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

Example 61 with ETLStage

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

the class DataPipelineTest method testActionFieldLineage.

private void testActionFieldLineage(Engine engine) throws Exception {
    String readDataset = "ActionReadDataset" + engine;
    String writeDataset = "ActionWriteDataset" + engine;
    List<String> srcFields = ImmutableList.of("srcField1", "srcField2", "srcField3");
    Set<String> destFields = ImmutableSet.of("destField1", "destField2", "destField3");
    List<Operation> operations = new ArrayList<>();
    /*
     *          |---------> srcField1 -> destField1----|
     *          |                                      |
     * ActionReadDataset -> srcField2 -> destField2 ---|-> ActionWriteDataset
     *          |                                      |
     *          |---------> srcField3 -> destField3 ---|
     */
    operations.add(new ReadOperation("Read", "1st operation", EndPoint.of("default", readDataset), srcFields));
    operations.add(new TransformOperation("Transform1", "2nd operation", Collections.singletonList(InputField.of("Read", "srcField1")), "destField1"));
    operations.add(new TransformOperation("Transform2", "3rd operation", Collections.singletonList(InputField.of("Read", "srcField2")), "destField2"));
    operations.add(new TransformOperation("Transform3", "4th operation", Collections.singletonList(InputField.of("Read", "srcField3")), "destField3"));
    operations.add(new WriteOperation("Write", "5th operation", EndPoint.of("default", writeDataset), ImmutableList.of(InputField.of("Transform1", "destField1"), InputField.of("Transform2", "destField2"), InputField.of("Transform3", "destField3"))));
    ETLStage action = new ETLStage("action", FieldLineageAction.getPlugin(readDataset, writeDataset, operations));
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().addStage(action).setEngine(engine).build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("ActionFieldLineage-" + engine);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.startAndWaitForGoodRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    FieldLineageAdmin fieldAdmin = getFieldLineageAdmin();
    // get field lineage for dest dataset
    DatasetFieldLineageSummary summary = fieldAdmin.getDatasetFieldLineage(Constants.FieldLineage.Direction.BOTH, EndPoint.of("default", writeDataset), 0, System.currentTimeMillis());
    Assert.assertEquals(NamespaceId.DEFAULT.dataset(writeDataset), summary.getDatasetId());
    Assert.assertEquals(destFields, summary.getFields());
    Assert.assertTrue(summary.getOutgoing().isEmpty());
    Assert.assertEquals(1, summary.getIncoming().size());
    Set<FieldRelation> fieldRelations = ImmutableSet.of(new FieldRelation("srcField1", "destField1"), new FieldRelation("srcField2", "destField2"), new FieldRelation("srcField3", "destField3"));
    DatasetFieldLineageSummary.FieldLineageRelations expectedRelations = new DatasetFieldLineageSummary.FieldLineageRelations(NamespaceId.DEFAULT.dataset(readDataset), 3, fieldRelations);
    Assert.assertEquals(expectedRelations, summary.getIncoming().iterator().next());
    // get field lineage for src dataset
    summary = fieldAdmin.getDatasetFieldLineage(Constants.FieldLineage.Direction.BOTH, EndPoint.of("default", readDataset), 0, System.currentTimeMillis());
    Assert.assertEquals(NamespaceId.DEFAULT.dataset(readDataset), summary.getDatasetId());
    Assert.assertEquals(new HashSet<>(srcFields), summary.getFields());
    Assert.assertTrue(summary.getIncoming().isEmpty());
    Assert.assertEquals(1, summary.getOutgoing().size());
    expectedRelations = new DatasetFieldLineageSummary.FieldLineageRelations(NamespaceId.DEFAULT.dataset(writeDataset), 3, fieldRelations);
    Assert.assertEquals(expectedRelations, summary.getOutgoing().iterator().next());
    LineageAdmin lineageAdmin = getLineageAdmin();
    ProgramId programId = appId.workflow(SmartWorkflow.NAME);
    RunId runId = RunIds.fromString(workflowManager.getHistory().iterator().next().getPid());
    // get dataset lineage for src dataset
    Tasks.waitFor(2, () -> {
        Lineage lineage = lineageAdmin.computeLineage(NamespaceId.DEFAULT.dataset(readDataset), 0, System.currentTimeMillis(), 1, "workflow");
        return lineage.getRelations().size();
    }, 10, TimeUnit.SECONDS);
    Lineage lineage = lineageAdmin.computeLineage(NamespaceId.DEFAULT.dataset(readDataset), 0, System.currentTimeMillis(), 1, "workflow");
    Set<Relation> expectedLineage = ImmutableSet.of(new Relation(NamespaceId.DEFAULT.dataset(readDataset), programId, AccessType.READ, runId), new Relation(NamespaceId.DEFAULT.dataset(writeDataset), programId, AccessType.WRITE, runId));
    Assert.assertEquals(expectedLineage, lineage.getRelations());
    // get dataset lineage for dest dataset, in this test they should be same
    lineage = lineageAdmin.computeLineage(NamespaceId.DEFAULT.dataset(writeDataset), 0, System.currentTimeMillis(), 1, "workflow");
    Assert.assertEquals(2, lineage.getRelations().size());
    Assert.assertEquals(expectedLineage, lineage.getRelations());
}

Example 62 with ETLStage

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

the class DataPipelineTest method testSimpleUpgradePipelinesWithArtifactScope.

/* Tests upgrade for a deployed application. Also tests artifact scope parameter for only considering artifacts in
       a given scope.
     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. Also deploy a snapshot version of plugin artifact 1.0.8 in USER scope.
     3. Upgrade the older deployed application with artifact scope set to USER for upgrade.
     4. Verify that after upgrading, application artifact and filter plugin artifact is upgraded to use latest version
        in its config and it uses snapshot plugin version with 1.0.8 from USER scope.
   */
@Test
public void testSimpleUpgradePipelinesWithArtifactScope() 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 with artifact scope as USER.
    appManager.upgrade(Collections.singleton(ArtifactScope.USER.toString()), false);
    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("1.0.8", upgradedPlugin.getArtifactConfig().getVersion());
    Assert.assertEquals(ArtifactScope.valueOf(upgradedPlugin.getArtifactConfig().getScope().toUpperCase()), ArtifactScope.USER);
}

Example 63 with ETLStage

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

the class DataPipelineTest method testSecureStorePipeline.

/**
 * Tests the secure storage macro function in a pipelines by creating datasets from the secure store data.
 */
private void testSecureStorePipeline(Engine engine, String prefix) throws Exception {
    /*
     * Trivial pipeline from batch source to batch sink.
     *
     * source --------- sink
     */
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().addStage(new ETLStage("source", MockRuntimeDatasetSource.getPlugin("input", "${secure(" + prefix + "source)}"))).addStage(new ETLStage("sink", MockRuntimeDatasetSink.getPlugin("output", "${secure(" + prefix + "sink)}"))).addConnection("source", "sink").setEngine(engine).build();
    // place dataset names into secure storage
    getSecureStoreManager().put("default", prefix + "source", prefix + "MockSecureSourceDataset", "secure source dataset name", new HashMap<>());
    getSecureStoreManager().put("default", prefix + "sink", prefix + "MockSecureSinkDataset", "secure dataset name", new HashMap<>());
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("App-" + engine);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    // make sure the datasets don't exist beforehand
    Assert.assertNull(getDataset(prefix + "MockSecureSourceDataset").get());
    Assert.assertNull(getDataset(prefix + "MockSecureSinkDataset").get());
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    // now the datasets should exist
    Assert.assertNotNull(getDataset(prefix + "MockSecureSourceDataset").get());
    Assert.assertNotNull(getDataset(prefix + "MockSecureSinkDataset").get());
}

Example 64 with ETLStage

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

the class DataPipelineTest method testAlertPublisher.

private void testAlertPublisher(Engine engine) throws Exception {
    String sourceName = "alertSource" + engine.name();
    String sinkName = "alertSink" + engine.name();
    String topic = "alertTopic" + engine.name();
    /*
     * source --> nullAlert --> sink
     *               |
     *               |--> TMS publisher
     */
    ETLBatchConfig config = ETLBatchConfig.builder().setEngine(engine).addStage(new ETLStage("source", MockSource.getPlugin(sourceName))).addStage(new ETLStage("nullAlert", NullAlertTransform.getPlugin("id"))).addStage(new ETLStage("sink", MockSink.getPlugin(sinkName))).addStage(new ETLStage("tms alert", TMSAlertPublisher.getPlugin(topic, NamespaceId.DEFAULT.getNamespace()))).addConnection("source", "nullAlert").addConnection("nullAlert", "sink").addConnection("nullAlert", "tms alert").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app("AlertTest-" + engine);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    Schema schema = Schema.recordOf("x", Schema.Field.of("id", Schema.nullableOf(Schema.of(Schema.Type.LONG))));
    StructuredRecord record1 = StructuredRecord.builder(schema).set("id", 1L).build();
    StructuredRecord record2 = StructuredRecord.builder(schema).set("id", 2L).build();
    StructuredRecord alertRecord = StructuredRecord.builder(schema).build();
    DataSetManager<Table> sourceTable = getDataset(sourceName);
    MockSource.writeInput(sourceTable, ImmutableList.of(record1, record2, alertRecord));
    WorkflowManager manager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    manager.start();
    manager.waitForRun(ProgramRunStatus.COMPLETED, 3, TimeUnit.MINUTES);
    DataSetManager<Table> sinkTable = getDataset(sinkName);
    Set<StructuredRecord> actual = new HashSet<>(MockSink.readOutput(sinkTable));
    Set<StructuredRecord> expected = ImmutableSet.of(record1, record2);
    Assert.assertEquals(expected, actual);
    MessageFetcher messageFetcher = getMessagingContext().getMessageFetcher();
    Set<Alert> actualMessages = new HashSet<>();
    try (CloseableIterator<Message> iter = messageFetcher.fetch(NamespaceId.DEFAULT.getNamespace(), topic, 5, 0)) {
        while (iter.hasNext()) {
            Message message = iter.next();
            Alert alert = message.decodePayload(r -> GSON.fromJson(r, Alert.class));
            actualMessages.add(alert);
        }
    }
    Set<Alert> expectedMessages = ImmutableSet.of(new Alert("nullAlert", new HashMap<>()));
    Assert.assertEquals(expectedMessages, actualMessages);
    validateMetric(3, appId, "source.records.out");
    validateMetric(3, appId, "nullAlert.records.in");
    validateMetric(2, appId, "nullAlert.records.out");
    validateMetric(1, appId, "nullAlert.records.alert");
    validateMetric(2, appId, "sink.records.in");
    validateMetric(1, appId, "tms alert.records.in");
}

Example 65 with ETLStage

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

the class DataPipelineTest method testSimpleConditionWithActions.

@Test
public void testSimpleConditionWithActions() throws Exception {
    Schema schema = Schema.recordOf("testRecord", Schema.Field.of("name", Schema.of(Schema.Type.STRING)));
    /*
     * action --> condition --> file ---> trueSink
     *              |
     *              |---file-->----> falseSink
     *
     */
    String appName = "SimpleConditionWithActions";
    String trueSource = "true" + appName + "Source";
    String falseSource = "false" + appName + "Source";
    String trueSink = "true" + appName + "Sink";
    String falseSink = "false" + appName + "Sink";
    String actionTable = "actionTable" + appName;
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().addStage(new ETLStage("trueSource", MockSource.getPlugin(trueSource, schema))).addStage(new ETLStage("falseSource", MockSource.getPlugin(falseSource, schema))).addStage(new ETLStage("trueSink", MockSink.getPlugin(trueSink))).addStage(new ETLStage("falseSink", MockSink.getPlugin(falseSink))).addStage(new ETLStage("condition", MockCondition.getPlugin("condition"))).addStage(new ETLStage("action", MockAction.getPlugin(actionTable, "row1", "key1", "val1"))).addConnection("action", "condition").addConnection("condition", "trueSource", true).addConnection("condition", "falseSource", false).addConnection("trueSource", "trueSink").addConnection("falseSource", "falseSink").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT_RANGE, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app(appName);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    StructuredRecord recordSamuel = StructuredRecord.builder(schema).set("name", "samuel").build();
    StructuredRecord recordBob = StructuredRecord.builder(schema).set("name", "bob").build();
    for (String branch : Arrays.asList("true", "false")) {
        String source = branch.equals("true") ? trueSource : falseSource;
        String sink = branch.equals("true") ? trueSink : falseSink;
        // write records to source
        DataSetManager<Table> inputManager = getDataset(NamespaceId.DEFAULT.dataset(source));
        MockSource.writeInput(inputManager, ImmutableList.of(recordSamuel, recordBob));
        WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
        workflowManager.start(ImmutableMap.of("condition.branch.to.execute", branch));
        if (branch.equals("true")) {
            workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
        } else {
            workflowManager.waitForRuns(ProgramRunStatus.COMPLETED, 2, 5, TimeUnit.MINUTES);
        }
        // check sink
        DataSetManager<Table> sinkManager = getDataset(sink);
        Set<StructuredRecord> expected = ImmutableSet.of(recordSamuel, recordBob);
        Set<StructuredRecord> actual = Sets.newHashSet(MockSink.readOutput(sinkManager));
        Assert.assertEquals(expected, actual);
        validateMetric(2, appId, branch + "Source.records.out");
        validateMetric(2, appId, branch + "Sink.records.in");
        // check Action is executed correctly
        DataSetManager<Table> actionTableDS = getDataset(actionTable);
        Assert.assertEquals("val1", MockAction.readOutput(actionTableDS, "row1", "key1"));
    }
}