Search in sources :

Example 1 with TimerReference

use of org.apache.beam.runners.core.construction.graph.TimerReference in project beam by apache.

the class ProcessBundleDescriptorsTest method testLengthPrefixingOfKeyCoderInStatefulExecutableStage.

/**
 * Tests that a stateful stage will wrap the key coder of a stateful transform in a
 * LengthPrefixCoder.
 */
@Test
public void testLengthPrefixingOfKeyCoderInStatefulExecutableStage() throws Exception {
    // Add another stateful stage with a non-standard key coder
    Pipeline p = Pipeline.create();
    Coder<Void> keycoder = VoidCoder.of();
    assertThat(ModelCoderRegistrar.isKnownCoder(keycoder), is(false));
    p.apply("impulse", Impulse.create()).apply("create", ParDo.of(new DoFn<byte[], KV<Void, String>>() {

        @ProcessElement
        public void process(ProcessContext ctxt) {
        }
    })).setCoder(KvCoder.of(keycoder, StringUtf8Coder.of())).apply("userState", ParDo.of(new DoFn<KV<Void, String>, KV<Void, String>>() {

        @StateId("stateId")
        private final StateSpec<BagState<String>> bufferState = StateSpecs.bag(StringUtf8Coder.of());

        @TimerId("timerId")
        private final TimerSpec timerSpec = TimerSpecs.timer(TimeDomain.EVENT_TIME);

        @ProcessElement
        public void processElement(@Element KV<Void, String> element, @StateId("stateId") BagState<String> state, @TimerId("timerId") Timer timer, OutputReceiver<KV<Void, String>> r) {
        }

        @OnTimer("timerId")
        public void onTimer() {
        }
    })).apply("gbk", GroupByKey.create());
    RunnerApi.Pipeline pipelineProto = PipelineTranslation.toProto(p);
    FusedPipeline fused = GreedyPipelineFuser.fuse(pipelineProto);
    Optional<ExecutableStage> optionalStage = Iterables.tryFind(fused.getFusedStages(), (ExecutableStage stage) -> stage.getUserStates().stream().anyMatch(spec -> spec.localName().equals("stateId")));
    checkState(optionalStage.isPresent(), "Expected a stage with user state.");
    ExecutableStage stage = optionalStage.get();
    PipelineNode.PCollectionNode inputPCollection = stage.getInputPCollection();
    // Ensure original key coder is not a LengthPrefixCoder
    Map<String, RunnerApi.Coder> stageCoderMap = stage.getComponents().getCodersMap();
    RunnerApi.Coder originalMainInputCoder = stageCoderMap.get(inputPCollection.getPCollection().getCoderId());
    String originalKeyCoderId = ModelCoders.getKvCoderComponents(originalMainInputCoder).keyCoderId();
    RunnerApi.Coder originalKeyCoder = stageCoderMap.get(originalKeyCoderId);
    assertThat(originalKeyCoder.getSpec().getUrn(), is(CoderTranslation.JAVA_SERIALIZED_CODER_URN));
    // Now create ProcessBundleDescriptor and check for the LengthPrefixCoder around the key coder
    BeamFnApi.ProcessBundleDescriptor pbd = ProcessBundleDescriptors.fromExecutableStage("test_stage", stage, Endpoints.ApiServiceDescriptor.getDefaultInstance()).getProcessBundleDescriptor();
    Map<String, RunnerApi.Coder> pbsCoderMap = pbd.getCodersMap();
    RunnerApi.Coder pbsMainInputCoder = pbsCoderMap.get(pbd.getPcollectionsOrThrow(inputPCollection.getId()).getCoderId());
    String keyCoderId = ModelCoders.getKvCoderComponents(pbsMainInputCoder).keyCoderId();
    RunnerApi.Coder keyCoder = pbsCoderMap.get(keyCoderId);
    ensureLengthPrefixed(keyCoder, originalKeyCoder, pbsCoderMap);
    TimerReference timerRef = Iterables.getOnlyElement(stage.getTimers());
    String timerTransformId = timerRef.transform().getId();
    RunnerApi.ParDoPayload parDoPayload = RunnerApi.ParDoPayload.parseFrom(pbd.getTransformsOrThrow(timerTransformId).getSpec().getPayload());
    RunnerApi.TimerFamilySpec timerSpec = parDoPayload.getTimerFamilySpecsOrThrow(timerRef.localName());
    RunnerApi.Coder timerCoder = pbsCoderMap.get(timerSpec.getTimerFamilyCoderId());
    String timerKeyCoderId = timerCoder.getComponentCoderIds(0);
    RunnerApi.Coder timerKeyCoder = pbsCoderMap.get(timerKeyCoderId);
    ensureLengthPrefixed(timerKeyCoder, originalKeyCoder, pbsCoderMap);
}
Also used : CoreMatchers.is(org.hamcrest.CoreMatchers.is) Endpoints(org.apache.beam.model.pipeline.v1.Endpoints) StateSpec(org.apache.beam.sdk.state.StateSpec) KV(org.apache.beam.sdk.values.KV) CoderTranslation(org.apache.beam.runners.core.construction.CoderTranslation) TimerSpecs(org.apache.beam.sdk.state.TimerSpecs) Coder(org.apache.beam.sdk.coders.Coder) Impulse(org.apache.beam.sdk.transforms.Impulse) GreedyPipelineFuser(org.apache.beam.runners.core.construction.graph.GreedyPipelineFuser) PipelineTranslation(org.apache.beam.runners.core.construction.PipelineTranslation) Optional(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Optional) StringUtf8Coder(org.apache.beam.sdk.coders.StringUtf8Coder) TimerSpec(org.apache.beam.sdk.state.TimerSpec) Map(java.util.Map) Iterables(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.Iterables) TimerReference(org.apache.beam.runners.core.construction.graph.TimerReference) MatcherAssert.assertThat(org.hamcrest.MatcherAssert.assertThat) ModelCoderRegistrar(org.apache.beam.runners.core.construction.ModelCoderRegistrar) Pipeline(org.apache.beam.sdk.Pipeline) ProcessContext(org.apache.beam.sdk.transforms.DoFn.ProcessContext) RestrictionTracker(org.apache.beam.sdk.transforms.splittabledofn.RestrictionTracker) RunnerApi(org.apache.beam.model.pipeline.v1.RunnerApi) FusedPipeline(org.apache.beam.runners.core.construction.graph.FusedPipeline) DoFn(org.apache.beam.sdk.transforms.DoFn) KvCoder(org.apache.beam.sdk.coders.KvCoder) GroupByKey(org.apache.beam.sdk.transforms.GroupByKey) PTransformTranslation(org.apache.beam.runners.core.construction.PTransformTranslation) ProcessElement(org.apache.beam.sdk.transforms.DoFn.ProcessElement) Test(org.junit.Test) BeamFnApi(org.apache.beam.model.fnexecution.v1.BeamFnApi) ExecutableStage(org.apache.beam.runners.core.construction.graph.ExecutableStage) ModelCoders(org.apache.beam.runners.core.construction.ModelCoders) Serializable(java.io.Serializable) SplittableParDoExpander(org.apache.beam.runners.core.construction.graph.SplittableParDoExpander) BagState(org.apache.beam.sdk.state.BagState) StateSpecs(org.apache.beam.sdk.state.StateSpecs) ParDo(org.apache.beam.sdk.transforms.ParDo) ProtoOverrides(org.apache.beam.runners.core.construction.graph.ProtoOverrides) Timer(org.apache.beam.sdk.state.Timer) Preconditions.checkState(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions.checkState) PipelineNode(org.apache.beam.runners.core.construction.graph.PipelineNode) VoidCoder(org.apache.beam.sdk.coders.VoidCoder) TimeDomain(org.apache.beam.sdk.state.TimeDomain) TimerReference(org.apache.beam.runners.core.construction.graph.TimerReference) ProcessContext(org.apache.beam.sdk.transforms.DoFn.ProcessContext) RunnerApi(org.apache.beam.model.pipeline.v1.RunnerApi) ExecutableStage(org.apache.beam.runners.core.construction.graph.ExecutableStage) BagState(org.apache.beam.sdk.state.BagState) TimerSpec(org.apache.beam.sdk.state.TimerSpec) Coder(org.apache.beam.sdk.coders.Coder) StringUtf8Coder(org.apache.beam.sdk.coders.StringUtf8Coder) KvCoder(org.apache.beam.sdk.coders.KvCoder) VoidCoder(org.apache.beam.sdk.coders.VoidCoder) BeamFnApi(org.apache.beam.model.fnexecution.v1.BeamFnApi) FusedPipeline(org.apache.beam.runners.core.construction.graph.FusedPipeline) KV(org.apache.beam.sdk.values.KV) PipelineNode(org.apache.beam.runners.core.construction.graph.PipelineNode) Pipeline(org.apache.beam.sdk.Pipeline) FusedPipeline(org.apache.beam.runners.core.construction.graph.FusedPipeline) Timer(org.apache.beam.sdk.state.Timer) ProcessElement(org.apache.beam.sdk.transforms.DoFn.ProcessElement) Test(org.junit.Test)

Example 2 with TimerReference

use of org.apache.beam.runners.core.construction.graph.TimerReference in project beam by apache.

the class ProcessBundleDescriptors method forTimerSpecs.

private static Map<String, Map<String, TimerSpec>> forTimerSpecs(ExecutableStage stage, Components.Builder components) throws IOException {
    ImmutableTable.Builder<String, String, TimerSpec> idsToSpec = ImmutableTable.builder();
    for (TimerReference timerReference : stage.getTimers()) {
        RunnerApi.ParDoPayload payload = RunnerApi.ParDoPayload.parseFrom(timerReference.transform().getTransform().getSpec().getPayload());
        RunnerApi.TimerFamilySpec timerFamilySpec = payload.getTimerFamilySpecsOrThrow(timerReference.localName());
        org.apache.beam.sdk.state.TimerSpec spec;
        switch(timerFamilySpec.getTimeDomain()) {
            case EVENT_TIME:
                spec = TimerSpecs.timer(TimeDomain.EVENT_TIME);
                break;
            case PROCESSING_TIME:
                spec = TimerSpecs.timer(TimeDomain.PROCESSING_TIME);
                break;
            default:
                throw new IllegalArgumentException(String.format("Unknown or unsupported time domain %s", timerFamilySpec.getTimeDomain()));
        }
        for (WireCoderSetting wireCoderSetting : stage.getWireCoderSettings()) {
            if (wireCoderSetting.hasTimer() && wireCoderSetting.getTimer().getTransformId().equals(timerReference.transform().getId()) && wireCoderSetting.getTimer().getLocalName().equals(timerReference.localName())) {
                throw new UnsupportedOperationException("WireCoderSetting for timer is yet to be supported.");
            }
        }
        String originalTimerCoderId = timerFamilySpec.getTimerFamilyCoderId();
        String sdkCoderId = LengthPrefixUnknownCoders.addLengthPrefixedCoder(originalTimerCoderId, components, false);
        String runnerCoderId = LengthPrefixUnknownCoders.addLengthPrefixedCoder(originalTimerCoderId, components, true);
        Coder<?> timerCoder = RehydratedComponents.forComponents(components.build()).getCoder(runnerCoderId);
        checkArgument(timerCoder instanceof Timer.Coder, "Expected a timer coder but received %s.", timerCoder);
        RunnerApi.FunctionSpec.Builder updatedSpec = components.getTransformsOrThrow(timerReference.transform().getId()).toBuilder().getSpecBuilder();
        RunnerApi.ParDoPayload.Builder updatedPayload = RunnerApi.ParDoPayload.parseFrom(updatedSpec.getPayload()).toBuilder();
        updatedPayload.putTimerFamilySpecs(timerReference.localName(), updatedPayload.getTimerFamilySpecsOrThrow(timerReference.localName()).toBuilder().setTimerFamilyCoderId(sdkCoderId).build());
        updatedSpec.setPayload(updatedPayload.build().toByteString());
        components.putTransforms(timerReference.transform().getId(), // and not the original
        components.getTransformsOrThrow(timerReference.transform().getId()).toBuilder().setSpec(updatedSpec).build());
        idsToSpec.put(timerReference.transform().getId(), timerReference.localName(), TimerSpec.of(timerReference.transform().getId(), timerReference.localName(), spec, (Coder) timerCoder));
    }
    return idsToSpec.build().rowMap();
}
Also used : Coder(org.apache.beam.sdk.coders.Coder) ByteStringCoder(org.apache.beam.runners.fnexecution.wire.ByteStringCoder) FullWindowedValueCoder(org.apache.beam.sdk.util.WindowedValue.FullWindowedValueCoder) TimerReference(org.apache.beam.runners.core.construction.graph.TimerReference) WireCoderSetting(org.apache.beam.model.pipeline.v1.RunnerApi.ExecutableStagePayload.WireCoderSetting) ImmutableTable(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.ImmutableTable) RunnerApi(org.apache.beam.model.pipeline.v1.RunnerApi) Timer(org.apache.beam.runners.core.construction.Timer)

Example 3 with TimerReference

use of org.apache.beam.runners.core.construction.graph.TimerReference in project beam by apache.

the class CreateExecutableStageNodeFunction method apply.

@Override
public Node apply(MutableNetwork<Node, Edge> input) {
    for (Node node : input.nodes()) {
        if (node instanceof RemoteGrpcPortNode || node instanceof ParallelInstructionNode || node instanceof InstructionOutputNode) {
            continue;
        }
        throw new IllegalArgumentException(String.format("Network contains unknown type of node: %s", input));
    }
    // Fix all non output nodes to have named edges.
    for (Node node : input.nodes()) {
        if (node instanceof InstructionOutputNode) {
            continue;
        }
        for (Node successor : input.successors(node)) {
            for (Edge edge : input.edgesConnecting(node, successor)) {
                if (edge instanceof DefaultEdge) {
                    input.removeEdge(edge);
                    input.addEdge(node, successor, MultiOutputInfoEdge.create(new MultiOutputInfo().setTag(idGenerator.getId())));
                }
            }
        }
    }
    RunnerApi.Components.Builder componentsBuilder = RunnerApi.Components.newBuilder();
    componentsBuilder.mergeFrom(this.pipeline.getComponents());
    // Default to use the Java environment if pipeline doesn't have environment specified.
    if (pipeline.getComponents().getEnvironmentsMap().isEmpty()) {
        String envId = Environments.JAVA_SDK_HARNESS_ENVIRONMENT.getUrn() + idGenerator.getId();
        componentsBuilder.putEnvironments(envId, Environments.JAVA_SDK_HARNESS_ENVIRONMENT);
    }
    // By default, use GlobalWindow for all languages.
    // For java, if there is a IntervalWindowCoder, then use FixedWindow instead.
    // TODO: should get real WindowingStategy from pipeline proto.
    String globalWindowingStrategyId = "generatedGlobalWindowingStrategy" + idGenerator.getId();
    String intervalWindowEncodingWindowingStrategyId = "generatedIntervalWindowEncodingWindowingStrategy" + idGenerator.getId();
    SdkComponents sdkComponents = SdkComponents.create(pipeline.getComponents(), null);
    try {
        registerWindowingStrategy(globalWindowingStrategyId, WindowingStrategy.globalDefault(), componentsBuilder, sdkComponents);
        registerWindowingStrategy(intervalWindowEncodingWindowingStrategyId, WindowingStrategy.of(FixedWindows.of(Duration.standardSeconds(1))), componentsBuilder, sdkComponents);
    } catch (IOException exc) {
        throw new RuntimeException("Could not convert default windowing stratey to proto", exc);
    }
    Map<Node, String> nodesToPCollections = new HashMap<>();
    ImmutableMap.Builder<String, NameContext> ptransformIdToNameContexts = ImmutableMap.builder();
    ImmutableMap.Builder<String, Iterable<SideInputInfo>> ptransformIdToSideInputInfos = ImmutableMap.builder();
    ImmutableMap.Builder<String, Iterable<PCollectionView<?>>> ptransformIdToPCollectionViews = ImmutableMap.builder();
    // A field of ExecutableStage which includes the PCollection goes to worker side.
    Set<PCollectionNode> executableStageOutputs = new HashSet<>();
    // A field of ExecutableStage which includes the PCollection goes to runner side.
    Set<PCollectionNode> executableStageInputs = new HashSet<>();
    for (InstructionOutputNode node : Iterables.filter(input.nodes(), InstructionOutputNode.class)) {
        InstructionOutput instructionOutput = node.getInstructionOutput();
        String coderId = "generatedCoder" + idGenerator.getId();
        String windowingStrategyId;
        try (ByteString.Output output = ByteString.newOutput()) {
            try {
                Coder<?> javaCoder = CloudObjects.coderFromCloudObject(CloudObject.fromSpec(instructionOutput.getCodec()));
                Coder<?> elementCoder = ((WindowedValueCoder<?>) javaCoder).getValueCoder();
                sdkComponents.registerCoder(elementCoder);
                RunnerApi.Coder coderProto = CoderTranslation.toProto(elementCoder, sdkComponents);
                componentsBuilder.putCoders(coderId, coderProto);
                // For now, Dataflow runner harness only deal with FixedWindow.
                if (javaCoder instanceof FullWindowedValueCoder) {
                    FullWindowedValueCoder<?> windowedValueCoder = (FullWindowedValueCoder<?>) javaCoder;
                    Coder<?> windowCoder = windowedValueCoder.getWindowCoder();
                    if (windowCoder instanceof IntervalWindowCoder) {
                        windowingStrategyId = intervalWindowEncodingWindowingStrategyId;
                    } else if (windowCoder instanceof GlobalWindow.Coder) {
                        windowingStrategyId = globalWindowingStrategyId;
                    } else {
                        throw new UnsupportedOperationException(String.format("Dataflow portable runner harness doesn't support windowing with %s", windowCoder));
                    }
                } else {
                    throw new UnsupportedOperationException("Dataflow portable runner harness only supports FullWindowedValueCoder");
                }
            } catch (IOException e) {
                throw new IllegalArgumentException(String.format("Unable to encode coder %s for output %s", instructionOutput.getCodec(), instructionOutput), e);
            } catch (Exception e) {
                // Coder probably wasn't a java coder
                OBJECT_MAPPER.writeValue(output, instructionOutput.getCodec());
                componentsBuilder.putCoders(coderId, RunnerApi.Coder.newBuilder().setSpec(RunnerApi.FunctionSpec.newBuilder().setPayload(output.toByteString())).build());
                // For non-java coder, hope it's GlobalWindows by default.
                // TODO(BEAM-6231): Actually discover the right windowing strategy.
                windowingStrategyId = globalWindowingStrategyId;
            }
        } catch (IOException e) {
            throw new IllegalArgumentException(String.format("Unable to encode coder %s for output %s", instructionOutput.getCodec(), instructionOutput), e);
        }
        // TODO(BEAM-6275): Set correct IsBounded on generated PCollections
        String pcollectionId = node.getPcollectionId();
        RunnerApi.PCollection pCollection = RunnerApi.PCollection.newBuilder().setCoderId(coderId).setWindowingStrategyId(windowingStrategyId).setIsBounded(RunnerApi.IsBounded.Enum.BOUNDED).build();
        nodesToPCollections.put(node, pcollectionId);
        componentsBuilder.putPcollections(pcollectionId, pCollection);
        // is set
        if (isExecutableStageOutputPCollection(input, node)) {
            executableStageOutputs.add(PipelineNode.pCollection(pcollectionId, pCollection));
        }
        if (isExecutableStageInputPCollection(input, node)) {
            executableStageInputs.add(PipelineNode.pCollection(pcollectionId, pCollection));
        }
    }
    componentsBuilder.putAllCoders(sdkComponents.toComponents().getCodersMap());
    Set<PTransformNode> executableStageTransforms = new HashSet<>();
    Set<TimerReference> executableStageTimers = new HashSet<>();
    List<UserStateId> userStateIds = new ArrayList<>();
    Set<SideInputReference> executableStageSideInputs = new HashSet<>();
    for (ParallelInstructionNode node : Iterables.filter(input.nodes(), ParallelInstructionNode.class)) {
        ImmutableMap.Builder<String, PCollectionNode> sideInputIds = ImmutableMap.builder();
        ParallelInstruction parallelInstruction = node.getParallelInstruction();
        String ptransformId = "generatedPtransform" + idGenerator.getId();
        ptransformIdToNameContexts.put(ptransformId, NameContext.create(null, parallelInstruction.getOriginalName(), parallelInstruction.getSystemName(), parallelInstruction.getName()));
        RunnerApi.PTransform.Builder pTransform = RunnerApi.PTransform.newBuilder();
        RunnerApi.FunctionSpec.Builder transformSpec = RunnerApi.FunctionSpec.newBuilder();
        List<String> timerIds = new ArrayList<>();
        if (parallelInstruction.getParDo() != null) {
            ParDoInstruction parDoInstruction = parallelInstruction.getParDo();
            CloudObject userFnSpec = CloudObject.fromSpec(parDoInstruction.getUserFn());
            String userFnClassName = userFnSpec.getClassName();
            if (userFnClassName.equals("CombineValuesFn") || userFnClassName.equals("KeyedCombineFn")) {
                transformSpec = transformCombineValuesFnToFunctionSpec(userFnSpec);
                ptransformIdToPCollectionViews.put(ptransformId, Collections.emptyList());
            } else {
                String parDoPTransformId = getString(userFnSpec, PropertyNames.SERIALIZED_FN);
                RunnerApi.PTransform parDoPTransform = pipeline.getComponents().getTransformsOrDefault(parDoPTransformId, null);
                // TODO: only the non-null branch should exist; for migration ease only
                if (parDoPTransform != null) {
                    checkArgument(parDoPTransform.getSpec().getUrn().equals(PTransformTranslation.PAR_DO_TRANSFORM_URN), "Found transform \"%s\" for ParallelDo instruction, " + " but that transform had unexpected URN \"%s\" (expected \"%s\")", parDoPTransformId, parDoPTransform.getSpec().getUrn(), PTransformTranslation.PAR_DO_TRANSFORM_URN);
                    RunnerApi.ParDoPayload parDoPayload;
                    try {
                        parDoPayload = RunnerApi.ParDoPayload.parseFrom(parDoPTransform.getSpec().getPayload());
                    } catch (InvalidProtocolBufferException exc) {
                        throw new RuntimeException("ParDo did not have a ParDoPayload", exc);
                    }
                    // user timers and user state.
                    for (Map.Entry<String, RunnerApi.TimerFamilySpec> entry : parDoPayload.getTimerFamilySpecsMap().entrySet()) {
                        timerIds.add(entry.getKey());
                    }
                    for (Map.Entry<String, RunnerApi.StateSpec> entry : parDoPayload.getStateSpecsMap().entrySet()) {
                        UserStateId.Builder builder = UserStateId.newBuilder();
                        builder.setTransformId(parDoPTransformId);
                        builder.setLocalName(entry.getKey());
                        userStateIds.add(builder.build());
                    }
                    // To facilitate the creation of Set executableStageSideInputs.
                    for (String sideInputTag : parDoPayload.getSideInputsMap().keySet()) {
                        String sideInputPCollectionId = parDoPTransform.getInputsOrThrow(sideInputTag);
                        RunnerApi.PCollection sideInputPCollection = pipeline.getComponents().getPcollectionsOrThrow(sideInputPCollectionId);
                        pTransform.putInputs(sideInputTag, sideInputPCollectionId);
                        PCollectionNode pCollectionNode = PipelineNode.pCollection(sideInputPCollectionId, sideInputPCollection);
                        sideInputIds.put(sideInputTag, pCollectionNode);
                    }
                    // To facilitate the creation of Map(ptransformId -> pCollectionView), which is
                    // required by constructing an ExecutableStageNode.
                    ImmutableList.Builder<PCollectionView<?>> pcollectionViews = ImmutableList.builder();
                    for (Map.Entry<String, RunnerApi.SideInput> sideInputEntry : parDoPayload.getSideInputsMap().entrySet()) {
                        pcollectionViews.add(RegisterNodeFunction.transformSideInputForRunner(pipeline, parDoPTransform, sideInputEntry.getKey(), sideInputEntry.getValue()));
                    }
                    ptransformIdToPCollectionViews.put(ptransformId, pcollectionViews.build());
                    transformSpec.setUrn(PTransformTranslation.PAR_DO_TRANSFORM_URN).setPayload(parDoPayload.toByteString());
                } else {
                    // legacy path - bytes are the FunctionSpec's payload field, basically, and
                    // SDKs expect it in the PTransform's payload field
                    byte[] userFnBytes = getBytes(userFnSpec, PropertyNames.SERIALIZED_FN);
                    transformSpec.setUrn(ParDoTranslation.CUSTOM_JAVA_DO_FN_URN).setPayload(ByteString.copyFrom(userFnBytes));
                }
                if (parDoInstruction.getSideInputs() != null) {
                    ptransformIdToSideInputInfos.put(ptransformId, forSideInputInfos(parDoInstruction.getSideInputs(), true));
                }
            }
        } else if (parallelInstruction.getRead() != null) {
            ReadInstruction readInstruction = parallelInstruction.getRead();
            CloudObject sourceSpec = CloudObject.fromSpec(CloudSourceUtils.flattenBaseSpecs(readInstruction.getSource()).getSpec());
            // TODO: Need to plumb through the SDK specific function spec.
            transformSpec.setUrn(JAVA_SOURCE_URN);
            try {
                byte[] serializedSource = Base64.getDecoder().decode(getString(sourceSpec, SERIALIZED_SOURCE));
                ByteString sourceByteString = ByteString.copyFrom(serializedSource);
                transformSpec.setPayload(sourceByteString);
            } catch (Exception e) {
                throw new IllegalArgumentException(String.format("Unable to process Read %s", parallelInstruction), e);
            }
        } else if (parallelInstruction.getFlatten() != null) {
            transformSpec.setUrn(PTransformTranslation.FLATTEN_TRANSFORM_URN);
        } else {
            throw new IllegalArgumentException(String.format("Unknown type of ParallelInstruction %s", parallelInstruction));
        }
        // predecessor in a ParDo. This PCollection is called the "main input".
        for (Node predecessorOutput : input.predecessors(node)) {
            pTransform.putInputs("generatedInput" + idGenerator.getId(), nodesToPCollections.get(predecessorOutput));
        }
        for (Edge edge : input.outEdges(node)) {
            Node nodeOutput = input.incidentNodes(edge).target();
            MultiOutputInfoEdge edge2 = (MultiOutputInfoEdge) edge;
            pTransform.putOutputs(edge2.getMultiOutputInfo().getTag(), nodesToPCollections.get(nodeOutput));
        }
        pTransform.setSpec(transformSpec);
        PTransformNode pTransformNode = PipelineNode.pTransform(ptransformId, pTransform.build());
        executableStageTransforms.add(pTransformNode);
        for (String timerId : timerIds) {
            executableStageTimers.add(TimerReference.of(pTransformNode, timerId));
        }
        ImmutableMap<String, PCollectionNode> sideInputIdToPCollectionNodes = sideInputIds.build();
        for (String sideInputTag : sideInputIdToPCollectionNodes.keySet()) {
            SideInputReference sideInputReference = SideInputReference.of(pTransformNode, sideInputTag, sideInputIdToPCollectionNodes.get(sideInputTag));
            executableStageSideInputs.add(sideInputReference);
        }
        executableStageTransforms.add(pTransformNode);
    }
    if (executableStageInputs.size() != 1) {
        throw new UnsupportedOperationException("ExecutableStage only support one input PCollection");
    }
    PCollectionNode executableInput = executableStageInputs.iterator().next();
    RunnerApi.Components executableStageComponents = componentsBuilder.build();
    // Get Environment from ptransform, otherwise, use JAVA_SDK_HARNESS_ENVIRONMENT as default.
    Environment executableStageEnv = getEnvironmentFromPTransform(executableStageComponents, executableStageTransforms);
    if (executableStageEnv == null) {
        executableStageEnv = Environments.JAVA_SDK_HARNESS_ENVIRONMENT;
    }
    Set<UserStateReference> executableStageUserStateReference = new HashSet<>();
    for (UserStateId userStateId : userStateIds) {
        executableStageUserStateReference.add(UserStateReference.fromUserStateId(userStateId, executableStageComponents));
    }
    ExecutableStage executableStage = ImmutableExecutableStage.ofFullComponents(executableStageComponents, executableStageEnv, executableInput, executableStageSideInputs, executableStageUserStateReference, executableStageTimers, executableStageTransforms, executableStageOutputs, DEFAULT_WIRE_CODER_SETTINGS);
    return ExecutableStageNode.create(executableStage, ptransformIdToNameContexts.build(), ptransformIdToSideInputInfos.build(), ptransformIdToPCollectionViews.build());
}
Also used : HashMap(java.util.HashMap) MultiOutputInfo(com.google.api.services.dataflow.model.MultiOutputInfo) ByteString(org.apache.beam.vendor.grpc.v1p43p2.com.google.protobuf.ByteString) PCollectionNode(org.apache.beam.runners.core.construction.graph.PipelineNode.PCollectionNode) InstructionOutputNode(org.apache.beam.runners.dataflow.worker.graph.Nodes.InstructionOutputNode) PTransformNode(org.apache.beam.runners.core.construction.graph.PipelineNode.PTransformNode) ParallelInstructionNode(org.apache.beam.runners.dataflow.worker.graph.Nodes.ParallelInstructionNode) Node(org.apache.beam.runners.dataflow.worker.graph.Nodes.Node) ExecutableStageNode(org.apache.beam.runners.dataflow.worker.graph.Nodes.ExecutableStageNode) PipelineNode(org.apache.beam.runners.core.construction.graph.PipelineNode) RemoteGrpcPortNode(org.apache.beam.runners.dataflow.worker.graph.Nodes.RemoteGrpcPortNode) InstructionOutput(com.google.api.services.dataflow.model.InstructionOutput) PTransformNode(org.apache.beam.runners.core.construction.graph.PipelineNode.PTransformNode) ArrayList(java.util.ArrayList) Structs.getString(org.apache.beam.runners.dataflow.util.Structs.getString) ByteString(org.apache.beam.vendor.grpc.v1p43p2.com.google.protobuf.ByteString) RehydratedComponents(org.apache.beam.runners.core.construction.RehydratedComponents) SdkComponents(org.apache.beam.runners.core.construction.SdkComponents) RunnerApi(org.apache.beam.model.pipeline.v1.RunnerApi) SideInputReference(org.apache.beam.runners.core.construction.graph.SideInputReference) ImmutableExecutableStage(org.apache.beam.runners.core.construction.graph.ImmutableExecutableStage) ExecutableStage(org.apache.beam.runners.core.construction.graph.ExecutableStage) HashSet(java.util.HashSet) DefaultEdge(org.apache.beam.runners.dataflow.worker.graph.Edges.DefaultEdge) MultiOutputInfoEdge(org.apache.beam.runners.dataflow.worker.graph.Edges.MultiOutputInfoEdge) ImmutableMap(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.ImmutableMap) ParDoInstruction(com.google.api.services.dataflow.model.ParDoInstruction) PCollectionView(org.apache.beam.sdk.values.PCollectionView) Environment(org.apache.beam.model.pipeline.v1.RunnerApi.Environment) GlobalWindow(org.apache.beam.sdk.transforms.windowing.GlobalWindow) ImmutableMap(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.ImmutableMap) Map(java.util.Map) HashMap(java.util.HashMap) IntervalWindowCoder(org.apache.beam.sdk.transforms.windowing.IntervalWindow.IntervalWindowCoder) RemoteGrpcPortNode(org.apache.beam.runners.dataflow.worker.graph.Nodes.RemoteGrpcPortNode) TimerReference(org.apache.beam.runners.core.construction.graph.TimerReference) ImmutableList(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.ImmutableList) ParallelInstructionNode(org.apache.beam.runners.dataflow.worker.graph.Nodes.ParallelInstructionNode) SdkComponents(org.apache.beam.runners.core.construction.SdkComponents) ReadInstruction(com.google.api.services.dataflow.model.ReadInstruction) InstructionOutputNode(org.apache.beam.runners.dataflow.worker.graph.Nodes.InstructionOutputNode) FullWindowedValueCoder(org.apache.beam.sdk.util.WindowedValue.FullWindowedValueCoder) NameContext(org.apache.beam.runners.dataflow.worker.counters.NameContext) InvalidProtocolBufferException(org.apache.beam.vendor.grpc.v1p43p2.com.google.protobuf.InvalidProtocolBufferException) IOException(java.io.IOException) PCollectionNode(org.apache.beam.runners.core.construction.graph.PipelineNode.PCollectionNode) InvalidProtocolBufferException(org.apache.beam.vendor.grpc.v1p43p2.com.google.protobuf.InvalidProtocolBufferException) IOException(java.io.IOException) ParallelInstruction(com.google.api.services.dataflow.model.ParallelInstruction) WindowedValueCoder(org.apache.beam.sdk.util.WindowedValue.WindowedValueCoder) FullWindowedValueCoder(org.apache.beam.sdk.util.WindowedValue.FullWindowedValueCoder) CloudObject(org.apache.beam.runners.dataflow.util.CloudObject) UserStateId(org.apache.beam.model.pipeline.v1.RunnerApi.ExecutableStagePayload.UserStateId) UserStateReference(org.apache.beam.runners.core.construction.graph.UserStateReference) Edge(org.apache.beam.runners.dataflow.worker.graph.Edges.Edge) MultiOutputInfoEdge(org.apache.beam.runners.dataflow.worker.graph.Edges.MultiOutputInfoEdge) DefaultEdge(org.apache.beam.runners.dataflow.worker.graph.Edges.DefaultEdge)

Aggregations

RunnerApi (org.apache.beam.model.pipeline.v1.RunnerApi)3 TimerReference (org.apache.beam.runners.core.construction.graph.TimerReference)3 Map (java.util.Map)2 ExecutableStage (org.apache.beam.runners.core.construction.graph.ExecutableStage)2 PipelineNode (org.apache.beam.runners.core.construction.graph.PipelineNode)2 Coder (org.apache.beam.sdk.coders.Coder)2 FullWindowedValueCoder (org.apache.beam.sdk.util.WindowedValue.FullWindowedValueCoder)2 InstructionOutput (com.google.api.services.dataflow.model.InstructionOutput)1 MultiOutputInfo (com.google.api.services.dataflow.model.MultiOutputInfo)1 ParDoInstruction (com.google.api.services.dataflow.model.ParDoInstruction)1 ParallelInstruction (com.google.api.services.dataflow.model.ParallelInstruction)1 ReadInstruction (com.google.api.services.dataflow.model.ReadInstruction)1 IOException (java.io.IOException)1 Serializable (java.io.Serializable)1 ArrayList (java.util.ArrayList)1 HashMap (java.util.HashMap)1 HashSet (java.util.HashSet)1 BeamFnApi (org.apache.beam.model.fnexecution.v1.BeamFnApi)1 Endpoints (org.apache.beam.model.pipeline.v1.Endpoints)1 Environment (org.apache.beam.model.pipeline.v1.RunnerApi.Environment)1