Search in sources :

Example 1 with AllowList

use of org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList 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();
}
Also used : Arrays(java.util.Arrays) PortablePipelineOptions(org.apache.beam.sdk.options.PortablePipelineOptions) ServerBuilder(org.apache.beam.vendor.grpc.v1p43p2.io.grpc.ServerBuilder) PipelineResult(org.apache.beam.sdk.PipelineResult) SchemaApi(org.apache.beam.model.pipeline.v1.SchemaApi) LoggerFactory(org.slf4j.LoggerFactory) SerializableFunction(org.apache.beam.sdk.transforms.SerializableFunction) ImmutableMap(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.ImmutableMap) RehydratedComponents(org.apache.beam.runners.core.construction.RehydratedComponents) Throwables(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Throwables) MonotonicNonNull(org.checkerframework.checker.nullness.qual.MonotonicNonNull) PCollectionList(org.apache.beam.sdk.values.PCollectionList) Map(java.util.Map) Iterables(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.Iterables) PCollectionTuple(org.apache.beam.sdk.values.PCollectionTuple) Method(java.lang.reflect.Method) SchemaCoder(org.apache.beam.sdk.schemas.SchemaCoder) Set(java.util.Set) ServiceLoader(java.util.ServiceLoader) Converter(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Converter) Collectors(java.util.stream.Collectors) InvocationTargetException(java.lang.reflect.InvocationTargetException) ExpansionMethods(org.apache.beam.model.pipeline.v1.ExternalTransforms.ExpansionMethods) POutput(org.apache.beam.sdk.values.POutput) List(java.util.List) VisibleForTesting(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting) ExpansionServiceGrpc(org.apache.beam.model.expansion.v1.ExpansionServiceGrpc) StreamObserver(org.apache.beam.vendor.grpc.v1p43p2.io.grpc.stub.StreamObserver) Optional(java.util.Optional) AllowList(org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList) ExternalTransformBuilder(org.apache.beam.sdk.transforms.ExternalTransformBuilder) SchemaTranslation(org.apache.beam.sdk.schemas.SchemaTranslation) NoSuchSchemaException(org.apache.beam.sdk.schemas.NoSuchSchemaException) ExperimentalOptions(org.apache.beam.sdk.options.ExperimentalOptions) CaseFormat(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.CaseFormat) Coder(org.apache.beam.sdk.coders.Coder) RowCoder(org.apache.beam.sdk.coders.RowCoder) PipelineTranslation(org.apache.beam.runners.core.construction.PipelineTranslation) PipelineOptionsFactory(org.apache.beam.sdk.options.PipelineOptionsFactory) Constructor(java.lang.reflect.Constructor) Environments(org.apache.beam.runners.core.construction.Environments) Server(org.apache.beam.vendor.grpc.v1p43p2.io.grpc.Server) Preconditions.checkArgumentNotNull(org.apache.beam.sdk.util.Preconditions.checkArgumentNotNull) PTransform(org.apache.beam.sdk.transforms.PTransform) ExpansionApi(org.apache.beam.model.expansion.v1.ExpansionApi) PipelineRunner(org.apache.beam.sdk.PipelineRunner) SchemaRegistry(org.apache.beam.sdk.schemas.SchemaRegistry) TupleTag(org.apache.beam.sdk.values.TupleTag) ByteString(org.apache.beam.vendor.grpc.v1p43p2.com.google.protobuf.ByteString) Pipeline(org.apache.beam.sdk.Pipeline) PInput(org.apache.beam.sdk.values.PInput) Row(org.apache.beam.sdk.values.Row) PipelineOptions(org.apache.beam.sdk.options.PipelineOptions) Nullable(org.checkerframework.checker.nullness.qual.Nullable) RunnerApi(org.apache.beam.model.pipeline.v1.RunnerApi) Field(org.apache.beam.sdk.schemas.Schema.Field) SdkComponents(org.apache.beam.runners.core.construction.SdkComponents) ExternalConfigurationPayload(org.apache.beam.model.pipeline.v1.ExternalTransforms.ExternalConfigurationPayload) PDone(org.apache.beam.sdk.values.PDone) Logger(org.slf4j.Logger) PipelineResources.detectClassPathResourcesToStage(org.apache.beam.runners.core.construction.resources.PipelineResources.detectClassPathResourcesToStage) IOException(java.io.IOException) SplittableParDo(org.apache.beam.runners.core.construction.SplittableParDo) ExternalTransformRegistrar(org.apache.beam.sdk.expansion.ExternalTransformRegistrar) PCollection(org.apache.beam.sdk.values.PCollection) Schema(org.apache.beam.sdk.schemas.Schema) AutoService(com.google.auto.service.AutoService) Preconditions(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions) ArtifactRetrievalService(org.apache.beam.runners.fnexecution.artifact.ArtifactRetrievalService) Collections(java.util.Collections) BeamUrns.getUrn(org.apache.beam.runners.core.construction.BeamUrns.getUrn) ExperimentalOptions(org.apache.beam.sdk.options.ExperimentalOptions) ByteString(org.apache.beam.vendor.grpc.v1p43p2.com.google.protobuf.ByteString) SdkComponents(org.apache.beam.runners.core.construction.SdkComponents) RunnerApi(org.apache.beam.model.pipeline.v1.RunnerApi) IOException(java.io.IOException) Pipeline(org.apache.beam.sdk.Pipeline) PCollection(org.apache.beam.sdk.values.PCollection) PortablePipelineOptions(org.apache.beam.sdk.options.PortablePipelineOptions) AllowList(org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList) RehydratedComponents(org.apache.beam.runners.core.construction.RehydratedComponents) ImmutableMap(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.collect.ImmutableMap) Map(java.util.Map) VisibleForTesting(org.apache.beam.vendor.guava.v26_0_jre.com.google.common.annotations.VisibleForTesting)

Example 2 with AllowList

use of org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList in project beam by apache.

the class JavaClassLookupTransformProviderTest method testEverythingAllowList.

@Test
public void testEverythingAllowList() {
    AllowList everything = AllowList.everything();
    AllowedClass allowedClass = everything.getAllowedClass("org.apache.beam.sdk.transforms.KvSwap");
    assertTrue(allowedClass.isAllowedBuilderMethod("builder"));
    assertTrue(allowedClass.isAllowedConstructorMethod("constructor"));
}
Also used : AllowedClass(org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowedClass) AllowList(org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList) Test(org.junit.Test)

Example 3 with AllowList

use of org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList in project beam by apache.

the class JavaClassLookupTransformProviderTest method testNothingAllowList.

@Test
public void testNothingAllowList() {
    AllowList nothing = AllowList.nothing();
    RuntimeException thrown = assertThrows(RuntimeException.class, () -> nothing.getAllowedClass("org.apache.beam.sdk.transforms.KvSwap"));
    assertTrue(thrown.getMessage(), thrown.getMessage().contains("allow list does not enable"));
    assertTrue(thrown.getMessage(), thrown.getMessage().contains("org.apache.beam.sdk.transforms.KvSwap"));
}
Also used : AllowList(org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList) Test(org.junit.Test)

Example 4 with AllowList

use of org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList in project beam by apache.

the class JavaClassLookupTransformProviderTest method testPackageAllowList.

@Test
public void testPackageAllowList() {
    AllowList allowList = AllowList.create(ALLOW_LIST_VERSION, Collections.singletonList(AllowedClass.create("good.package.*", Collections.singletonList("goodBuilder"), AllowedClass.WILDCARD)));
    assertThrows(RuntimeException.class, () -> allowList.getAllowedClass("bad.package.Transform"));
    AllowedClass allowedClass = allowList.getAllowedClass("good.package.Transform");
    assertTrue(allowedClass.isAllowedBuilderMethod("goodBuilder"));
    assertFalse(allowedClass.isAllowedBuilderMethod("badBuilder"));
    assertTrue(allowedClass.isAllowedConstructorMethod("anyConstructor"));
}
Also used : AllowedClass(org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowedClass) AllowList(org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList) Test(org.junit.Test)

Aggregations

AllowList (org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowList)4 Test (org.junit.Test)3 AllowedClass (org.apache.beam.sdk.expansion.service.JavaClassLookupTransformProvider.AllowedClass)2 AutoService (com.google.auto.service.AutoService)1 IOException (java.io.IOException)1 Constructor (java.lang.reflect.Constructor)1 InvocationTargetException (java.lang.reflect.InvocationTargetException)1 Method (java.lang.reflect.Method)1 Arrays (java.util.Arrays)1 Collections (java.util.Collections)1 List (java.util.List)1 Map (java.util.Map)1 Optional (java.util.Optional)1 ServiceLoader (java.util.ServiceLoader)1 Set (java.util.Set)1 Collectors (java.util.stream.Collectors)1 ExpansionApi (org.apache.beam.model.expansion.v1.ExpansionApi)1 ExpansionServiceGrpc (org.apache.beam.model.expansion.v1.ExpansionServiceGrpc)1 ExpansionMethods (org.apache.beam.model.pipeline.v1.ExternalTransforms.ExpansionMethods)1 ExternalConfigurationPayload (org.apache.beam.model.pipeline.v1.ExternalTransforms.ExternalConfigurationPayload)1