use of com.facebook.presto.hive.datasink.OutputStreamDataSinkFactory 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));
}
use of com.facebook.presto.hive.datasink.OutputStreamDataSinkFactory 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())));
}
use of com.facebook.presto.hive.datasink.OutputStreamDataSinkFactory in project presto by prestodb.
the class TestHiveMetadataFileFormatEncryptionSettings method setup.
@BeforeClass
public void setup() {
baseDirectory = new File(Files.createTempDir(), "metastore");
metastore = new BridgingHiveMetastore(new InMemoryHiveMetastore(baseDirectory), new HivePartitionMutator());
executor = newCachedThreadPool(daemonThreadsNamed("hive-encryption-test-%s"));
transactionManager = new HiveTransactionManager();
metadataFactory = new HiveMetadataFactory(metastore, HDFS_ENVIRONMENT, new HivePartitionManager(FUNCTION_AND_TYPE_MANAGER, HIVE_CLIENT_CONFIG), DateTimeZone.forTimeZone(TimeZone.getTimeZone(ZoneId.of(HIVE_CLIENT_CONFIG.getTimeZone()))), true, false, false, false, true, true, HIVE_CLIENT_CONFIG.getMaxPartitionBatchSize(), HIVE_CLIENT_CONFIG.getMaxPartitionsPerScan(), false, FUNCTION_AND_TYPE_MANAGER, new HiveLocationService(HDFS_ENVIRONMENT), FUNCTION_RESOLUTION, ROW_EXPRESSION_SERVICE, FILTER_STATS_CALCULATOR_SERVICE, new TableParameterCodec(), PARTITION_UPDATE_CODEC, PARTITION_UPDATE_SMILE_CODEC, listeningDecorator(executor), new HiveTypeTranslator(), new HiveStagingFileCommitter(HDFS_ENVIRONMENT, listeningDecorator(executor)), new HiveZeroRowFileCreator(HDFS_ENVIRONMENT, new OutputStreamDataSinkFactory(), listeningDecorator(executor)), TEST_SERVER_VERSION, new HivePartitionObjectBuilder(), new HiveEncryptionInformationProvider(ImmutableList.of(new TestDwrfEncryptionInformationSource())), new HivePartitionStats(), new HiveFileRenamer(), HiveColumnConverterProvider.DEFAULT_COLUMN_CONVERTER_PROVIDER);
metastore.createDatabase(METASTORE_CONTEXT, Database.builder().setDatabaseName(TEST_DB_NAME).setOwnerName("public").setOwnerType(PrincipalType.ROLE).build());
}
use of com.facebook.presto.hive.datasink.OutputStreamDataSinkFactory in project presto by prestodb.
the class TestHiveSplitManager method assertRedundantColumnDomains.
private void assertRedundantColumnDomains(Range predicateRange, PartitionStatistics partitionStatistics, List<Set<ColumnHandle>> expectedRedundantColumnDomains, HiveColumnHandle columnHandle) throws Exception {
// Prepare query predicate tuple domain
TupleDomain<ColumnHandle> queryTupleDomain = TupleDomain.fromColumnDomains(Optional.of(ImmutableList.of(new ColumnDomain<>(columnHandle, Domain.create(SortedRangeSet.copyOf(predicateRange.getType(), ImmutableList.of(predicateRange)), false)))));
// Prepare partition with stats
PartitionWithStatistics partitionWithStatistics = new PartitionWithStatistics(new Partition("test_db", "test_table", ImmutableList.of(PARTITION_VALUE), new Storage(fromHiveStorageFormat(ORC), "location", Optional.empty(), true, ImmutableMap.of(), ImmutableMap.of()), COLUMNS, ImmutableMap.of(), Optional.empty(), false, true, 0), PARTITION_NAME, partitionStatistics);
HiveClientConfig hiveClientConfig = new HiveClientConfig().setPartitionStatisticsBasedOptimizationEnabled(true);
HdfsEnvironment hdfsEnvironment = new HdfsEnvironment(new HiveHdfsConfiguration(new HdfsConfigurationInitializer(hiveClientConfig, new MetastoreClientConfig()), ImmutableSet.of()), new MetastoreClientConfig(), new NoHdfsAuthentication());
HiveMetadataFactory metadataFactory = new HiveMetadataFactory(new TestingExtendedHiveMetastore(TEST_TABLE, partitionWithStatistics), hdfsEnvironment, new HivePartitionManager(FUNCTION_AND_TYPE_MANAGER, hiveClientConfig), DateTimeZone.forOffsetHours(1), true, false, false, false, true, true, hiveClientConfig.getMaxPartitionBatchSize(), hiveClientConfig.getMaxPartitionsPerScan(), false, FUNCTION_AND_TYPE_MANAGER, new HiveLocationService(hdfsEnvironment), FUNCTION_RESOLUTION, ROW_EXPRESSION_SERVICE, FILTER_STATS_CALCULATOR_SERVICE, new TableParameterCodec(), HiveTestUtils.PARTITION_UPDATE_CODEC, HiveTestUtils.PARTITION_UPDATE_SMILE_CODEC, executor, new HiveTypeTranslator(), new HiveStagingFileCommitter(hdfsEnvironment, executor), new HiveZeroRowFileCreator(hdfsEnvironment, new OutputStreamDataSinkFactory(), executor), TEST_SERVER_VERSION, new HivePartitionObjectBuilder(), new HiveEncryptionInformationProvider(ImmutableList.of()), new HivePartitionStats(), new HiveFileRenamer(), HiveColumnConverterProvider.DEFAULT_COLUMN_CONVERTER_PROVIDER);
HiveSplitManager splitManager = new HiveSplitManager(new TestingHiveTransactionManager(metadataFactory), new NamenodeStats(), hdfsEnvironment, new TestingDirectoryLister(), directExecutor(), new HiveCoercionPolicy(FUNCTION_AND_TYPE_MANAGER), new CounterStat(), 100, hiveClientConfig.getMaxOutstandingSplitsSize(), hiveClientConfig.getMinPartitionBatchSize(), hiveClientConfig.getMaxPartitionBatchSize(), hiveClientConfig.getSplitLoaderConcurrency(), false, new ConfigBasedCacheQuotaRequirementProvider(new CacheConfig()), new HiveEncryptionInformationProvider(ImmutableList.of()));
HiveColumnHandle partitionColumn = new HiveColumnHandle("ds", HIVE_STRING, parseTypeSignature(VARCHAR), MAX_PARTITION_KEY_COLUMN_INDEX, PARTITION_KEY, Optional.empty(), Optional.empty());
List<HivePartition> partitions = ImmutableList.of(new HivePartition(new SchemaTableName("test_schema", "test_table"), PARTITION_NAME, ImmutableMap.of(partitionColumn, NullableValue.of(createUnboundedVarcharType(), utf8Slice(PARTITION_VALUE)))));
TupleDomain<Subfield> domainPredicate = queryTupleDomain.transform(HiveColumnHandle.class::cast).transform(column -> new Subfield(column.getName(), ImmutableList.of()));
ConnectorSplitSource splitSource = splitManager.getSplits(new HiveTransactionHandle(), new TestingConnectorSession(new HiveSessionProperties(hiveClientConfig, new OrcFileWriterConfig(), new ParquetFileWriterConfig(), new CacheConfig()).getSessionProperties()), new HiveTableLayoutHandle(new SchemaTableName("test_schema", "test_table"), "test_path", ImmutableList.of(partitionColumn), COLUMNS, ImmutableMap.of(), partitions, domainPredicate, TRUE_CONSTANT, ImmutableMap.of(partitionColumn.getName(), partitionColumn, columnHandle.getName(), columnHandle), queryTupleDomain, Optional.empty(), Optional.empty(), false, "layout", Optional.empty(), false), SPLIT_SCHEDULING_CONTEXT);
List<Set<ColumnHandle>> actualRedundantColumnDomains = splitSource.getNextBatch(NOT_PARTITIONED, 100).get().getSplits().stream().map(HiveSplit.class::cast).map(HiveSplit::getRedundantColumnDomains).collect(toImmutableList());
assertEquals(actualRedundantColumnDomains, expectedRedundantColumnDomains);
}
use of com.facebook.presto.hive.datasink.OutputStreamDataSinkFactory in project presto by prestodb.
the class AbstractTestHiveFileSystem method setup.
protected void setup(String host, int port, String databaseName, BiFunction<HiveClientConfig, MetastoreClientConfig, HdfsConfiguration> hdfsConfigurationProvider, boolean s3SelectPushdownEnabled) {
database = databaseName;
table = new SchemaTableName(database, "presto_test_external_fs");
String random = UUID.randomUUID().toString().toLowerCase(ENGLISH).replace("-", "");
temporaryCreateTable = new SchemaTableName(database, "tmp_presto_test_create_" + random);
config = new HiveClientConfig().setS3SelectPushdownEnabled(s3SelectPushdownEnabled);
cacheConfig = new CacheConfig();
metastoreClientConfig = new MetastoreClientConfig();
String proxy = System.getProperty("hive.metastore.thrift.client.socks-proxy");
if (proxy != null) {
metastoreClientConfig.setMetastoreSocksProxy(HostAndPort.fromString(proxy));
}
HiveCluster hiveCluster = new TestingHiveCluster(metastoreClientConfig, host, port);
ExecutorService executor = newCachedThreadPool(daemonThreadsNamed("hive-%s"));
HivePartitionManager hivePartitionManager = new HivePartitionManager(FUNCTION_AND_TYPE_MANAGER, config);
HdfsConfiguration hdfsConfiguration = hdfsConfigurationProvider.apply(config, metastoreClientConfig);
hdfsEnvironment = new HdfsEnvironment(hdfsConfiguration, metastoreClientConfig, new NoHdfsAuthentication());
ColumnConverterProvider columnConverterProvider = HiveColumnConverterProvider.DEFAULT_COLUMN_CONVERTER_PROVIDER;
metastoreClient = new TestingHiveMetastore(new BridgingHiveMetastore(new ThriftHiveMetastore(hiveCluster, metastoreClientConfig), new HivePartitionMutator()), executor, metastoreClientConfig, getBasePath(), hdfsEnvironment);
locationService = new HiveLocationService(hdfsEnvironment);
metadataFactory = new HiveMetadataFactory(config, metastoreClientConfig, metastoreClient, hdfsEnvironment, hivePartitionManager, newDirectExecutorService(), FUNCTION_AND_TYPE_MANAGER, locationService, FUNCTION_RESOLUTION, ROW_EXPRESSION_SERVICE, FILTER_STATS_CALCULATOR_SERVICE, new TableParameterCodec(), HiveTestUtils.PARTITION_UPDATE_CODEC, HiveTestUtils.PARTITION_UPDATE_SMILE_CODEC, new HiveTypeTranslator(), new HiveStagingFileCommitter(hdfsEnvironment, listeningDecorator(executor)), new HiveZeroRowFileCreator(hdfsEnvironment, new OutputStreamDataSinkFactory(), listeningDecorator(executor)), new NodeVersion("test_version"), new HivePartitionObjectBuilder(), new HiveEncryptionInformationProvider(ImmutableList.of()), new HivePartitionStats(), new HiveFileRenamer(), columnConverterProvider);
transactionManager = new HiveTransactionManager();
splitManager = new HiveSplitManager(transactionManager, new NamenodeStats(), hdfsEnvironment, new CachingDirectoryLister(new HadoopDirectoryLister(), new HiveClientConfig()), new BoundedExecutor(executor, config.getMaxSplitIteratorThreads()), new HiveCoercionPolicy(FUNCTION_AND_TYPE_MANAGER), new CounterStat(), config.getMaxOutstandingSplits(), config.getMaxOutstandingSplitsSize(), config.getMinPartitionBatchSize(), config.getMaxPartitionBatchSize(), config.getSplitLoaderConcurrency(), config.getRecursiveDirWalkerEnabled(), new ConfigBasedCacheQuotaRequirementProvider(cacheConfig), new HiveEncryptionInformationProvider(ImmutableSet.of()));
pageSinkProvider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(config, metastoreClientConfig), hdfsEnvironment, PAGE_SORTER, metastoreClient, new GroupByHashPageIndexerFactory(new JoinCompiler(MetadataManager.createTestMetadataManager(), new FeaturesConfig())), FUNCTION_AND_TYPE_MANAGER, config, metastoreClientConfig, locationService, HiveTestUtils.PARTITION_UPDATE_CODEC, HiveTestUtils.PARTITION_UPDATE_SMILE_CODEC, new TestingNodeManager("fake-environment"), new HiveEventClient(), new HiveSessionProperties(config, new OrcFileWriterConfig(), new ParquetFileWriterConfig(), new CacheConfig()), new HiveWriterStats(), getDefaultOrcFileWriterFactory(config, metastoreClientConfig), columnConverterProvider);
pageSourceProvider = new HivePageSourceProvider(config, hdfsEnvironment, getDefaultHiveRecordCursorProvider(config, metastoreClientConfig), getDefaultHiveBatchPageSourceFactories(config, metastoreClientConfig), getDefaultHiveSelectivePageSourceFactories(config, metastoreClientConfig), FUNCTION_AND_TYPE_MANAGER, ROW_EXPRESSION_SERVICE);
}
Aggregations