Examples with ColumnMetadata io.prestosql.orc.metadata.ColumnMetadata used on opensource projects

Example 1 with ColumnMetadata

use of io.prestosql.orc.metadata.ColumnMetadata in project hetu-core by openlookeng.

the class StripeReader method getRowGroupStatistics.

private static ColumnMetadata<ColumnStatistics> getRowGroupStatistics(ColumnMetadata<OrcType> types, Map<StreamId, List<RowGroupIndex>> columnIndexes, int rowGroup) {
    requireNonNull(columnIndexes, "columnIndexes is null");
    checkArgument(rowGroup >= 0, "rowGroup is negative");
    Map<Integer, List<RowGroupIndex>> rowGroupIndexesByColumn = columnIndexes.entrySet().stream().collect(toImmutableMap(entry -> entry.getKey().getColumnId().getId(), Entry::getValue));
    List<ColumnStatistics> statistics = new ArrayList<>(types.size());
    for (int columnIndex = 0; columnIndex < types.size(); columnIndex++) {
        List<RowGroupIndex> rowGroupIndexes = rowGroupIndexesByColumn.get(columnIndex);
        if (rowGroupIndexes != null) {
            statistics.add(rowGroupIndexes.get(rowGroup).getColumnStatistics());
        } else {
            statistics.add(null);
        }
    }
    return new ColumnMetadata<>(statistics);
}

Example 2 with ColumnMetadata

use of io.prestosql.orc.metadata.ColumnMetadata in project hetu-core by openlookeng.

the class AbstractOrcRecordReader method close.

@Override
public void close() throws IOException {
    try (Closer closer = Closer.create()) {
        closer.register(orcDataSource);
        for (AbstractColumnReader column : columnReaders) {
            if (column != null) {
                closer.register(column::close);
            }
        }
    }
    if (writeChecksumBuilder.isPresent()) {
        OrcWriteValidation.WriteChecksum actualChecksum = writeChecksumBuilder.get().build();
        validateWrite(validation -> validation.getChecksum().getTotalRowCount() == actualChecksum.getTotalRowCount(), "Invalid row count");
        List<Long> columnHashes = actualChecksum.getColumnHashes();
        for (int i = 0; i < columnHashes.size(); i++) {
            int columnIndex = i;
            validateWrite(validation -> validation.getChecksum().getColumnHashes().get(columnIndex).equals(columnHashes.get(columnIndex)), "Invalid checksum for column %s", columnIndex);
        }
        validateWrite(validation -> validation.getChecksum().getStripeHash() == actualChecksum.getStripeHash(), "Invalid stripes checksum");
    }
    if (fileStatisticsValidation.isPresent()) {
        Optional<ColumnMetadata<ColumnStatistics>> columnStatistics = fileStatisticsValidation.get().build();
        writeValidation.get().validateFileStatistics(orcDataSource.getId(), columnStatistics);
    }
}

Example 3 with ColumnMetadata

use of io.prestosql.orc.metadata.ColumnMetadata in project hetu-core by openlookeng.

the class TestOrcBloomFilters method testMatches.

@Test
public // simulate query on a 2 columns where 1 is used as part of the where, with and without bloom filter
void testMatches() {
    TupleDomainOrcPredicate predicate = TupleDomainOrcPredicate.builder().setBloomFiltersEnabled(true).addColumn(ROOT_COLUMN, Domain.singleValue(BIGINT, 1234L)).build();
    TupleDomainOrcPredicate emptyPredicate = TupleDomainOrcPredicate.builder().build();
    // assemble a matching and a non-matching bloom filter
    HashableBloomFilter bloomFilter = new HashableBloomFilter(1000, 0.01);
    OrcProto.BloomFilter emptyOrcBloomFilter = toOrcBloomFilter(bloomFilter);
    bloomFilter.addLong(1234);
    OrcProto.BloomFilter orcBloomFilter = toOrcBloomFilter(bloomFilter);
    ColumnMetadata<ColumnStatistics> matchingStatisticsByColumnIndex = new ColumnMetadata<>(ImmutableList.of(new ColumnStatistics(null, 0, null, new IntegerStatistics(10L, 2000L, null), null, null, null, null, null, toBloomFilter(orcBloomFilter))));
    ColumnMetadata<ColumnStatistics> nonMatchingStatisticsByColumnIndex = new ColumnMetadata<>(ImmutableList.of(new ColumnStatistics(null, 0, null, new IntegerStatistics(10L, 2000L, null), null, null, null, null, null, toBloomFilter(emptyOrcBloomFilter))));
    ColumnMetadata<ColumnStatistics> withoutBloomFilterStatisticsByColumnIndex = new ColumnMetadata<>(ImmutableList.of(new ColumnStatistics(null, 0, null, new IntegerStatistics(10L, 2000L, null), null, null, null, null, null, null)));
    assertTrue(predicate.matches(1L, matchingStatisticsByColumnIndex));
    assertTrue(predicate.matches(1L, withoutBloomFilterStatisticsByColumnIndex));
    assertFalse(predicate.matches(1L, nonMatchingStatisticsByColumnIndex));
    assertTrue(emptyPredicate.matches(1L, matchingStatisticsByColumnIndex));
}

Example 4 with ColumnMetadata

use of io.prestosql.orc.metadata.ColumnMetadata in project hetu-core by openlookeng.

the class OrcWriter method toFileStats.

private static Optional<ColumnMetadata<ColumnStatistics>> toFileStats(List<ColumnMetadata<ColumnStatistics>> stripes) {
    if (stripes.isEmpty()) {
        return Optional.empty();
    }
    int columnCount = stripes.get(0).size();
    checkArgument(stripes.stream().allMatch(stripe -> columnCount == stripe.size()));
    ImmutableList.Builder<ColumnStatistics> fileStats = ImmutableList.builder();
    for (int i = 0; i < columnCount; i++) {
        OrcColumnId columnId = new OrcColumnId(i);
        fileStats.add(ColumnStatistics.mergeColumnStatistics(stripes.stream().map(stripe -> stripe.get(columnId)).collect(toList())));
    }
    return Optional.of(new ColumnMetadata<>(fileStats.build()));
}

Example 5 with ColumnMetadata

use of io.prestosql.orc.metadata.ColumnMetadata in project hetu-core by openlookeng.

the class OrcWriter method bufferFileFooter.

/**
 * Collect the data for for the file footer.  This is not the actual data, but
 * instead are functions that know how to write the data.
 */
private List<OrcDataOutput> bufferFileFooter() throws IOException {
    if (preCloseCallback.isPresent()) {
        try {
            preCloseCallback.get().call();
        } catch (Exception e) {
            log.debug("Call pre close call back error");
        }
    }
    List<OrcDataOutput> outputData = new ArrayList<>();
    Metadata metadata = new Metadata(closedStripes.stream().map(ClosedStripe::getStatistics).map(Optional::of).collect(toList()));
    Slice metadataSlice = metadataWriter.writeMetadata(metadata);
    outputData.add(createDataOutput(metadataSlice));
    long numberOfRows = closedStripes.stream().mapToLong(stripe -> stripe.getStripeInformation().getNumberOfRows()).sum();
    Optional<ColumnMetadata<ColumnStatistics>> fileStats = toFileStats(closedStripes.stream().map(ClosedStripe::getStatistics).map(StripeStatistics::getColumnStatistics).collect(toList()));
    recordValidation(validation -> validation.setFileStatistics(fileStats));
    Map<String, Slice> localUserMetadata = this.userMetadata.entrySet().stream().collect(Collectors.toMap(Entry::getKey, entry -> utf8Slice(entry.getValue())));
    Footer footer = new Footer(numberOfRows, rowGroupMaxRowCount, closedStripes.stream().map(ClosedStripe::getStripeInformation).collect(toImmutableList()), orcTypes, fileStats, localUserMetadata);
    closedStripes.clear();
    closedStripesRetainedBytes = 0;
    Slice footerSlice = metadataWriter.writeFooter(footer);
    outputData.add(createDataOutput(footerSlice));
    recordValidation(validation -> validation.setVersion(metadataWriter.getOrcMetadataVersion()));
    Slice postscriptSlice = metadataWriter.writePostscript(footerSlice.length(), metadataSlice.length(), compression, maxCompressionBufferSize);
    outputData.add(createDataOutput(postscriptSlice));
    outputData.add(createDataOutput(Slices.wrappedBuffer(UnsignedBytes.checkedCast(postscriptSlice.length()))));
    return outputData;
}