Examples with StreamDataOutput com.facebook.presto.orc.stream.StreamDataOutput used on opensource projects

Example 1 with StreamDataOutput

use of com.facebook.presto.orc.stream.StreamDataOutput in project presto by prestodb.

the class ListColumnWriter method getIndexStreams.

@Override
public List<StreamDataOutput> getIndexStreams() throws IOException {
    checkState(closed);
    ImmutableList.Builder<RowGroupIndex> rowGroupIndexes = ImmutableList.builder();
    List<LongStreamCheckpoint> lengthCheckpoints = lengthStream.getCheckpoints();
    Optional<List<BooleanStreamCheckpoint>> presentCheckpoints = presentStream.getCheckpoints();
    for (int i = 0; i < rowGroupColumnStatistics.size(); i++) {
        int groupId = i;
        ColumnStatistics columnStatistics = rowGroupColumnStatistics.get(groupId);
        LongStreamCheckpoint lengthCheckpoint = lengthCheckpoints.get(groupId);
        Optional<BooleanStreamCheckpoint> presentCheckpoint = presentCheckpoints.map(checkpoints -> checkpoints.get(groupId));
        List<Integer> positions = createArrayColumnPositionList(compressed, lengthCheckpoint, presentCheckpoint);
        rowGroupIndexes.add(new RowGroupIndex(positions, columnStatistics));
    }
    Slice slice = metadataWriter.writeRowIndexes(rowGroupIndexes.build());
    Stream stream = new Stream(column, StreamKind.ROW_INDEX, slice.length(), false);
    ImmutableList.Builder<StreamDataOutput> indexStreams = ImmutableList.builder();
    indexStreams.add(new StreamDataOutput(slice, stream));
    indexStreams.addAll(elementWriter.getIndexStreams());
    return indexStreams.build();
}

Example 2 with StreamDataOutput

use of com.facebook.presto.orc.stream.StreamDataOutput in project presto by prestodb.

the class MapColumnWriter method getIndexStreams.

@Override
public List<StreamDataOutput> getIndexStreams() throws IOException {
    checkState(closed);
    ImmutableList.Builder<RowGroupIndex> rowGroupIndexes = ImmutableList.builder();
    List<LongStreamCheckpoint> lengthCheckpoints = lengthStream.getCheckpoints();
    Optional<List<BooleanStreamCheckpoint>> presentCheckpoints = presentStream.getCheckpoints();
    for (int i = 0; i < rowGroupColumnStatistics.size(); i++) {
        int groupId = i;
        ColumnStatistics columnStatistics = rowGroupColumnStatistics.get(groupId);
        LongStreamCheckpoint lengthCheckpoint = lengthCheckpoints.get(groupId);
        Optional<BooleanStreamCheckpoint> presentCheckpoint = presentCheckpoints.map(checkpoints -> checkpoints.get(groupId));
        List<Integer> positions = createArrayColumnPositionList(compressed, lengthCheckpoint, presentCheckpoint);
        rowGroupIndexes.add(new RowGroupIndex(positions, columnStatistics));
    }
    Slice slice = metadataWriter.writeRowIndexes(rowGroupIndexes.build());
    Stream stream = new Stream(column, StreamKind.ROW_INDEX, slice.length(), false);
    ImmutableList.Builder<StreamDataOutput> indexStreams = ImmutableList.builder();
    indexStreams.add(new StreamDataOutput(slice, stream));
    indexStreams.addAll(keyWriter.getIndexStreams());
    indexStreams.addAll(valueWriter.getIndexStreams());
    return indexStreams.build();
}

Example 3 with StreamDataOutput

use of com.facebook.presto.orc.stream.StreamDataOutput in project presto by prestodb.

the class SliceDirectColumnWriter method getIndexStreams.

@Override
public List<StreamDataOutput> getIndexStreams() throws IOException {
    checkState(closed);
    ImmutableList.Builder<RowGroupIndex> rowGroupIndexes = ImmutableList.builder();
    List<LongStreamCheckpoint> lengthCheckpoints = lengthStream.getCheckpoints();
    List<ByteArrayStreamCheckpoint> dataCheckpoints = dataStream.getCheckpoints();
    Optional<List<BooleanStreamCheckpoint>> presentCheckpoints = presentStream.getCheckpoints();
    for (int i = 0; i < rowGroupColumnStatistics.size(); i++) {
        int groupId = i;
        ColumnStatistics columnStatistics = rowGroupColumnStatistics.get(groupId);
        LongStreamCheckpoint lengthCheckpoint = lengthCheckpoints.get(groupId);
        ByteArrayStreamCheckpoint dataCheckpoint = dataCheckpoints.get(groupId);
        Optional<BooleanStreamCheckpoint> presentCheckpoint = presentCheckpoints.map(checkpoints -> checkpoints.get(groupId));
        List<Integer> positions = createSliceColumnPositionList(compressed, lengthCheckpoint, dataCheckpoint, presentCheckpoint);
        rowGroupIndexes.add(new RowGroupIndex(positions, columnStatistics));
    }
    Slice slice = metadataWriter.writeRowIndexes(rowGroupIndexes.build());
    Stream stream = new Stream(column, StreamKind.ROW_INDEX, slice.length(), false);
    return ImmutableList.of(new StreamDataOutput(slice, stream));
}

Example 4 with StreamDataOutput

use of com.facebook.presto.orc.stream.StreamDataOutput in project presto by prestodb.

the class ByteColumnWriter method getIndexStreams.

@Override
public List<StreamDataOutput> getIndexStreams() throws IOException {
    checkState(closed);
    ImmutableList.Builder<RowGroupIndex> rowGroupIndexes = ImmutableList.builder();
    List<ByteStreamCheckpoint> dataCheckpoints = dataStream.getCheckpoints();
    Optional<List<BooleanStreamCheckpoint>> presentCheckpoints = presentStream.getCheckpoints();
    for (int i = 0; i < rowGroupColumnStatistics.size(); i++) {
        int groupId = i;
        ColumnStatistics columnStatistics = rowGroupColumnStatistics.get(groupId);
        ByteStreamCheckpoint dataCheckpoint = dataCheckpoints.get(groupId);
        Optional<BooleanStreamCheckpoint> presentCheckpoint = presentCheckpoints.map(checkpoints -> checkpoints.get(groupId));
        List<Integer> positions = createByteColumnPositionList(compressed, dataCheckpoint, presentCheckpoint);
        rowGroupIndexes.add(new RowGroupIndex(positions, columnStatistics));
    }
    Slice slice = metadataWriter.writeRowIndexes(rowGroupIndexes.build());
    Stream stream = new Stream(column, StreamKind.ROW_INDEX, slice.length(), false);
    return ImmutableList.of(new StreamDataOutput(slice, stream));
}

Example 5 with StreamDataOutput

use of com.facebook.presto.orc.stream.StreamDataOutput in project presto by prestodb.

the class TestSliceDictionaryColumnWriter method getDictionaryKeys.

private List<String> getDictionaryKeys(List<String> values, OrcEncoding orcEncoding, boolean sortDictionaryKeys) throws IOException {
    DictionaryColumnWriter writer = getDictionaryColumnWriter(orcEncoding, sortDictionaryKeys);
    for (int index = 0; index < values.size(); ) {
        int endIndex = Math.min(index + 10_000, values.size());
        BlockBuilder blockBuilder = VARCHAR.createBlockBuilder(null, 10_000);
        while (index < endIndex) {
            VARCHAR.writeSlice(blockBuilder, utf8Slice(values.get(index++)));
        }
        writer.beginRowGroup();
        writer.writeBlock(blockBuilder);
        writer.finishRowGroup();
    }
    writer.close();
    List<StreamDataOutput> streams = writer.getDataStreams();
    int dictionarySize = writer.getColumnEncodings().get(COLUMN_ID).getDictionarySize();
    ByteArrayInputStream dictionaryDataStream = new ByteArrayInputStream(getOrcInputStream(streams, DICTIONARY_DATA));
    LongInputStream dictionaryLengthStream = getDictionaryLengthStream(streams, orcEncoding);
    List<String> dictionaryKeys = new ArrayList<>(dictionarySize);
    for (int i = 0; i < dictionarySize; i++) {
        int length = toIntExact(dictionaryLengthStream.next());
        String dictionaryKey = new String(dictionaryDataStream.next(length), UTF_8);
        dictionaryKeys.add(dictionaryKey);
    }
    return dictionaryKeys;
}