodeLongitude(box.minLon), minLon2, 0);
    } else {
        NumericUtils.intToSortableBytes(encodeLongitude(box.minLon), minLon, 0);
        NumericUtils.intToSortableBytes(encodeLongitude(box.maxLon), maxLon, 0);
        // disable box2
        NumericUtils.intToSortableBytes(Integer.MAX_VALUE, minLon2, 0);
    }
    // compute exact sort key: avoid any asin() computations
    final double sortKey = GeoUtils.distanceQuerySortKey(radiusMeters);
    final double axisLat = Rectangle.axisLat(latitude, radiusMeters);
    return new ConstantScoreWeight(this, boost) {

        final GeoEncodingUtils.DistancePredicate distancePredicate = GeoEncodingUtils.createDistancePredicate(latitude, longitude, radiusMeters);

        @Override
        public Scorer scorer(LeafReaderContext context) throws IOException {
            ScorerSupplier scorerSupplier = scorerSupplier(context);
            if (scorerSupplier == null) {
                return null;
            }
            return scorerSupplier.get(false);
        }

        @Override
        public ScorerSupplier scorerSupplier(LeafReaderContext context) throws IOException {
            LeafReader reader = context.reader();
            PointValues values = reader.getPointValues(field);
            if (values == null) {
                // No docs in this segment had any points fields
                return null;
            }
            FieldInfo fieldInfo = reader.getFieldInfos().fieldInfo(field);
            if (fieldInfo == null) {
                // No docs in this segment indexed this field at all
                return null;
            }
            LatLonPoint.checkCompatible(fieldInfo);
            // matching docids
            DocIdSetBuilder result = new DocIdSetBuilder(reader.maxDoc(), values, field);
            final IntersectVisitor visitor = new IntersectVisitor() {

                DocIdSetBuilder.BulkAdder adder;

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

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

                @Override
                public void visit(int docID, byte[] packedValue) {
                    // bounding box check
                    if (StringHelper.compare(Integer.BYTES, packedValue, 0, maxLat, 0) > 0 || StringHelper.compare(Integer.BYTES, packedValue, 0, minLat, 0) < 0) {
                        // latitude out of bounding box range
                        return;
                    }
                    if ((StringHelper.compare(Integer.BYTES, packedValue, Integer.BYTES, maxLon, 0) > 0 || StringHelper.compare(Integer.BYTES, packedValue, Integer.BYTES, minLon, 0) < 0) && StringHelper.compare(Integer.BYTES, packedValue, Integer.BYTES, minLon2, 0) < 0) {
                        // longitude out of bounding box range
                        return;
                    }
                    int docLatitude = NumericUtils.sortableBytesToInt(packedValue, 0);
                    int docLongitude = NumericUtils.sortableBytesToInt(packedValue, Integer.BYTES);
                    if (distancePredicate.test(docLatitude, docLongitude)) {
                        adder.add(docID);
                    }
                }

                // algorithm: we create a bounding box (two bounding boxes if we cross the dateline).
                // 1. check our bounding box(es) first. if the subtree is entirely outside of those, bail.
                // 2. check if the subtree is disjoint. it may cross the bounding box but not intersect with circle
                // 3. see if the subtree is fully contained. if the subtree is enormous along the x axis, wrapping half way around the world, etc: then this can't work, just go to step 4.
                // 4. recurse naively (subtrees crossing over circle edge)
                @Override
                public Relation compare(byte[] minPackedValue, byte[] maxPackedValue) {
                    if (StringHelper.compare(Integer.BYTES, minPackedValue, 0, maxLat, 0) > 0 || StringHelper.compare(Integer.BYTES, maxPackedValue, 0, minLat, 0) < 0) {
                        // latitude out of bounding box range
                        return Relation.CELL_OUTSIDE_QUERY;
                    }
                    if ((StringHelper.compare(Integer.BYTES, minPackedValue, Integer.BYTES, maxLon, 0) > 0 || StringHelper.compare(Integer.BYTES, maxPackedValue, Integer.BYTES, minLon, 0) < 0) && StringHelper.compare(Integer.BYTES, maxPackedValue, Integer.BYTES, minLon2, 0) < 0) {
                        // longitude out of bounding box range
                        return Relation.CELL_OUTSIDE_QUERY;
                    }
                    double latMin = decodeLatitude(minPackedValue, 0);
                    double lonMin = decodeLongitude(minPackedValue, Integer.BYTES);
                    double latMax = decodeLatitude(maxPackedValue, 0);
                    double lonMax = decodeLongitude(maxPackedValue, Integer.BYTES);
                    return GeoUtils.relate(latMin, latMax, lonMin, lonMax, latitude, longitude, sortKey, axisLat);
                }
            };
            final Weight weight = this;
            return new ScorerSupplier() {

                long cost = -1;

                @Override
                public Scorer get(boolean randomAccess) throws IOException {
                    values.intersect(visitor);
                    return new ConstantScoreScorer(weight, score(), result.build().iterator());
                }

                @Override
                public long cost() {
                    if (cost == -1) {
                        cost = values.estimatePointCount(visitor);
                    }
                    assert cost >= 0;
                    return cost;
                }
            };
        }
    };
}

Example 18 with PointValues

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

the class LatLonPointInPolygonQuery method createWeight.

@Override
public Weight createWeight(IndexSearcher searcher, boolean needsScores, float boost) throws IOException {
    // I don't use RandomAccessWeight here: it's no good to approximate with "match all docs"; this is an inverted structure and should be
    // used in the first pass:
    // bounding box over all polygons, this can speed up tree intersection/cheaply improve approximation for complex multi-polygons
    // these are pre-encoded with LatLonPoint's encoding
    final Rectangle box = Rectangle.fromPolygon(polygons);
    final byte[] minLat = new byte[Integer.BYTES];
    final byte[] maxLat = new byte[Integer.BYTES];
    final byte[] minLon = new byte[Integer.BYTES];
    final byte[] maxLon = new byte[Integer.BYTES];
    NumericUtils.intToSortableBytes(encodeLatitude(box.minLat), minLat, 0);
    NumericUtils.intToSortableBytes(encodeLatitude(box.maxLat), maxLat, 0);
    NumericUtils.intToSortableBytes(encodeLongitude(box.minLon), minLon, 0);
    NumericUtils.intToSortableBytes(encodeLongitude(box.maxLon), maxLon, 0);
    final Polygon2D tree = Polygon2D.create(polygons);
    final GeoEncodingUtils.PolygonPredicate polygonPredicate = GeoEncodingUtils.createPolygonPredicate(polygons, tree);
    return new ConstantScoreWeight(this, boost) {

        @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 had any points fields
                return null;
            }
            FieldInfo fieldInfo = reader.getFieldInfos().fieldInfo(field);
            if (fieldInfo == null) {
                // No docs in this segment indexed this field at all
                return null;
            }
            LatLonPoint.checkCompatible(fieldInfo);
            // matching docids
            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) {
                    adder.add(docID);
                }

                @Override
                public void visit(int docID, byte[] packedValue) {
                    if (polygonPredicate.test(NumericUtils.sortableBytesToInt(packedValue, 0), NumericUtils.sortableBytesToInt(packedValue, Integer.BYTES))) {
                        adder.add(docID);
                    }
                }

                @Override
                public Relation compare(byte[] minPackedValue, byte[] maxPackedValue) {
                    if (StringHelper.compare(Integer.BYTES, minPackedValue, 0, maxLat, 0) > 0 || StringHelper.compare(Integer.BYTES, maxPackedValue, 0, minLat, 0) < 0 || StringHelper.compare(Integer.BYTES, minPackedValue, Integer.BYTES, maxLon, 0) > 0 || StringHelper.compare(Integer.BYTES, maxPackedValue, Integer.BYTES, minLon, 0) < 0) {
                        // outside of global bounding box range
                        return Relation.CELL_OUTSIDE_QUERY;
                    }
                    double cellMinLat = decodeLatitude(minPackedValue, 0);
                    double cellMinLon = decodeLongitude(minPackedValue, Integer.BYTES);
                    double cellMaxLat = decodeLatitude(maxPackedValue, 0);
                    double cellMaxLon = decodeLongitude(maxPackedValue, Integer.BYTES);
                    return tree.relate(cellMinLat, cellMaxLat, cellMinLon, cellMaxLon);
                }
            });
            return new ConstantScoreScorer(this, score(), result.build().iterator());
        }
    };
}

Example 19 with PointValues

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

the class TestDocIdSetBuilder method testLeverageStats.

public void testLeverageStats() throws IOException {
    // single-valued points
    PointValues values = new DummyPointValues(42, 42);
    DocIdSetBuilder builder = new DocIdSetBuilder(100, values, "foo");
    assertEquals(1d, builder.numValuesPerDoc, 0d);
    assertFalse(builder.multivalued);
    DocIdSetBuilder.BulkAdder adder = builder.grow(2);
    adder.add(5);
    adder.add(7);
    DocIdSet set = builder.build();
    assertTrue(set instanceof BitDocIdSet);
    assertEquals(2, set.iterator().cost());
    // multi-valued points
    values = new DummyPointValues(42, 63);
    builder = new DocIdSetBuilder(100, values, "foo");
    assertEquals(1.5, builder.numValuesPerDoc, 0d);
    assertTrue(builder.multivalued);
    adder = builder.grow(2);
    adder.add(5);
    adder.add(7);
    set = builder.build();
    assertTrue(set instanceof BitDocIdSet);
    // it thinks the same doc was added twice
    assertEquals(1, set.iterator().cost());
    // incomplete stats
    values = new DummyPointValues(42, -1);
    builder = new DocIdSetBuilder(100, values, "foo");
    assertEquals(1d, builder.numValuesPerDoc, 0d);
    assertTrue(builder.multivalued);
    values = new DummyPointValues(-1, 84);
    builder = new DocIdSetBuilder(100, values, "foo");
    assertEquals(1d, builder.numValuesPerDoc, 0d);
    assertTrue(builder.multivalued);
    // single-valued terms
    Terms terms = new DummyTerms(42, 42);
    builder = new DocIdSetBuilder(100, terms);
    assertEquals(1d, builder.numValuesPerDoc, 0d);
    assertFalse(builder.multivalued);
    adder = builder.grow(2);
    adder.add(5);
    adder.add(7);
    set = builder.build();
    assertTrue(set instanceof BitDocIdSet);
    assertEquals(2, set.iterator().cost());
    // multi-valued terms
    terms = new DummyTerms(42, 63);
    builder = new DocIdSetBuilder(100, terms);
    assertEquals(1.5, builder.numValuesPerDoc, 0d);
    assertTrue(builder.multivalued);
    adder = builder.grow(2);
    adder.add(5);
    adder.add(7);
    set = builder.build();
    assertTrue(set instanceof BitDocIdSet);
    // it thinks the same doc was added twice
    assertEquals(1, set.iterator().cost());
    // incomplete stats
    terms = new DummyTerms(42, -1);
    builder = new DocIdSetBuilder(100, terms);
    assertEquals(1d, builder.numValuesPerDoc, 0d);
    assertTrue(builder.multivalued);
    terms = new DummyTerms(-1, 84);
    builder = new DocIdSetBuilder(100, terms);
    assertEquals(1d, builder.numValuesPerDoc, 0d);
    assertTrue(builder.multivalued);
}