Examples with GroupByHashPageIndexerFactory com.facebook.presto.GroupByHashPageIndexerFactory used on opensource projects

Example 1 with GroupByHashPageIndexerFactory

use of com.facebook.presto.GroupByHashPageIndexerFactory in project presto by prestodb.

the class TestHivePageSink method createPageSink.

private static ConnectorPageSink createPageSink(HiveTransactionHandle transaction, HiveClientConfig config, ExtendedHiveMetastore metastore, Path outputPath) {
    LocationHandle locationHandle = new LocationHandle(outputPath, Optional.of(outputPath), false);
    HiveOutputTableHandle handle = new HiveOutputTableHandle(CLIENT_ID, SCHEMA_NAME, TABLE_NAME, getColumnHandles(), "test", new HivePageSinkMetadata(new SchemaTableName(SCHEMA_NAME, TABLE_NAME), metastore.getTable(SCHEMA_NAME, TABLE_NAME), ImmutableMap.of()), locationHandle, config.getHiveStorageFormat(), config.getHiveStorageFormat(), ImmutableList.of(), Optional.empty(), "test", ImmutableMap.of());
    JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
    HdfsEnvironment hdfsEnvironment = createTestHdfsEnvironment(config);
    HivePageSinkProvider provider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(config), hdfsEnvironment, metastore, new GroupByHashPageIndexerFactory(new JoinCompiler()), TYPE_MANAGER, config, new HiveLocationService(hdfsEnvironment), partitionUpdateCodec);
    return provider.createPageSink(transaction, getSession(config), handle);
}

Example 2 with GroupByHashPageIndexerFactory

use of com.facebook.presto.GroupByHashPageIndexerFactory in project presto by prestodb.

the class AbstractTestHiveClient method setup.

protected final void setup(String databaseName, HiveClientConfig hiveClientConfig, ExtendedHiveMetastore hiveMetastore) {
    HiveConnectorId connectorId = new HiveConnectorId("hive-test");
    setupHive(connectorId.toString(), databaseName, hiveClientConfig.getTimeZone());
    metastoreClient = hiveMetastore;
    HdfsConfiguration hdfsConfiguration = new HiveHdfsConfiguration(new HdfsConfigurationUpdater(hiveClientConfig, new HiveS3Config()));
    hdfsEnvironment = new HdfsEnvironment(hdfsConfiguration, hiveClientConfig, new NoHdfsAuthentication());
    locationService = new HiveLocationService(hdfsEnvironment);
    TypeManager typeManager = new TypeRegistry();
    JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
    metadataFactory = new HiveMetadataFactory(connectorId, metastoreClient, hdfsEnvironment, new HivePartitionManager(connectorId, TYPE_MANAGER, hiveClientConfig), timeZone, 10, true, true, false, true, HiveStorageFormat.RCBINARY, 1000, typeManager, locationService, new TableParameterCodec(), partitionUpdateCodec, newFixedThreadPool(2), new HiveTypeTranslator(), TEST_SERVER_VERSION);
    transactionManager = new HiveTransactionManager();
    splitManager = new HiveSplitManager(connectorId, transactionHandle -> ((HiveMetadata) transactionManager.get(transactionHandle)).getMetastore(), new NamenodeStats(), hdfsEnvironment, new HadoopDirectoryLister(), newDirectExecutorService(), new HiveCoercionPolicy(typeManager), 100, hiveClientConfig.getMinPartitionBatchSize(), hiveClientConfig.getMaxPartitionBatchSize(), hiveClientConfig.getMaxInitialSplits(), false);
    pageSinkProvider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(hiveClientConfig), hdfsEnvironment, metastoreClient, new GroupByHashPageIndexerFactory(JOIN_COMPILER), typeManager, new HiveClientConfig(), locationService, partitionUpdateCodec);
    pageSourceProvider = new HivePageSourceProvider(hiveClientConfig, hdfsEnvironment, getDefaultHiveRecordCursorProvider(hiveClientConfig), getDefaultHiveDataStreamFactories(hiveClientConfig), TYPE_MANAGER);
}

Example 3 with GroupByHashPageIndexerFactory

use of com.facebook.presto.GroupByHashPageIndexerFactory in project presto by prestodb.

the class AbstractTestHiveClientS3 method setup.

protected void setup(String host, int port, String databaseName, String awsAccessKey, String awsSecretKey, String writableBucket) {
    this.writableBucket = writableBucket;
    setupHive(databaseName);
    HiveS3Config s3Config = new HiveS3Config().setS3AwsAccessKey(awsAccessKey).setS3AwsSecretKey(awsSecretKey);
    HiveClientConfig hiveClientConfig = new HiveClientConfig();
    String proxy = System.getProperty("hive.metastore.thrift.client.socks-proxy");
    if (proxy != null) {
        hiveClientConfig.setMetastoreSocksProxy(HostAndPort.fromString(proxy));
    }
    HiveConnectorId connectorId = new HiveConnectorId("hive-test");
    HiveCluster hiveCluster = new TestingHiveCluster(hiveClientConfig, host, port);
    ExecutorService executor = newCachedThreadPool(daemonThreadsNamed("hive-s3-%s"));
    HdfsConfiguration hdfsConfiguration = new HiveHdfsConfiguration(new HdfsConfigurationUpdater(hiveClientConfig, s3Config));
    HivePartitionManager hivePartitionManager = new HivePartitionManager(connectorId, TYPE_MANAGER, hiveClientConfig);
    hdfsEnvironment = new HdfsEnvironment(hdfsConfiguration, hiveClientConfig, new NoHdfsAuthentication());
    metastoreClient = new TestingHiveMetastore(new BridgingHiveMetastore(new ThriftHiveMetastore(hiveCluster)), executor, hiveClientConfig, writableBucket, hdfsEnvironment);
    locationService = new HiveLocationService(hdfsEnvironment);
    TypeRegistry typeManager = new TypeRegistry();
    JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
    metadataFactory = new HiveMetadataFactory(connectorId, hiveClientConfig, metastoreClient, hdfsEnvironment, hivePartitionManager, newDirectExecutorService(), typeManager, locationService, new TableParameterCodec(), partitionUpdateCodec, new HiveTypeTranslator(), new NodeVersion("test_version"));
    transactionManager = new HiveTransactionManager();
    splitManager = new HiveSplitManager(connectorId, transactionHandle -> ((HiveMetadata) transactionManager.get(transactionHandle)).getMetastore(), new NamenodeStats(), hdfsEnvironment, new HadoopDirectoryLister(), new BoundedExecutor(executor, hiveClientConfig.getMaxSplitIteratorThreads()), new HiveCoercionPolicy(typeManager), hiveClientConfig.getMaxOutstandingSplits(), hiveClientConfig.getMinPartitionBatchSize(), hiveClientConfig.getMaxPartitionBatchSize(), hiveClientConfig.getMaxInitialSplits(), hiveClientConfig.getRecursiveDirWalkerEnabled());
    pageSinkProvider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(hiveClientConfig), hdfsEnvironment, metastoreClient, new GroupByHashPageIndexerFactory(new JoinCompiler()), typeManager, new HiveClientConfig(), locationService, partitionUpdateCodec);
    pageSourceProvider = new HivePageSourceProvider(hiveClientConfig, hdfsEnvironment, getDefaultHiveRecordCursorProvider(hiveClientConfig), getDefaultHiveDataStreamFactories(hiveClientConfig), TYPE_MANAGER);
}