Examples with HashBuilderOperatorFactory io.trino.operator.join.HashBuilderOperator.HashBuilderOperatorFactory used on opensource projects

Example 1 with HashBuilderOperatorFactory

use of io.trino.operator.join.HashBuilderOperator.HashBuilderOperatorFactory in project trino by trinodb.

the class BenchmarkHashBuildAndJoinOperators method buildHash.

private static void buildHash(BuildContext buildContext, JoinBridgeManager<PartitionedLookupSourceFactory> lookupSourceFactoryManager, List<Integer> outputChannels, int partitionCount) {
    HashBuilderOperatorFactory hashBuilderOperatorFactory = new HashBuilderOperatorFactory(HASH_BUILD_OPERATOR_ID, TEST_PLAN_NODE_ID, lookupSourceFactoryManager, outputChannels, buildContext.getHashChannels(), buildContext.getHashChannel(), Optional.empty(), Optional.empty(), ImmutableList.of(), 10_000, new PagesIndex.TestingFactory(false), false, SingleStreamSpillerFactory.unsupportedSingleStreamSpillerFactory(), incrementalLoadFactorHashArraySizeSupplier(buildContext.getSession()));
    Operator[] operators = IntStream.range(0, partitionCount).mapToObj(i -> buildContext.createTaskContext().addPipelineContext(0, true, true, partitionCount > 1).addDriverContext()).map(hashBuilderOperatorFactory::createOperator).toArray(Operator[]::new);
    if (partitionCount == 1) {
        for (Page page : buildContext.getBuildPages()) {
            operators[0].addInput(page);
        }
    } else {
        PartitionFunction partitionGenerator = new LocalPartitionGenerator(new InterpretedHashGenerator(buildContext.getHashChannels().stream().map(channel -> buildContext.getTypes().get(channel)).collect(toImmutableList()), buildContext.getHashChannels(), TYPE_OPERATOR_FACTORY), partitionCount);
        for (Page page : buildContext.getBuildPages()) {
            Page[] partitionedPages = partitionPages(page, buildContext.getTypes(), partitionCount, partitionGenerator);
            for (int i = 0; i < partitionCount; i++) {
                operators[i].addInput(partitionedPages[i]);
            }
        }
    }
    LookupSourceFactory lookupSourceFactory = lookupSourceFactoryManager.getJoinBridge(Lifespan.taskWide());
    ListenableFuture<LookupSourceProvider> lookupSourceProvider = lookupSourceFactory.createLookupSourceProvider();
    for (Operator operator : operators) {
        operator.finish();
    }
    if (!lookupSourceProvider.isDone()) {
        throw new AssertionError("Expected lookup source provider to be ready");
    }
    getFutureValue(lookupSourceProvider).close();
}

Example 2 with HashBuilderOperatorFactory

use of io.trino.operator.join.HashBuilderOperator.HashBuilderOperatorFactory in project trino by trinodb.

the class HashBuildAndJoinBenchmark method createDrivers.

/*
    select orderkey, quantity, totalprice
    from lineitem join orders using (orderkey)
     */
@Override
protected List<Driver> createDrivers(TaskContext taskContext) {
    ImmutableList.Builder<OperatorFactory> driversBuilder = ImmutableList.builder();
    driversBuilder.add(ordersTableScan);
    List<Type> sourceTypes = ordersTableTypes;
    OptionalInt hashChannel = OptionalInt.empty();
    if (hashEnabled) {
        driversBuilder.add(createHashProjectOperator(1, new PlanNodeId("test"), sourceTypes));
        sourceTypes = ImmutableList.<Type>builder().addAll(sourceTypes).add(BIGINT).build();
        hashChannel = OptionalInt.of(sourceTypes.size() - 1);
    }
    // hash build
    BlockTypeOperators blockTypeOperators = new BlockTypeOperators(new TypeOperators());
    JoinBridgeManager<PartitionedLookupSourceFactory> lookupSourceFactoryManager = JoinBridgeManager.lookupAllAtOnce(new PartitionedLookupSourceFactory(sourceTypes, ImmutableList.of(0, 1).stream().map(sourceTypes::get).collect(toImmutableList()), Ints.asList(0).stream().map(sourceTypes::get).collect(toImmutableList()), 1, false, blockTypeOperators));
    HashBuilderOperatorFactory hashBuilder = new HashBuilderOperatorFactory(2, new PlanNodeId("test"), lookupSourceFactoryManager, ImmutableList.of(0, 1), Ints.asList(0), hashChannel, Optional.empty(), Optional.empty(), ImmutableList.of(), 1_500_000, new PagesIndex.TestingFactory(false), false, SingleStreamSpillerFactory.unsupportedSingleStreamSpillerFactory(), incrementalLoadFactorHashArraySizeSupplier(session));
    driversBuilder.add(hashBuilder);
    DriverFactory hashBuildDriverFactory = new DriverFactory(0, true, false, driversBuilder.build(), OptionalInt.empty(), UNGROUPED_EXECUTION);
    // join
    ImmutableList.Builder<OperatorFactory> joinDriversBuilder = ImmutableList.builder();
    joinDriversBuilder.add(lineItemTableScan);
    sourceTypes = lineItemTableTypes;
    hashChannel = OptionalInt.empty();
    if (hashEnabled) {
        joinDriversBuilder.add(createHashProjectOperator(1, new PlanNodeId("test"), sourceTypes));
        sourceTypes = ImmutableList.<Type>builder().addAll(sourceTypes).add(BIGINT).build();
        hashChannel = OptionalInt.of(sourceTypes.size() - 1);
    }
    OperatorFactory joinOperator = operatorFactories.innerJoin(2, new PlanNodeId("test"), lookupSourceFactoryManager, false, false, false, sourceTypes, Ints.asList(0), hashChannel, Optional.empty(), OptionalInt.empty(), unsupportedPartitioningSpillerFactory(), blockTypeOperators);
    joinDriversBuilder.add(joinOperator);
    joinDriversBuilder.add(new NullOutputOperatorFactory(3, new PlanNodeId("test")));
    DriverFactory joinDriverFactory = new DriverFactory(1, true, true, joinDriversBuilder.build(), OptionalInt.empty(), UNGROUPED_EXECUTION);
    Driver hashBuildDriver = hashBuildDriverFactory.createDriver(taskContext.addPipelineContext(0, true, false, false).addDriverContext());
    hashBuildDriverFactory.noMoreDrivers();
    Driver joinDriver = joinDriverFactory.createDriver(taskContext.addPipelineContext(1, true, true, false).addDriverContext());
    joinDriverFactory.noMoreDrivers();
    return ImmutableList.of(hashBuildDriver, joinDriver);
}

Example 3 with HashBuilderOperatorFactory

use of io.trino.operator.join.HashBuilderOperator.HashBuilderOperatorFactory in project trino by trinodb.

the class JoinTestUtils method setupBuildSide.

public static BuildSideSetup setupBuildSide(NodePartitioningManager nodePartitioningManager, boolean parallelBuild, TaskContext taskContext, RowPagesBuilder buildPages, Optional<InternalJoinFilterFunction> filterFunction, boolean spillEnabled, SingleStreamSpillerFactory singleStreamSpillerFactory) {
    Optional<JoinFilterFunctionCompiler.JoinFilterFunctionFactory> filterFunctionFactory = filterFunction.map(function -> (session, addresses, pages) -> new StandardJoinFilterFunction(function, addresses, pages));
    int partitionCount = parallelBuild ? PARTITION_COUNT : 1;
    List<Integer> hashChannels = buildPages.getHashChannels().orElseThrow();
    LocalExchange.LocalExchangeFactory localExchangeFactory = new LocalExchange.LocalExchangeFactory(nodePartitioningManager, taskContext.getSession(), FIXED_HASH_DISTRIBUTION, partitionCount, buildPages.getTypes(), hashChannels, buildPages.getHashChannel(), UNGROUPED_EXECUTION, DataSize.of(32, DataSize.Unit.MEGABYTE), TYPE_OPERATOR_FACTORY);
    LocalExchange.LocalExchangeSinkFactoryId localExchangeSinkFactoryId = localExchangeFactory.newSinkFactoryId();
    localExchangeFactory.noMoreSinkFactories();
    // collect input data into the partitioned exchange
    DriverContext collectDriverContext = taskContext.addPipelineContext(0, true, true, false).addDriverContext();
    ValuesOperator.ValuesOperatorFactory valuesOperatorFactory = new ValuesOperator.ValuesOperatorFactory(0, new PlanNodeId("values"), buildPages.build());
    LocalExchangeSinkOperator.LocalExchangeSinkOperatorFactory sinkOperatorFactory = new LocalExchangeSinkOperator.LocalExchangeSinkOperatorFactory(localExchangeFactory, 1, new PlanNodeId("sink"), localExchangeSinkFactoryId, Function.identity());
    Driver sourceDriver = Driver.createDriver(collectDriverContext, valuesOperatorFactory.createOperator(collectDriverContext), sinkOperatorFactory.createOperator(collectDriverContext));
    valuesOperatorFactory.noMoreOperators();
    sinkOperatorFactory.noMoreOperators();
    while (!sourceDriver.isFinished()) {
        sourceDriver.process();
    }
    // build side operator factories
    LocalExchangeSourceOperatorFactory sourceOperatorFactory = new LocalExchangeSourceOperatorFactory(0, new PlanNodeId("source"), localExchangeFactory);
    JoinBridgeManager<PartitionedLookupSourceFactory> lookupSourceFactoryManager = JoinBridgeManager.lookupAllAtOnce(new PartitionedLookupSourceFactory(buildPages.getTypes(), rangeList(buildPages.getTypes().size()).stream().map(buildPages.getTypes()::get).collect(toImmutableList()), hashChannels.stream().map(buildPages.getTypes()::get).collect(toImmutableList()), partitionCount, false, TYPE_OPERATOR_FACTORY));
    HashBuilderOperatorFactory buildOperatorFactory = new HashBuilderOperatorFactory(1, new PlanNodeId("build"), lookupSourceFactoryManager, rangeList(buildPages.getTypes().size()), hashChannels, buildPages.getHashChannel().map(OptionalInt::of).orElse(OptionalInt.empty()), filterFunctionFactory, Optional.empty(), ImmutableList.of(), 100, new PagesIndex.TestingFactory(false), spillEnabled, singleStreamSpillerFactory, incrementalLoadFactorHashArraySizeSupplier(taskContext.getSession()));
    return new BuildSideSetup(lookupSourceFactoryManager, buildOperatorFactory, sourceOperatorFactory, partitionCount);
}

Example 4 with HashBuilderOperatorFactory

use of io.trino.operator.join.HashBuilderOperator.HashBuilderOperatorFactory in project trino by trinodb.

the class HashBuildBenchmark method createDrivers.

@Override
protected List<Driver> createDrivers(TaskContext taskContext) {
    // hash build
    List<Type> ordersTypes = getColumnTypes("orders", "orderkey", "totalprice");
    OperatorFactory ordersTableScan = createTableScanOperator(0, new PlanNodeId("test"), "orders", "orderkey", "totalprice");
    BlockTypeOperators blockTypeOperators = new BlockTypeOperators(new TypeOperators());
    JoinBridgeManager<PartitionedLookupSourceFactory> lookupSourceFactoryManager = JoinBridgeManager.lookupAllAtOnce(new PartitionedLookupSourceFactory(ordersTypes, ImmutableList.of(0, 1).stream().map(ordersTypes::get).collect(toImmutableList()), Ints.asList(0).stream().map(ordersTypes::get).collect(toImmutableList()), 1, false, blockTypeOperators));
    HashBuilderOperatorFactory hashBuilder = new HashBuilderOperatorFactory(1, new PlanNodeId("test"), lookupSourceFactoryManager, ImmutableList.of(0, 1), Ints.asList(0), OptionalInt.empty(), Optional.empty(), Optional.empty(), ImmutableList.of(), 1_500_000, new PagesIndex.TestingFactory(false), false, SingleStreamSpillerFactory.unsupportedSingleStreamSpillerFactory(), incrementalLoadFactorHashArraySizeSupplier(session));
    DriverFactory hashBuildDriverFactory = new DriverFactory(0, true, true, ImmutableList.of(ordersTableScan, hashBuilder), OptionalInt.empty(), UNGROUPED_EXECUTION);
    // empty join so build finishes
    ImmutableList.Builder<OperatorFactory> joinDriversBuilder = ImmutableList.builder();
    joinDriversBuilder.add(new ValuesOperatorFactory(0, new PlanNodeId("values"), ImmutableList.of()));
    OperatorFactory joinOperator = operatorFactories.innerJoin(2, new PlanNodeId("test"), lookupSourceFactoryManager, false, false, false, ImmutableList.of(BIGINT), Ints.asList(0), OptionalInt.empty(), Optional.empty(), OptionalInt.empty(), unsupportedPartitioningSpillerFactory(), blockTypeOperators);
    joinDriversBuilder.add(joinOperator);
    joinDriversBuilder.add(new NullOutputOperatorFactory(3, new PlanNodeId("test")));
    DriverFactory joinDriverFactory = new DriverFactory(1, true, true, joinDriversBuilder.build(), OptionalInt.empty(), UNGROUPED_EXECUTION);
    Driver hashBuildDriver = hashBuildDriverFactory.createDriver(taskContext.addPipelineContext(0, true, true, false).addDriverContext());
    hashBuildDriverFactory.noMoreDrivers();
    Driver joinDriver = joinDriverFactory.createDriver(taskContext.addPipelineContext(1, true, true, false).addDriverContext());
    joinDriverFactory.noMoreDrivers();
    return ImmutableList.of(hashBuildDriver, joinDriver);
}

Example 5 with HashBuilderOperatorFactory

use of io.trino.operator.join.HashBuilderOperator.HashBuilderOperatorFactory in project trino by trinodb.

the class HashJoinBenchmark method createDrivers.

/*
    select orderkey, quantity, totalprice
    from lineitem join orders using (orderkey)
     */
@Override
protected List<Driver> createDrivers(TaskContext taskContext) {
    if (probeDriverFactory == null) {
        List<Type> ordersTypes = getColumnTypes("orders", "orderkey", "totalprice");
        OperatorFactory ordersTableScan = createTableScanOperator(0, new PlanNodeId("test"), "orders", "orderkey", "totalprice");
        BlockTypeOperators blockTypeOperators = new BlockTypeOperators(new TypeOperators());
        JoinBridgeManager<PartitionedLookupSourceFactory> lookupSourceFactoryManager = JoinBridgeManager.lookupAllAtOnce(new PartitionedLookupSourceFactory(ordersTypes, ImmutableList.of(0, 1).stream().map(ordersTypes::get).collect(toImmutableList()), Ints.asList(0).stream().map(ordersTypes::get).collect(toImmutableList()), 1, false, blockTypeOperators));
        HashBuilderOperatorFactory hashBuilder = new HashBuilderOperatorFactory(1, new PlanNodeId("test"), lookupSourceFactoryManager, ImmutableList.of(0, 1), Ints.asList(0), OptionalInt.empty(), Optional.empty(), Optional.empty(), ImmutableList.of(), 1_500_000, new PagesIndex.TestingFactory(false), false, SingleStreamSpillerFactory.unsupportedSingleStreamSpillerFactory(), incrementalLoadFactorHashArraySizeSupplier(session));
        DriverContext driverContext = taskContext.addPipelineContext(0, false, false, false).addDriverContext();
        DriverFactory buildDriverFactory = new DriverFactory(0, false, false, ImmutableList.of(ordersTableScan, hashBuilder), OptionalInt.empty(), UNGROUPED_EXECUTION);
        List<Type> lineItemTypes = getColumnTypes("lineitem", "orderkey", "quantity");
        OperatorFactory lineItemTableScan = createTableScanOperator(0, new PlanNodeId("test"), "lineitem", "orderkey", "quantity");
        OperatorFactory joinOperator = operatorFactories.innerJoin(1, new PlanNodeId("test"), lookupSourceFactoryManager, false, false, false, lineItemTypes, Ints.asList(0), OptionalInt.empty(), Optional.empty(), OptionalInt.empty(), unsupportedPartitioningSpillerFactory(), blockTypeOperators);
        NullOutputOperatorFactory output = new NullOutputOperatorFactory(2, new PlanNodeId("test"));
        this.probeDriverFactory = new DriverFactory(1, true, true, ImmutableList.of(lineItemTableScan, joinOperator, output), OptionalInt.empty(), UNGROUPED_EXECUTION);
        Driver driver = buildDriverFactory.createDriver(driverContext);
        Future<LookupSourceProvider> lookupSourceProvider = lookupSourceFactoryManager.getJoinBridge(Lifespan.taskWide()).createLookupSourceProvider();
        while (!lookupSourceProvider.isDone()) {
            driver.process();
        }
        getFutureValue(lookupSourceProvider).close();
    }
    DriverContext driverContext = taskContext.addPipelineContext(1, true, true, false).addDriverContext();
    Driver driver = probeDriverFactory.createDriver(driverContext);
    return ImmutableList.of(driver);
}