Examples with PTransformRunnerFactory org.apache.beam.fn.harness.PTransformRunnerFactory used on opensource projects

Example 6 with PTransformRunnerFactory

use of org.apache.beam.fn.harness.PTransformRunnerFactory in project beam by apache.

the class ProcessBundleHandlerTest method testOrderOfSetupTeardownCalls.

@Test
public void testOrderOfSetupTeardownCalls() throws Exception {
    DoFnWithExecutionInformation doFnWithExecutionInformation = DoFnWithExecutionInformation.of(new TestDoFn(), TestDoFn.mainOutput, Collections.emptyMap(), DoFnSchemaInformation.create());
    RunnerApi.FunctionSpec functionSpec = RunnerApi.FunctionSpec.newBuilder().setUrn(ParDoTranslation.CUSTOM_JAVA_DO_FN_URN).setPayload(ByteString.copyFrom(SerializableUtils.serializeToByteArray(doFnWithExecutionInformation))).build();
    RunnerApi.ParDoPayload parDoPayload = RunnerApi.ParDoPayload.newBuilder().setDoFn(functionSpec).build();
    BeamFnApi.ProcessBundleDescriptor processBundleDescriptor = BeamFnApi.ProcessBundleDescriptor.newBuilder().putTransforms("2L", PTransform.newBuilder().setSpec(RunnerApi.FunctionSpec.newBuilder().setUrn(DATA_INPUT_URN).build()).putOutputs("2L-output", "2L-output-pc").build()).putTransforms("3L", PTransform.newBuilder().setSpec(RunnerApi.FunctionSpec.newBuilder().setUrn(PTransformTranslation.PAR_DO_TRANSFORM_URN).setPayload(parDoPayload.toByteString())).putInputs("3L-input", "2L-output-pc").build()).putPcollections("2L-output-pc", PCollection.newBuilder().setWindowingStrategyId("window-strategy").setCoderId("2L-output-coder").setIsBounded(IsBounded.Enum.BOUNDED).build()).putWindowingStrategies("window-strategy", WindowingStrategy.newBuilder().setWindowCoderId("window-strategy-coder").setWindowFn(RunnerApi.FunctionSpec.newBuilder().setUrn("beam:window_fn:global_windows:v1")).setOutputTime(RunnerApi.OutputTime.Enum.END_OF_WINDOW).setAccumulationMode(RunnerApi.AccumulationMode.Enum.ACCUMULATING).setTrigger(RunnerApi.Trigger.newBuilder().setAlways(RunnerApi.Trigger.Always.getDefaultInstance())).setClosingBehavior(RunnerApi.ClosingBehavior.Enum.EMIT_ALWAYS).setOnTimeBehavior(RunnerApi.OnTimeBehavior.Enum.FIRE_ALWAYS).build()).putCoders("2L-output-coder", CoderTranslation.toProto(StringUtf8Coder.of()).getCoder()).putCoders("window-strategy-coder", Coder.newBuilder().setSpec(RunnerApi.FunctionSpec.newBuilder().setUrn(ModelCoders.GLOBAL_WINDOW_CODER_URN).build()).build()).build();
    Map<String, BeamFnApi.ProcessBundleDescriptor> fnApiRegistry = ImmutableMap.of("1L", processBundleDescriptor);
    Map<String, PTransformRunnerFactory> urnToPTransformRunnerFactoryMap = Maps.newHashMap(REGISTERED_RUNNER_FACTORIES);
    urnToPTransformRunnerFactoryMap.put(DATA_INPUT_URN, (context) -> null);
    ProcessBundleHandler handler = new ProcessBundleHandler(PipelineOptionsFactory.create(), Collections.emptySet(), fnApiRegistry::get, beamFnDataClient, null, /* beamFnStateClient */
    null, /* finalizeBundleHandler */
    new ShortIdMap(), urnToPTransformRunnerFactoryMap, Caches.noop(), new BundleProcessorCache());
    handler.processBundle(BeamFnApi.InstructionRequest.newBuilder().setInstructionId("998L").setProcessBundle(BeamFnApi.ProcessBundleRequest.newBuilder().setProcessBundleDescriptorId("1L")).build());
    handler.processBundle(BeamFnApi.InstructionRequest.newBuilder().setInstructionId("999L").setProcessBundle(BeamFnApi.ProcessBundleRequest.newBuilder().setProcessBundleDescriptorId("1L")).build());
    handler.shutdown();
    // setup and teardown should occur only once when processing multiple bundles for the same
    // descriptor
    assertThat(TestDoFn.orderOfOperations, contains("setUp", "startBundle", "finishBundle", "startBundle", "finishBundle", "tearDown"));
}

Example 7 with PTransformRunnerFactory

use of org.apache.beam.fn.harness.PTransformRunnerFactory in project beam by apache.

the class ProcessBundleHandlerTest method testPendingStateCallsBlockTillCompletion.

@Test
public void testPendingStateCallsBlockTillCompletion() throws Exception {
    BeamFnApi.ProcessBundleDescriptor processBundleDescriptor = BeamFnApi.ProcessBundleDescriptor.newBuilder().putTransforms("2L", RunnerApi.PTransform.newBuilder().setSpec(RunnerApi.FunctionSpec.newBuilder().setUrn(DATA_INPUT_URN).build()).build()).setStateApiServiceDescriptor(ApiServiceDescriptor.getDefaultInstance()).build();
    Map<String, BeamFnApi.ProcessBundleDescriptor> fnApiRegistry = ImmutableMap.of("1L", processBundleDescriptor);
    CompletableFuture<StateResponse>[] successfulResponse = new CompletableFuture[1];
    CompletableFuture<StateResponse>[] unsuccessfulResponse = new CompletableFuture[1];
    BeamFnStateGrpcClientCache mockBeamFnStateGrpcClient = Mockito.mock(BeamFnStateGrpcClientCache.class);
    BeamFnStateClient mockBeamFnStateClient = Mockito.mock(BeamFnStateClient.class);
    when(mockBeamFnStateGrpcClient.forApiServiceDescriptor(any())).thenReturn(mockBeamFnStateClient);
    doAnswer(invocation -> {
        StateRequest.Builder stateRequestBuilder = (StateRequest.Builder) invocation.getArguments()[0];
        CompletableFuture<StateResponse> completableFuture = new CompletableFuture<>();
        new Thread(() -> {
            // Simulate sleeping which introduces a race which most of the time requires
            // the ProcessBundleHandler to block.
            Uninterruptibles.sleepUninterruptibly(500, TimeUnit.MILLISECONDS);
            switch(stateRequestBuilder.getInstructionId()) {
                case "SUCCESS":
                    completableFuture.complete(StateResponse.getDefaultInstance());
                    break;
                case "FAIL":
                    completableFuture.completeExceptionally(new RuntimeException("TEST ERROR"));
            }
        }).start();
        return completableFuture;
    }).when(mockBeamFnStateClient).handle(any());
    ProcessBundleHandler handler = new ProcessBundleHandler(PipelineOptionsFactory.create(), Collections.emptySet(), fnApiRegistry::get, beamFnDataClient, mockBeamFnStateGrpcClient, null, /* finalizeBundleHandler */
    new ShortIdMap(), ImmutableMap.of(DATA_INPUT_URN, new PTransformRunnerFactory<Object>() {

        @Override
        public Object createRunnerForPTransform(Context context) throws IOException {
            BeamFnStateClient beamFnStateClient = context.getBeamFnStateClient();
            context.addStartBundleFunction(() -> doStateCalls(beamFnStateClient));
            return null;
        }

        private void doStateCalls(BeamFnStateClient beamFnStateClient) {
            successfulResponse[0] = beamFnStateClient.handle(StateRequest.newBuilder().setInstructionId("SUCCESS"));
            unsuccessfulResponse[0] = beamFnStateClient.handle(StateRequest.newBuilder().setInstructionId("FAIL"));
        }
    }), Caches.noop(), new BundleProcessorCache());
    handler.processBundle(BeamFnApi.InstructionRequest.newBuilder().setProcessBundle(BeamFnApi.ProcessBundleRequest.newBuilder().setProcessBundleDescriptorId("1L")).build());
    assertTrue(successfulResponse[0].isDone());
    assertTrue(unsuccessfulResponse[0].isDone());
}

Example 8 with PTransformRunnerFactory

use of org.apache.beam.fn.harness.PTransformRunnerFactory in project beam by apache.

the class ProcessBundleHandlerTest method testTimerRegistrationsFailIfNoTimerApiServiceDescriptorSpecified.

@Test
public void testTimerRegistrationsFailIfNoTimerApiServiceDescriptorSpecified() throws Exception {
    BeamFnApi.ProcessBundleDescriptor processBundleDescriptor = BeamFnApi.ProcessBundleDescriptor.newBuilder().putTransforms("2L", RunnerApi.PTransform.newBuilder().setSpec(RunnerApi.FunctionSpec.newBuilder().setUrn(DATA_INPUT_URN).build()).build()).build();
    Map<String, BeamFnApi.ProcessBundleDescriptor> fnApiRegistry = ImmutableMap.of("1L", processBundleDescriptor);
    ProcessBundleHandler handler = new ProcessBundleHandler(PipelineOptionsFactory.create(), Collections.emptySet(), fnApiRegistry::get, beamFnDataClient, null, /* beamFnStateGrpcClientCache */
    null, /* finalizeBundleHandler */
    new ShortIdMap(), ImmutableMap.of(DATA_INPUT_URN, new PTransformRunnerFactory<Object>() {

        @Override
        public Object createRunnerForPTransform(Context context) throws IOException {
            context.addOutgoingTimersEndpoint("timer", Timer.Coder.of(StringUtf8Coder.of(), GlobalWindow.Coder.INSTANCE));
            return null;
        }
    }), Caches.noop(), new BundleProcessorCache());
    assertThrows("Timers are unsupported", IllegalStateException.class, () -> handler.processBundle(BeamFnApi.InstructionRequest.newBuilder().setProcessBundle(BeamFnApi.ProcessBundleRequest.newBuilder().setProcessBundleDescriptorId("1L")).build()));
}

Example 9 with PTransformRunnerFactory

use of org.apache.beam.fn.harness.PTransformRunnerFactory in project beam by apache.

the class ProcessBundleHandlerTest method testDataProcessingExceptionsArePropagated.

@Test
public void testDataProcessingExceptionsArePropagated() throws Exception {
    BeamFnApi.ProcessBundleDescriptor processBundleDescriptor = BeamFnApi.ProcessBundleDescriptor.newBuilder().putTransforms("2L", RunnerApi.PTransform.newBuilder().setSpec(RunnerApi.FunctionSpec.newBuilder().setUrn(DATA_INPUT_URN).build()).build()).build();
    Map<String, BeamFnApi.ProcessBundleDescriptor> fnApiRegistry = ImmutableMap.of("1L", processBundleDescriptor);
    Mockito.doAnswer((invocation) -> {
        ByteString.Output encodedData = ByteString.newOutput();
        StringUtf8Coder.of().encode("A", encodedData);
        String instructionId = invocation.getArgument(0, String.class);
        CloseableFnDataReceiver<BeamFnApi.Elements> data = invocation.getArgument(2, CloseableFnDataReceiver.class);
        data.accept(BeamFnApi.Elements.newBuilder().addData(BeamFnApi.Elements.Data.newBuilder().setInstructionId(instructionId).setTransformId("2L").setData(encodedData.toByteString()).setIsLast(true)).build());
        return null;
    }).when(beamFnDataClient).registerReceiver(any(), any(), any());
    ProcessBundleHandler handler = new ProcessBundleHandler(PipelineOptionsFactory.create(), Collections.emptySet(), fnApiRegistry::get, beamFnDataClient, null, /* beamFnStateGrpcClientCache */
    null, /* finalizeBundleHandler */
    new ShortIdMap(), ImmutableMap.of(DATA_INPUT_URN, (PTransformRunnerFactory<Object>) (context) -> {
        context.addIncomingDataEndpoint(ApiServiceDescriptor.getDefaultInstance(), StringUtf8Coder.of(), (input) -> {
            throw new IllegalStateException("TestException");
        });
        return null;
    }), Caches.noop(), new BundleProcessorCache());
    assertThrows("TestException", IllegalStateException.class, () -> handler.processBundle(BeamFnApi.InstructionRequest.newBuilder().setInstructionId("instructionId").setProcessBundle(BeamFnApi.ProcessBundleRequest.newBuilder().setProcessBundleDescriptorId("1L")).build()));
    // Ensure that we unregister during successful processing
    verify(beamFnDataClient).registerReceiver(eq("instructionId"), any(), any());
    verifyNoMoreInteractions(beamFnDataClient);
}

Example 10 with PTransformRunnerFactory

use of org.apache.beam.fn.harness.PTransformRunnerFactory in project beam by apache.

the class ProcessBundleHandlerTest method setupProcessBundleHandlerForSimpleRecordingDoFn.

private ProcessBundleHandler setupProcessBundleHandlerForSimpleRecordingDoFn(List<String> dataOutput, List<Timers> timerOutput, boolean enableOutputEmbedding) throws Exception {
    DoFnWithExecutionInformation doFnWithExecutionInformation = DoFnWithExecutionInformation.of(new SimpleDoFn(), SimpleDoFn.MAIN_OUTPUT_TAG, Collections.emptyMap(), DoFnSchemaInformation.create());
    RunnerApi.FunctionSpec functionSpec = RunnerApi.FunctionSpec.newBuilder().setUrn(ParDoTranslation.CUSTOM_JAVA_DO_FN_URN).setPayload(ByteString.copyFrom(SerializableUtils.serializeToByteArray(doFnWithExecutionInformation))).build();
    RunnerApi.ParDoPayload parDoPayload = ParDoPayload.newBuilder().setDoFn(functionSpec).putTimerFamilySpecs("tfs-" + SimpleDoFn.TIMER_FAMILY_ID, TimerFamilySpec.newBuilder().setTimeDomain(RunnerApi.TimeDomain.Enum.EVENT_TIME).setTimerFamilyCoderId("timer-coder").build()).build();
    BeamFnApi.ProcessBundleDescriptor processBundleDescriptor = ProcessBundleDescriptor.newBuilder().putTransforms("2L", PTransform.newBuilder().setSpec(FunctionSpec.newBuilder().setUrn(DATA_INPUT_URN).build()).putOutputs("2L-output", "2L-output-pc").build()).putTransforms("3L", PTransform.newBuilder().setSpec(FunctionSpec.newBuilder().setUrn(PTransformTranslation.PAR_DO_TRANSFORM_URN).setPayload(parDoPayload.toByteString())).putInputs("3L-input", "2L-output-pc").build()).putPcollections("2L-output-pc", PCollection.newBuilder().setWindowingStrategyId("window-strategy").setCoderId("2L-output-coder").setIsBounded(IsBounded.Enum.BOUNDED).build()).putWindowingStrategies("window-strategy", WindowingStrategy.newBuilder().setWindowCoderId("window-strategy-coder").setWindowFn(FunctionSpec.newBuilder().setUrn("beam:window_fn:global_windows:v1")).setOutputTime(OutputTime.Enum.END_OF_WINDOW).setAccumulationMode(AccumulationMode.Enum.ACCUMULATING).setTrigger(Trigger.newBuilder().setAlways(Always.getDefaultInstance())).setClosingBehavior(ClosingBehavior.Enum.EMIT_ALWAYS).setOnTimeBehavior(OnTimeBehavior.Enum.FIRE_ALWAYS).build()).setTimerApiServiceDescriptor(ApiServiceDescriptor.newBuilder().setUrl("url").build()).putCoders("string_coder", CoderTranslation.toProto(StringUtf8Coder.of()).getCoder()).putCoders("2L-output-coder", Coder.newBuilder().setSpec(FunctionSpec.newBuilder().setUrn(ModelCoders.KV_CODER_URN).build()).addComponentCoderIds("string_coder").addComponentCoderIds("string_coder").build()).putCoders("window-strategy-coder", Coder.newBuilder().setSpec(FunctionSpec.newBuilder().setUrn(ModelCoders.GLOBAL_WINDOW_CODER_URN).build()).build()).putCoders("timer-coder", Coder.newBuilder().setSpec(FunctionSpec.newBuilder().setUrn(ModelCoders.TIMER_CODER_URN)).addComponentCoderIds("string_coder").addComponentCoderIds("window-strategy-coder").build()).build();
    Map<String, BeamFnApi.ProcessBundleDescriptor> fnApiRegistry = ImmutableMap.of("1L", processBundleDescriptor);
    Map<String, PTransformRunnerFactory> urnToPTransformRunnerFactoryMap = Maps.newHashMap(REGISTERED_RUNNER_FACTORIES);
    urnToPTransformRunnerFactoryMap.put(DATA_INPUT_URN, (PTransformRunnerFactory<Object>) (context) -> {
        context.addIncomingDataEndpoint(ApiServiceDescriptor.getDefaultInstance(), KvCoder.of(StringUtf8Coder.of(), StringUtf8Coder.of()), (input) -> {
            dataOutput.add(input.getValue());
        });
        return null;
    });
    Mockito.doAnswer((invocation) -> new BeamFnDataOutboundAggregator(PipelineOptionsFactory.create(), invocation.getArgument(1), new StreamObserver<Elements>() {

        @Override
        public void onNext(Elements elements) {
            for (Timers timer : elements.getTimersList()) {
                timerOutput.addAll(elements.getTimersList());
            }
        }

        @Override
        public void onError(Throwable throwable) {
        }

        @Override
        public void onCompleted() {
        }
    }, invocation.getArgument(2))).when(beamFnDataClient).createOutboundAggregator(any(), any(), anyBoolean());
    return new ProcessBundleHandler(PipelineOptionsFactory.create(), enableOutputEmbedding ? Collections.singleton(BeamUrns.getUrn(StandardRunnerProtocols.Enum.CONTROL_RESPONSE_ELEMENTS_EMBEDDING)) : Collections.emptySet(), fnApiRegistry::get, beamFnDataClient, null, /* beamFnStateClient */
    null, /* finalizeBundleHandler */
    new ShortIdMap(), urnToPTransformRunnerFactoryMap, Caches.noop(), new BundleProcessorCache());
}