Examples with OrcRecordReader com.facebook.presto.orc.OrcRecordReader used on opensource projects

Example 1 with OrcRecordReader

use of com.facebook.presto.orc.OrcRecordReader in project presto by prestodb.

the class OrcPageSourceFactory method createOrcPageSource.

public static OrcPageSource createOrcPageSource(MetadataReader metadataReader, HdfsEnvironment hdfsEnvironment, String sessionUser, Configuration configuration, Path path, long start, long length, List<HiveColumnHandle> columns, boolean useOrcColumnNames, TupleDomain<HiveColumnHandle> effectivePredicate, DateTimeZone hiveStorageTimeZone, TypeManager typeManager, DataSize maxMergeDistance, DataSize maxBufferSize, DataSize streamBufferSize, boolean orcBloomFiltersEnabled) {
    OrcDataSource orcDataSource;
    try {
        FileSystem fileSystem = hdfsEnvironment.getFileSystem(sessionUser, path, configuration);
        long size = fileSystem.getFileStatus(path).getLen();
        FSDataInputStream inputStream = fileSystem.open(path);
        orcDataSource = new HdfsOrcDataSource(path.toString(), size, maxMergeDistance, maxBufferSize, streamBufferSize, inputStream);
    } catch (Exception e) {
        if (nullToEmpty(e.getMessage()).trim().equals("Filesystem closed") || e instanceof FileNotFoundException) {
            throw new PrestoException(HIVE_CANNOT_OPEN_SPLIT, e);
        }
        throw new PrestoException(HIVE_CANNOT_OPEN_SPLIT, splitError(e, path, start, length), e);
    }
    AggregatedMemoryContext systemMemoryUsage = new AggregatedMemoryContext();
    try {
        OrcReader reader = new OrcReader(orcDataSource, metadataReader, maxMergeDistance, maxBufferSize);
        List<HiveColumnHandle> physicalColumns = getPhysicalHiveColumnHandles(columns, useOrcColumnNames, reader, path);
        ImmutableMap.Builder<Integer, Type> includedColumns = ImmutableMap.builder();
        ImmutableList.Builder<ColumnReference<HiveColumnHandle>> columnReferences = ImmutableList.builder();
        for (HiveColumnHandle column : physicalColumns) {
            if (column.getColumnType() == REGULAR) {
                Type type = typeManager.getType(column.getTypeSignature());
                includedColumns.put(column.getHiveColumnIndex(), type);
                columnReferences.add(new ColumnReference<>(column, column.getHiveColumnIndex(), type));
            }
        }
        OrcPredicate predicate = new TupleDomainOrcPredicate<>(effectivePredicate, columnReferences.build(), orcBloomFiltersEnabled);
        OrcRecordReader recordReader = reader.createRecordReader(includedColumns.build(), predicate, start, length, hiveStorageTimeZone, systemMemoryUsage);
        return new OrcPageSource(recordReader, orcDataSource, physicalColumns, typeManager, systemMemoryUsage);
    } catch (Exception e) {
        try {
            orcDataSource.close();
        } catch (IOException ignored) {
        }
        if (e instanceof PrestoException) {
            throw (PrestoException) e;
        }
        String message = splitError(e, path, start, length);
        if (e.getClass().getSimpleName().equals("BlockMissingException")) {
            throw new PrestoException(HIVE_MISSING_DATA, message, e);
        }
        throw new PrestoException(HIVE_CANNOT_OPEN_SPLIT, message, e);
    }
}

Example 2 with OrcRecordReader

use of com.facebook.presto.orc.OrcRecordReader in project presto by prestodb.

the class OrcStorageManager method getPageSource.

@Override
public ConnectorPageSource getPageSource(UUID shardUuid, OptionalInt bucketNumber, List<Long> columnIds, List<Type> columnTypes, TupleDomain<RaptorColumnHandle> effectivePredicate, ReaderAttributes readerAttributes, OptionalLong transactionId) {
    OrcDataSource dataSource = openShard(shardUuid, readerAttributes);
    AggregatedMemoryContext systemMemoryUsage = new AggregatedMemoryContext();
    try {
        OrcReader reader = new OrcReader(dataSource, new OrcMetadataReader(), readerAttributes.getMaxMergeDistance(), readerAttributes.getMaxReadSize());
        Map<Long, Integer> indexMap = columnIdIndex(reader.getColumnNames());
        ImmutableMap.Builder<Integer, Type> includedColumns = ImmutableMap.builder();
        ImmutableList.Builder<Integer> columnIndexes = ImmutableList.builder();
        for (int i = 0; i < columnIds.size(); i++) {
            long columnId = columnIds.get(i);
            if (isHiddenColumn(columnId)) {
                columnIndexes.add(toSpecialIndex(columnId));
                continue;
            }
            Integer index = indexMap.get(columnId);
            if (index == null) {
                columnIndexes.add(NULL_COLUMN);
            } else {
                columnIndexes.add(index);
                includedColumns.put(index, columnTypes.get(i));
            }
        }
        OrcPredicate predicate = getPredicate(effectivePredicate, indexMap);
        OrcRecordReader recordReader = reader.createRecordReader(includedColumns.build(), predicate, UTC, systemMemoryUsage);
        Optional<ShardRewriter> shardRewriter = Optional.empty();
        if (transactionId.isPresent()) {
            shardRewriter = Optional.of(createShardRewriter(transactionId.getAsLong(), bucketNumber, shardUuid));
        }
        return new OrcPageSource(shardRewriter, recordReader, dataSource, columnIds, columnTypes, columnIndexes.build(), shardUuid, bucketNumber, systemMemoryUsage);
    } catch (IOException | RuntimeException e) {
        closeQuietly(dataSource);
        throw new PrestoException(RAPTOR_ERROR, "Failed to create page source for shard " + shardUuid, e);
    } catch (Throwable t) {
        closeQuietly(dataSource);
        throw t;
    }
}

Example 3 with OrcRecordReader

use of com.facebook.presto.orc.OrcRecordReader in project presto by prestodb.

the class TestShardWriter method testWriterZeroRows.

@SuppressWarnings("EmptyTryBlock")
@Test
public void testWriterZeroRows() throws Exception {
    List<Long> columnIds = ImmutableList.of(1L);
    List<Type> columnTypes = ImmutableList.of(BIGINT);
    File file = new File(directory, System.nanoTime() + ".orc");
    try (OrcFileWriter ignored = new OrcFileWriter(columnIds, columnTypes, file)) {
    // no rows
    }
    try (OrcDataSource dataSource = fileOrcDataSource(file)) {
        OrcRecordReader reader = createReaderNoRows(dataSource);
        assertEquals(reader.getReaderRowCount(), 0);
        assertEquals(reader.getReaderPosition(), 0);
        assertEquals(reader.nextBatch(), -1);
    }
}

Example 4 with OrcRecordReader

use of com.facebook.presto.orc.OrcRecordReader in project presto by prestodb.

the class TestOrcFileRewriter method testRewriteWithoutMetadata.

@Test
public void testRewriteWithoutMetadata() throws Exception {
    List<Long> columnIds = ImmutableList.of(3L, 7L);
    List<Type> columnTypes = ImmutableList.of(BIGINT, createVarcharType(20));
    File file = new File(temporary, randomUUID().toString());
    try (OrcFileWriter writer = new OrcFileWriter(columnIds, columnTypes, file, false)) {
        List<Page> pages = rowPagesBuilder(columnTypes).row(123L, "hello").row(777L, "sky").build();
        writer.appendPages(pages);
    }
    try (OrcDataSource dataSource = fileOrcDataSource(file)) {
        OrcRecordReader reader = createReader(dataSource, columnIds, columnTypes);
        assertEquals(reader.getReaderRowCount(), 2);
        assertEquals(reader.getFileRowCount(), 2);
        assertEquals(reader.getSplitLength(), file.length());
        assertEquals(reader.nextBatch(), 2);
        Block column0 = reader.readBlock(BIGINT, 0);
        assertEquals(column0.getPositionCount(), 2);
        for (int i = 0; i < 2; i++) {
            assertEquals(column0.isNull(i), false);
        }
        assertEquals(BIGINT.getLong(column0, 0), 123L);
        assertEquals(BIGINT.getLong(column0, 1), 777L);
        Block column1 = reader.readBlock(createVarcharType(20), 1);
        assertEquals(column1.getPositionCount(), 2);
        for (int i = 0; i < 2; i++) {
            assertEquals(column1.isNull(i), false);
        }
        assertEquals(createVarcharType(20).getSlice(column1, 0), utf8Slice("hello"));
        assertEquals(createVarcharType(20).getSlice(column1, 1), utf8Slice("sky"));
        assertFalse(reader.getUserMetadata().containsKey(OrcFileMetadata.KEY));
    }
    BitSet rowsToDelete = new BitSet(5);
    rowsToDelete.set(1);
    File newFile = new File(temporary, randomUUID().toString());
    OrcFileInfo info = OrcFileRewriter.rewrite(file, newFile, rowsToDelete);
    assertEquals(info.getRowCount(), 1);
    assertEquals(info.getUncompressedSize(), 13);
    try (OrcDataSource dataSource = fileOrcDataSource(newFile)) {
        OrcRecordReader reader = createReader(dataSource, columnIds, columnTypes);
        assertEquals(reader.getReaderRowCount(), 1);
        assertEquals(reader.getFileRowCount(), 1);
        assertEquals(reader.getSplitLength(), newFile.length());
        assertEquals(reader.nextBatch(), 1);
        Block column0 = reader.readBlock(BIGINT, 0);
        assertEquals(column0.getPositionCount(), 1);
        assertEquals(column0.isNull(0), false);
        assertEquals(BIGINT.getLong(column0, 0), 123L);
        Block column1 = reader.readBlock(createVarcharType(20), 1);
        assertEquals(column1.getPositionCount(), 1);
        assertEquals(column1.isNull(0), false);
        assertEquals(createVarcharType(20).getSlice(column1, 0), utf8Slice("hello"));
        assertFalse(reader.getUserMetadata().containsKey(OrcFileMetadata.KEY));
    }
}

Example 5 with OrcRecordReader

use of com.facebook.presto.orc.OrcRecordReader in project presto by prestodb.

the class TestOrcFileRewriter method testRewrite.

@Test
public void testRewrite() throws Exception {
    ArrayType arrayType = new ArrayType(BIGINT);
    ArrayType arrayOfArrayType = new ArrayType(arrayType);
    MapType mapType = new MapType(createVarcharType(5), BOOLEAN);
    List<Long> columnIds = ImmutableList.of(3L, 7L, 9L, 10L, 11L);
    List<Type> columnTypes = ImmutableList.of(BIGINT, createVarcharType(20), arrayType, mapType, arrayOfArrayType);
    File file = new File(temporary, randomUUID().toString());
    try (OrcFileWriter writer = new OrcFileWriter(columnIds, columnTypes, file)) {
        List<Page> pages = rowPagesBuilder(columnTypes).row(123L, "hello", arrayBlockOf(BIGINT, 1, 2), mapBlockOf(createVarcharType(5), BOOLEAN, "k1", true), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 5))).row(777L, "sky", arrayBlockOf(BIGINT, 3, 4), mapBlockOf(createVarcharType(5), BOOLEAN, "k2", false), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 6))).row(456L, "bye", arrayBlockOf(BIGINT, 5, 6), mapBlockOf(createVarcharType(5), BOOLEAN, "k3", true), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 7))).row(888L, "world", arrayBlockOf(BIGINT, 7, 8), mapBlockOf(createVarcharType(5), BOOLEAN, "k4", true), arrayBlockOf(arrayType, null, arrayBlockOf(BIGINT, 8), null)).row(999L, "done", arrayBlockOf(BIGINT, 9, 10), mapBlockOf(createVarcharType(5), BOOLEAN, "k5", true), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 9, 10))).build();
        writer.appendPages(pages);
    }
    try (OrcDataSource dataSource = fileOrcDataSource(file)) {
        OrcRecordReader reader = createReader(dataSource, columnIds, columnTypes);
        assertEquals(reader.getReaderRowCount(), 5);
        assertEquals(reader.getFileRowCount(), 5);
        assertEquals(reader.getSplitLength(), file.length());
        assertEquals(reader.nextBatch(), 5);
        Block column0 = reader.readBlock(BIGINT, 0);
        assertEquals(column0.getPositionCount(), 5);
        for (int i = 0; i < 5; i++) {
            assertEquals(column0.isNull(i), false);
        }
        assertEquals(BIGINT.getLong(column0, 0), 123L);
        assertEquals(BIGINT.getLong(column0, 1), 777L);
        assertEquals(BIGINT.getLong(column0, 2), 456L);
        assertEquals(BIGINT.getLong(column0, 3), 888L);
        assertEquals(BIGINT.getLong(column0, 4), 999L);
        Block column1 = reader.readBlock(createVarcharType(20), 1);
        assertEquals(column1.getPositionCount(), 5);
        for (int i = 0; i < 5; i++) {
            assertEquals(column1.isNull(i), false);
        }
        assertEquals(createVarcharType(20).getSlice(column1, 0), utf8Slice("hello"));
        assertEquals(createVarcharType(20).getSlice(column1, 1), utf8Slice("sky"));
        assertEquals(createVarcharType(20).getSlice(column1, 2), utf8Slice("bye"));
        assertEquals(createVarcharType(20).getSlice(column1, 3), utf8Slice("world"));
        assertEquals(createVarcharType(20).getSlice(column1, 4), utf8Slice("done"));
        Block column2 = reader.readBlock(arrayType, 2);
        assertEquals(column2.getPositionCount(), 5);
        for (int i = 0; i < 5; i++) {
            assertEquals(column2.isNull(i), false);
        }
        assertTrue(arrayBlocksEqual(BIGINT, arrayType.getObject(column2, 0), arrayBlockOf(BIGINT, 1, 2)));
        assertTrue(arrayBlocksEqual(BIGINT, arrayType.getObject(column2, 1), arrayBlockOf(BIGINT, 3, 4)));
        assertTrue(arrayBlocksEqual(BIGINT, arrayType.getObject(column2, 2), arrayBlockOf(BIGINT, 5, 6)));
        assertTrue(arrayBlocksEqual(BIGINT, arrayType.getObject(column2, 3), arrayBlockOf(BIGINT, 7, 8)));
        assertTrue(arrayBlocksEqual(BIGINT, arrayType.getObject(column2, 4), arrayBlockOf(BIGINT, 9, 10)));
        Block column3 = reader.readBlock(mapType, 3);
        assertEquals(column3.getPositionCount(), 5);
        for (int i = 0; i < 5; i++) {
            assertEquals(column3.isNull(i), false);
        }
        assertTrue(mapBlocksEqual(createVarcharType(5), BOOLEAN, arrayType.getObject(column3, 0), mapBlockOf(createVarcharType(5), BOOLEAN, "k1", true)));
        assertTrue(mapBlocksEqual(createVarcharType(5), BOOLEAN, arrayType.getObject(column3, 1), mapBlockOf(createVarcharType(5), BOOLEAN, "k2", false)));
        assertTrue(mapBlocksEqual(createVarcharType(5), BOOLEAN, arrayType.getObject(column3, 2), mapBlockOf(createVarcharType(5), BOOLEAN, "k3", true)));
        assertTrue(mapBlocksEqual(createVarcharType(5), BOOLEAN, arrayType.getObject(column3, 3), mapBlockOf(createVarcharType(5), BOOLEAN, "k4", true)));
        assertTrue(mapBlocksEqual(createVarcharType(5), BOOLEAN, arrayType.getObject(column3, 4), mapBlockOf(createVarcharType(5), BOOLEAN, "k5", true)));
        Block column4 = reader.readBlock(arrayOfArrayType, 4);
        assertEquals(column4.getPositionCount(), 5);
        for (int i = 0; i < 5; i++) {
            assertEquals(column4.isNull(i), false);
        }
        assertTrue(arrayBlocksEqual(arrayType, arrayOfArrayType.getObject(column4, 0), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 5))));
        assertTrue(arrayBlocksEqual(arrayType, arrayOfArrayType.getObject(column4, 1), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 6))));
        assertTrue(arrayBlocksEqual(arrayType, arrayOfArrayType.getObject(column4, 2), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 7))));
        assertTrue(arrayBlocksEqual(arrayType, arrayOfArrayType.getObject(column4, 3), arrayBlockOf(arrayType, null, arrayBlockOf(BIGINT, 8), null)));
        assertTrue(arrayBlocksEqual(arrayType, arrayOfArrayType.getObject(column4, 4), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 9, 10))));
        assertEquals(reader.nextBatch(), -1);
        OrcFileMetadata orcFileMetadata = METADATA_CODEC.fromJson(reader.getUserMetadata().get(OrcFileMetadata.KEY).getBytes());
        assertEquals(orcFileMetadata, new OrcFileMetadata(ImmutableMap.<Long, TypeSignature>builder().put(3L, BIGINT.getTypeSignature()).put(7L, createVarcharType(20).getTypeSignature()).put(9L, arrayType.getTypeSignature()).put(10L, mapType.getTypeSignature()).put(11L, arrayOfArrayType.getTypeSignature()).build()));
    }
    BitSet rowsToDelete = new BitSet(5);
    rowsToDelete.set(1);
    rowsToDelete.set(3);
    rowsToDelete.set(4);
    File newFile = new File(temporary, randomUUID().toString());
    OrcFileInfo info = OrcFileRewriter.rewrite(file, newFile, rowsToDelete);
    assertEquals(info.getRowCount(), 2);
    assertEquals(info.getUncompressedSize(), 78);
    try (OrcDataSource dataSource = fileOrcDataSource(newFile)) {
        OrcRecordReader reader = createReader(dataSource, columnIds, columnTypes);
        assertEquals(reader.getReaderRowCount(), 2);
        assertEquals(reader.getFileRowCount(), 2);
        assertEquals(reader.getSplitLength(), newFile.length());
        assertEquals(reader.nextBatch(), 2);
        Block column0 = reader.readBlock(BIGINT, 0);
        assertEquals(column0.getPositionCount(), 2);
        for (int i = 0; i < 2; i++) {
            assertEquals(column0.isNull(i), false);
        }
        assertEquals(BIGINT.getLong(column0, 0), 123L);
        assertEquals(BIGINT.getLong(column0, 1), 456L);
        Block column1 = reader.readBlock(createVarcharType(20), 1);
        assertEquals(column1.getPositionCount(), 2);
        for (int i = 0; i < 2; i++) {
            assertEquals(column1.isNull(i), false);
        }
        assertEquals(createVarcharType(20).getSlice(column1, 0), utf8Slice("hello"));
        assertEquals(createVarcharType(20).getSlice(column1, 1), utf8Slice("bye"));
        Block column2 = reader.readBlock(arrayType, 2);
        assertEquals(column2.getPositionCount(), 2);
        for (int i = 0; i < 2; i++) {
            assertEquals(column2.isNull(i), false);
        }
        assertTrue(arrayBlocksEqual(BIGINT, arrayType.getObject(column2, 0), arrayBlockOf(BIGINT, 1, 2)));
        assertTrue(arrayBlocksEqual(BIGINT, arrayType.getObject(column2, 1), arrayBlockOf(BIGINT, 5, 6)));
        Block column3 = reader.readBlock(mapType, 3);
        assertEquals(column3.getPositionCount(), 2);
        for (int i = 0; i < 2; i++) {
            assertEquals(column3.isNull(i), false);
        }
        assertTrue(mapBlocksEqual(createVarcharType(5), BOOLEAN, arrayType.getObject(column3, 0), mapBlockOf(createVarcharType(5), BOOLEAN, "k1", true)));
        assertTrue(mapBlocksEqual(createVarcharType(5), BOOLEAN, arrayType.getObject(column3, 1), mapBlockOf(createVarcharType(5), BOOLEAN, "k3", true)));
        Block column4 = reader.readBlock(arrayOfArrayType, 4);
        assertEquals(column4.getPositionCount(), 2);
        for (int i = 0; i < 2; i++) {
            assertEquals(column4.isNull(i), false);
        }
        assertTrue(arrayBlocksEqual(arrayType, arrayOfArrayType.getObject(column4, 0), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 5))));
        assertTrue(arrayBlocksEqual(arrayType, arrayOfArrayType.getObject(column4, 1), arrayBlockOf(arrayType, arrayBlockOf(BIGINT, 7))));
        assertEquals(reader.nextBatch(), -1);
        OrcFileMetadata orcFileMetadata = METADATA_CODEC.fromJson(reader.getUserMetadata().get(OrcFileMetadata.KEY).getBytes());
        assertEquals(orcFileMetadata, new OrcFileMetadata(ImmutableMap.<Long, TypeSignature>builder().put(3L, BIGINT.getTypeSignature()).put(7L, createVarcharType(20).getTypeSignature()).put(9L, arrayType.getTypeSignature()).put(10L, mapType.getTypeSignature()).put(11L, arrayOfArrayType.getTypeSignature()).build()));
    }
}