Examples with BatchPipelineSpec io.cdap.cdap.etl.batch.BatchPipelineSpec used on opensource projects

Example 1 with BatchPipelineSpec

use of io.cdap.cdap.etl.batch.BatchPipelineSpec in project cdap by caskdata.

the class ValidationUtils method validate.

/**
 * Validate plugin based on the {@link StageValidationRequest}
 *
 * @param validationRequest {@link StageValidationRequest} with plugin properties
 * @param pluginConfigurer  {@link PluginConfigurer} for using the plugin
 * @param macroFn           {@link Function} for evaluating macros
 * @return {@link StageValidationResponse} in json format
 */
public static StageValidationResponse validate(String namespace, StageValidationRequest validationRequest, PluginConfigurer pluginConfigurer, Function<Map<String, String>, Map<String, String>> macroFn, FeatureFlagsProvider featureFlagsProvider) {
    ETLStage stageConfig = validationRequest.getStage();
    ValidatingConfigurer validatingConfigurer = new ValidatingConfigurer(pluginConfigurer, featureFlagsProvider);
    // Batch or Streaming doesn't matter for a single stage.
    PipelineSpecGenerator<ETLBatchConfig, BatchPipelineSpec> pipelineSpecGenerator = new BatchPipelineSpecGenerator(namespace, validatingConfigurer, null, Collections.emptySet(), Collections.emptySet(), Engine.SPARK, featureFlagsProvider);
    DefaultStageConfigurer stageConfigurer = new DefaultStageConfigurer(stageConfig.getName());
    for (StageSchema stageSchema : validationRequest.getInputSchemas()) {
        stageConfigurer.addInputSchema(stageSchema.getStage(), stageSchema.getSchema());
        stageConfigurer.addInputStage(stageSchema.getStage());
    }
    DefaultPipelineConfigurer pipelineConfigurer = new DefaultPipelineConfigurer(validatingConfigurer, stageConfig.getName(), Engine.SPARK, stageConfigurer, featureFlagsProvider);
    // evaluate macros
    Map<String, String> evaluatedProperties = macroFn.apply(stageConfig.getPlugin().getProperties());
    ETLPlugin originalConfig = stageConfig.getPlugin();
    ETLPlugin evaluatedConfig = new ETLPlugin(originalConfig.getName(), originalConfig.getType(), evaluatedProperties, originalConfig.getArtifactConfig());
    try {
        StageSpec spec = pipelineSpecGenerator.configureStage(stageConfig.getName(), evaluatedConfig, pipelineConfigurer).build();
        return new StageValidationResponse(spec);
    } catch (ValidationException e) {
        return new StageValidationResponse(e.getFailures());
    }
}

Example 2 with BatchPipelineSpec

use of io.cdap.cdap.etl.batch.BatchPipelineSpec in project cdap by caskdata.

the class PipelineSpecGeneratorTest method testGenerateSpec.

@Test
public void testGenerateSpec() throws ValidationException {
    /*
     *           ---- t1 ------------
     *           |            |      |
     * source ---             |      |--- t3 --- sink1
     *           |            |      |
     *           ------------ t2 --------------- sink2
     *           |                        |
     *           |                        |
     *           -------------------------
     */
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().setTimeSchedule("* * * * *").addStage(new ETLStage("source", MOCK_SOURCE)).addStage(new ETLStage("sink1", MOCK_SINK)).addStage(new ETLStage("sink2", MOCK_SINK)).addStage(new ETLStage("t1", MOCK_TRANSFORM_A)).addStage(new ETLStage("t2", MOCK_TRANSFORM_A)).addStage(new ETLStage("t3", MOCK_TRANSFORM_B)).addConnection("source", "t1").addConnection("source", "t2").addConnection("source", "sink2").addConnection("t1", "t2").addConnection("t1", "t3").addConnection("t1", "sink2").addConnection("t2", "sink2").addConnection("t2", "t3").addConnection("t3", "sink1").setNumOfRecordsPreview(100).build();
    // test the spec generated is correct, with the right input and output schemas and artifact information.
    BatchPipelineSpec actual = specGenerator.generateSpec(etlConfig);
    Map<String, String> emptyMap = ImmutableMap.of();
    PipelineSpec expected = BatchPipelineSpec.builder().addStage(StageSpec.builder("source", new PluginSpec(BatchSource.PLUGIN_TYPE, "mocksource", emptyMap, ARTIFACT_ID)).addOutput(SCHEMA_A, "t1", "t2", "sink2").build()).addStage(StageSpec.builder("sink1", new PluginSpec(BatchSink.PLUGIN_TYPE, "mocksink", emptyMap, ARTIFACT_ID)).addInputSchema("t3", SCHEMA_B).setErrorSchema(SCHEMA_B).build()).addStage(StageSpec.builder("sink2", new PluginSpec(BatchSink.PLUGIN_TYPE, "mocksink", emptyMap, ARTIFACT_ID)).addInputSchemas(ImmutableMap.of("t1", SCHEMA_A, "t2", SCHEMA_A, "source", SCHEMA_A)).setErrorSchema(SCHEMA_A).build()).addStage(StageSpec.builder("t1", new PluginSpec(Transform.PLUGIN_TYPE, "mockA", emptyMap, ARTIFACT_ID)).addInputSchema("source", SCHEMA_A).addOutput(SCHEMA_A, "t2", "t3", "sink2").setErrorSchema(SCHEMA_B).build()).addStage(StageSpec.builder("t2", new PluginSpec(Transform.PLUGIN_TYPE, "mockA", emptyMap, ARTIFACT_ID)).addInputSchemas(ImmutableMap.of("source", SCHEMA_A, "t1", SCHEMA_A)).addOutput(SCHEMA_A, "t3", "sink2").setErrorSchema(SCHEMA_B).build()).addStage(StageSpec.builder("t3", new PluginSpec(Transform.PLUGIN_TYPE, "mockB", emptyMap, ARTIFACT_ID)).addInputSchemas(ImmutableMap.of("t1", SCHEMA_A, "t2", SCHEMA_A)).addOutput(SCHEMA_B, "sink1").setErrorSchema(SCHEMA_A).build()).addConnections(etlConfig.getConnections()).setResources(etlConfig.getResources()).setDriverResources(new Resources(1024, 1)).setClientResources(new Resources(1024, 1)).setStageLoggingEnabled(etlConfig.isStageLoggingEnabled()).setNumOfRecordsPreview(etlConfig.getNumOfRecordsPreview()).build();
    Assert.assertEquals(expected, actual);
}