com.google.cloud.dataflow.sdk.values.PCollectionView Java Examples

@Override
protected void testProgram() throws Exception {


	Pipeline p = FlinkTestPipeline.createForBatch();


	final PCollectionView<String> sidesInput = p
			.apply(Create.of(expected))
			.apply(View.<String>asSingleton());

	p.apply(Create.of("bli"))
			.apply(ParDo.of(new DoFn<String, String>() {
				@Override
				public void processElement(ProcessContext c) throws Exception {
					String s = c.sideInput(sidesInput);
					c.output(s);
				}
			}).withSideInputs(sidesInput)).apply(TextIO.Write.to(resultPath));

	p.run();
}
 

private FlinkInMemoryKeyedCombiningValue(ByteString stateKey,
                                         CombineWithContext.KeyedCombineFnWithContext<? super K, InputT, AccumT, OutputT> combineFn,
                                         Coder<AccumT> accumCoder,
                                         final StateContext<?> stateContext) {
	Preconditions.checkNotNull(combineFn);
	Preconditions.checkNotNull(accumCoder);

	this.stateKey = stateKey;
	this.combineFn = combineFn;
	this.accumCoder = accumCoder;
	this.context = new CombineWithContext.Context() {
		@Override
		public PipelineOptions getPipelineOptions() {
			return stateContext.getPipelineOptions();
		}

		@Override
		public <T> T sideInput(PCollectionView<T> view) {
			return stateContext.sideInput(view);
		}
	};
	accum = combineFn.createAccumulator(key, context);
}
 

/**
 * @param config Configuration for the correlations
 * @param sideInput Sideinput to be used for the transform 
 * @param matrix Key Matrix
 */
public PartitionedComputeCorrelationsDoFn(CorrelationParDoConfig config,
    PCollectionView<Map<String, WorkPacket>> sideInput,
    PCollectionView<Map<Integer, List<WorkPacketKey>>> matrix) {
  
  this.sideInput = sideInput;
  this.keyMatrix = matrix;
  this.config = config;
}
 

@Override
public WindowingInternals<IN, OUT> windowingInternals() {
	return new WindowingInternals<IN, OUT>() {
		@Override
		public StateInternals stateInternals() {
			return null;
		}

		@Override
		public void outputWindowedValue(OUT output, Instant timestamp, Collection<? extends BoundedWindow> windows, PaneInfo pane) {

		}

		@Override
		public TimerInternals timerInternals() {
			return null;
		}

		@Override
		public Collection<? extends BoundedWindow> windows() {
			return ImmutableList.of(GlobalWindow.INSTANCE);
		}

		@Override
		public PaneInfo pane() {
			return PaneInfo.NO_FIRING;
		}

		@Override
		public <T> void writePCollectionViewData(TupleTag<?> tag, Iterable<WindowedValue<T>> data, Coder<T> elemCoder) throws IOException {
		}

		@Override
		public <T> T sideInput(PCollectionView<T> view, BoundedWindow mainInputWindow) {
			throw new RuntimeException("sideInput() not implemented.");
		}
	};
}
 

private static void transformSideInputs(List<PCollectionView<?>> sideInputs,
                                        MapPartitionOperator<?, ?> outputDataSet,
                                        FlinkBatchTranslationContext context) {
	// get corresponding Flink broadcast DataSets
	for(PCollectionView<?> input : sideInputs) {
		DataSet<?> broadcastSet = context.getSideInputDataSet(input);
		outputDataSet.withBroadcastSet(broadcastSet, input.getTagInternal().getId());
	}
}
 

@Override
public WindowingInternals<IN, OUT> windowingInternalsHelper(final WindowedValue<IN> inElement, final Collector<WindowedValue<OUT>> collector) {
	return new WindowingInternals<IN, OUT>() {
		@Override
		public StateInternals stateInternals() {
			throw new NullPointerException("StateInternals are not available for ParDo.Bound().");
		}

		@Override
		public void outputWindowedValue(OUT output, Instant timestamp, Collection<? extends BoundedWindow> windows, PaneInfo pane) {
			collector.collect(makeWindowedValue(output, timestamp, windows, pane));
		}

		@Override
		public TimerInternals timerInternals() {
			throw new NullPointerException("TimeInternals are not available for ParDo.Bound().");
		}

		@Override
		public Collection<? extends BoundedWindow> windows() {
			return inElement.getWindows();
		}

		@Override
		public PaneInfo pane() {
			return inElement.getPane();
		}

		@Override
		public <T> void writePCollectionViewData(TupleTag<?> tag, Iterable<WindowedValue<T>> data, Coder<T> elemCoder) throws IOException {
			throw new RuntimeException("writePCollectionViewData() not supported in Streaming mode.");
		}

		@Override
		public <T> T sideInput(PCollectionView<T> view, BoundedWindow mainInputWindow) {
			throw new RuntimeException("sideInput() not implemented.");
		}
	};
}
 

@Test
public void testFilterStateOneMismatch() {
  GCPProject project = getSampleProject("");
  GCPResourceState checkedPolicy = getSampleGCPResourcePolicy(project, 1);
  GCPResourceState livePolicy = getSampleGCPResourcePolicy(project, 2);
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> sideInputList =
      Arrays.asList(KV.of((GCPResource) project, KV.of(StateSource.DESIRED, checkedPolicy)));
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> mainInputList =
      Arrays.asList(KV.of((GCPResource) project, KV.of(StateSource.LIVE, livePolicy)));

  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> collection =
      pipeline.apply(Create.of(sideInputList)).setCoder(
          KvCoder.of(SerializableCoder.of(GCPResource.class),
              KvCoder.of(SerializableCoder.of(StateSource.class),
                  SerializableCoder.of(GCPResourceState.class))));
  PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view =
      View.<GCPResource, KV<StateSource, GCPResourceState>>asMap().apply(collection);

  JoinKnownGoodAndLiveStates function = new JoinKnownGoodAndLiveStates(view);
  DoFnTester<KV<GCPResource, KV<StateSource, GCPResourceState>>,
      KV<GCPResource, Map<StateSource, GCPResourceState>>> tester = DoFnTester.of(function);
  tester.setSideInputInGlobalWindow(view, sideInputList);

  Map<StateSource, GCPResourceState> outputMap = new HashMap<>(2);
  outputMap.put(StateSource.DESIRED, checkedPolicy);
  outputMap.put(StateSource.LIVE, livePolicy);
  List<KV<GCPResource, Map<StateSource, GCPResourceState>>> expectedOutput =
      Arrays.asList(KV.of((GCPResource) project, outputMap));
  List<KV<GCPResource, Map<StateSource, GCPResourceState>>> results =
      tester.processBatch(mainInputList);

  assertEquals(expectedOutput, results);
}
 

@Test
public void testFilterStateNoMismatches() {
  GCPProject project = getSampleProject("");
  GCPResourceState checkedPolicy = getSampleGCPResourcePolicy(project, 1);
  GCPResourceState livePolicy = checkedPolicy;
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> sideInputList =
      Arrays.asList(KV.of((GCPResource) project, KV.of(StateSource.DESIRED, checkedPolicy)));
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> mainInputList =
      Arrays.asList(KV.of((GCPResource) project, KV.of(StateSource.LIVE, livePolicy)));

  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> collection =
      pipeline.apply(Create.of(sideInputList)).setCoder(
          KvCoder.of(SerializableCoder.of(GCPResource.class),
              KvCoder.of(SerializableCoder.of(StateSource.class),
                  SerializableCoder.of(GCPResourceState.class))));
  PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view =
      View.<GCPResource, KV<StateSource, GCPResourceState>>asMap().apply(collection);

  JoinKnownGoodAndLiveStates function = new JoinKnownGoodAndLiveStates(view);
  DoFnTester<KV<GCPResource, KV<StateSource, GCPResourceState>>,
      KV<GCPResource, Map<StateSource, GCPResourceState>>> tester = DoFnTester.of(function);
  tester.setSideInputInGlobalWindow(view, sideInputList);

  List<KV<GCPResource, Map<StateSource, GCPResourceState>>> results =
      tester.processBatch(mainInputList);
  assertEquals(1, results.size());
}
 

@Test
public void testFilterStateNoMatchingResources() {
  GCPProject checkedProject = getSampleProject("_checked");
  GCPProject liveProject = getSampleProject("_live");
  GCPResourceState checkedPolicy = getSampleGCPResourcePolicy(checkedProject, 1);
  GCPResourceState livePolicy = getSampleGCPResourcePolicy(liveProject, 2);
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> sideInputList =
      Arrays.asList(KV.of((GCPResource) checkedProject, KV.of(StateSource.DESIRED, checkedPolicy)));
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> mainInputList =
      Arrays.asList(KV.of((GCPResource) liveProject, KV.of(StateSource.LIVE, livePolicy)));

  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> collection =
      pipeline.apply(Create.of(sideInputList)).setCoder(
          KvCoder.of(SerializableCoder.of(GCPResource.class),
              KvCoder.of(SerializableCoder.of(StateSource.class),
                  SerializableCoder.of(GCPResourceState.class))));
  PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view =
      View.<GCPResource, KV<StateSource, GCPResourceState>>asMap().apply(collection);

  JoinKnownGoodAndLiveStates function = new JoinKnownGoodAndLiveStates(view);
  DoFnTester<KV<GCPResource, KV<StateSource, GCPResourceState>>,
      KV<GCPResource, Map<StateSource, GCPResourceState>>> tester = DoFnTester.of(function);
  tester.setSideInputInGlobalWindow(view, sideInputList);

  List<KV<GCPResource, Map<StateSource, GCPResourceState>>> results =
      tester.processBatch(mainInputList);
  assertEquals(0, results.size());
}
 

@Test
public void testFilterStateOneMismatch() {
  GCPProject project = getSampleProject("");
  GCPResourceState checkedPolicy = getSampleGCPResourcePolicy(project, 1);
  GCPResourceState livePolicy = getSampleGCPResourcePolicy(project, 2);
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> sideInputList =
      Arrays.asList(KV.of((GCPResource) project, KV.of(StateSource.DESIRED, checkedPolicy)));
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> mainInputList =
      Arrays.asList(KV.of((GCPResource) project, KV.of(StateSource.LIVE, livePolicy)));

  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> collection =
      pipeline.apply(Create.of(sideInputList)).setCoder(
          KvCoder.of(SerializableCoder.of(GCPResource.class),
              KvCoder.of(SerializableCoder.of(StateSource.class),
                  SerializableCoder.of(GCPResourceState.class))));
  PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view =
      View.<GCPResource, KV<StateSource, GCPResourceState>>asMap().apply(collection);

  FilterOutMatchingState function = new FilterOutMatchingState(view);
  DoFnTester<KV<GCPResource, KV<StateSource, GCPResourceState>>,
      KV<GCPResource, Map<StateSource, GCPResourceState>>> tester = DoFnTester.of(function);
  tester.setSideInputInGlobalWindow(view, sideInputList);

  Map<StateSource, GCPResourceState> outputMap = new HashMap<>(2);
  outputMap.put(StateSource.DESIRED, checkedPolicy);
  outputMap.put(StateSource.LIVE, livePolicy);
  List<KV<GCPResource, Map<StateSource, GCPResourceState>>> expectedOutput =
      Arrays.asList(KV.of((GCPResource) project, outputMap));
  List<KV<GCPResource, Map<StateSource, GCPResourceState>>> results =
      tester.processBatch(mainInputList);

  assertEquals(expectedOutput, results);
}
 

@Test
public void testFilterStateNoMismatches() {
  GCPProject project = getSampleProject("");
  GCPResourceState checkedPolicy = getSampleGCPResourcePolicy(project, 1);
  GCPResourceState livePolicy = checkedPolicy;
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> sideInputList =
      Arrays.asList(KV.of((GCPResource) project, KV.of(StateSource.DESIRED, checkedPolicy)));
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> mainInputList =
      Arrays.asList(KV.of((GCPResource) project, KV.of(StateSource.LIVE, livePolicy)));

  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> collection =
      pipeline.apply(Create.of(sideInputList)).setCoder(
          KvCoder.of(SerializableCoder.of(GCPResource.class),
              KvCoder.of(SerializableCoder.of(StateSource.class),
                  SerializableCoder.of(GCPResourceState.class))));
  PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view =
      View.<GCPResource, KV<StateSource, GCPResourceState>>asMap().apply(collection);

  FilterOutMatchingState function = new FilterOutMatchingState(view);
  DoFnTester<KV<GCPResource, KV<StateSource, GCPResourceState>>,
      KV<GCPResource, Map<StateSource, GCPResourceState>>> tester = DoFnTester.of(function);
  tester.setSideInputInGlobalWindow(view, sideInputList);

  List<KV<GCPResource, Map<StateSource, GCPResourceState>>> results =
      tester.processBatch(mainInputList);
  assertEquals(0, results.size());
}
 

@Test
public void testFilterStateNoMatchingResources() {
  GCPProject checkedProject = getSampleProject("_checked");
  GCPProject liveProject = getSampleProject("_live");
  GCPResourceState checkedPolicy = getSampleGCPResourcePolicy(checkedProject, 1);
  GCPResourceState livePolicy = getSampleGCPResourcePolicy(liveProject, 2);
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> sideInputList =
      Arrays.asList(KV.of((GCPResource) checkedProject, KV.of(StateSource.DESIRED, checkedPolicy)));
  List<KV<GCPResource, KV<StateSource, GCPResourceState>>> mainInputList =
      Arrays.asList(KV.of((GCPResource) liveProject, KV.of(StateSource.LIVE, livePolicy)));

  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> collection =
      pipeline.apply(Create.of(sideInputList)).setCoder(
          KvCoder.of(SerializableCoder.of(GCPResource.class),
              KvCoder.of(SerializableCoder.of(StateSource.class),
                  SerializableCoder.of(GCPResourceState.class))));
  PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view =
      View.<GCPResource, KV<StateSource, GCPResourceState>>asMap().apply(collection);

  FilterOutMatchingState function = new FilterOutMatchingState(view);
  DoFnTester<KV<GCPResource, KV<StateSource, GCPResourceState>>,
      KV<GCPResource, Map<StateSource, GCPResourceState>>> tester = DoFnTester.of(function);
  tester.setSideInputInGlobalWindow(view, sideInputList);

  List<KV<GCPResource, Map<StateSource, GCPResourceState>>> results =
      tester.processBatch(mainInputList);
  assertEquals(0, results.size());
}
 

private PCollection<String> constructPipeline(Pipeline pipeline,
    BoundedSource<KV<List<String>, String>> knownGoodSource) {
  // Read files from GCS.
  PCollection<KV<List<String>, String>> knownGoodFiles =
      pipeline.apply("Read known-good data", Read.from(knownGoodSource));
  // Convert files to GCPResourceState objects.
  PCollection<KV<GCPResource, GCPResourceState>> knownGoodStates =
      knownGoodFiles.apply(ParDo.named("Convert file data to Java objects")
          .of(new FileToState()));
  // Tag the state objects to indicate they're from a checked-in repo and not live.
  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> taggedKnownGoodStates =
      knownGoodStates.apply(ParDo.named("Mark states as being known-good")
          .of(new TagStateWithSource(StateSource.DESIRED)));

  // Extract a list of checked-in projects from GCS.
  PCollection<List<String>> allFilePaths = knownGoodFiles
      .apply("Extract just the file paths", ParDo.of(new FilePathFromPair()));
  // Read the live version of the states of the checked-in projects.
  PCollection<KV<GCPResource, GCPResourceState>> liveStates =
      allFilePaths.apply(ParDo.named("Get live resource and states from file path")
          .of(new FilePathToLiveState()));
  // Tag the states to indicate they're live and not from a checked-in source.
  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> taggedLiveStates =
      liveStates.apply(ParDo.named("Mark states as being live")
          .of(new TagStateWithSource(StateSource.LIVE)));

  // Join the two known-good and the live halves.
  PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> liveStatesView =
      taggedLiveStates.apply(View.<GCPResource, KV<StateSource, GCPResourceState>>asMap());
  PCollection<KV<GCPResource, Map<StateSource, GCPResourceState>>> mismatchedStates =
      taggedKnownGoodStates.apply(ParDo.named("Find states that don't match")
          .withSideInputs(liveStatesView)
          .of(new FilterOutMatchingState(liveStatesView)));
  // Construct an alert message for all the discrepancies found.
  return mismatchedStates.apply(ParDo
      .named("Generate notification messages")
      .of(new StateDiscrepancyMessenger()));
}
 

private PCollection<String> constructPipeline(Pipeline pipeline, String org,
    BoundedSource<KV<List<String>, String>> knownGoodSource) {
  // Read files from GCS.
  PCollection<KV<List<String>, String>> knownGoodFiles =
      pipeline.apply("Read known-good data", Read.from(knownGoodSource));
  // Convert files to GCPResourceState objects.
  PCollection<KV<GCPResource, GCPResourceState>> knownGoodStates =
      knownGoodFiles.apply(ParDo.named("Convert file data to Java Objects")
          .of(new FileToState()));
  // Tag the state objects to indicate they're from a checked-in repo and not live.
  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> taggedKnownGoodStates =
      knownGoodStates.apply(ParDo.named("Mark states as being known-good")
          .of(new TagStateWithSource(StateSource.DESIRED)));

  // Read projects from the CRM API.
  PCollection<GCPProject> allProjects =
      pipeline.apply("Read live projects", Read.from(new LiveProjectSource(org)));
  // Extract project states.
  PCollection<KV<GCPResource, GCPResourceState>> liveStates =
      allProjects
          .apply(ParDo.named("Extract project policies").of(new ExtractState()));
  // Tag the states to indicate they're live and not from a checked-in source.
  PCollection<KV<GCPResource, KV<StateSource, GCPResourceState>>> taggedLiveStates =
      liveStates.apply(ParDo.named("Mark states as being live")
          .of(new TagStateWithSource(StateSource.LIVE)));

  // Join the two known-good and the live halves.
  PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> knownGoodStatesView =
      taggedKnownGoodStates.apply(View.<GCPResource, KV<StateSource, GCPResourceState>>asMap());
  PCollection<KV<GCPResource, Map<StateSource, GCPResourceState>>> mismatchedStates =
      taggedLiveStates.apply(ParDo.named("Find states that don't match")
          .withSideInputs(knownGoodStatesView)
          .of(new FilterOutMatchingState(knownGoodStatesView)));

  // Construct an alert message for all the discrepancies found and fix the discrepancies.
  return mismatchedStates
      .apply(ParDo.named("Fix discrepancies").of(discrepancyAutoFixMessenger));
}
 

/**
 * Constructor for the JoinKnownGoodAndLiveStates DoFn.
 * @param view The PCollectionView which contains the side-input elements.
 */
public JoinKnownGoodAndLiveStates
    (PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view) {
  this.view = view;
}
 

public CreatePartitionWorkPacketsDoFn(WorkPacketConfig workPacketView, PCollectionView<Map<String,WorkPacket>>  sideInput) {
  this.workPacketView = workPacketView;
  this.sideInput=sideInput;
}
 

@SuppressWarnings("unchecked")
public <T> DataSet<T> getSideInputDataSet(PCollectionView<?> value) {
	return (DataSet<T>) broadcastDataSets.get(value);
}
 

public void setSideInputDataSet(PCollectionView<?> value, DataSet<?> set) {
	if (!broadcastDataSets.containsKey(value)) {
		broadcastDataSets.put(value, set);
	}
}
 

@Override
public <T> T sideInput(PCollectionView<T> view) {
	throw new RuntimeException("sideInput() is not supported in Streaming mode.");
}
 

@Override
public void translateNode(View.CreatePCollectionView<R, T> transform, FlinkBatchTranslationContext context) {
	DataSet<T> inputDataSet = context.getInputDataSet(context.getInput(transform));
	PCollectionView<T> input = transform.apply(null);
	context.setSideInputDataSet(input, inputDataSet);
}
 

public FindUnmatchedStates(
    PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view) {
  this.view = view;
}
 

/**
 * Constructor for the FilterOutMatchingState DoFn.
 * @param view The PCollectionView which contains the side-input elements.
 */
public FilterOutMatchingState
    (PCollectionView<Map<GCPResource, KV<StateSource, GCPResourceState>>> view) {
  this.view = view;
}