Search in sources :

Example 31 with ImmutableBitmap

use of io.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.

the class Filters method estimateSelectivity.

/**
   * Return an estimated selectivity for bitmaps given by an iterator.
   *
   * @param bitmaps      iterator of bitmaps
   * @param totalNumRows number of rows in the column associated with this bitmap index
   *
   * @return estimated selectivity
   */
public static double estimateSelectivity(final Iterator<ImmutableBitmap> bitmaps, final long totalNumRows) {
    long numMatchedRows = 0;
    while (bitmaps.hasNext()) {
        final ImmutableBitmap bitmap = bitmaps.next();
        numMatchedRows += bitmap.size();
    }
    return Math.min(1., (double) numMatchedRows / totalNumRows);
}
Also used : ImmutableBitmap(io.druid.collections.bitmap.ImmutableBitmap)

Example 32 with ImmutableBitmap

use of io.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.

the class DictionaryEncodedColumnPartSerde method getDeserializer.

@Override
public Deserializer getDeserializer() {
    return new Deserializer() {

        @Override
        public void read(ByteBuffer buffer, ColumnBuilder builder, ColumnConfig columnConfig) {
            final VERSION rVersion = VERSION.fromByte(buffer.get());
            final int rFlags;
            if (rVersion.compareTo(VERSION.COMPRESSED) >= 0) {
                rFlags = buffer.getInt();
            } else {
                rFlags = rVersion.equals(VERSION.UNCOMPRESSED_MULTI_VALUE) ? Feature.MULTI_VALUE.getMask() : NO_FLAGS;
            }
            final boolean hasMultipleValues = Feature.MULTI_VALUE.isSet(rFlags) || Feature.MULTI_VALUE_V3.isSet(rFlags);
            final GenericIndexed<String> rDictionary = GenericIndexed.read(buffer, GenericIndexed.STRING_STRATEGY, builder.getFileMapper());
            builder.setType(ValueType.STRING);
            final WritableSupplier<IndexedInts> rSingleValuedColumn;
            final WritableSupplier<IndexedMultivalue<IndexedInts>> rMultiValuedColumn;
            if (hasMultipleValues) {
                rMultiValuedColumn = readMultiValuedColum(rVersion, buffer, rFlags, builder.getFileMapper());
                rSingleValuedColumn = null;
            } else {
                rSingleValuedColumn = readSingleValuedColumn(rVersion, buffer, builder.getFileMapper());
                rMultiValuedColumn = null;
            }
            builder.setHasMultipleValues(hasMultipleValues).setDictionaryEncodedColumn(new DictionaryEncodedColumnSupplier(rDictionary, rSingleValuedColumn, rMultiValuedColumn, columnConfig.columnCacheSizeBytes()));
            GenericIndexed<ImmutableBitmap> rBitmaps = GenericIndexed.read(buffer, bitmapSerdeFactory.getObjectStrategy(), builder.getFileMapper());
            builder.setBitmapIndex(new BitmapIndexColumnPartSupplier(bitmapSerdeFactory.getBitmapFactory(), rBitmaps, rDictionary));
            ImmutableRTree rSpatialIndex = null;
            if (buffer.hasRemaining()) {
                rSpatialIndex = ByteBufferSerializer.read(buffer, new IndexedRTree.ImmutableRTreeObjectStrategy(bitmapSerdeFactory.getBitmapFactory()));
                builder.setSpatialIndex(new SpatialIndexColumnPartSupplier(rSpatialIndex));
            }
        }

        private WritableSupplier<IndexedInts> readSingleValuedColumn(VERSION version, ByteBuffer buffer, SmooshedFileMapper fileMapper) {
            switch(version) {
                case UNCOMPRESSED_SINGLE_VALUE:
                    return VSizeIndexedInts.readFromByteBuffer(buffer).asWritableSupplier();
                case COMPRESSED:
                    return CompressedVSizeIntsIndexedSupplier.fromByteBuffer(buffer, byteOrder, fileMapper);
            }
            throw new IAE("Unsupported single-value version[%s]", version);
        }

        private WritableSupplier<IndexedMultivalue<IndexedInts>> readMultiValuedColum(VERSION version, ByteBuffer buffer, int flags, SmooshedFileMapper fileMapper) {
            switch(version) {
                case UNCOMPRESSED_MULTI_VALUE:
                    return VSizeIndexed.readFromByteBuffer(buffer).asWritableSupplier();
                case COMPRESSED:
                    if (Feature.MULTI_VALUE.isSet(flags)) {
                        return CompressedVSizeIndexedSupplier.fromByteBuffer(buffer, byteOrder, fileMapper);
                    } else if (Feature.MULTI_VALUE_V3.isSet(flags)) {
                        return CompressedVSizeIndexedV3Supplier.fromByteBuffer(buffer, byteOrder, fileMapper);
                    } else {
                        throw new IAE("Unrecognized multi-value flag[%d]", flags);
                    }
            }
            throw new IAE("Unsupported multi-value version[%s]", version);
        }
    };
}
Also used : ColumnConfig(io.druid.segment.column.ColumnConfig) ImmutableBitmap(io.druid.collections.bitmap.ImmutableBitmap) IAE(io.druid.java.util.common.IAE) ByteBuffer(java.nio.ByteBuffer) ImmutableRTree(io.druid.collections.spatial.ImmutableRTree) VSizeIndexedInts(io.druid.segment.data.VSizeIndexedInts) IndexedInts(io.druid.segment.data.IndexedInts) IndexedMultivalue(io.druid.segment.data.IndexedMultivalue) ColumnBuilder(io.druid.segment.column.ColumnBuilder) SmooshedFileMapper(io.druid.java.util.common.io.smoosh.SmooshedFileMapper)

Example 33 with ImmutableBitmap

use of io.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.

the class BitmapIterationBenchmark method constructAndIter.

/**
   * Benchmark of cumulative cost of construction of an immutable bitmap and then iterating over it. This is a pattern
   * from realtime nodes, see {@link io.druid.segment.StringDimensionIndexer#fillBitmapsFromUnsortedEncodedKeyComponent}.
   * However this benchmark is yet approximate and to be improved to better reflect actual workloads of realtime nodes.
   */
@Benchmark
public int constructAndIter(ConstructAndIterState state) {
    int dataSize = state.dataSize;
    int[] data = state.data;
    MutableBitmap mutableBitmap = factory.makeEmptyMutableBitmap();
    for (int i = 0; i < dataSize; i++) {
        mutableBitmap.add(data[i]);
    }
    ImmutableBitmap bitmap = factory.makeImmutableBitmap(mutableBitmap);
    return iter(bitmap);
}
Also used : ImmutableBitmap(io.druid.collections.bitmap.ImmutableBitmap) MutableBitmap(io.druid.collections.bitmap.MutableBitmap) Benchmark(org.openjdk.jmh.annotations.Benchmark)

Example 34 with ImmutableBitmap

use of io.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.

the class BoundFilterBenchmark method matchHalfLexicographic.

@Benchmark
@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.MICROSECONDS)
public void matchHalfLexicographic() {
    final ImmutableBitmap bitmapIndex = HALF_LEXICOGRAPHIC.getBitmapIndex(selector);
    Preconditions.checkState(bitmapIndex.size() > 0 && bitmapIndex.size() < cardinality);
}
Also used : ImmutableBitmap(io.druid.collections.bitmap.ImmutableBitmap) BenchmarkMode(org.openjdk.jmh.annotations.BenchmarkMode) Benchmark(org.openjdk.jmh.annotations.Benchmark) OutputTimeUnit(org.openjdk.jmh.annotations.OutputTimeUnit)

Example 35 with ImmutableBitmap

use of io.druid.collections.bitmap.ImmutableBitmap in project druid by druid-io.

the class ExtractionDimFilterTest method testNull.

@Test
public void testNull() {
    Filter extractionFilter = new SelectorDimFilter("FDHJSFFHDS", "extractDimVal", DIM_EXTRACTION_FN).toFilter();
    ImmutableBitmap immutableBitmap = extractionFilter.getBitmapIndex(BITMAP_INDEX_SELECTOR);
    Assert.assertEquals(0, immutableBitmap.size());
}
Also used : SelectorDimFilter(io.druid.query.filter.SelectorDimFilter) ExtractionDimFilter(io.druid.query.filter.ExtractionDimFilter) Filter(io.druid.query.filter.Filter) ImmutableBitmap(io.druid.collections.bitmap.ImmutableBitmap) SelectorDimFilter(io.druid.query.filter.SelectorDimFilter) Test(org.junit.Test)

Aggregations

ImmutableBitmap (io.druid.collections.bitmap.ImmutableBitmap)51 BitmapFactory (io.druid.collections.bitmap.BitmapFactory)23 Test (org.junit.Test)23 RoaringBitmapFactory (io.druid.collections.bitmap.RoaringBitmapFactory)22 ConciseBitmapFactory (io.druid.collections.bitmap.ConciseBitmapFactory)19 LinearGutmanSplitStrategy (io.druid.collections.spatial.split.LinearGutmanSplitStrategy)18 IntIterator (org.roaringbitmap.IntIterator)16 RadiusBound (io.druid.collections.spatial.search.RadiusBound)13 Benchmark (org.openjdk.jmh.annotations.Benchmark)13 Random (java.util.Random)12 BenchmarkMode (org.openjdk.jmh.annotations.BenchmarkMode)12 OutputTimeUnit (org.openjdk.jmh.annotations.OutputTimeUnit)12 MutableBitmap (io.druid.collections.bitmap.MutableBitmap)7 Filter (io.druid.query.filter.Filter)6 BitmapIndexSelector (io.druid.query.filter.BitmapIndexSelector)5 BitmapIndex (io.druid.segment.column.BitmapIndex)5 ImmutableRTree (io.druid.collections.spatial.ImmutableRTree)4 BitmapSerdeFactory (io.druid.segment.data.BitmapSerdeFactory)4 RoaringBitmapSerdeFactory (io.druid.segment.data.RoaringBitmapSerdeFactory)4 Function (com.google.common.base.Function)3