Search in sources :

Example 1 with ImmutableRoaringBitmap

use of org.roaringbitmap.buffer.ImmutableRoaringBitmap in project druid by druid-io.

the class WrappedImmutableRoaringBitmap method intersection.

@Override
public ImmutableBitmap intersection(ImmutableBitmap otherBitmap) {
    WrappedImmutableRoaringBitmap other = (WrappedImmutableRoaringBitmap) otherBitmap;
    ImmutableRoaringBitmap unwrappedOtherBitmap = other.bitmap;
    return new WrappedImmutableRoaringBitmap(ImmutableRoaringBitmap.and(bitmap, unwrappedOtherBitmap));
}
Also used : ImmutableRoaringBitmap(org.roaringbitmap.buffer.ImmutableRoaringBitmap)

Example 2 with ImmutableRoaringBitmap

use of org.roaringbitmap.buffer.ImmutableRoaringBitmap in project pinot by linkedin.

the class AndBlockDocIdSet method slowIterator.

public BlockDocIdIterator slowIterator() {
    List<BlockDocIdIterator> rawIterators = new ArrayList<>();
    boolean useBitmapBasedIntersection = false;
    for (BlockDocIdSet docIdSet : blockDocIdSets) {
        if (docIdSet instanceof BitmapDocIdSet) {
            useBitmapBasedIntersection = true;
        }
    }
    BlockDocIdIterator[] docIdIterators;
    if (useBitmapBasedIntersection) {
        List<ImmutableRoaringBitmap> allBitmaps = new ArrayList<ImmutableRoaringBitmap>();
        for (BlockDocIdSet docIdSet : blockDocIdSets) {
            if (docIdSet instanceof SortedDocIdSet) {
                MutableRoaringBitmap bitmap = new MutableRoaringBitmap();
                SortedDocIdSet sortedDocIdSet = (SortedDocIdSet) docIdSet;
                List<IntPair> pairs = sortedDocIdSet.getRaw();
                for (IntPair pair : pairs) {
                    // add takes [start, end) i.e inclusive
                    bitmap.add(pair.getLeft(), pair.getRight() + 1);
                // start, exclusive end.
                }
                allBitmaps.add(bitmap);
            } else if (docIdSet instanceof BitmapDocIdSet) {
                BitmapDocIdSet bitmapDocIdSet = (BitmapDocIdSet) docIdSet;
                ImmutableRoaringBitmap childBitmap = bitmapDocIdSet.getRaw();
                allBitmaps.add(childBitmap);
            } else {
                BlockDocIdIterator iterator = docIdSet.iterator();
                rawIterators.add(iterator);
            }
        }
        IntIterator intIterator;
        if (allBitmaps.size() > 1) {
            MutableRoaringBitmap answer = (MutableRoaringBitmap) allBitmaps.get(0).clone();
            for (int i = 1; i < allBitmaps.size(); i++) {
                answer.and(allBitmaps.get(i));
            }
            intIterator = answer.getIntIterator();
        } else {
            intIterator = allBitmaps.get(0).getIntIterator();
        }
        BitmapDocIdIterator singleBitmapBlockIdIterator = new BitmapDocIdIterator(intIterator);
        singleBitmapBlockIdIterator.setStartDocId(minDocId);
        singleBitmapBlockIdIterator.setEndDocId(maxDocId);
        rawIterators.add(0, singleBitmapBlockIdIterator);
        docIdIterators = new BlockDocIdIterator[rawIterators.size()];
        rawIterators.toArray(docIdIterators);
    } else {
        docIdIterators = new BlockDocIdIterator[blockDocIdSets.size()];
        for (int srcId = 0; srcId < blockDocIdSets.size(); srcId++) {
            docIdIterators[srcId] = blockDocIdSets.get(srcId).iterator();
        }
    }
    return new AndDocIdIterator(docIdIterators);
}
Also used : IntIterator(org.roaringbitmap.IntIterator) MutableRoaringBitmap(org.roaringbitmap.buffer.MutableRoaringBitmap) ArrayList(java.util.ArrayList) IntPair(com.linkedin.pinot.common.utils.Pairs.IntPair) BlockDocIdIterator(com.linkedin.pinot.core.common.BlockDocIdIterator) BlockDocIdSet(com.linkedin.pinot.core.common.BlockDocIdSet) AndDocIdIterator(com.linkedin.pinot.core.operator.dociditerators.AndDocIdIterator) ImmutableRoaringBitmap(org.roaringbitmap.buffer.ImmutableRoaringBitmap) RangelessBitmapDocIdIterator(com.linkedin.pinot.core.operator.dociditerators.RangelessBitmapDocIdIterator) BitmapDocIdIterator(com.linkedin.pinot.core.operator.dociditerators.BitmapDocIdIterator)

Example 3 with ImmutableRoaringBitmap

use of org.roaringbitmap.buffer.ImmutableRoaringBitmap in project pinot by linkedin.

the class AndBlockDocIdSet method fastIterator.

public BlockDocIdIterator fastIterator() {
    long start = System.currentTimeMillis();
    List<List<IntPair>> sortedRangeSets = new ArrayList<>();
    List<ImmutableRoaringBitmap> childBitmaps = new ArrayList<ImmutableRoaringBitmap>();
    List<FilterBlockDocIdSet> scanBasedDocIdSets = new ArrayList<>();
    List<BlockDocIdIterator> remainingIterators = new ArrayList<>();
    for (BlockDocIdSet docIdSet : blockDocIdSets) {
        if (docIdSet instanceof SortedDocIdSet) {
            SortedDocIdSet sortedDocIdSet = (SortedDocIdSet) docIdSet;
            List<IntPair> pairs = sortedDocIdSet.getRaw();
            sortedRangeSets.add(pairs);
        } else if (docIdSet instanceof BitmapDocIdSet) {
            BitmapDocIdSet bitmapDocIdSet = (BitmapDocIdSet) docIdSet;
            ImmutableRoaringBitmap childBitmap = bitmapDocIdSet.getRaw();
            childBitmaps.add(childBitmap);
        } else if (docIdSet instanceof ScanBasedSingleValueDocIdSet) {
            scanBasedDocIdSets.add((ScanBasedSingleValueDocIdSet) docIdSet);
        } else if (docIdSet instanceof ScanBasedMultiValueDocIdSet) {
            scanBasedDocIdSets.add((ScanBasedMultiValueDocIdSet) docIdSet);
        } else {
            // TODO:handle child OR/AND as bitmap if possible
            remainingIterators.add(docIdSet.iterator());
        }
    }
    if (childBitmaps.size() == 0 && sortedRangeSets.size() == 0) {
        // When one or more of the operands are operators themselves, then we don't have a sorted or
        // bitmap index. In that case, just use the AndDocIdIterator to iterate over all of of the subtree.
        BlockDocIdIterator[] docIdIterators = new BlockDocIdIterator[blockDocIdSets.size()];
        for (int srcId = 0; srcId < blockDocIdSets.size(); srcId++) {
            docIdIterators[srcId] = blockDocIdSets.get(srcId).iterator();
        }
        return new AndDocIdIterator(docIdIterators);
    } else {
        // TODO: will be nice to re-order sorted and bitmap index based on size
        if (sortedRangeSets.size() > 0) {
            List<IntPair> pairList;
            pairList = SortedRangeIntersection.intersectSortedRangeSets(sortedRangeSets);
            answer = new MutableRoaringBitmap();
            for (IntPair pair : pairList) {
                // end is exclusive
                answer.add(pair.getLeft(), pair.getRight() + 1);
            }
        }
        // handle bitmaps
        if (childBitmaps.size() > 0) {
            if (answer == null) {
                answer = childBitmaps.get(0).toMutableRoaringBitmap();
                for (int i = 1; i < childBitmaps.size(); i++) {
                    answer.and(childBitmaps.get(i));
                }
            } else {
                for (int i = 0; i < childBitmaps.size(); i++) {
                    answer.and(childBitmaps.get(i));
                }
            }
        }
        // At this point, we must have 'answer' to be non-null.
        assert (answer != null) : "sortedRangeSets=" + sortedRangeSets.size() + ",childBitmaps=" + childBitmaps.size();
        // handle raw iterators
        for (FilterBlockDocIdSet scanBasedDocIdSet : scanBasedDocIdSets) {
            ScanBasedDocIdIterator iterator = (ScanBasedDocIdIterator) scanBasedDocIdSet.iterator();
            MutableRoaringBitmap scanAnswer = iterator.applyAnd(answer);
            answer.and(scanAnswer);
        }
        long end = System.currentTimeMillis();
        LOGGER.debug("Time to evaluate and Filter:{}", (end - start));
        // if other iterators exists resort to iterator style intersection
        BlockDocIdIterator answerDocIdIterator = new RangelessBitmapDocIdIterator(answer.getIntIterator());
        if (remainingIterators.size() == 0) {
            return answerDocIdIterator;
        } else {
            BlockDocIdIterator[] docIdIterators = new BlockDocIdIterator[remainingIterators.size() + 1];
            docIdIterators[0] = answerDocIdIterator;
            for (int i = 0; i < remainingIterators.size(); i++) {
                docIdIterators[i + 1] = remainingIterators.get(i);
            }
            return new AndDocIdIterator(docIdIterators);
        }
    }
}
Also used : RangelessBitmapDocIdIterator(com.linkedin.pinot.core.operator.dociditerators.RangelessBitmapDocIdIterator) MutableRoaringBitmap(org.roaringbitmap.buffer.MutableRoaringBitmap) ArrayList(java.util.ArrayList) ScanBasedDocIdIterator(com.linkedin.pinot.core.operator.dociditerators.ScanBasedDocIdIterator) IntPair(com.linkedin.pinot.common.utils.Pairs.IntPair) BlockDocIdIterator(com.linkedin.pinot.core.common.BlockDocIdIterator) BlockDocIdSet(com.linkedin.pinot.core.common.BlockDocIdSet) AndDocIdIterator(com.linkedin.pinot.core.operator.dociditerators.AndDocIdIterator) ImmutableRoaringBitmap(org.roaringbitmap.buffer.ImmutableRoaringBitmap) ArrayList(java.util.ArrayList) List(java.util.List)

Example 4 with ImmutableRoaringBitmap

use of org.roaringbitmap.buffer.ImmutableRoaringBitmap in project pinot by linkedin.

the class BitmapInvertedIndexCreatorTest method validate.

private void validate(String colName, File bitmapIndexFile, int cardinality, Map<Integer, Set<Integer>> postingListMap) throws IOException {
    Assert.assertTrue(bitmapIndexFile.exists());
    PinotDataBuffer dataBuffer = PinotDataBuffer.fromFile(bitmapIndexFile, ReadMode.mmap, FileChannel.MapMode.READ_ONLY, "testing");
    BitmapInvertedIndexReader reader = new BitmapInvertedIndexReader(dataBuffer, cardinality);
    for (int i = 0; i < cardinality; i++) {
        ImmutableRoaringBitmap bitmap = reader.getImmutable(i);
        Set<Integer> expected = postingListMap.get(i);
        Assert.assertEquals(bitmap.getCardinality(), expected.size());
        int[] actual = bitmap.toArray();
        List<Integer> actualList = Ints.asList(actual);
        Assert.assertEquals(actualList, expected);
    }
    dataBuffer.close();
}
Also used : PinotDataBuffer(com.linkedin.pinot.core.segment.memory.PinotDataBuffer) ImmutableRoaringBitmap(org.roaringbitmap.buffer.ImmutableRoaringBitmap) BitmapInvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader)

Example 5 with ImmutableRoaringBitmap

use of org.roaringbitmap.buffer.ImmutableRoaringBitmap in project pinot by linkedin.

the class BitmapPerformanceBenchmark method iterationSpeed.

public static void iterationSpeed(String indexSegmentDir, String column) throws Exception {
    File indexSegment = new File(indexSegmentDir);
    SegmentMetadataImpl segmentMetadata = new SegmentMetadataImpl(indexSegment);
    Map<String, BitmapInvertedIndexReader> bitMapIndexMap = new HashMap<String, BitmapInvertedIndexReader>();
    Map<String, Integer> cardinalityMap = new HashMap<String, Integer>();
    Map<String, ImmutableDictionaryReader> dictionaryMap = new HashMap<String, ImmutableDictionaryReader>();
    File bitMapIndexFile = new File(indexSegmentDir, column + ".bitmap.inv");
    ColumnMetadata columnMetadata = segmentMetadata.getColumnMetadataFor(column);
    int cardinality = columnMetadata.getCardinality();
    cardinalityMap.put(column, cardinality);
    PinotDataBuffer bitMapDataBuffer = PinotDataBuffer.fromFile(bitMapIndexFile, ReadMode.mmap, FileChannel.MapMode.READ_ONLY, "testing");
    BitmapInvertedIndexReader bitmapInvertedIndex = new BitmapInvertedIndexReader(bitMapDataBuffer, cardinality);
    File dictionaryFile = new File(indexSegmentDir + "/" + column + ".dict");
    SegmentDirectory segmentDirectory = SegmentDirectory.createFromLocalFS(indexSegment, segmentMetadata, ReadMode.mmap);
    SegmentDirectory.Reader segmentReader = segmentDirectory.createReader();
    ColumnIndexContainer container = ColumnIndexContainer.init(segmentReader, columnMetadata, null);
    ImmutableDictionaryReader dictionary = container.getDictionary();
    dictionaryMap.put(column, dictionary);
    // System.out.println(column + ":\t" + MemoryUtil.deepMemoryUsageOf(bitmapInvertedIndex));
    bitMapIndexMap.put(column, bitmapInvertedIndex);
    int dictId = dictionary.indexOf("na.us");
    ImmutableRoaringBitmap immutable = bitmapInvertedIndex.getImmutable(dictId);
    Iterator<Integer> iterator = immutable.iterator();
    int count = 0;
    long start = System.currentTimeMillis();
    while (iterator.hasNext()) {
        iterator.next();
        count = count + 1;
    }
    long end = System.currentTimeMillis();
    System.out.println(" matched: " + count + " Time to iterate:" + (end - start));
    bitMapDataBuffer.close();
}
Also used : ColumnMetadata(com.linkedin.pinot.core.segment.index.ColumnMetadata) HashMap(java.util.HashMap) ImmutableDictionaryReader(com.linkedin.pinot.core.segment.index.readers.ImmutableDictionaryReader) BitmapInvertedIndexReader(com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader) SegmentDirectory(com.linkedin.pinot.core.segment.store.SegmentDirectory) ColumnIndexContainer(com.linkedin.pinot.core.segment.index.column.ColumnIndexContainer) PinotDataBuffer(com.linkedin.pinot.core.segment.memory.PinotDataBuffer) ImmutableRoaringBitmap(org.roaringbitmap.buffer.ImmutableRoaringBitmap) SegmentMetadataImpl(com.linkedin.pinot.core.segment.index.SegmentMetadataImpl) File(java.io.File)

Aggregations

ImmutableRoaringBitmap (org.roaringbitmap.buffer.ImmutableRoaringBitmap)18 ArrayList (java.util.ArrayList)5 MutableRoaringBitmap (org.roaringbitmap.buffer.MutableRoaringBitmap)5 BlockDocIdIterator (com.linkedin.pinot.core.common.BlockDocIdIterator)4 BlockDocIdSet (com.linkedin.pinot.core.common.BlockDocIdSet)3 BitmapInvertedIndexReader (com.linkedin.pinot.core.segment.index.readers.BitmapInvertedIndexReader)3 PinotDataBuffer (com.linkedin.pinot.core.segment.memory.PinotDataBuffer)3 Test (org.junit.Test)3 IntPair (com.linkedin.pinot.common.utils.Pairs.IntPair)2 AndDocIdIterator (com.linkedin.pinot.core.operator.dociditerators.AndDocIdIterator)2 BitmapDocIdIterator (com.linkedin.pinot.core.operator.dociditerators.BitmapDocIdIterator)2 RangelessBitmapDocIdIterator (com.linkedin.pinot.core.operator.dociditerators.RangelessBitmapDocIdIterator)2 ColumnMetadata (com.linkedin.pinot.core.segment.index.ColumnMetadata)2 SegmentMetadataImpl (com.linkedin.pinot.core.segment.index.SegmentMetadataImpl)2 ColumnIndexContainer (com.linkedin.pinot.core.segment.index.column.ColumnIndexContainer)2 ImmutableDictionaryReader (com.linkedin.pinot.core.segment.index.readers.ImmutableDictionaryReader)2 InvertedIndexReader (com.linkedin.pinot.core.segment.index.readers.InvertedIndexReader)2 SegmentDirectory (com.linkedin.pinot.core.segment.store.SegmentDirectory)2 ByteArrayOutputStream (java.io.ByteArrayOutputStream)2 DataOutputStream (java.io.DataOutputStream)2