Examples with ProcessContext org.apache.beam.sdk.transforms.DoFn.ProcessContext used on opensource projects

Example 1 with ProcessContext

use of org.apache.beam.sdk.transforms.DoFn.ProcessContext 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);
}

Example 2 with ProcessContext

use of org.apache.beam.sdk.transforms.DoFn.ProcessContext in project beam by apache.

the class ProcessBundleDescriptorsTest method testLengthPrefixingOfInputCoderExecutableStage.

@Test
public void testLengthPrefixingOfInputCoderExecutableStage() throws Exception {
    Pipeline p = Pipeline.create();
    Coder<Void> voidCoder = VoidCoder.of();
    assertThat(ModelCoderRegistrar.isKnownCoder(voidCoder), is(false));
    p.apply("impulse", Impulse.create()).apply(ParDo.of(new DoFn<byte[], Void>() {

        @ProcessElement
        public void process(ProcessContext ctxt) {
        }
    })).setCoder(voidCoder).apply(ParDo.of(new DoFn<Void, Void>() {

        @ProcessElement
        public void processElement(ProcessContext context, RestrictionTracker<Void, Void> tracker) {
        }

        @GetInitialRestriction
        public Void getInitialRestriction() {
            return null;
        }

        @NewTracker
        public SomeTracker newTracker(@Restriction Void restriction) {
            return null;
        }
    })).setCoder(voidCoder);
    RunnerApi.Pipeline pipelineProto = PipelineTranslation.toProto(p);
    RunnerApi.Pipeline pipelineWithSdfExpanded = ProtoOverrides.updateTransform(PTransformTranslation.PAR_DO_TRANSFORM_URN, pipelineProto, SplittableParDoExpander.createSizedReplacement());
    FusedPipeline fused = GreedyPipelineFuser.fuse(pipelineWithSdfExpanded);
    Optional<ExecutableStage> optionalStage = Iterables.tryFind(fused.getFusedStages(), (ExecutableStage stage) -> stage.getTransforms().stream().anyMatch(transform -> transform.getTransform().getSpec().getUrn().equals(PTransformTranslation.SPLITTABLE_PROCESS_SIZED_ELEMENTS_AND_RESTRICTIONS_URN)));
    checkState(optionalStage.isPresent(), "Expected a stage with SPLITTABLE_PROCESS_SIZED_ELEMENTS_AND_RESTRICTIONS_URN.");
    ExecutableStage stage = optionalStage.get();
    PipelineNode.PCollectionNode inputPCollection = stage.getInputPCollection();
    Map<String, RunnerApi.Coder> stageCoderMap = stage.getComponents().getCodersMap();
    RunnerApi.Coder originalMainInputCoder = stageCoderMap.get(inputPCollection.getPCollection().getCoderId());
    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());
    RunnerApi.Coder kvCoder = pbsCoderMap.get(ModelCoders.getKvCoderComponents(pbsMainInputCoder).keyCoderId());
    RunnerApi.Coder keyCoder = pbsCoderMap.get(ModelCoders.getKvCoderComponents(kvCoder).keyCoderId());
    RunnerApi.Coder valueKvCoder = pbsCoderMap.get(ModelCoders.getKvCoderComponents(kvCoder).valueCoderId());
    RunnerApi.Coder valueCoder = pbsCoderMap.get(ModelCoders.getKvCoderComponents(valueKvCoder).keyCoderId());
    RunnerApi.Coder originalKvCoder = stageCoderMap.get(ModelCoders.getKvCoderComponents(originalMainInputCoder).keyCoderId());
    RunnerApi.Coder originalKeyCoder = stageCoderMap.get(ModelCoders.getKvCoderComponents(originalKvCoder).keyCoderId());
    RunnerApi.Coder originalvalueKvCoder = stageCoderMap.get(ModelCoders.getKvCoderComponents(originalKvCoder).valueCoderId());
    RunnerApi.Coder originalvalueCoder = stageCoderMap.get(ModelCoders.getKvCoderComponents(originalvalueKvCoder).keyCoderId());
    ensureLengthPrefixed(keyCoder, originalKeyCoder, pbsCoderMap);
    ensureLengthPrefixed(valueCoder, originalvalueCoder, pbsCoderMap);
}