Examples with TrackedTransform co.cask.cdap.etl.common.TrackedTransform used on opensource projects

Example 1 with TrackedTransform

use of co.cask.cdap.etl.common.TrackedTransform in project cdap by caskdata.

the class ETLWorker method initializeTransforms.

private void initializeTransforms(WorkerContext context, Map<String, TransformDetail> transformDetailMap, PipelinePhase pipeline) throws Exception {
    Set<StageInfo> transformInfos = pipeline.getStagesOfType(Transform.PLUGIN_TYPE);
    Preconditions.checkArgument(transformInfos != null);
    tranformIdToDatasetName = new HashMap<>(transformInfos.size());
    for (StageInfo transformInfo : transformInfos) {
        String transformName = transformInfo.getName();
        try {
            Transform<?, ?> transform = context.newPluginInstance(transformName);
            ;
            transform = new WrappedTransform<>(transform, Caller.DEFAULT);
            WorkerRealtimeContext transformContext = new WorkerRealtimeContext(context, metrics, new TxLookupProvider(context), transformInfo);
            LOG.debug("Transform Class : {}", transform.getClass().getName());
            transform.initialize(transformContext);
            StageMetrics stageMetrics = new DefaultStageMetrics(metrics, transformName);
            transformDetailMap.put(transformName, new TransformDetail(new TrackedTransform<>(transform, stageMetrics, context.getDataTracer(transformName)), pipeline.getStageOutputs(transformName)));
            if (transformInfo.getErrorDatasetName() != null) {
                tranformIdToDatasetName.put(transformName, transformInfo.getErrorDatasetName());
            }
        } catch (InstantiationException e) {
            LOG.error("Unable to instantiate Transform", e);
            Throwables.propagate(e);
        }
    }
}

Example 2 with TrackedTransform

use of co.cask.cdap.etl.common.TrackedTransform in project cdap by caskdata.

the class ETLWorker method initializeSinks.

@SuppressWarnings("unchecked")
private void initializeSinks(WorkerContext context, Map<String, TransformDetail> transformationMap, PipelinePhase pipeline) throws Exception {
    Set<StageInfo> sinkInfos = pipeline.getStagesOfType(RealtimeSink.PLUGIN_TYPE);
    sinks = new HashMap<>(sinkInfos.size());
    for (StageInfo sinkInfo : sinkInfos) {
        String sinkName = sinkInfo.getName();
        RealtimeSink sink = context.newPluginInstance(sinkName);
        sink = new LoggedRealtimeSink(sinkName, sink);
        WorkerRealtimeContext sinkContext = new WorkerRealtimeContext(context, metrics, new TxLookupProvider(context), sinkInfo);
        LOG.debug("Sink Class : {}", sink.getClass().getName());
        sink.initialize(sinkContext);
        sink = new TrackedRealtimeSink(sink, new DefaultStageMetrics(metrics, sinkName));
        Transformation identityTransformation = new Transformation() {

            @Override
            public void transform(Object input, Emitter emitter) throws Exception {
                emitter.emit(input);
            }
        };
        // we use identity transformation to simplify executing transformation in pipeline (similar to ETLMapreduce),
        // since we want to emit metrics during write to sink and not during this transformation, we use NoOpMetrics.
        TrackedTransform trackedTransform = new TrackedTransform(identityTransformation, new DefaultStageMetrics(metrics, sinkName), TrackedTransform.RECORDS_IN, null, context.getDataTracer(sinkName));
        transformationMap.put(sinkInfo.getName(), new TransformDetail(trackedTransform, new HashSet<String>()));
        sinks.put(sinkInfo.getName(), sink);
    }
}

Example 3 with TrackedTransform

use of co.cask.cdap.etl.common.TrackedTransform in project cdap by caskdata.

the class MapReduceTransformExecutorFactory method getTransformation.

@SuppressWarnings("unchecked")
@Override
protected TrackedTransform getTransformation(StageInfo stageInfo) throws Exception {
    DefaultMacroEvaluator macroEvaluator = new DefaultMacroEvaluator(taskContext.getWorkflowToken(), taskContext.getRuntimeArguments(), taskContext.getLogicalStartTime(), taskContext, taskContext.getNamespace());
    String stageName = stageInfo.getName();
    String pluginType = stageInfo.getPluginType();
    StageMetrics stageMetrics = new DefaultStageMetrics(metrics, stageName);
    if (BatchAggregator.PLUGIN_TYPE.equals(pluginType)) {
        BatchAggregator<?, ?, ?> batchAggregator = pluginInstantiator.newPluginInstance(stageName, macroEvaluator);
        BatchRuntimeContext runtimeContext = createRuntimeContext(stageInfo);
        batchAggregator.initialize(runtimeContext);
        if (isMapPhase) {
            return getTrackedEmitKeyStep(new MapperAggregatorTransformation(batchAggregator, mapOutputKeyClassName, mapOutputValClassName), stageMetrics, taskContext.getDataTracer(stageName));
        } else {
            return getTrackedAggregateStep(new ReducerAggregatorTransformation(batchAggregator, mapOutputKeyClassName, mapOutputValClassName), stageMetrics, taskContext.getDataTracer(stageName));
        }
    } else if (BatchJoiner.PLUGIN_TYPE.equals(pluginType)) {
        BatchJoiner<?, ?, ?> batchJoiner = pluginInstantiator.newPluginInstance(stageName, macroEvaluator);
        BatchJoinerRuntimeContext runtimeContext = createRuntimeContext(stageInfo);
        batchJoiner.initialize(runtimeContext);
        if (isMapPhase) {
            return getTrackedEmitKeyStep(new MapperJoinerTransformation(batchJoiner, mapOutputKeyClassName, mapOutputValClassName), stageMetrics, taskContext.getDataTracer(stageName));
        } else {
            return getTrackedMergeStep(new ReducerJoinerTransformation(batchJoiner, mapOutputKeyClassName, mapOutputValClassName, runtimeContext.getInputSchemas().size()), stageMetrics, taskContext.getDataTracer(stageName));
        }
    }
    Transformation transformation = getInitializedTransformation(stageInfo);
    boolean isLimitingSource = taskContext.getDataTracer(stageName).isEnabled() && BatchSource.PLUGIN_TYPE.equals(pluginType) && isMapPhase;
    return new TrackedTransform(isLimitingSource ? new LimitingTransform(transformation, numberOfRecordsPreview) : transformation, stageMetrics, taskContext.getDataTracer(stageName));
}