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

Example 16 with DoFn

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

the class RemoteExecutionTest method testExecutionWithMultipleStages.

@Test
public void testExecutionWithMultipleStages() throws Exception {
    launchSdkHarness(PipelineOptionsFactory.create());
    Pipeline p = Pipeline.create();
    Function<String, PCollection<String>> pCollectionGenerator = suffix -> p.apply("impulse" + suffix, Impulse.create()).apply("create" + suffix, ParDo.of(new DoFn<byte[], String>() {

        @ProcessElement
        public void process(ProcessContext c) {
            try {
                c.output(CoderUtils.decodeFromByteArray(StringUtf8Coder.of(), c.element()));
            } catch (CoderException e) {
                throw new RuntimeException(e);
            }
        }
    })).setCoder(StringUtf8Coder.of()).apply(ParDo.of(new DoFn<String, String>() {

        @ProcessElement
        public void processElement(ProcessContext c) {
            c.output("stream" + suffix + c.element());
        }
    }));
    PCollection<String> input1 = pCollectionGenerator.apply("1");
    PCollection<String> input2 = pCollectionGenerator.apply("2");
    PCollection<String> outputMerged = PCollectionList.of(input1).and(input2).apply(Flatten.pCollections());
    outputMerged.apply("createKV", ParDo.of(new DoFn<String, KV<String, String>>() {

        @ProcessElement
        public void process(ProcessContext c) {
            c.output(KV.of(c.element(), ""));
        }
    })).setCoder(KvCoder.of(StringUtf8Coder.of(), StringUtf8Coder.of())).apply("gbk", GroupByKey.create());
    RunnerApi.Pipeline pipelineProto = PipelineTranslation.toProto(p);
    FusedPipeline fused = GreedyPipelineFuser.fuse(pipelineProto);
    Set<ExecutableStage> stages = fused.getFusedStages();
    assertThat(stages.size(), equalTo(2));
    List<WindowedValue<?>> outputValues = Collections.synchronizedList(new ArrayList<>());
    for (ExecutableStage stage : stages) {
        ExecutableProcessBundleDescriptor descriptor = ProcessBundleDescriptors.fromExecutableStage(stage.toString(), stage, dataServer.getApiServiceDescriptor(), stateServer.getApiServiceDescriptor());
        BundleProcessor processor = controlClient.getProcessor(descriptor.getProcessBundleDescriptor(), descriptor.getRemoteInputDestinations(), stateDelegator);
        Map<String, Coder> remoteOutputCoders = descriptor.getRemoteOutputCoders();
        Map<String, RemoteOutputReceiver<?>> outputReceivers = new HashMap<>();
        for (Entry<String, Coder> remoteOutputCoder : remoteOutputCoders.entrySet()) {
            outputReceivers.putIfAbsent(remoteOutputCoder.getKey(), RemoteOutputReceiver.of((Coder<WindowedValue<?>>) remoteOutputCoder.getValue(), outputValues::add));
        }
        try (RemoteBundle bundle = processor.newBundle(outputReceivers, StateRequestHandler.unsupported(), BundleProgressHandler.ignored())) {
            Iterables.getOnlyElement(bundle.getInputReceivers().values()).accept(valueInGlobalWindow(CoderUtils.encodeToByteArray(StringUtf8Coder.of(), "X")));
        }
    }
    assertThat(outputValues, containsInAnyOrder(valueInGlobalWindow(KV.of("stream1X", "")), valueInGlobalWindow(KV.of("stream2X", ""))));
}

Example 17 with DoFn

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

the class ConfigGeneratorTest method testDuplicateStateIdConfig.

@Test
public void testDuplicateStateIdConfig() {
    SamzaPipelineOptions options = PipelineOptionsFactory.create().as(SamzaPipelineOptions.class);
    options.setJobName("TestStoreConfig");
    options.setRunner(SamzaRunner.class);
    Pipeline pipeline = Pipeline.create(options);
    pipeline.apply(Create.empty(TypeDescriptors.kvs(TypeDescriptors.strings(), TypeDescriptors.strings()))).apply(ParDo.of(new DoFn<KV<String, String>, KV<String, String>>() {

        private static final String testState = "testState";

        @StateId(testState)
        private final StateSpec<ValueState<Integer>> state = StateSpecs.value();

        @ProcessElement
        public void processElement(ProcessContext context, @StateId(testState) ValueState<Integer> state) {
            context.output(context.element());
        }
    })).apply(ParDo.of(new DoFn<KV<String, String>, Void>() {

        private static final String testState = "testState";

        @StateId(testState)
        private final StateSpec<ValueState<Integer>> state = StateSpecs.value();

        @ProcessElement
        public void processElement(ProcessContext context, @StateId(testState) ValueState<Integer> state) {
        }
    }));
    final Map<PValue, String> idMap = PViewToIdMapper.buildIdMap(pipeline);
    final ConfigBuilder configBuilder = new ConfigBuilder(options);
    assertThrows(IllegalStateException.class, () -> SamzaPipelineTranslator.createConfig(pipeline, options, idMap, configBuilder));
}

Example 18 with DoFn

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

the class DoFnSignatures method analyzeSplitRestrictionMethod.

@VisibleForTesting
static DoFnSignature.SplitRestrictionMethod analyzeSplitRestrictionMethod(ErrorReporter errors, TypeDescriptor<? extends DoFn<?, ?>> fnT, Method m, TypeDescriptor<?> inputT, TypeDescriptor<?> outputT, TypeDescriptor<?> restrictionT, FnAnalysisContext fnContext) {
    // Method is of the form:
    // @SplitRestriction
    // void splitRestriction(... parameters ...);
    errors.checkArgument(void.class.equals(m.getReturnType()), "Must return void");
    Type[] params = m.getGenericParameterTypes();
    MethodAnalysisContext methodContext = MethodAnalysisContext.create();
    TypeDescriptor<? extends BoundedWindow> windowT = getWindowType(fnT, m);
    for (int i = 0; i < params.length; ++i) {
        Parameter extraParam = analyzeExtraParameter(errors, fnContext, methodContext, ParameterDescription.of(m, i, fnT.resolveType(params[i]), Arrays.asList(m.getParameterAnnotations()[i])), inputT, restrictionT);
        if (extraParam instanceof SchemaElementParameter) {
            errors.throwIllegalArgument("Schema @%s are not supported for @%s method. Found %s, did you mean to use %s?", format(DoFn.Element.class), format(DoFn.SplitRestriction.class), format(((SchemaElementParameter) extraParam).elementT()), format(inputT));
        } else if (extraParam instanceof RestrictionParameter) {
            errors.checkArgument(restrictionT.equals(((RestrictionParameter) extraParam).restrictionT()), "Uses restriction type %s, but @%s method uses restriction type %s", format(((RestrictionParameter) extraParam).restrictionT()), format(DoFn.GetInitialRestriction.class), format(restrictionT));
        }
        methodContext.addParameter(extraParam);
    }
    for (Parameter parameter : methodContext.getExtraParameters()) {
        checkParameterOneOf(errors, parameter, ALLOWED_SPLIT_RESTRICTION_PARAMETERS);
    }
    return DoFnSignature.SplitRestrictionMethod.create(m, windowT, methodContext.getExtraParameters());
}

Example 19 with DoFn

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

the class DoFnSignatures method analyzeProcessElementMethod.

@VisibleForTesting
static DoFnSignature.ProcessElementMethod analyzeProcessElementMethod(ErrorReporter errors, TypeDescriptor<? extends DoFn<?, ?>> fnClass, Method m, TypeDescriptor<?> inputT, TypeDescriptor<?> outputT, FnAnalysisContext fnContext) {
    errors.checkArgument(void.class.equals(m.getReturnType()) || DoFn.ProcessContinuation.class.equals(m.getReturnType()), "Must return void or %s", format(DoFn.ProcessContinuation.class));
    MethodAnalysisContext methodContext = MethodAnalysisContext.create();
    boolean requiresStableInput = m.isAnnotationPresent(DoFn.RequiresStableInput.class);
    boolean requiresTimeSortedInput = m.isAnnotationPresent(DoFn.RequiresTimeSortedInput.class);
    TypeDescriptor<? extends BoundedWindow> windowT = getWindowType(fnClass, m);
    Type[] params = m.getGenericParameterTypes();
    for (int i = 0; i < params.length; ++i) {
        Parameter extraParam = analyzeExtraParameter(errors.forMethod(DoFn.ProcessElement.class, m), fnContext, methodContext, ParameterDescription.of(m, i, fnClass.resolveType(params[i]), Arrays.asList(m.getParameterAnnotations()[i])), inputT, outputT);
        methodContext.addParameter(extraParam);
    }
    int schemaElementIndex = 0;
    for (int i = 0; i < methodContext.getExtraParameters().size(); ++i) {
        Parameter parameter = methodContext.getExtraParameters().get(i);
        if (parameter instanceof SchemaElementParameter) {
            SchemaElementParameter schemaParameter = (SchemaElementParameter) parameter;
            schemaParameter = schemaParameter.toBuilder().setIndex(schemaElementIndex).build();
            methodContext.setParameter(i, schemaParameter);
            ++schemaElementIndex;
        }
    }
    TypeDescriptor<?> trackerT = methodContext.findParameter(RestrictionTrackerParameter.class).map(p -> p.trackerT()).orElse(null);
    TypeDescriptor<?> watermarkEstimatorT = methodContext.findParameter(WatermarkEstimatorParameter.class).map(p -> p.estimatorT()).orElse(null);
    // The allowed parameters depend on whether this DoFn is splittable
    if (trackerT != null) {
        for (Parameter parameter : methodContext.getExtraParameters()) {
            checkParameterOneOf(errors, parameter, ALLOWED_SPLITTABLE_PROCESS_ELEMENT_PARAMETERS);
        }
    } else {
        for (Parameter parameter : methodContext.getExtraParameters()) {
            checkParameterOneOf(errors, parameter, ALLOWED_NON_SPLITTABLE_PROCESS_ELEMENT_PARAMETERS);
        }
    }
    return DoFnSignature.ProcessElementMethod.create(m, methodContext.getExtraParameters(), requiresStableInput, requiresTimeSortedInput, trackerT, watermarkEstimatorT, windowT, DoFn.ProcessContinuation.class.equals(m.getReturnType()));
}

Example 20 with DoFn

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

the class DoFnSignatures method analyzeOnTimerFamilyMethod.

@VisibleForTesting
static DoFnSignature.OnTimerFamilyMethod analyzeOnTimerFamilyMethod(ErrorReporter errors, TypeDescriptor<? extends DoFn<?, ?>> fnClass, Method m, String timerFamilyId, TypeDescriptor<?> inputT, TypeDescriptor<?> outputT, FnAnalysisContext fnContext) {
    errors.checkArgument(void.class.equals(m.getReturnType()), "Must return void");
    Type[] params = m.getGenericParameterTypes();
    MethodAnalysisContext methodContext = MethodAnalysisContext.create();
    boolean requiresStableInput = m.isAnnotationPresent(DoFn.RequiresStableInput.class);
    @Nullable TypeDescriptor<? extends BoundedWindow> windowT = getWindowType(fnClass, m);
    List<DoFnSignature.Parameter> extraParameters = new ArrayList<>();
    ErrorReporter onTimerErrors = errors.forMethod(DoFn.OnTimerFamily.class, m);
    for (int i = 0; i < params.length; ++i) {
        Parameter parameter = analyzeExtraParameter(onTimerErrors, fnContext, methodContext, ParameterDescription.of(m, i, fnClass.resolveType(params[i]), Arrays.asList(m.getParameterAnnotations()[i])), inputT, outputT);
        checkParameterOneOf(errors, parameter, ALLOWED_ON_TIMER_FAMILY_PARAMETERS);
        extraParameters.add(parameter);
    }
    return DoFnSignature.OnTimerFamilyMethod.create(m, timerFamilyId, requiresStableInput, windowT, extraParameters);
}