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

Example 71 with Pipeline

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

the class QueryablePipelineTest method transformWithSideAndMainInputs.

/**
 * Tests that inputs that are only side inputs are not returned from {@link
 * QueryablePipeline#getPerElementConsumers(PCollectionNode)} and are returned from {@link
 * QueryablePipeline#getSideInputs(PTransformNode)}.
 */
@Test
public void transformWithSideAndMainInputs() {
    Pipeline p = Pipeline.create();
    PCollection<byte[]> impulse = p.apply("Impulse", Impulse.create());
    PCollectionView<String> view = p.apply("Create", Create.of("foo")).apply("View", View.asSingleton());
    impulse.apply("par_do", ParDo.of(new TestFn()).withSideInputs(view).withOutputTags(new TupleTag<>(), TupleTagList.empty()));
    Components components = PipelineTranslation.toProto(p).getComponents();
    QueryablePipeline qp = QueryablePipeline.forPrimitivesIn(components);
    String mainInputName = getOnlyElement(PipelineNode.pTransform("Impulse", components.getTransformsOrThrow("Impulse")).getTransform().getOutputsMap().values());
    PCollectionNode mainInput = PipelineNode.pCollection(mainInputName, components.getPcollectionsOrThrow(mainInputName));
    PTransform parDoTransform = components.getTransformsOrThrow("par_do");
    String sideInputLocalName = getOnlyElement(parDoTransform.getInputsMap().entrySet().stream().filter(entry -> !entry.getValue().equals(mainInputName)).map(Map.Entry::getKey).collect(Collectors.toSet()));
    String sideInputCollectionId = parDoTransform.getInputsOrThrow(sideInputLocalName);
    PCollectionNode sideInput = PipelineNode.pCollection(sideInputCollectionId, components.getPcollectionsOrThrow(sideInputCollectionId));
    PTransformNode parDoNode = PipelineNode.pTransform("par_do", components.getTransformsOrThrow("par_do"));
    SideInputReference sideInputRef = SideInputReference.of(parDoNode, sideInputLocalName, sideInput);
    assertThat(qp.getSideInputs(parDoNode), contains(sideInputRef));
    assertThat(qp.getPerElementConsumers(mainInput), contains(parDoNode));
    assertThat(qp.getPerElementConsumers(sideInput), not(contains(parDoNode)));
}

Example 72 with Pipeline

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

the class QueryablePipelineTest method retainOnlyPrimitivesComposites.

@Test
public void retainOnlyPrimitivesComposites() {
    Pipeline p = Pipeline.create();
    p.apply(new org.apache.beam.sdk.transforms.PTransform<PBegin, PCollection<Long>>() {

        @Override
        public PCollection<Long> expand(PBegin input) {
            return input.apply(Impulse.create()).apply(Window.into(FixedWindows.of(Duration.standardMinutes(5L)))).apply(MapElements.into(TypeDescriptors.longs()).via(l -> 1L));
        }
    });
    Components originalComponents = PipelineTranslation.toProto(p).getComponents();
    Collection<String> primitiveComponents = QueryablePipeline.getPrimitiveTransformIds(originalComponents);
    // Read, Window.Assign, ParDo. This will need to be updated if the expansions change.
    assertThat(primitiveComponents, hasSize(3));
    for (String transformId : primitiveComponents) {
        assertThat(originalComponents.getTransformsMap(), hasKey(transformId));
    }
}

Example 73 with Pipeline

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

the class QueryablePipelineTest method getEnvironmentWithEnvironment.

@Test
public void getEnvironmentWithEnvironment() {
    Pipeline p = Pipeline.create();
    PCollection<Long> longs = p.apply("Impulse", Impulse.create()).apply("ParDo", ParDo.of(new TestFn()));
    longs.apply(WithKeys.of("a")).apply("groupByKey", GroupByKey.create());
    Components components = PipelineTranslation.toProto(p).getComponents();
    QueryablePipeline qp = QueryablePipeline.forPrimitivesIn(components);
    PTransformNode environmentalTransform = PipelineNode.pTransform("ParDo-ParMultiDo-Test-", components.getTransformsOrThrow("ParDo-ParMultiDo-Test-"));
    PTransformNode nonEnvironmentalTransform = PipelineNode.pTransform("groupByKey", components.getTransformsOrThrow("groupByKey"));
    assertThat(qp.getEnvironment(environmentalTransform).isPresent(), is(true));
    assertThat(qp.getEnvironment(environmentalTransform).get().getUrn(), equalTo(Environments.JAVA_SDK_HARNESS_ENVIRONMENT.getUrn()));
    assertThat(qp.getEnvironment(environmentalTransform).get().getPayload(), equalTo(Environments.JAVA_SDK_HARNESS_ENVIRONMENT.getPayload()));
    assertThat(qp.getEnvironment(nonEnvironmentalTransform).isPresent(), is(false));
}

Example 74 with Pipeline

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

the class QueryablePipelineTest method getProducer.

@Test
public void getProducer() {
    Pipeline p = Pipeline.create();
    PCollection<Long> longs = p.apply("BoundedRead", Read.from(CountingSource.upTo(100L)));
    PCollectionList.of(longs).and(longs).and(longs).apply("flatten", Flatten.pCollections());
    Components components = PipelineTranslation.toProto(p).getComponents();
    QueryablePipeline qp = QueryablePipeline.forPrimitivesIn(components);
    String impulseOutputName = getOnlyElement(PipelineNode.pTransform("BoundedRead-Impulse", components.getTransformsOrThrow("BoundedRead-Impulse")).getTransform().getOutputsMap().values());
    PTransformNode impulseProducer = PipelineNode.pTransform("BoundedRead-Impulse", components.getTransformsOrThrow("BoundedRead-Impulse"));
    PCollectionNode impulseOutput = PipelineNode.pCollection(impulseOutputName, components.getPcollectionsOrThrow(impulseOutputName));
    String flattenOutputName = getOnlyElement(PipelineNode.pTransform("flatten", components.getTransformsOrThrow("flatten")).getTransform().getOutputsMap().values());
    PTransformNode flattenProducer = PipelineNode.pTransform("flatten", components.getTransformsOrThrow("flatten"));
    PCollectionNode flattenOutput = PipelineNode.pCollection(flattenOutputName, components.getPcollectionsOrThrow(flattenOutputName));
    assertThat(qp.getProducer(impulseOutput), equalTo(impulseProducer));
    assertThat(qp.getProducer(flattenOutput), equalTo(flattenProducer));
}

Example 75 with Pipeline

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

the class QueryablePipelineTest method retainOnlyPrimitivesWithOnlyPrimitivesUnchanged.

@Test
public void retainOnlyPrimitivesWithOnlyPrimitivesUnchanged() {
    Pipeline p = Pipeline.create();
    p.apply("Impulse", Impulse.create()).apply("multi-do", ParDo.of(new TestFn()).withOutputTags(new TupleTag<>(), TupleTagList.empty()));
    Components originalComponents = PipelineTranslation.toProto(p).getComponents();
    Collection<String> primitiveComponents = QueryablePipeline.getPrimitiveTransformIds(originalComponents);
    assertThat(primitiveComponents, equalTo(originalComponents.getTransformsMap().keySet()));
}