Examples with PCollectionNode org.apache.beam.runners.core.construction.graph.PipelineNode.PCollectionNode used on opensource projects

Example 11 with PCollectionNode

use of org.apache.beam.runners.core.construction.graph.PipelineNode.PCollectionNode in project beam by apache.

the class ExecutableStage method fromPayload.

/**
 * Return an {@link ExecutableStage} constructed from the provided {@link FunctionSpec}
 * representation.
 *
 * <p>See {@link #toPTransform} for how the payload is constructed.
 *
 * <p>Note: The payload contains some information redundant with the {@link PTransform} it is the
 * payload of. The {@link ExecutableStagePayload} should be sufficiently rich to construct a
 * {@code ProcessBundleDescriptor} using only the payload.
 */
static ExecutableStage fromPayload(ExecutableStagePayload payload) {
    Components components = payload.getComponents();
    Environment environment = payload.getEnvironment();
    Collection<WireCoderSetting> wireCoderSettings = payload.getWireCoderSettingsList();
    PCollectionNode input = PipelineNode.pCollection(payload.getInput(), components.getPcollectionsOrThrow(payload.getInput()));
    List<SideInputReference> sideInputs = payload.getSideInputsList().stream().map(sideInputId -> SideInputReference.fromSideInputId(sideInputId, components)).collect(Collectors.toList());
    List<UserStateReference> userStates = payload.getUserStatesList().stream().map(userStateId -> UserStateReference.fromUserStateId(userStateId, components)).collect(Collectors.toList());
    List<TimerReference> timers = payload.getTimersList().stream().map(timerId -> TimerReference.fromTimerId(timerId, components)).collect(Collectors.toList());
    List<PTransformNode> transforms = payload.getTransformsList().stream().map(id -> PipelineNode.pTransform(id, components.getTransformsOrThrow(id))).collect(Collectors.toList());
    List<PCollectionNode> outputs = payload.getOutputsList().stream().map(id -> PipelineNode.pCollection(id, components.getPcollectionsOrThrow(id))).collect(Collectors.toList());
    return ImmutableExecutableStage.of(components, environment, input, sideInputs, userStates, timers, transforms, outputs, wireCoderSettings);
}

Example 12 with PCollectionNode

use of org.apache.beam.runners.core.construction.graph.PipelineNode.PCollectionNode in project beam by apache.

the class QueryablePipeline method buildNetwork.

private MutableNetwork<PipelineNode, PipelineEdge> buildNetwork(Collection<String> transformIds, Components components) {
    MutableNetwork<PipelineNode, PipelineEdge> network = NetworkBuilder.directed().allowsParallelEdges(true).allowsSelfLoops(false).build();
    Set<PCollectionNode> unproducedCollections = new HashSet<>();
    for (String transformId : transformIds) {
        PTransform transform = components.getTransformsOrThrow(transformId);
        PTransformNode transformNode = PipelineNode.pTransform(transformId, this.components.getTransformsOrThrow(transformId));
        network.addNode(transformNode);
        for (String produced : transform.getOutputsMap().values()) {
            PCollectionNode producedNode = PipelineNode.pCollection(produced, components.getPcollectionsOrThrow(produced));
            network.addNode(producedNode);
            network.addEdge(transformNode, producedNode, new PerElementEdge());
            checkArgument(network.inDegree(producedNode) == 1, "A %s should have exactly one producing %s, but found %s:\nPCollection:\n%s\nProducers:\n%s", PCollectionNode.class.getSimpleName(), PTransformNode.class.getSimpleName(), network.predecessors(producedNode).size(), producedNode, network.predecessors(producedNode));
            unproducedCollections.remove(producedNode);
        }
        for (Map.Entry<String, String> consumed : transform.getInputsMap().entrySet()) {
            // This loop may add an edge between the consumed PCollection and the current PTransform.
            // The local name of the transform must be used to determine the type of edge.
            String pcollectionId = consumed.getValue();
            PCollectionNode consumedNode = PipelineNode.pCollection(pcollectionId, this.components.getPcollectionsOrThrow(pcollectionId));
            if (network.addNode(consumedNode)) {
                // This node has been added to the network for the first time, so it has no producer.
                unproducedCollections.add(consumedNode);
            }
            if (getLocalSideInputNames(transform).contains(consumed.getKey())) {
                network.addEdge(consumedNode, transformNode, new SingletonEdge());
            } else {
                network.addEdge(consumedNode, transformNode, new PerElementEdge());
            }
        }
    }
    checkArgument(unproducedCollections.isEmpty(), "%ss %s were consumed but never produced", PCollectionNode.class.getSimpleName(), unproducedCollections);
    return network;
}

Example 13 with PCollectionNode

use of org.apache.beam.runners.core.construction.graph.PipelineNode.PCollectionNode in project beam by apache.

the class OutputDeduplicator method deduplicateStageOutput.

/**
 * Returns an {@link ExecutableStage} where all of the {@link PCollectionNode PCollections}
 * matching the original are replaced with the introduced partial {@link PCollection} in all
 * references made within the {@link ExecutableStage}.
 */
private static ExecutableStage deduplicateStageOutput(ExecutableStage stage, Map<String, PCollectionNode> originalToPartial) {
    Collection<PTransformNode> updatedTransforms = new ArrayList<>();
    for (PTransformNode transform : stage.getTransforms()) {
        PTransform updatedTransform = updateOutputs(transform.getTransform(), originalToPartial);
        updatedTransforms.add(PipelineNode.pTransform(transform.getId(), updatedTransform));
    }
    Collection<PCollectionNode> updatedOutputs = new ArrayList<>();
    for (PCollectionNode output : stage.getOutputPCollections()) {
        updatedOutputs.add(originalToPartial.getOrDefault(output.getId(), output));
    }
    RunnerApi.Components updatedStageComponents = stage.getComponents().toBuilder().clearTransforms().putAllTransforms(updatedTransforms.stream().collect(Collectors.toMap(PTransformNode::getId, PTransformNode::getTransform))).putAllPcollections(originalToPartial.values().stream().collect(Collectors.toMap(PCollectionNode::getId, PCollectionNode::getPCollection))).build();
    return ImmutableExecutableStage.of(updatedStageComponents, stage.getEnvironment(), stage.getInputPCollection(), stage.getSideInputs(), stage.getUserStates(), stage.getTimers(), updatedTransforms, updatedOutputs, stage.getWireCoderSettings());
}

Example 14 with PCollectionNode

use of org.apache.beam.runners.core.construction.graph.PipelineNode.PCollectionNode in project beam by apache.

the class OutputDeduplicator method ensureSingleProducer.

/**
 * Ensure that no {@link PCollection} output by any of the {@code stages} or {@code
 * unfusedTransforms} is produced by more than one of those stages or transforms.
 *
 * <p>For each {@link PCollection} output by multiple stages and/or transforms, each producer is
 * rewritten to produce a partial {@link PCollection}, which are then flattened together via an
 * introduced Flatten node which produces the original output.
 */
static DeduplicationResult ensureSingleProducer(QueryablePipeline pipeline, Collection<ExecutableStage> stages, Collection<PTransformNode> unfusedTransforms) {
    RunnerApi.Components.Builder unzippedComponents = pipeline.getComponents().toBuilder();
    Multimap<PCollectionNode, StageOrTransform> pcollectionProducers = getProducers(pipeline, stages, unfusedTransforms);
    Multimap<StageOrTransform, PCollectionNode> requiresNewOutput = HashMultimap.create();
    // ExecutableStage must also be rewritten to have updated outputs and transforms.
    for (Map.Entry<PCollectionNode, Collection<StageOrTransform>> collectionProducer : pcollectionProducers.asMap().entrySet()) {
        if (collectionProducer.getValue().size() > 1) {
            for (StageOrTransform producer : collectionProducer.getValue()) {
                requiresNewOutput.put(producer, collectionProducer.getKey());
            }
        }
    }
    Map<ExecutableStage, ExecutableStage> updatedStages = new LinkedHashMap<>();
    Map<String, PTransformNode> updatedTransforms = new LinkedHashMap<>();
    Multimap<String, PCollectionNode> originalToPartial = HashMultimap.create();
    for (Map.Entry<StageOrTransform, Collection<PCollectionNode>> deduplicationTargets : requiresNewOutput.asMap().entrySet()) {
        if (deduplicationTargets.getKey().getStage() != null) {
            StageDeduplication deduplication = deduplicatePCollections(deduplicationTargets.getKey().getStage(), deduplicationTargets.getValue(), unzippedComponents::containsPcollections);
            for (Entry<String, PCollectionNode> originalToPartialReplacement : deduplication.getOriginalToPartialPCollections().entrySet()) {
                originalToPartial.put(originalToPartialReplacement.getKey(), originalToPartialReplacement.getValue());
                unzippedComponents.putPcollections(originalToPartialReplacement.getValue().getId(), originalToPartialReplacement.getValue().getPCollection());
            }
            updatedStages.put(deduplicationTargets.getKey().getStage(), deduplication.getUpdatedStage());
        } else if (deduplicationTargets.getKey().getTransform() != null) {
            PTransformDeduplication deduplication = deduplicatePCollections(deduplicationTargets.getKey().getTransform(), deduplicationTargets.getValue(), unzippedComponents::containsPcollections);
            for (Entry<String, PCollectionNode> originalToPartialReplacement : deduplication.getOriginalToPartialPCollections().entrySet()) {
                originalToPartial.put(originalToPartialReplacement.getKey(), originalToPartialReplacement.getValue());
                unzippedComponents.putPcollections(originalToPartialReplacement.getValue().getId(), originalToPartialReplacement.getValue().getPCollection());
            }
            updatedTransforms.put(deduplicationTargets.getKey().getTransform().getId(), deduplication.getUpdatedTransform());
        } else {
            throw new IllegalStateException(String.format("%s with no %s or %s", StageOrTransform.class.getSimpleName(), ExecutableStage.class.getSimpleName(), PTransformNode.class.getSimpleName()));
        }
    }
    Set<PTransformNode> introducedFlattens = new LinkedHashSet<>();
    for (Map.Entry<String, Collection<PCollectionNode>> partialFlattenTargets : originalToPartial.asMap().entrySet()) {
        String flattenId = SyntheticComponents.uniqueId("unzipped_flatten", unzippedComponents::containsTransforms);
        PTransform flattenPartialPCollections = createFlattenOfPartials(flattenId, partialFlattenTargets.getKey(), partialFlattenTargets.getValue());
        unzippedComponents.putTransforms(flattenId, flattenPartialPCollections);
        introducedFlattens.add(PipelineNode.pTransform(flattenId, flattenPartialPCollections));
    }
    Components components = unzippedComponents.build();
    return DeduplicationResult.of(components, introducedFlattens, updatedStages, updatedTransforms);
}

Example 15 with PCollectionNode

use of org.apache.beam.runners.core.construction.graph.PipelineNode.PCollectionNode in project beam by apache.

the class OutputDeduplicator method deduplicatePCollections.

private static PTransformDeduplication deduplicatePCollections(PTransformNode transform, Collection<PCollectionNode> duplicates, Predicate<String> existingPCollectionIds) {
    Map<String, PCollectionNode> unzippedOutputs = createPartialPCollections(duplicates, existingPCollectionIds);
    PTransform pTransform = updateOutputs(transform.getTransform(), unzippedOutputs);
    return PTransformDeduplication.of(PipelineNode.pTransform(transform.getId(), pTransform), unzippedOutputs);
}