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

Example 61 with ETLBatchConfig

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

the class DataPipelineTest method testSinglePhase.

@Test
public void testSinglePhase() throws Exception {
    Schema schema = Schema.recordOf("testRecord", Schema.Field.of("name", Schema.of(Schema.Type.STRING)));
    /*
     * source --> sink
     */
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().addStage(new ETLStage("source", MockSource.getPlugin("singleInput", schema))).addStage(new ETLStage("sink", MockSink.getPlugin("singleOutput"))).addConnection("source", "sink").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT_RANGE, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("SinglePhaseApp");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    StructuredRecord recordSamuel = StructuredRecord.builder(schema).set("name", "samuel").build();
    StructuredRecord recordBob = StructuredRecord.builder(schema).set("name", "bob").build();
    // write records to source
    DataSetManager<Table> inputManager = getDataset(NamespaceId.DEFAULT.dataset("singleInput"));
    MockSource.writeInput(inputManager, ImmutableList.of(recordSamuel, recordBob));
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    // check sink
    DataSetManager<Table> sinkManager = getDataset("singleOutput");
    Set<StructuredRecord> expected = ImmutableSet.of(recordSamuel, recordBob);
    Set<StructuredRecord> actual = Sets.newHashSet(MockSink.readOutput(sinkManager));
    Assert.assertEquals(expected, actual);
    validateMetric(2, appId, "source.records.out");
    validateMetric(2, appId, "sink.records.in");
}

Example 62 with ETLBatchConfig

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

the class DataPipelineTest method testPostAction.

@Test
public void testPostAction() throws Exception {
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().addStage(new ETLStage("source", MockSource.getPlugin("actionInput"))).addStage(new ETLStage("sink", MockSink.getPlugin("actionOutput"))).addPostAction(new ETLStage("tokenWriter", NodeStatesAction.getPlugin("tokenTable"))).addConnection("source", "sink").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("ActionApp");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    Schema schema = Schema.recordOf("testRecord", Schema.Field.of("name", Schema.of(Schema.Type.STRING)));
    StructuredRecord recordSamuel = StructuredRecord.builder(schema).set("name", "samuel").build();
    StructuredRecord recordBob = StructuredRecord.builder(schema).set("name", "bob").build();
    StructuredRecord recordJane = StructuredRecord.builder(schema).set("name", "jane").build();
    DataSetManager<Table> inputManager = getDataset(NamespaceId.DEFAULT.dataset("actionInput"));
    MockSource.writeInput(inputManager, ImmutableList.of(recordSamuel, recordBob, recordJane));
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    DataSetManager<Table> tokenTableManager = getDataset(NamespaceId.DEFAULT.dataset("tokenTable"));
    Table tokenTable = tokenTableManager.get();
    NodeStatus status = NodeStatus.valueOf(Bytes.toString(tokenTable.get(Bytes.toBytes("phase-1"), Bytes.toBytes("status"))));
    Assert.assertEquals(NodeStatus.COMPLETED, status);
}

Example 63 with ETLBatchConfig

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

the class DataPipelineTest method testMultiConditionControlOnlyDag.

@Test
public void testMultiConditionControlOnlyDag() throws Exception {
    // 
    // 
    // action1---
    // |---condition1---action3
    // action2---      |
    // action4-----condition2-----condition3----action5
    // 
    String appName = "MultiConditionControlOnlyDag";
    String actionTable = "actionTable" + appName;
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().addStage(new ETLStage("condition1", MockCondition.getPlugin("condition1"))).addStage(new ETLStage("condition2", MockCondition.getPlugin("condition2"))).addStage(new ETLStage("condition3", MockCondition.getPlugin("condition3"))).addStage(new ETLStage("action1", MockAction.getPlugin(actionTable, "row1", "key1", "val1"))).addStage(new ETLStage("action2", MockAction.getPlugin(actionTable, "row2", "key2", "val2"))).addStage(new ETLStage("action3", MockAction.getPlugin(actionTable, "row3", "key3", "val3"))).addStage(new ETLStage("action4", MockAction.getPlugin(actionTable, "row4", "key4", "val4"))).addStage(new ETLStage("action5", MockAction.getPlugin(actionTable, "row5", "key5", "val5"))).addConnection("action1", "condition1").addConnection("action2", "condition1").addConnection("condition1", "action3", true).addConnection("condition1", "action4", false).addConnection("action4", "condition2").addConnection("condition2", "condition3", true).addConnection("condition3", "action5", true).build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT_RANGE, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app(appName);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start(ImmutableMap.of("condition2.branch.to.execute", "true", "condition3.branch.to.execute", "true"));
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    DataSetManager<Table> actionTableDS = getDataset(actionTable);
    Assert.assertEquals("val1", MockAction.readOutput(actionTableDS, "row1", "key1"));
    Assert.assertEquals("val2", MockAction.readOutput(actionTableDS, "row2", "key2"));
    Assert.assertEquals("val4", MockAction.readOutput(actionTableDS, "row4", "key4"));
    Assert.assertEquals("val5", MockAction.readOutput(actionTableDS, "row5", "key5"));
    Assert.assertNull(MockAction.readOutput(actionTableDS, "row3", "key3"));
}

Example 64 with ETLBatchConfig

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

the class DataPipelineTest method testSimpleUpgradePipelines.

/* Tests upgrade for a deployed application.
     1. Deploy an application with older application artifact (1.0.0) and older filter plugin version (1.0.0).
     2. Add new versions of application artifacts (0.0.9, 1.1.0, 1.2.0) and filter plugin artifacts (1.0.5, 1.1.0) in
        SYSTEM scope (in test class setup).
     3. Upgrade the older deployed application.
     4. Verify that after upgrading, application artifact and filter plugin artifact is upgraded to use latest version
        in its config.
   */
@Test
public void testSimpleUpgradePipelines() throws Exception {
    ArtifactSelectorConfig currentArtifactSelector = new ArtifactSelectorConfig(ArtifactScope.USER.name(), "test-plugins", "1.0.0");
    Engine engine = Engine.MAPREDUCE;
    String sourceName = "testSource" + engine.name();
    String sinkName = "testSink" + engine.name();
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().setEngine(engine).addStage(new ETLStage("source", MockSource.getPlugin(sourceName))).addStage(new ETLStage("filter", PluggableFilterTransform.getPlugin(ValueFilter.NAME, ValueFilter.getProperties("${field}", "${value}"), currentArtifactSelector))).addStage(new ETLStage("sink", MockSink.getPlugin(sinkName))).addConnection("source", "filter").addConnection("filter", "sink").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("sparkProgramTest");
    // Deploy app with artifact version 1.0.0.
    ApplicationManager appManager = deployApplication(appId, appRequest);
    ApplicationDetail oldAppDetail = getAppDetail(appId);
    ETLBatchConfig oldBatchConfig = GSON.fromJson(oldAppDetail.getConfiguration(), ETLBatchConfig.class);
    Map<String, ETLStage> oldStageMap = oldBatchConfig.getStages().stream().collect(Collectors.toMap(ETLStage::getName, e -> e));
    // Upgrade application.
    appManager.upgrade();
    ApplicationDetail upgradedAppDetail = getAppDetail(appId);
    ETLBatchConfig newBatchConfig = GSON.fromJson(upgradedAppDetail.getConfiguration(), ETLBatchConfig.class);
    Map<String, ETLStage> newStageMap = newBatchConfig.getStages().stream().collect(Collectors.toMap(ETLStage::getName, e -> e));
    // Compare stages that should be same after upgrade.
    Assert.assertEquals(oldStageMap.get("source"), newStageMap.get("source"));
    Assert.assertEquals(oldStageMap.get("sink"), newStageMap.get("sink"));
    // Verify that after upgrade, application upgrades artifact version to latest version available.
    Assert.assertEquals(UPGRADE_APP_ARTIFACT_ID_2.getVersion(), upgradedAppDetail.getArtifact().getVersion());
    // Check if the filter stage, for which version should be upgraded to desired version in SYSTEM scope.
    ETLPlugin upgradedPlugin = newStageMap.get("filter").getPlugin();
    Assert.assertEquals(upgradedPlugin.getArtifactConfig().getVersion(), "1.1.0");
    Assert.assertEquals(ArtifactScope.valueOf(upgradedPlugin.getArtifactConfig().getScope().toUpperCase()), ArtifactScope.SYSTEM);
}

Example 65 with ETLBatchConfig

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

the class DataPipelineTest method testSinglePhaseWithSparkSink.

private void testSinglePhaseWithSparkSink() throws Exception {
    /*
     * source1 ---|
     *            |--> sparksink
     * source2 ---|
     */
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().addStage(new ETLStage("source1", MockSource.getPlugin("messages1", SpamMessage.SCHEMA))).addStage(new ETLStage("source2", MockSource.getPlugin("messages2", SpamMessage.SCHEMA))).addStage(new ETLStage("customsink", new ETLPlugin(NaiveBayesTrainer.PLUGIN_NAME, SparkSink.PLUGIN_TYPE, ImmutableMap.of("fileSetName", "modelFileSet", "path", "output", "fieldToClassify", SpamMessage.TEXT_FIELD, "predictionField", SpamMessage.SPAM_PREDICTION_FIELD), null))).addConnection("source1", "customsink").addConnection("source2", "customsink").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("SparkSinkApp");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // set up five spam messages and five non-spam messages to be used for classification
    List<StructuredRecord> messagesToWrite = new ArrayList<>();
    messagesToWrite.add(new SpamMessage("buy our clothes", 1.0).toStructuredRecord());
    messagesToWrite.add(new SpamMessage("sell your used books to us", 1.0).toStructuredRecord());
    messagesToWrite.add(new SpamMessage("earn money for free", 1.0).toStructuredRecord());
    messagesToWrite.add(new SpamMessage("this is definitely not spam", 1.0).toStructuredRecord());
    messagesToWrite.add(new SpamMessage("you won the lottery", 1.0).toStructuredRecord());
    // write records to source1
    DataSetManager<Table> inputManager = getDataset(NamespaceId.DEFAULT.dataset("messages1"));
    MockSource.writeInput(inputManager, messagesToWrite);
    messagesToWrite.clear();
    messagesToWrite.add(new SpamMessage("how was your day", 0.0).toStructuredRecord());
    messagesToWrite.add(new SpamMessage("what are you up to", 0.0).toStructuredRecord());
    messagesToWrite.add(new SpamMessage("this is a genuine message", 0.0).toStructuredRecord());
    messagesToWrite.add(new SpamMessage("this is an even more genuine message", 0.0).toStructuredRecord());
    messagesToWrite.add(new SpamMessage("could you send me the report", 0.0).toStructuredRecord());
    // write records to source2
    inputManager = getDataset(NamespaceId.DEFAULT.dataset("messages2"));
    MockSource.writeInput(inputManager, messagesToWrite);
    // ingest in some messages to be classified
    DataSetManager<FileSet> fileSetManager = getDataset(NaiveBayesTrainer.TEXTS_TO_CLASSIFY);
    FileSet fileSet = fileSetManager.get();
    try (PrintStream out = new PrintStream(fileSet.getLocation("inputTexts").getOutputStream(), true, "UTF-8")) {
        out.println("how are you doing today");
        out.println("free money money");
        out.println("what are you doing today");
        out.println("genuine report");
    }
    // manually trigger the pipeline
    Map<String, String> runtimeArgs = new HashMap<>();
    FileSetArguments.setInputPath(runtimeArgs, "inputTexts");
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start(runtimeArgs);
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    DataSetManager<KeyValueTable> classifiedTexts = getDataset(NaiveBayesTrainer.CLASSIFIED_TEXTS);
    Assert.assertEquals(0.0d, Bytes.toDouble(classifiedTexts.get().read("how are you doing today")), 0.01d);
    // only 'free money money' should be predicated as spam
    Assert.assertEquals(1.0d, Bytes.toDouble(classifiedTexts.get().read("free money money")), 0.01d);
    Assert.assertEquals(0.0d, Bytes.toDouble(classifiedTexts.get().read("what are you doing today")), 0.01d);
    Assert.assertEquals(0.0d, Bytes.toDouble(classifiedTexts.get().read("genuine report")), 0.01d);
    validateMetric(5, appId, "source1.records.out");
    validateMetric(5, appId, "source2.records.out");
    validateMetric(10, appId, "customsink.records.in");
}