Examples with FUNCTION_AND_TYPE_MANAGER com.facebook.presto.hive.HiveTestUtils.FUNCTION_AND_TYPE_MANAGER used on opensource projects

Example 1 with FUNCTION_AND_TYPE_MANAGER

use of com.facebook.presto.hive.HiveTestUtils.FUNCTION_AND_TYPE_MANAGER in project presto by prestodb.

the class TestHiveFileFormats method testRcBinaryPageSource.

@Test(dataProvider = "rowCount")
public void testRcBinaryPageSource(int rowCount) throws Exception {
    // RCBinary does not support complex type as key of a map and interprets empty VARCHAR as nulls
    List<TestColumn> testColumns = TEST_COLUMNS.stream().filter(testColumn -> !testColumn.getName().equals("t_empty_varchar")).collect(toList());
    assertThatFileFormat(RCBINARY).withColumns(testColumns).withRowsCount(rowCount).isReadableByPageSource(new RcFilePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, HDFS_ENVIRONMENT, STATS));
}

Example 2 with FUNCTION_AND_TYPE_MANAGER

use of com.facebook.presto.hive.HiveTestUtils.FUNCTION_AND_TYPE_MANAGER in project presto by prestodb.

the class TestHiveFileFormats method testDwrfOptimizedWriter.

@Test(dataProvider = "rowCount")
public void testDwrfOptimizedWriter(int rowCount) throws Exception {
    TestingConnectorSession session = new TestingConnectorSession(new HiveSessionProperties(new HiveClientConfig().setOrcOptimizedWriterEnabled(true).setOrcWriterValidationPercentage(100.0), new OrcFileWriterConfig(), new ParquetFileWriterConfig(), new CacheConfig()).getSessionProperties());
    // DWRF does not support modern Hive types
    // A Presto page can not contain a map with null keys, so a page based writer can not write null keys
    List<TestColumn> testColumns = TEST_COLUMNS.stream().filter(testColumn -> !hasType(testColumn.getObjectInspector(), PrimitiveCategory.DATE, PrimitiveCategory.VARCHAR, PrimitiveCategory.CHAR, PrimitiveCategory.DECIMAL)).filter(testColumn -> !testColumn.getName().equals("t_map_null_key") && !testColumn.getName().equals("t_map_null_key_complex_value") && !testColumn.getName().equals("t_map_null_key_complex_key_value")).collect(toList());
    assertThatFileFormat(DWRF).withColumns(testColumns).withRowsCount(rowCount).withSession(session).withFileWriterFactory(new OrcFileWriterFactory(HDFS_ENVIRONMENT, new OutputStreamDataSinkFactory(), FUNCTION_AND_TYPE_MANAGER, new NodeVersion("test"), HIVE_STORAGE_TIME_ZONE, STATS, new OrcFileWriterConfig(), NO_ENCRYPTION)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT)).isReadableByPageSource(new DwrfBatchPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, HIVE_CLIENT_CONFIG, HDFS_ENVIRONMENT, STATS, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource()), NO_ENCRYPTION));
}

Example 3 with FUNCTION_AND_TYPE_MANAGER

use of com.facebook.presto.hive.HiveTestUtils.FUNCTION_AND_TYPE_MANAGER in project presto by prestodb.

the class TestHiveFileFormats method testOrcOptimizedWriter.

@Test(dataProvider = "rowCount")
public void testOrcOptimizedWriter(int rowCount) throws Exception {
    TestingConnectorSession session = new TestingConnectorSession(new HiveSessionProperties(new HiveClientConfig().setOrcOptimizedWriterEnabled(true).setOrcWriterValidationPercentage(100.0), new OrcFileWriterConfig(), new ParquetFileWriterConfig(), new CacheConfig()).getSessionProperties());
    // A Presto page can not contain a map with null keys, so a page based writer can not write null keys
    List<TestColumn> testColumns = TEST_COLUMNS.stream().filter(testColumn -> !testColumn.getName().equals("t_map_null_key") && !testColumn.getName().equals("t_map_null_key_complex_value") && !testColumn.getName().equals("t_map_null_key_complex_key_value")).collect(toList());
    assertThatFileFormat(ORC).withColumns(testColumns).withRowsCount(rowCount).withSession(session).withFileWriterFactory(new OrcFileWriterFactory(HDFS_ENVIRONMENT, new OutputStreamDataSinkFactory(), FUNCTION_AND_TYPE_MANAGER, new NodeVersion("test"), HIVE_STORAGE_TIME_ZONE, STATS, new OrcFileWriterConfig(), NO_ENCRYPTION)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT)).isReadableByPageSource(new OrcBatchPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, false, HDFS_ENVIRONMENT, STATS, 100, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource())));
}

Example 4 with FUNCTION_AND_TYPE_MANAGER

use of com.facebook.presto.hive.HiveTestUtils.FUNCTION_AND_TYPE_MANAGER in project presto by prestodb.

the class TestHivePageSink method writeTestFile.

private static long writeTestFile(HiveClientConfig config, MetastoreClientConfig metastoreClientConfig, ExtendedHiveMetastore metastore, String outputPath) {
    HiveTransactionHandle transaction = new HiveTransactionHandle();
    HiveWriterStats stats = new HiveWriterStats();
    ConnectorPageSink pageSink = createPageSink(transaction, config, metastoreClientConfig, metastore, new Path("file:///" + outputPath), stats);
    List<LineItemColumn> columns = getTestColumns();
    List<Type> columnTypes = columns.stream().map(LineItemColumn::getType).map(TestHivePageSink::getHiveType).map(hiveType -> hiveType.getType(FUNCTION_AND_TYPE_MANAGER)).collect(toList());
    PageBuilder pageBuilder = new PageBuilder(columnTypes);
    int rows = 0;
    for (LineItem lineItem : new LineItemGenerator(0.01, 1, 1)) {
        rows++;
        if (rows >= NUM_ROWS) {
            break;
        }
        pageBuilder.declarePosition();
        for (int i = 0; i < columns.size(); i++) {
            LineItemColumn column = columns.get(i);
            BlockBuilder blockBuilder = pageBuilder.getBlockBuilder(i);
            switch(column.getType().getBase()) {
                case IDENTIFIER:
                    BIGINT.writeLong(blockBuilder, column.getIdentifier(lineItem));
                    break;
                case INTEGER:
                    INTEGER.writeLong(blockBuilder, column.getInteger(lineItem));
                    break;
                case DATE:
                    DATE.writeLong(blockBuilder, column.getDate(lineItem));
                    break;
                case DOUBLE:
                    DOUBLE.writeDouble(blockBuilder, column.getDouble(lineItem));
                    break;
                case VARCHAR:
                    createUnboundedVarcharType().writeSlice(blockBuilder, Slices.utf8Slice(column.getString(lineItem)));
                    break;
                default:
                    throw new IllegalArgumentException("Unsupported type " + column.getType());
            }
        }
    }
    Page page = pageBuilder.build();
    pageSink.appendPage(page);
    getFutureValue(pageSink.finish());
    File outputDir = new File(outputPath);
    List<File> files = ImmutableList.copyOf(outputDir.listFiles((dir, name) -> !name.endsWith(".crc")));
    File outputFile = getOnlyElement(files);
    long length = outputFile.length();
    ConnectorPageSource pageSource = createPageSource(transaction, config, metastoreClientConfig, outputFile);
    List<Page> pages = new ArrayList<>();
    while (!pageSource.isFinished()) {
        Page nextPage = pageSource.getNextPage();
        if (nextPage != null) {
            pages.add(nextPage.getLoadedPage());
        }
    }
    MaterializedResult expectedResults = toMaterializedResult(getSession(config), columnTypes, ImmutableList.of(page));
    MaterializedResult results = toMaterializedResult(getSession(config), columnTypes, pages);
    assertEquals(results, expectedResults);
    assertEquals(stats.getInputPageSizeInBytes().getAllTime().getMax(), page.getRetainedSizeInBytes());
    return length;
}

Example 5 with FUNCTION_AND_TYPE_MANAGER

use of com.facebook.presto.hive.HiveTestUtils.FUNCTION_AND_TYPE_MANAGER in project presto by prestodb.

the class AbstractTestHiveFileFormats method createTestFile.

public static FileSplit createTestFile(String filePath, HiveStorageFormat storageFormat, HiveCompressionCodec compressionCodec, List<TestColumn> testColumns, ConnectorSession session, int numRows, HiveFileWriterFactory fileWriterFactory) {
    // filter out partition keys, which are not written to the file
    testColumns = ImmutableList.copyOf(filter(testColumns, not(TestColumn::isPartitionKey)));
    List<Type> types = testColumns.stream().map(TestColumn::getType).map(HiveType::valueOf).map(type -> type.getType(FUNCTION_AND_TYPE_MANAGER)).collect(toList());
    PageBuilder pageBuilder = new PageBuilder(types);
    for (int rowNumber = 0; rowNumber < numRows; rowNumber++) {
        pageBuilder.declarePosition();
        for (int columnNumber = 0; columnNumber < testColumns.size(); columnNumber++) {
            serializeObject(types.get(columnNumber), pageBuilder.getBlockBuilder(columnNumber), testColumns.get(columnNumber).getWriteValue(), testColumns.get(columnNumber).getObjectInspector(), false);
        }
    }
    Page page = pageBuilder.build();
    JobConf jobConf = configureCompression(new JobConf(), compressionCodec);
    Properties tableProperties = new Properties();
    tableProperties.setProperty("columns", Joiner.on(',').join(transform(testColumns, TestColumn::getName)));
    tableProperties.setProperty("columns.types", Joiner.on(',').join(transform(testColumns, TestColumn::getType)));
    Optional<HiveFileWriter> fileWriter = fileWriterFactory.createFileWriter(new Path(filePath), testColumns.stream().map(TestColumn::getName).collect(toList()), StorageFormat.fromHiveStorageFormat(storageFormat), tableProperties, jobConf, session, Optional.empty());
    HiveFileWriter hiveFileWriter = fileWriter.orElseThrow(() -> new IllegalArgumentException("fileWriterFactory"));
    hiveFileWriter.appendRows(page);
    Optional<Page> fileStatistics = hiveFileWriter.commit();
    assertFileStatistics(fileStatistics, hiveFileWriter.getFileSizeInBytes(), storageFormat);
    return new FileSplit(new Path(filePath), 0, new File(filePath).length(), new String[0]);
}