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

Example 66 with ETLStage

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

the class DataPipelineTest method testSplitterToJoiner.

private void testSplitterToJoiner(Engine engine) throws Exception {
    Schema schema = Schema.recordOf("user", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("name", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    Schema infoSchema = Schema.recordOf("userInfo", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("fname", Schema.of(Schema.Type.STRING)));
    Schema joinedSchema = Schema.recordOf("join.output", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("name", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("fname", Schema.of(Schema.Type.STRING)));
    StructuredRecord user0 = StructuredRecord.builder(schema).set("id", 0L).build();
    StructuredRecord user1 = StructuredRecord.builder(schema).set("id", 1L).set("name", "one").build();
    StructuredRecord user0Info = StructuredRecord.builder(infoSchema).set("id", 0L).set("fname", "zero").build();
    StructuredRecord user0Joined = StructuredRecord.builder(joinedSchema).set("id", 0L).set("fname", "zero").build();
    String signupsName = "splitjoinSignups" + engine.name();
    String userInfoName = "splitjoinUserInfo" + engine.name();
    String sink1Name = "splitjoinSink1" + engine.name();
    String sink2Name = "splitjoinSink2" + engine.name();
    /*
     * userInfo --------------------------|
     *                                    |--> joiner --> sink1
     *                            |null --|
     * signups --> namesplitter --|
     *                            |non-null --> sink2
     */
    ETLBatchConfig config = ETLBatchConfig.builder().setEngine(engine).addStage(new ETLStage("signups", MockSource.getPlugin(signupsName, schema))).addStage(new ETLStage("userInfo", MockSource.getPlugin(userInfoName, infoSchema))).addStage(new ETLStage("namesplitter", NullFieldSplitterTransform.getPlugin("name"))).addStage(new ETLStage("joiner", MockJoiner.getPlugin("namesplitter.id=userInfo.id", "namesplitter,userInfo", ""))).addStage(new ETLStage("sink1", MockSink.getPlugin(sink1Name))).addStage(new ETLStage("sink2", MockSink.getPlugin(sink2Name))).addConnection("signups", "namesplitter").addConnection("namesplitter", "sink2", "non-null").addConnection("namesplitter", "joiner", "null").addConnection("userInfo", "joiner").addConnection("joiner", "sink1").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app("SplitJoinTest-" + engine);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // write signups data
    DataSetManager<Table> signupsManager = getDataset(signupsName);
    MockSource.writeInput(signupsManager, ImmutableList.of(user0, user1));
    // write to userInfo the name for user0 to join against
    DataSetManager<Table> userInfoManager = getDataset(userInfoName);
    MockSource.writeInput(userInfoManager, ImmutableList.of(user0Info));
    // run pipeline
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    // check output
    DataSetManager<Table> sinkManager = getDataset(sink2Name);
    Set<StructuredRecord> expected = ImmutableSet.of(user1);
    Set<StructuredRecord> actual = Sets.newHashSet(MockSink.readOutput(sinkManager));
    Assert.assertEquals(expected, actual);
    sinkManager = getDataset(sink1Name);
    expected = ImmutableSet.of(user0Joined);
    actual = Sets.newHashSet(MockSink.readOutput(sinkManager));
    Assert.assertEquals(expected, actual);
    validateMetric(2, appId, "signups.records.out");
    validateMetric(1, appId, "userInfo.records.out");
    validateMetric(2, appId, "namesplitter.records.in");
    validateMetric(1, appId, "namesplitter.records.out.null");
    validateMetric(1, appId, "namesplitter.records.out.non-null");
    validateMetric(2, appId, "joiner.records.in");
    validateMetric(1, appId, "joiner.records.out");
    validateMetric(1, appId, "sink1.records.in");
    validateMetric(1, appId, "sink2.records.in");
}

Example 67 with ETLStage

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

the class DataPipelineTest method testMacrosSparkPipeline.

/**
 * Tests that if macros are provided
 */
@Test
public void testMacrosSparkPipeline() throws Exception {
    /*
     * Trivial Spark pipeline from batch source to batch sink.
     *
     * source --------- sink
     */
    ETLBatchConfig etlConfig = ETLBatchConfig.builder().setEngine(Engine.SPARK).addStage(new ETLStage("source", MockRuntimeDatasetSource.getPlugin("sparkinput", "${runtime${source}}"))).addStage(new ETLStage("sink", MockRuntimeDatasetSink.getPlugin("sparkoutput", "${runtime}${sink}"))).addConnection("source", "sink").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("SparkApp");
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // set runtime arguments for macro substitution
    Map<String, String> runtimeArguments = ImmutableMap.of("runtime", "mockRuntime", "sink", "SparkSinkDataset", "source", "Source", "runtimeSource", "mockRuntimeSparkSourceDataset");
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.setRuntimeArgs(runtimeArguments);
    // make sure the datasets don't exist beforehand
    Assert.assertNull(getDataset("mockRuntimeSparkSourceDataset").get());
    Assert.assertNull(getDataset("mockRuntimeSparkSinkDataset").get());
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    // now the datasets should exist
    Assert.assertNotNull(getDataset("mockRuntimeSparkSourceDataset").get());
    Assert.assertNotNull(getDataset("mockRuntimeSparkSinkDataset").get());
}

Example 68 with ETLStage

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

the class DataPipelineTest method testSplitterToConnector.

private void testSplitterToConnector(Engine engine) throws Exception {
    Schema schema = Schema.recordOf("user", Schema.Field.of("id", Schema.of(Schema.Type.LONG)), Schema.Field.of("name", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("email", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    StructuredRecord user0 = StructuredRecord.builder(schema).set("id", 0L).build();
    StructuredRecord user1 = StructuredRecord.builder(schema).set("id", 1L).set("email", "one@example.com").build();
    StructuredRecord user2 = StructuredRecord.builder(schema).set("id", 2L).set("name", "two").build();
    StructuredRecord user3 = StructuredRecord.builder(schema).set("id", 3L).set("name", "three").set("email", "three@example.com").build();
    String sourceName = "splitconSource" + engine.name();
    String sink1Name = "splitconSink1" + engine.name();
    String sink2Name = "splitconSink2" + engine.name();
    /*
     *
     *                                                             |null --> sink1
     *                       |null--> identity-agg --> splitter2 --|
     * source --> splitter1--|                                     |non-null --|
     *                       |                                                 |--> sink2
     *                       |non-null-----------------------------------------|
     */
    ETLBatchConfig config = ETLBatchConfig.builder().setEngine(engine).addStage(new ETLStage("source", MockSource.getPlugin(sourceName))).addStage(new ETLStage("splitter1", NullFieldSplitterTransform.getPlugin("name"))).addStage(new ETLStage("splitter2", NullFieldSplitterTransform.getPlugin("email"))).addStage(new ETLStage("identity", IdentityAggregator.getPlugin())).addStage(new ETLStage("sink1", MockSink.getPlugin(sink1Name))).addStage(new ETLStage("sink2", MockSink.getPlugin(sink2Name))).addConnection("source", "splitter1").addConnection("splitter1", "identity", "null").addConnection("splitter1", "sink2", "non-null").addConnection("identity", "splitter2").addConnection("splitter2", "sink1", "null").addConnection("splitter2", "sink2", "non-null").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app("SplitConTest-" + engine);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // write input data
    DataSetManager<Table> inputManager = getDataset(sourceName);
    MockSource.writeInput(inputManager, ImmutableList.of(user0, user1, user2, user3));
    // run pipeline
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    // check output
    // sink1 should only have records where both name and email are null (user0)
    DataSetManager<Table> sinkManager = getDataset(sink1Name);
    Set<StructuredRecord> expected = ImmutableSet.of(user0);
    Set<StructuredRecord> actual = Sets.newHashSet(MockSink.readOutput(sinkManager));
    Assert.assertEquals(expected, actual);
    // sink2 should have anything with a non-null name or non-null email
    sinkManager = getDataset(sink2Name);
    expected = ImmutableSet.of(user1, user2, user3);
    actual = Sets.newHashSet(MockSink.readOutput(sinkManager));
    Assert.assertEquals(expected, actual);
    validateMetric(4, appId, "source.records.out");
    validateMetric(1, appId, "sink1.records.in");
    validateMetric(3, appId, "sink2.records.in");
}

Example 69 with ETLStage

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

the class DataPipelineTest method testErrorTransform.

private void testErrorTransform(Engine engine) throws Exception {
    String source1TableName = "errTestIn1-" + engine;
    String source2TableName = "errTestIn2-" + engine;
    Schema inputSchema = Schema.recordOf("user", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("id", Schema.of(Schema.Type.INT)));
    File baseOutput = TMP_FOLDER.newFolder();
    String outputDir1 = new File(baseOutput, "output1").getAbsolutePath();
    String outputDir2 = new File(baseOutput, "output2").getAbsolutePath();
    /*
     *
     * source1 --> filter1 --> filter2 --> agg1 --> agg2
     *                |           |         |        |
     *                |-----------|---------|--------|--------|--> errorflatten --> sink1
     *                |                                       |
     *                |                                       |--> errorfilter --> sink2
     *                |
     * source2 --> dropnull
     *
     * arrows coming out the right represent output records
     * arrows coming out the bottom represent error records
     * this will test multiple stages from multiple phases emitting errors to the same stage
     * as well as errors from one stage going to multiple stages
     * and transforms that have an error schema different from their output schema
     */
    ETLBatchConfig config = ETLBatchConfig.builder().setEngine(engine).addStage(new ETLStage("source1", MockSource.getPlugin(source1TableName, inputSchema))).addStage(new ETLStage("source2", MockSource.getPlugin(source2TableName, inputSchema))).addStage(new ETLStage("filter1", StringValueFilterTransform.getPlugin("name", "Leo"))).addStage(new ETLStage("filter2", StringValueFilterTransform.getPlugin("name", "Ralph"))).addStage(new ETLStage("agg1", GroupFilterAggregator.getPlugin("name", "Don"))).addStage(new ETLStage("agg2", GroupFilterAggregator.getPlugin("name", "Mike"))).addStage(new ETLStage("errorflatten", FlattenErrorTransform.getPlugin())).addStage(new ETLStage("errorfilter", FilterErrorTransform.getPlugin(3))).addStage(new ETLStage("dropnull", DropNullTransform.getPlugin("name"))).addStage(new ETLStage("sink1", MockExternalSink.getPlugin("sink1", "sink1", outputDir1))).addStage(new ETLStage("sink2", MockExternalSink.getPlugin("sink2", "sink2", outputDir2))).addConnection("source1", "filter1").addConnection("source2", "dropnull").addConnection("filter1", "filter2").addConnection("filter2", "agg1").addConnection("agg1", "agg2").addConnection("filter1", "errorflatten").addConnection("filter1", "errorfilter").addConnection("filter2", "errorflatten").addConnection("filter2", "errorfilter").addConnection("agg1", "errorflatten").addConnection("agg1", "errorfilter").addConnection("agg2", "errorflatten").addConnection("agg2", "errorfilter").addConnection("dropnull", "errorflatten").addConnection("dropnull", "errorfilter").addConnection("errorflatten", "sink1").addConnection("errorfilter", "sink2").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app("ErrTransformTest-" + engine);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    List<StructuredRecord> input = ImmutableList.of(StructuredRecord.builder(inputSchema).set("name", "Leo").set("id", 1).build(), StructuredRecord.builder(inputSchema).set("name", "Ralph").set("id", 2).build(), StructuredRecord.builder(inputSchema).set("name", "Don").set("id", 3).build(), StructuredRecord.builder(inputSchema).set("name", "Mike").set("id", 4).build());
    DataSetManager<Table> source1Table = getDataset(source1TableName);
    MockSource.writeInput(source1Table, input);
    input = ImmutableList.of(StructuredRecord.builder(inputSchema).set("name", "April").set("id", 5).build());
    DataSetManager<Table> source2Table = getDataset(source2TableName);
    MockSource.writeInput(source2Table, input);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    Map<String, String> args = Collections.singletonMap(io.cdap.cdap.etl.common.Constants.CONSOLIDATE_STAGES, "true");
    workflowManager.startAndWaitForGoodRun(args, ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    Schema flattenSchema = Schema.recordOf("erroruser", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("id", Schema.of(Schema.Type.INT)), Schema.Field.of("errMsg", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("errCode", Schema.nullableOf(Schema.of(Schema.Type.INT))), Schema.Field.of("errStage", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    Set<StructuredRecord> expected = ImmutableSet.of(StructuredRecord.builder(flattenSchema).set("name", "Leo").set("id", 1).set("errMsg", "bad string value").set("errCode", 1).set("errStage", "filter1").build(), StructuredRecord.builder(flattenSchema).set("name", "Ralph").set("id", 2).set("errMsg", "bad string value").set("errCode", 1).set("errStage", "filter2").build(), StructuredRecord.builder(flattenSchema).set("name", "Don").set("id", 3).set("errMsg", "bad val").set("errCode", 3).set("errStage", "agg1").build(), StructuredRecord.builder(flattenSchema).set("name", "Mike").set("id", 4).set("errMsg", "bad val").set("errCode", 3).set("errStage", "agg2").build(), StructuredRecord.builder(flattenSchema).set("name", "April").set("id", 5).set("errMsg", "Field name was not null").set("errCode", 5).set("errStage", "dropnull").build());
    Set<StructuredRecord> actual = new HashSet<>(MockExternalSink.readOutput(outputDir1, flattenSchema));
    Assert.assertEquals(expected, actual);
    expected = ImmutableSet.of(StructuredRecord.builder(inputSchema).set("name", "Leo").set("id", 1).build(), StructuredRecord.builder(inputSchema).set("name", "Ralph").set("id", 2).build(), StructuredRecord.builder(inputSchema).set("name", "April").set("id", 5).build());
    actual = new HashSet<>(MockExternalSink.readOutput(outputDir2, inputSchema));
    Assert.assertEquals(expected, actual);
    /*
     *
     * source1 (4) --> filter1 (3) --> filter2 (2) --> agg1 (1) --> agg2
     *                   |                |              |            |
     *                  (1)              (1)            (1)          (1)
     *                   |----------------|--------------|------------|--------|--> errorflatten (5) --> sink1
     *                   |                                                     |
     *                  (1)                                                    |--> errorfilter (3) --> sink2
     *                   |
     * source2 --> dropnull
     */
    validateMetric(4, appId, "source1.records.out");
    validateMetric(1, appId, "source2.records.out");
    validateMetric(5, appId, "errorflatten.records.out");
    validateMetric(3, appId, "errorfilter.records.out");
    validateMetric(5, appId, "sink1.records.in");
    validateMetric(3, appId, "sink2.records.in");
}

Example 70 with ETLStage

use of io.cdap.cdap.etl.proto.v2.ETLStage 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");
}