Examples with ETLPlugin io.cdap.cdap.etl.proto.v2.ETLPlugin used on opensource projects

Example 16 with ETLPlugin

use of io.cdap.cdap.etl.proto.v2.ETLPlugin in project cdap by caskdata.

the class DataPipelineServiceTest method testValidateMultiInputInvalidInputField.

@Test
public void testValidateMultiInputInvalidInputField() throws Exception {
    // StringValueFilterTransform will be configured to filter records where field x has value 'y'
    // it will be invalid because the type of field x will be an int instead of the required string
    String stageName = "tx";
    Map<String, String> properties = new HashMap<>();
    properties.put("field", "x");
    properties.put("value", "y");
    ETLStage stage = new ETLStage(stageName, new ETLPlugin(StringValueFilterTransform.NAME, Transform.PLUGIN_TYPE, properties));
    Schema inputSchema = Schema.recordOf("x", Schema.Field.of("x", Schema.of(Schema.Type.INT)));
    StageValidationRequest requestBody = new StageValidationRequest(stage, ImmutableList.of(new StageSchema("input1", inputSchema), new StageSchema("input2", inputSchema)), false);
    StageValidationResponse actual = sendRequest(requestBody);
    List<String> expectedInputs = ImmutableList.of("input1", "input2");
    Assert.assertNull(actual.getSpec());
    Assert.assertEquals(1, actual.getFailures().size());
    ValidationFailure failure = actual.getFailures().iterator().next();
    // the stage will add 3 causes. Two are related to input field and one is related to config property.
    Assert.assertEquals(3, failure.getCauses().size());
    Assert.assertEquals("field", failure.getCauses().get(0).getAttribute(CauseAttributes.STAGE_CONFIG));
    Assert.assertEquals(stageName, failure.getCauses().get(0).getAttribute(STAGE));
    Assert.assertEquals(stageName, failure.getCauses().get(1).getAttribute(STAGE));
    Assert.assertEquals(stageName, failure.getCauses().get(2).getAttribute(STAGE));
    Assert.assertTrue(expectedInputs.contains(failure.getCauses().get(1).getAttribute(CauseAttributes.INPUT_STAGE)));
    Assert.assertTrue(expectedInputs.contains(failure.getCauses().get(2).getAttribute(CauseAttributes.INPUT_STAGE)));
}

Example 17 with ETLPlugin

use of io.cdap.cdap.etl.proto.v2.ETLPlugin in project cdap by caskdata.

the class PipelineSpecGenerator method configureStage.

/**
 * Configures a plugin and returns the spec for it.
 *
 * @param stageName the unique plugin id
 * @param etlPlugin user provided configuration for the plugin
 * @param pipelineConfigurer default pipeline configurer to configure the plugin
 * @return the spec for the plugin
 * @throws IllegalArgumentException if the plugin with same id is already deployed
 * @throws ValidationException if the plugin threw an exception during configuration
 */
public StageSpec.Builder configureStage(String stageName, ETLPlugin etlPlugin, DefaultPipelineConfigurer pipelineConfigurer) throws ValidationException {
    TrackedPluginSelector pluginSelector = new TrackedPluginSelector(new ArtifactSelectorProvider().getPluginSelector(etlPlugin.getArtifactConfig()));
    String type = etlPlugin.getType();
    String pluginName = etlPlugin.getName();
    DefaultStageConfigurer stageConfigurer = pipelineConfigurer.getStageConfigurer();
    FailureCollector collector = stageConfigurer.getFailureCollector();
    Object plugin = getPlugin(stageName, etlPlugin, pluginSelector, type, pluginName, collector);
    try {
        if (type.equals(BatchJoiner.PLUGIN_TYPE)) {
            MultiInputPipelineConfigurable multiPlugin = (MultiInputPipelineConfigurable) plugin;
            multiPlugin.configurePipeline(pipelineConfigurer);
            // to the BatchAutoJoiner while preserving backwards compatibility in the pipeline config.
            if (plugin instanceof AutoJoiner) {
                configureAutoJoiner(stageName, (AutoJoiner) plugin, stageConfigurer, collector);
            }
        } else if (type.equals(SplitterTransform.PLUGIN_TYPE)) {
            MultiOutputPipelineConfigurable multiOutputPlugin = (MultiOutputPipelineConfigurable) plugin;
            multiOutputPlugin.configurePipeline(pipelineConfigurer);
        } else if (!type.equals(Constants.SPARK_PROGRAM_PLUGIN_TYPE)) {
            PipelineConfigurable singlePlugin = (PipelineConfigurable) plugin;
            singlePlugin.configurePipeline(pipelineConfigurer);
            // evaluate macros and find out if there is connection used
            if ((sourcePluginTypes.contains(type) || BatchSink.PLUGIN_TYPE.equals(type)) && runtimeEvaluator == null) {
                pluginConfigurer.evaluateMacros(etlPlugin.getProperties(), connectionEvaluator, options);
            }
        }
    } catch (InvalidConfigPropertyException e) {
        collector.addFailure(e.getMessage(), String.format("Provide valid value for config property '%s'.", e.getProperty())).withConfigProperty(e.getProperty());
    } catch (InvalidStageException e) {
        if (e.getReasons().isEmpty()) {
            collector.addFailure(e.getMessage(), null);
        }
        for (InvalidStageException reason : e.getReasons()) {
            if (reason instanceof InvalidConfigPropertyException) {
                InvalidConfigPropertyException configException = (InvalidConfigPropertyException) reason;
                collector.addFailure(configException.getMessage(), String.format("Provide valid value for config property '%s'.", configException.getProperty())).withConfigProperty(configException.getProperty());
            } else {
                collector.addFailure(reason.getMessage(), null);
            }
        }
    } catch (ValidationException e) {
        throw e;
    } catch (NullPointerException e) {
        // handle the case where plugin throws null pointer exception, this is to avoid having 'null' as error message
        collector.addFailure(String.format("Null error occurred while configuring the stage %s.", stageName), null).withStacktrace(e.getStackTrace());
    } catch (ArrayIndexOutOfBoundsException e) {
        // handle the case where plugin throws index out of bounds exception,
        // this is to avoid having a number like '2', '8' etc as error message
        collector.addFailure(String.format("Index out of bounds error occurred while configuring the stage %s.", stageName), null).withStacktrace(e.getStackTrace());
    } catch (ConnectionBadRequestException e) {
        collector.addFailure(e.getMessage(), "Provide a valid connection name.");
    } catch (Exception e) {
        collector.addFailure(String.format("Error encountered while configuring the stage: '%s'", e.getMessage()), null).withStacktrace(e.getStackTrace());
    }
    // throw validation exception if there are any errors being carried by failure collector
    collector.getOrThrowException();
    PluginSpec pluginSpec = new PluginSpec(type, pluginName, etlPlugin.getProperties(), pluginSelector.getSelectedArtifact());
    StageSpec.Builder specBuilder = StageSpec.builder(stageName, pluginSpec).addInputSchemas(pipelineConfigurer.getStageConfigurer().getInputSchemas()).setErrorSchema(stageConfigurer.getErrorSchema());
    if (type.equals(SplitterTransform.PLUGIN_TYPE)) {
        specBuilder.setPortSchemas(stageConfigurer.getOutputPortSchemas());
    } else {
        specBuilder.setOutputSchema(stageConfigurer.getOutputSchema());
    }
    return specBuilder;
}

Example 18 with ETLPlugin

use of io.cdap.cdap.etl.proto.v2.ETLPlugin in project cdap by caskdata.

the class PipelineTest method testWordCount.

public void testWordCount(String pluginType) throws Exception {
    String inputName = "wcInput-" + pluginType;
    String outputName = "wcOutput-" + pluginType;
    // create the pipeline config
    ETLStage source = new ETLStage("wcInput", MockSource.getPlugin(inputName));
    ETLStage sink = new ETLStage("wcOutput", MockSink.getPlugin(outputName));
    Map<String, String> aggProperties = new HashMap<>();
    aggProperties.put("field", "text");
    ETLStage agg = new ETLStage("middle", new ETLPlugin("WordCount", pluginType, aggProperties, null));
    ETLBatchConfig pipelineConfig = ETLBatchConfig.builder("* * * * *").addStage(source).addStage(sink).addStage(agg).addConnection(source.getName(), agg.getName()).addConnection(agg.getName(), sink.getName()).build();
    // create the pipeline
    ApplicationId pipelineId = NamespaceId.DEFAULT.app("wcTestPipeline-" + pluginType);
    ApplicationManager appManager = deployApplication(pipelineId, new AppRequest<>(APP_ARTIFACT, pipelineConfig));
    // write the input
    Schema inputSchema = Schema.recordOf("text", Schema.Field.of("text", Schema.of(Schema.Type.STRING)));
    DataSetManager<Table> inputManager = getDataset(inputName);
    List<StructuredRecord> inputRecords = new ArrayList<>();
    inputRecords.add(StructuredRecord.builder(inputSchema).set("text", "Hello World").build());
    inputRecords.add(StructuredRecord.builder(inputSchema).set("text", "Hello my name is Hal").build());
    inputRecords.add(StructuredRecord.builder(inputSchema).set("text", "Hello my name is Sam").build());
    MockSource.writeInput(inputManager, inputRecords);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 4, TimeUnit.MINUTES);
    DataSetManager<Table> outputManager = getDataset(outputName);
    Set<StructuredRecord> outputRecords = new HashSet<>();
    outputRecords.addAll(MockSink.readOutput(outputManager));
    Set<StructuredRecord> expected = new HashSet<>();
    expected.add(StructuredRecord.builder(WordCountAggregator.OUTPUT_SCHEMA).set("word", "Hello").set("count", 3L).build());
    expected.add(StructuredRecord.builder(WordCountAggregator.OUTPUT_SCHEMA).set("word", "World").set("count", 1L).build());
    expected.add(StructuredRecord.builder(WordCountAggregator.OUTPUT_SCHEMA).set("word", "my").set("count", 2L).build());
    expected.add(StructuredRecord.builder(WordCountAggregator.OUTPUT_SCHEMA).set("word", "name").set("count", 2L).build());
    expected.add(StructuredRecord.builder(WordCountAggregator.OUTPUT_SCHEMA).set("word", "is").set("count", 2L).build());
    expected.add(StructuredRecord.builder(WordCountAggregator.OUTPUT_SCHEMA).set("word", "Hal").set("count", 1L).build());
    expected.add(StructuredRecord.builder(WordCountAggregator.OUTPUT_SCHEMA).set("word", "Sam").set("count", 1L).build());
    Assert.assertEquals(expected, outputRecords);
}

Example 19 with ETLPlugin

use of io.cdap.cdap.etl.proto.v2.ETLPlugin in project cdap by cdapio.

the class PipelineSpecGenerator method configureStage.

/**
 * Configures a stage and returns the spec for it.
 *
 * @param stage the user provided configuration for the stage
 * @param validatedPipeline the validated pipeline config
 * @param pluginConfigurer configurer used to configure the stage
 * @return the spec for the stage
 * @throws ValidationException if the plugin threw an exception during configuration
 */
protected ConfiguredStage configureStage(ETLStage stage, ValidatedPipeline validatedPipeline, DefaultPipelineConfigurer pluginConfigurer) throws ValidationException {
    String stageName = stage.getName();
    ETLPlugin stagePlugin = stage.getPlugin();
    StageSpec.Builder specBuilder = configureStage(stageName, stagePlugin, pluginConfigurer);
    DefaultStageConfigurer stageConfigurer = pluginConfigurer.getStageConfigurer();
    String pluginType = stage.getPlugin().getType();
    if (pluginType.equals(SplitterTransform.PLUGIN_TYPE)) {
        Map<String, Schema> outputPortSchemas = stageConfigurer.getOutputPortSchemas();
        for (Map.Entry<String, String> outputEntry : validatedPipeline.getOutputPorts(stageName).entrySet()) {
            String outputStage = outputEntry.getKey();
            String outputPort = outputEntry.getValue();
            if (outputPort == null) {
                throw new IllegalArgumentException(String.format("Connection from Splitter '%s' to '%s' must specify a port.", stageName, outputStage));
            }
            specBuilder.addOutput(outputStage, outputPort, outputPortSchemas.get(outputPort));
        }
    } else {
        Schema outputSchema = stageConfigurer.getOutputSchema();
        // all the same
        if (Condition.PLUGIN_TYPE.equals(pluginType)) {
            outputSchema = null;
            for (Schema schema : stageConfigurer.getInputSchemas().values()) {
                if (schema != null) {
                    // todo: fix this cleanly and fully
                    if (outputSchema != null && !Schemas.equalsIgnoringRecordName(outputSchema, schema)) {
                        throw new IllegalArgumentException("Cannot have different input schemas going into stage " + stageName);
                    }
                    outputSchema = schema;
                }
            }
        }
        for (String outputStage : validatedPipeline.getOutputs(stageName)) {
            specBuilder.addOutput(outputStage, null, outputSchema);
        }
    }
    StageSpec stageSpec = specBuilder.setProcessTimingEnabled(validatedPipeline.isProcessTimingEnabled()).setStageLoggingEnabled(validatedPipeline.isStageLoggingEnabled()).setMaxPreviewRecords(validatedPipeline.getMaxPreviewRecords()).build();
    return new ConfiguredStage(stageSpec, pluginConfigurer.getPipelineProperties());
}

Example 20 with ETLPlugin

use of io.cdap.cdap.etl.proto.v2.ETLPlugin in project cdap by cdapio.

the class PipelineTest method testTextFileSinkAndDeletePostAction.

@Test
public void testTextFileSinkAndDeletePostAction() throws Exception {
    // create the pipeline config
    String inputName = "sinkTestInput";
    String outputName = "sinkTestOutput";
    String outputDirName = "users";
    ETLStage source = new ETLStage("source", MockSource.getPlugin(inputName));
    Map<String, String> sinkProperties = new HashMap<>();
    sinkProperties.put(TextFileSetSink.Conf.FILESET_NAME, outputName);
    sinkProperties.put(TextFileSetSink.Conf.FIELD_SEPARATOR, "|");
    sinkProperties.put(TextFileSetSink.Conf.OUTPUT_DIR, "${dir}");
    ETLStage sink = new ETLStage("sink", new ETLPlugin(TextFileSetSink.NAME, BatchSink.PLUGIN_TYPE, sinkProperties, null));
    Map<String, String> actionProperties = new HashMap<>();
    actionProperties.put(FilesetDeletePostAction.Conf.FILESET_NAME, outputName);
    // mapreduce writes multiple files to the output directory. Along with the actual output,
    // there are various .crc files that do not contain any of the output content.
    actionProperties.put(FilesetDeletePostAction.Conf.DELETE_REGEX, ".*\\.crc|_SUCCESS");
    actionProperties.put(FilesetDeletePostAction.Conf.DIRECTORY, outputDirName);
    ETLStage postAction = new ETLStage("cleanup", new ETLPlugin(FilesetDeletePostAction.NAME, PostAction.PLUGIN_TYPE, actionProperties, null));
    ETLBatchConfig pipelineConfig = ETLBatchConfig.builder("* * * * *").addStage(source).addStage(sink).addPostAction(postAction).addConnection(source.getName(), sink.getName()).build();
    // create the pipeline
    ApplicationId pipelineId = NamespaceId.DEFAULT.app("textSinkTestPipeline");
    ApplicationManager appManager = deployApplication(pipelineId, new AppRequest<>(APP_ARTIFACT, pipelineConfig));
    // write some data to the input fileset
    Schema inputSchema = Schema.recordOf("test", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("item", Schema.of(Schema.Type.STRING)));
    Map<String, String> users = new HashMap<>();
    users.put("samuel", "wallet");
    users.put("dwayne", "rock");
    users.put("christopher", "cowbell");
    List<StructuredRecord> inputRecords = new ArrayList<>();
    for (Map.Entry<String, String> userEntry : users.entrySet()) {
        String name = userEntry.getKey();
        String item = userEntry.getValue();
        inputRecords.add(StructuredRecord.builder(inputSchema).set("name", name).set("item", item).build());
    }
    DataSetManager<Table> inputManager = getDataset(inputName);
    MockSource.writeInput(inputManager, inputRecords);
    // run the pipeline
    Map<String, String> runtimeArgs = new HashMap<>();
    // the ${dir} macro will be substituted with "users" for our pipeline run
    runtimeArgs.put("dir", outputDirName);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start(runtimeArgs);
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 4, TimeUnit.MINUTES);
    // check the pipeline output
    DataSetManager<FileSet> outputManager = getDataset(outputName);
    FileSet output = outputManager.get();
    Location outputDir = output.getBaseLocation().append(outputDirName);
    Map<String, String> actual = new HashMap<>();
    for (Location outputFile : outputDir.list()) {
        if (outputFile.getName().endsWith(".crc") || "_SUCCESS".equals(outputFile.getName())) {
            Assert.fail("Post action did not delete file " + outputFile.getName());
        }
        try (BufferedReader reader = new BufferedReader(new InputStreamReader(outputFile.getInputStream()))) {
            String line;
            while ((line = reader.readLine()) != null) {
                String[] parts = line.split("\\|");
                actual.put(parts[0], parts[1]);
            }
        }
    }
    Assert.assertEquals(actual, users);
}