Examples with JoinCompiler io.trino.sql.gen.JoinCompiler used on opensource projects

Example 11 with JoinCompiler

use of io.trino.sql.gen.JoinCompiler in project trino by trinodb.

the class TestHivePageSink method createPageSink.

private static ConnectorPageSink createPageSink(HiveTransactionHandle transaction, HiveConfig config, HiveMetastore metastore, Path outputPath, HiveWriterStats stats) {
    LocationHandle locationHandle = new LocationHandle(outputPath, outputPath, false, DIRECT_TO_TARGET_NEW_DIRECTORY);
    HiveOutputTableHandle handle = new HiveOutputTableHandle(SCHEMA_NAME, TABLE_NAME, getColumnHandles(), 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(), NO_ACID_TRANSACTION, false, false);
    JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
    TypeOperators typeOperators = new TypeOperators();
    BlockTypeOperators blockTypeOperators = new BlockTypeOperators(typeOperators);
    HivePageSinkProvider provider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(config, HDFS_ENVIRONMENT), HDFS_ENVIRONMENT, PAGE_SORTER, HiveMetastoreFactory.ofInstance(metastore), new GroupByHashPageIndexerFactory(new JoinCompiler(typeOperators), blockTypeOperators), TESTING_TYPE_MANAGER, config, new HiveLocationService(HDFS_ENVIRONMENT), partitionUpdateCodec, new TestingNodeManager("fake-environment"), new HiveEventClient(), getHiveSessionProperties(config), stats);
    return provider.createPageSink(transaction, getHiveSession(config), handle);
}

Example 12 with JoinCompiler

use of io.trino.sql.gen.JoinCompiler in project trino by trinodb.

the class TestDeltaLakePageSink method createPageSink.

private static ConnectorPageSink createPageSink(Path outputPath, DeltaLakeWriterStats stats) {
    HiveTransactionHandle transaction = new HiveTransactionHandle(false);
    DeltaLakeConfig deltaLakeConfig = new DeltaLakeConfig();
    DeltaLakeOutputTableHandle tableHandle = new DeltaLakeOutputTableHandle(SCHEMA_NAME, TABLE_NAME, getColumnHandles(), outputPath.toString(), Optional.of(deltaLakeConfig.getDefaultCheckpointWritingInterval()), true);
    DeltaLakePageSinkProvider provider = new DeltaLakePageSinkProvider(new GroupByHashPageIndexerFactory(new JoinCompiler(new TypeOperators()), new BlockTypeOperators()), HDFS_ENVIRONMENT, JsonCodec.jsonCodec(DataFileInfo.class), stats, deltaLakeConfig, new TestingTypeManager(), new NodeVersion("test-version"));
    return provider.createPageSink(transaction, SESSION, tableHandle);
}

Example 13 with JoinCompiler

use of io.trino.sql.gen.JoinCompiler in project trino by trinodb.

the class AbstractTestHiveFileSystem method setup.

protected void setup(String host, int port, String databaseName, boolean s3SelectPushdownEnabled, HdfsConfiguration hdfsConfiguration) {
    database = databaseName;
    table = new SchemaTableName(database, "trino_test_external_fs");
    tableWithHeader = new SchemaTableName(database, "trino_test_external_fs_with_header");
    tableWithHeaderAndFooter = new SchemaTableName(database, "trino_test_external_fs_with_header_and_footer");
    String random = randomUUID().toString().toLowerCase(ENGLISH).replace("-", "");
    temporaryCreateTable = new SchemaTableName(database, "tmp_trino_test_create_" + random);
    config = new HiveConfig().setS3SelectPushdownEnabled(s3SelectPushdownEnabled);
    Optional<HostAndPort> proxy = Optional.ofNullable(System.getProperty("hive.metastore.thrift.client.socks-proxy")).map(HostAndPort::fromString);
    MetastoreLocator metastoreLocator = new TestingMetastoreLocator(proxy, HostAndPort.fromParts(host, port));
    HivePartitionManager hivePartitionManager = new HivePartitionManager(config);
    hdfsEnvironment = new HdfsEnvironment(hdfsConfiguration, new HdfsConfig(), new NoHdfsAuthentication());
    MetastoreConfig metastoreConfig = new MetastoreConfig();
    metastoreClient = new TestingHiveMetastore(new BridgingHiveMetastore(new ThriftHiveMetastore(metastoreLocator, new HiveConfig(), metastoreConfig, new ThriftMetastoreConfig(), hdfsEnvironment, false), new HiveIdentity(getHiveSession(config).getIdentity())), getBasePath(), hdfsEnvironment);
    locationService = new HiveLocationService(hdfsEnvironment);
    JsonCodec<PartitionUpdate> partitionUpdateCodec = JsonCodec.jsonCodec(PartitionUpdate.class);
    metadataFactory = new HiveMetadataFactory(new CatalogName("hive"), config, metastoreConfig, HiveMetastoreFactory.ofInstance(metastoreClient), hdfsEnvironment, hivePartitionManager, newDirectExecutorService(), heartbeatService, TESTING_TYPE_MANAGER, NOOP_METADATA_PROVIDER, locationService, partitionUpdateCodec, new NodeVersion("test_version"), new NoneHiveRedirectionsProvider(), ImmutableSet.of(new PartitionsSystemTableProvider(hivePartitionManager, TESTING_TYPE_MANAGER), new PropertiesSystemTableProvider()), new DefaultHiveMaterializedViewMetadataFactory(), SqlStandardAccessControlMetadata::new, NO_REDIRECTIONS, TableInvalidationCallback.NOOP);
    transactionManager = new HiveTransactionManager(metadataFactory);
    splitManager = new HiveSplitManager(transactionManager, hivePartitionManager, new NamenodeStats(), hdfsEnvironment, new CachingDirectoryLister(new HiveConfig()), new BoundedExecutor(executor, config.getMaxSplitIteratorThreads()), new CounterStat(), config.getMaxOutstandingSplits(), config.getMaxOutstandingSplitsSize(), config.getMinPartitionBatchSize(), config.getMaxPartitionBatchSize(), config.getMaxInitialSplits(), config.getSplitLoaderConcurrency(), config.getMaxSplitsPerSecond(), config.getRecursiveDirWalkerEnabled(), TESTING_TYPE_MANAGER);
    TypeOperators typeOperators = new TypeOperators();
    BlockTypeOperators blockTypeOperators = new BlockTypeOperators(typeOperators);
    pageSinkProvider = new HivePageSinkProvider(getDefaultHiveFileWriterFactories(config, hdfsEnvironment), hdfsEnvironment, PAGE_SORTER, HiveMetastoreFactory.ofInstance(metastoreClient), new GroupByHashPageIndexerFactory(new JoinCompiler(typeOperators), blockTypeOperators), TESTING_TYPE_MANAGER, config, locationService, partitionUpdateCodec, new TestingNodeManager("fake-environment"), new HiveEventClient(), getHiveSessionProperties(config), new HiveWriterStats());
    pageSourceProvider = new HivePageSourceProvider(TESTING_TYPE_MANAGER, hdfsEnvironment, config, getDefaultHivePageSourceFactories(hdfsEnvironment, config), getDefaultHiveRecordCursorProviders(config, hdfsEnvironment), new GenericHiveRecordCursorProvider(hdfsEnvironment, config), Optional.empty());
    onSetupComplete();
}

Example 14 with JoinCompiler

use of io.trino.sql.gen.JoinCompiler in project trino by trinodb.

the class TestHashSemiJoinOperator method testProbeAndBuildNulls.

@Test(dataProvider = "hashEnabledValues")
public void testProbeAndBuildNulls(boolean hashEnabled) {
    DriverContext driverContext = taskContext.addPipelineContext(0, true, true, false).addDriverContext();
    // build
    OperatorContext operatorContext = driverContext.addOperatorContext(0, new PlanNodeId("test"), ValuesOperator.class.getSimpleName());
    List<Type> buildTypes = ImmutableList.of(BIGINT);
    RowPagesBuilder rowPagesBuilder = rowPagesBuilder(hashEnabled, Ints.asList(0), buildTypes);
    Operator buildOperator = new ValuesOperator(operatorContext, rowPagesBuilder.row(0L).row(1L).row((Object) null).row(3L).build());
    SetBuilderOperatorFactory setBuilderOperatorFactory = new SetBuilderOperatorFactory(1, new PlanNodeId("test"), buildTypes.get(0), 0, rowPagesBuilder.getHashChannel(), 10, new JoinCompiler(typeOperators), blockTypeOperators);
    Operator setBuilderOperator = setBuilderOperatorFactory.createOperator(driverContext);
    Driver driver = Driver.createDriver(driverContext, buildOperator, setBuilderOperator);
    while (!driver.isFinished()) {
        driver.process();
    }
    // probe
    List<Type> probeTypes = ImmutableList.of(BIGINT);
    RowPagesBuilder rowPagesBuilderProbe = rowPagesBuilder(hashEnabled, Ints.asList(0), probeTypes);
    List<Page> probeInput = rowPagesBuilderProbe.row(0L).row((Object) null).row(1L).row(2L).build();
    Optional<Integer> probeHashChannel = hashEnabled ? Optional.of(probeTypes.size()) : Optional.empty();
    OperatorFactory joinOperatorFactory = HashSemiJoinOperator.createOperatorFactory(2, new PlanNodeId("test"), setBuilderOperatorFactory.getSetProvider(), rowPagesBuilderProbe.getTypes(), 0, probeHashChannel);
    // expected
    MaterializedResult expected = resultBuilder(driverContext.getSession(), concat(probeTypes, ImmutableList.of(BOOLEAN))).row(0L, true).row(null, null).row(1L, true).row(2L, null).build();
    OperatorAssertion.assertOperatorEquals(joinOperatorFactory, driverContext, probeInput, expected, hashEnabled, ImmutableList.of(probeTypes.size()));
}

Example 15 with JoinCompiler

use of io.trino.sql.gen.JoinCompiler in project trino by trinodb.

the class TestHashSemiJoinOperator method testBuildSideNulls.

@Test(dataProvider = "hashEnabledValues")
public void testBuildSideNulls(boolean hashEnabled) {
    DriverContext driverContext = taskContext.addPipelineContext(0, true, true, false).addDriverContext();
    // build
    OperatorContext operatorContext = driverContext.addOperatorContext(0, new PlanNodeId("test"), ValuesOperator.class.getSimpleName());
    List<Type> buildTypes = ImmutableList.of(BIGINT);
    RowPagesBuilder rowPagesBuilder = rowPagesBuilder(hashEnabled, Ints.asList(0), buildTypes);
    Operator buildOperator = new ValuesOperator(operatorContext, rowPagesBuilder.row(0L).row(1L).row(2L).row(2L).row(3L).row((Object) null).build());
    SetBuilderOperatorFactory setBuilderOperatorFactory = new SetBuilderOperatorFactory(1, new PlanNodeId("test"), buildTypes.get(0), 0, rowPagesBuilder.getHashChannel(), 10, new JoinCompiler(typeOperators), blockTypeOperators);
    Operator setBuilderOperator = setBuilderOperatorFactory.createOperator(driverContext);
    Driver driver = Driver.createDriver(driverContext, buildOperator, setBuilderOperator);
    while (!driver.isFinished()) {
        driver.process();
    }
    // probe
    List<Type> probeTypes = ImmutableList.of(BIGINT);
    RowPagesBuilder rowPagesBuilderProbe = rowPagesBuilder(hashEnabled, Ints.asList(0), probeTypes);
    List<Page> probeInput = rowPagesBuilderProbe.addSequencePage(4, 1).build();
    Optional<Integer> probeHashChannel = hashEnabled ? Optional.of(probeTypes.size()) : Optional.empty();
    OperatorFactory joinOperatorFactory = HashSemiJoinOperator.createOperatorFactory(2, new PlanNodeId("test"), setBuilderOperatorFactory.getSetProvider(), rowPagesBuilderProbe.getTypes(), 0, probeHashChannel);
    // expected
    MaterializedResult expected = resultBuilder(driverContext.getSession(), concat(probeTypes, ImmutableList.of(BOOLEAN))).row(1L, true).row(2L, true).row(3L, true).row(4L, null).build();
    OperatorAssertion.assertOperatorEquals(joinOperatorFactory, driverContext, probeInput, expected, hashEnabled, ImmutableList.of(probeTypes.size()));
}