Examples with RcFilePageSourceFactory com.facebook.presto.hive.rcfile.RcFilePageSourceFactory used on opensource projects

Example 1 with RcFilePageSourceFactory

use of com.facebook.presto.hive.rcfile.RcFilePageSourceFactory in project presto by prestodb.

the class HiveTestUtils method getDefaultHiveBatchPageSourceFactories.

public static Set<HiveBatchPageSourceFactory> getDefaultHiveBatchPageSourceFactories(HiveClientConfig hiveClientConfig, MetastoreClientConfig metastoreClientConfig) {
    FileFormatDataSourceStats stats = new FileFormatDataSourceStats();
    HdfsEnvironment testHdfsEnvironment = createTestHdfsEnvironment(hiveClientConfig, metastoreClientConfig);
    return ImmutableSet.<HiveBatchPageSourceFactory>builder().add(new RcFilePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, testHdfsEnvironment, stats)).add(new OrcBatchPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, hiveClientConfig, testHdfsEnvironment, stats, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource()))).add(new DwrfBatchPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, hiveClientConfig, testHdfsEnvironment, stats, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource()), NO_ENCRYPTION)).add(new ParquetPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, testHdfsEnvironment, stats, new MetadataReader())).add(new PageFilePageSourceFactory(testHdfsEnvironment, new BlockEncodingManager())).build();
}

Example 2 with RcFilePageSourceFactory

use of com.facebook.presto.hive.rcfile.RcFilePageSourceFactory 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 3 with RcFilePageSourceFactory

use of com.facebook.presto.hive.rcfile.RcFilePageSourceFactory in project presto by prestodb.

the class TestHiveFileFormats method testRcTextOptimizedWriter.

@Test(dataProvider = "rowCount")
public void testRcTextOptimizedWriter(int rowCount) throws Exception {
    List<TestColumn> testColumns = TEST_COLUMNS.stream().filter(TestHiveFileFormats::withoutNullMapKeyTests).collect(toImmutableList());
    TestingConnectorSession session = new TestingConnectorSession(new HiveSessionProperties(new HiveClientConfig().setRcfileOptimizedWriterEnabled(true), new OrcFileWriterConfig(), new ParquetFileWriterConfig(), new CacheConfig()).getSessionProperties());
    assertThatFileFormat(RCTEXT).withColumns(testColumns).withRowsCount(rowCount).withSession(session).withFileWriterFactory(new RcFileFileWriterFactory(HDFS_ENVIRONMENT, FUNCTION_AND_TYPE_MANAGER, new NodeVersion("test"), HIVE_STORAGE_TIME_ZONE, STATS)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT)).isReadableByPageSource(new RcFilePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, HDFS_ENVIRONMENT, STATS));
}

Example 4 with RcFilePageSourceFactory

use of com.facebook.presto.hive.rcfile.RcFilePageSourceFactory in project presto by prestodb.

the class TestHiveFileFormats method testRcBinaryOptimizedWriter.

@Test(dataProvider = "rowCount")
public void testRcBinaryOptimizedWriter(int rowCount) throws Exception {
    List<TestColumn> testColumns = TEST_COLUMNS.stream().filter(testColumn -> !testColumn.getName().equals("t_empty_varchar")).filter(TestHiveFileFormats::withoutNullMapKeyTests).collect(toList());
    TestingConnectorSession session = new TestingConnectorSession(new HiveSessionProperties(new HiveClientConfig().setRcfileOptimizedWriterEnabled(true), new OrcFileWriterConfig(), new ParquetFileWriterConfig(), new CacheConfig()).getSessionProperties());
    assertThatFileFormat(RCBINARY).withColumns(testColumns).withRowsCount(rowCount).withSession(session).withFileWriterFactory(new RcFileFileWriterFactory(HDFS_ENVIRONMENT, FUNCTION_AND_TYPE_MANAGER, new NodeVersion("test"), HIVE_STORAGE_TIME_ZONE, STATS)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT)).isReadableByPageSource(new RcFilePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, HDFS_ENVIRONMENT, STATS));
}

Example 5 with RcFilePageSourceFactory

use of com.facebook.presto.hive.rcfile.RcFilePageSourceFactory in project presto by prestodb.

the class TestHiveFileFormats method testTruncateVarcharColumn.

@Test
public void testTruncateVarcharColumn() throws Exception {
    TestColumn writeColumn = new TestColumn("varchar_column", getPrimitiveJavaObjectInspector(new VarcharTypeInfo(4)), new HiveVarchar("test", 4), utf8Slice("test"));
    TestColumn readColumn = new TestColumn("varchar_column", getPrimitiveJavaObjectInspector(new VarcharTypeInfo(3)), new HiveVarchar("tes", 3), utf8Slice("tes"));
    assertThatFileFormat(RCTEXT).withWriteColumns(ImmutableList.of(writeColumn)).withReadColumns(ImmutableList.of(readColumn)).isReadableByPageSource(new RcFilePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, HDFS_ENVIRONMENT, STATS)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT));
    assertThatFileFormat(RCBINARY).withWriteColumns(ImmutableList.of(writeColumn)).withReadColumns(ImmutableList.of(readColumn)).isReadableByPageSource(new RcFilePageSourceFactory(FUNCTION_AND_TYPE_MANAGER, HDFS_ENVIRONMENT, STATS)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT));
    assertThatFileFormat(ORC).withWriteColumns(ImmutableList.of(writeColumn)).withReadColumns(ImmutableList.of(readColumn)).isReadableByPageSource(new OrcBatchPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, false, HDFS_ENVIRONMENT, STATS, 100, new StorageOrcFileTailSource(), StripeMetadataSourceFactory.of(new StorageStripeMetadataSource())));
    assertThatFileFormat(PARQUET).withWriteColumns(ImmutableList.of(writeColumn)).withReadColumns(ImmutableList.of(readColumn)).withSession(parquetPageSourceSession).isReadableByPageSource(new ParquetPageSourceFactory(FUNCTION_AND_TYPE_MANAGER, FUNCTION_RESOLUTION, HDFS_ENVIRONMENT, STATS, METADATA_READER));
    assertThatFileFormat(AVRO).withWriteColumns(ImmutableList.of(writeColumn)).withReadColumns(ImmutableList.of(readColumn)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT));
    assertThatFileFormat(SEQUENCEFILE).withWriteColumns(ImmutableList.of(writeColumn)).withReadColumns(ImmutableList.of(readColumn)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT));
    assertThatFileFormat(TEXTFILE).withWriteColumns(ImmutableList.of(writeColumn)).withReadColumns(ImmutableList.of(readColumn)).isReadableByRecordCursor(new GenericHiveRecordCursorProvider(HDFS_ENVIRONMENT));
}