Examples with PCollection org.apache.beam.model.pipeline.v1.RunnerApi.PCollection used on opensource projects

Example 11 with PCollection

use of org.apache.beam.model.pipeline.v1.RunnerApi.PCollection in project beam by apache.

the class PipelineTranslatorUtils method getWindowedValueCoder.

public static <T> WindowedValue.WindowedValueCoder<T> getWindowedValueCoder(String pCollectionId, RunnerApi.Components components) {
    RunnerApi.PCollection pCollection = components.getPcollectionsOrThrow(pCollectionId);
    PipelineNode.PCollectionNode pCollectionNode = PipelineNode.pCollection(pCollectionId, pCollection);
    WindowedValue.WindowedValueCoder<T> coder;
    try {
        coder = (WindowedValue.WindowedValueCoder) WireCoders.instantiateRunnerWireCoder(pCollectionNode, components);
    } catch (IOException e) {
        throw new RuntimeException(e);
    }
    return coder;
}

Example 12 with PCollection

use of org.apache.beam.model.pipeline.v1.RunnerApi.PCollection in project beam by apache.

the class PubSubReadPayloadTranslationTest method testTranslateSourceToFunctionSpec.

@Test
public void testTranslateSourceToFunctionSpec() throws Exception {
    PCollection<byte[]> output = pipeline.apply(readFromPubSub);
    AppliedPTransform<?, ?, Read.Unbounded<byte[]>> appliedPTransform = AppliedPTransform.of("ReadFromPubsub", PValues.expandInput(pipeline.begin()), PValues.expandOutput(output), readFromPubSub, ResourceHints.create(), pipeline);
    SdkComponents components = SdkComponents.create();
    components.registerEnvironment(Environments.createDockerEnvironment("java"));
    RunnerApi.FunctionSpec spec = sourceTranslator.translate((AppliedPTransform) appliedPTransform, components);
    assertEquals(PTransformTranslation.PUBSUB_READ, spec.getUrn());
    PubSubReadPayload result = PubSubReadPayload.parseFrom(spec.getPayload());
    assertEquals(pubsubReadPayload, result);
}

Example 13 with PCollection

use of org.apache.beam.model.pipeline.v1.RunnerApi.PCollection in project beam by apache.

the class ExpansionService method expand.

@VisibleForTesting
/*package*/
ExpansionApi.ExpansionResponse expand(ExpansionApi.ExpansionRequest request) {
    LOG.info("Expanding '{}' with URN '{}'", request.getTransform().getUniqueName(), request.getTransform().getSpec().getUrn());
    LOG.debug("Full transform: {}", request.getTransform());
    Set<String> existingTransformIds = request.getComponents().getTransformsMap().keySet();
    Pipeline pipeline = createPipeline();
    boolean isUseDeprecatedRead = ExperimentalOptions.hasExperiment(pipelineOptions, "use_deprecated_read") || ExperimentalOptions.hasExperiment(pipelineOptions, "beam_fn_api_use_deprecated_read");
    if (!isUseDeprecatedRead) {
        ExperimentalOptions.addExperiment(pipeline.getOptions().as(ExperimentalOptions.class), "beam_fn_api");
        // TODO(BEAM-10670): Remove this when we address performance issue.
        ExperimentalOptions.addExperiment(pipeline.getOptions().as(ExperimentalOptions.class), "use_sdf_read");
    } else {
        LOG.warn("Using use_depreacted_read in portable runners is runner-dependent. The " + "ExpansionService will respect that, but if your runner does not have support for " + "native Read transform, your Pipeline will fail during Pipeline submission.");
    }
    RehydratedComponents rehydratedComponents = RehydratedComponents.forComponents(request.getComponents()).withPipeline(pipeline);
    Map<String, PCollection<?>> inputs = request.getTransform().getInputsMap().entrySet().stream().collect(Collectors.toMap(Map.Entry::getKey, input -> {
        try {
            return rehydratedComponents.getPCollection(input.getValue());
        } catch (IOException exn) {
            throw new RuntimeException(exn);
        }
    }));
    String urn = request.getTransform().getSpec().getUrn();
    TransformProvider transformProvider = null;
    if (getUrn(ExpansionMethods.Enum.JAVA_CLASS_LOOKUP).equals(urn)) {
        AllowList allowList = pipelineOptions.as(ExpansionServiceOptions.class).getJavaClassLookupAllowlist();
        assert allowList != null;
        transformProvider = new JavaClassLookupTransformProvider(allowList);
    } else {
        transformProvider = getRegisteredTransforms().get(urn);
        if (transformProvider == null) {
            throw new UnsupportedOperationException("Unknown urn: " + request.getTransform().getSpec().getUrn());
        }
    }
    List<String> classpathResources = transformProvider.getDependencies(request.getTransform().getSpec(), pipeline.getOptions());
    pipeline.getOptions().as(PortablePipelineOptions.class).setFilesToStage(classpathResources);
    Map<String, PCollection<?>> outputs = transformProvider.apply(pipeline, request.getTransform().getUniqueName(), request.getTransform().getSpec(), inputs);
    // Needed to find which transform was new...
    SdkComponents sdkComponents = rehydratedComponents.getSdkComponents(Collections.emptyList()).withNewIdPrefix(request.getNamespace());
    sdkComponents.registerEnvironment(Environments.createOrGetDefaultEnvironment(pipeline.getOptions().as(PortablePipelineOptions.class)));
    Map<String, String> outputMap = outputs.entrySet().stream().collect(Collectors.toMap(Map.Entry::getKey, output -> {
        try {
            return sdkComponents.registerPCollection(output.getValue());
        } catch (IOException exn) {
            throw new RuntimeException(exn);
        }
    }));
    if (isUseDeprecatedRead) {
        SplittableParDo.convertReadBasedSplittableDoFnsToPrimitiveReadsIfNecessary(pipeline);
    }
    RunnerApi.Pipeline pipelineProto = PipelineTranslation.toProto(pipeline, sdkComponents);
    String expandedTransformId = Iterables.getOnlyElement(pipelineProto.getRootTransformIdsList().stream().filter(id -> !existingTransformIds.contains(id)).collect(Collectors.toList()));
    RunnerApi.Components components = pipelineProto.getComponents();
    RunnerApi.PTransform expandedTransform = components.getTransformsOrThrow(expandedTransformId).toBuilder().setUniqueName(expandedTransformId).clearOutputs().putAllOutputs(outputMap).build();
    LOG.debug("Expanded to {}", expandedTransform);
    return ExpansionApi.ExpansionResponse.newBuilder().setComponents(components.toBuilder().removeTransforms(expandedTransformId)).setTransform(expandedTransform).addAllRequirements(pipelineProto.getRequirementsList()).build();
}

Example 14 with PCollection

use of org.apache.beam.model.pipeline.v1.RunnerApi.PCollection in project beam by apache.

the class ParDoTranslation method translateParDo.

public static ParDoPayload translateParDo(AppliedPTransform<?, ?, ParDo.MultiOutput<?, ?>> appliedPTransform, SdkComponents components) throws IOException {
    final ParDo.MultiOutput<?, ?> parDo = appliedPTransform.getTransform();
    final Pipeline pipeline = appliedPTransform.getPipeline();
    final DoFn<?, ?> doFn = parDo.getFn();
    // Get main input.
    Set<String> allInputs = appliedPTransform.getInputs().keySet().stream().map(TupleTag::getId).collect(Collectors.toSet());
    Set<String> sideInputs = parDo.getSideInputs().values().stream().map(s -> s.getTagInternal().getId()).collect(Collectors.toSet());
    String mainInputName = Iterables.getOnlyElement(Sets.difference(allInputs, sideInputs));
    PCollection<?> mainInput = (PCollection<?>) appliedPTransform.getInputs().get(new TupleTag<>(mainInputName));
    final DoFnSchemaInformation doFnSchemaInformation = ParDo.getDoFnSchemaInformation(doFn, mainInput);
    return translateParDo((ParDo.MultiOutput) parDo, mainInput, doFnSchemaInformation, pipeline, components);
}

Example 15 with PCollection

use of org.apache.beam.model.pipeline.v1.RunnerApi.PCollection in project beam by apache.

the class ParDoTranslation method translateParDo.

/**
 * Translate a ParDo.
 */
public static <InputT> ParDoPayload translateParDo(ParDo.MultiOutput<InputT, ?> parDo, PCollection<InputT> mainInput, DoFnSchemaInformation doFnSchemaInformation, Pipeline pipeline, SdkComponents components) throws IOException {
    final DoFn<?, ?> doFn = parDo.getFn();
    final DoFnSignature signature = DoFnSignatures.getSignature(doFn.getClass());
    final String restrictionCoderId;
    if (signature.processElement().isSplittable()) {
        DoFnInvoker<?, ?> doFnInvoker = DoFnInvokers.invokerFor(doFn);
        final Coder<?> restrictionAndWatermarkStateCoder = KvCoder.of(doFnInvoker.invokeGetRestrictionCoder(pipeline.getCoderRegistry()), doFnInvoker.invokeGetWatermarkEstimatorStateCoder(pipeline.getCoderRegistry()));
        restrictionCoderId = components.registerCoder(restrictionAndWatermarkStateCoder);
    } else {
        restrictionCoderId = "";
    }
    Coder<BoundedWindow> windowCoder = (Coder<BoundedWindow>) mainInput.getWindowingStrategy().getWindowFn().windowCoder();
    Coder<?> keyCoder;
    if (signature.usesState() || signature.usesTimers()) {
        checkArgument(mainInput.getCoder() instanceof KvCoder, "DoFn's that use state or timers must have an input PCollection with a KvCoder but received %s", mainInput.getCoder());
        keyCoder = ((KvCoder) mainInput.getCoder()).getKeyCoder();
    } else {
        keyCoder = null;
    }
    return payloadForParDoLike(new ParDoLike() {

        @Override
        public FunctionSpec translateDoFn(SdkComponents newComponents) {
            return ParDoTranslation.translateDoFn(parDo.getFn(), parDo.getMainOutputTag(), parDo.getSideInputs(), doFnSchemaInformation, newComponents);
        }

        @Override
        public Map<String, SideInput> translateSideInputs(SdkComponents components) {
            Map<String, SideInput> sideInputs = new HashMap<>();
            for (PCollectionView<?> sideInput : parDo.getSideInputs().values()) {
                sideInputs.put(sideInput.getTagInternal().getId(), translateView(sideInput, components));
            }
            return sideInputs;
        }

        @Override
        public Map<String, RunnerApi.StateSpec> translateStateSpecs(SdkComponents components) throws IOException {
            Map<String, RunnerApi.StateSpec> stateSpecs = new HashMap<>();
            for (Map.Entry<String, StateDeclaration> state : signature.stateDeclarations().entrySet()) {
                RunnerApi.StateSpec spec = translateStateSpec(getStateSpecOrThrow(state.getValue(), doFn), components);
                stateSpecs.put(state.getKey(), spec);
            }
            return stateSpecs;
        }

        @Override
        public ParDoLikeTimerFamilySpecs translateTimerFamilySpecs(SdkComponents newComponents) {
            Map<String, RunnerApi.TimerFamilySpec> timerFamilySpecs = new HashMap<>();
            for (Map.Entry<String, TimerDeclaration> timer : signature.timerDeclarations().entrySet()) {
                RunnerApi.TimerFamilySpec spec = translateTimerFamilySpec(getTimerSpecOrThrow(timer.getValue(), doFn), newComponents, keyCoder, windowCoder);
                timerFamilySpecs.put(timer.getKey(), spec);
            }
            for (Map.Entry<String, DoFnSignature.TimerFamilyDeclaration> timerFamily : signature.timerFamilyDeclarations().entrySet()) {
                RunnerApi.TimerFamilySpec spec = translateTimerFamilySpec(DoFnSignatures.getTimerFamilySpecOrThrow(timerFamily.getValue(), doFn), newComponents, keyCoder, windowCoder);
                timerFamilySpecs.put(timerFamily.getKey(), spec);
            }
            String onWindowExpirationTimerFamilySpec = null;
            if (signature.onWindowExpiration() != null) {
                RunnerApi.TimerFamilySpec spec = RunnerApi.TimerFamilySpec.newBuilder().setTimeDomain(translateTimeDomain(TimeDomain.EVENT_TIME)).setTimerFamilyCoderId(registerCoderOrThrow(components, Timer.Coder.of(keyCoder, windowCoder))).build();
                for (int i = 0; i < Integer.MAX_VALUE; ++i) {
                    onWindowExpirationTimerFamilySpec = "onWindowExpiration" + i;
                    if (!timerFamilySpecs.containsKey(onWindowExpirationTimerFamilySpec)) {
                        break;
                    }
                }
                timerFamilySpecs.put(onWindowExpirationTimerFamilySpec, spec);
            }
            return ParDoLikeTimerFamilySpecs.create(timerFamilySpecs, onWindowExpirationTimerFamilySpec);
        }

        @Override
        public boolean isStateful() {
            return !signature.stateDeclarations().isEmpty() || !signature.timerDeclarations().isEmpty() || !signature.timerFamilyDeclarations().isEmpty() || signature.onWindowExpiration() != null;
        }

        @Override
        public boolean isSplittable() {
            return signature.processElement().isSplittable();
        }

        @Override
        public boolean isRequiresStableInput() {
            return signature.processElement().requiresStableInput();
        }

        @Override
        public boolean isRequiresTimeSortedInput() {
            return signature.processElement().requiresTimeSortedInput();
        }

        @Override
        public boolean requestsFinalization() {
            return (signature.startBundle() != null && signature.startBundle().extraParameters().contains(Parameter.bundleFinalizer())) || (signature.processElement() != null && signature.processElement().extraParameters().contains(Parameter.bundleFinalizer())) || (signature.finishBundle() != null && signature.finishBundle().extraParameters().contains(Parameter.bundleFinalizer()));
        }

        @Override
        public String translateRestrictionCoderId(SdkComponents newComponents) {
            return restrictionCoderId;
        }
    }, components);
}