Examples with StructuredRecord co.cask.cdap.api.data.format.StructuredRecord used on opensource projects

Example 1 with StructuredRecord

use of co.cask.cdap.api.data.format.StructuredRecord in project cdap by caskdata.

the class DataPipelineTest method testTableLookup.

@Test
public void testTableLookup() throws Exception {
    addDatasetInstance(Table.class.getName(), "personTable");
    DataSetManager<Table> lookupTableManager = getDataset("personTable");
    Table lookupTable = lookupTableManager.get();
    lookupTable.put("samuel".getBytes(Charsets.UTF_8), "age".getBytes(Charsets.UTF_8), "12".getBytes(Charsets.UTF_8));
    lookupTable.put("samuel".getBytes(Charsets.UTF_8), "gender".getBytes(Charsets.UTF_8), "m".getBytes(Charsets.UTF_8));
    lookupTable.put("bob".getBytes(Charsets.UTF_8), "age".getBytes(Charsets.UTF_8), "36".getBytes(Charsets.UTF_8));
    lookupTable.put("bob".getBytes(Charsets.UTF_8), "gender".getBytes(Charsets.UTF_8), "m".getBytes(Charsets.UTF_8));
    lookupTable.put("jane".getBytes(Charsets.UTF_8), "age".getBytes(Charsets.UTF_8), "25".getBytes(Charsets.UTF_8));
    lookupTable.put("jane".getBytes(Charsets.UTF_8), "gender".getBytes(Charsets.UTF_8), "f".getBytes(Charsets.UTF_8));
    lookupTableManager.flush();
    ETLBatchConfig etlConfig = ETLBatchConfig.builder("* * * * *").addStage(new ETLStage("source", MockSource.getPlugin("inputTable"))).addStage(new ETLStage("transform", LookupTransform.getPlugin("person", "age", "personTable"))).addStage(new ETLStage("sink", MockSink.getPlugin("outputTable"))).addConnection("source", "transform").addConnection("transform", "sink").build();
    ApplicationId appId = NamespaceId.DEFAULT.app("testTableLookup");
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // set up input data
    Schema inputSchema = Schema.recordOf("person", Schema.Field.of("person", Schema.of(Schema.Type.STRING)));
    StructuredRecord recordSamuel = StructuredRecord.builder(inputSchema).set("person", "samuel").build();
    StructuredRecord recordBob = StructuredRecord.builder(inputSchema).set("person", "bob").build();
    StructuredRecord recordJane = StructuredRecord.builder(inputSchema).set("person", "jane").build();
    DataSetManager<Table> inputTable = getDataset("inputTable");
    MockSource.writeInput(inputTable, ImmutableList.of(recordSamuel, recordBob, recordJane));
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME).start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    Schema schema = Schema.recordOf("person", Schema.Field.of("person", Schema.of(Schema.Type.STRING)), Schema.Field.of("age", Schema.of(Schema.Type.STRING)), Schema.Field.of("gender", Schema.of(Schema.Type.STRING)));
    Set<StructuredRecord> expected = new HashSet<>();
    expected.add(StructuredRecord.builder(schema).set("person", "samuel").set("age", "12").set("gender", "m").build());
    expected.add(StructuredRecord.builder(schema).set("person", "bob").set("age", "36").set("gender", "m").build());
    expected.add(StructuredRecord.builder(schema).set("person", "jane").set("age", "25").set("gender", "f").build());
    DataSetManager<Table> outputTable = getDataset("outputTable");
    Set<StructuredRecord> actual = new HashSet<>(MockSink.readOutput(outputTable));
    Assert.assertEquals(expected, actual);
    validateMetric(3, appId, "source.records.out");
    validateMetric(3, appId, "sink.records.in");
    deleteDatasetInstance(NamespaceId.DEFAULT.dataset("inputTable"));
    deleteDatasetInstance(NamespaceId.DEFAULT.dataset("outputTable"));
}

Example 2 with StructuredRecord

use of co.cask.cdap.api.data.format.StructuredRecord 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;
    String sink1TableName = "errTestOut1-" + engine;
    String sink2TableName = "errTestOut2-" + engine;
    Schema inputSchema = Schema.recordOf("user", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("id", Schema.of(Schema.Type.INT)));
    /*
     *
     * 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", MockSink.getPlugin(sink1TableName))).addStage(new ETLStage("sink2", MockSink.getPlugin(sink2TableName))).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.toId(), 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);
    workflowManager.start();
    workflowManager.waitForRun(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());
    DataSetManager<Table> sink1Table = getDataset(sink1TableName);
    Assert.assertEquals(expected, ImmutableSet.copyOf(MockSink.readOutput(sink1Table)));
    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());
    DataSetManager<Table> sink2Table = getDataset(sink2TableName);
    Assert.assertEquals(expected, ImmutableSet.copyOf(MockSink.readOutput(sink2Table)));
    /*
     *
     * 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(1, appId, "dropnull.records.error");
    validateMetric(3, appId, "filter1.records.out");
    validateMetric(1, appId, "filter1.records.error");
    validateMetric(2, appId, "filter2.records.out");
    validateMetric(1, appId, "filter2.records.error");
    validateMetric(1, appId, "agg1.records.out");
    validateMetric(1, appId, "agg1.records.error");
    validateMetric(1, appId, "agg2.records.error");
    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 3 with StructuredRecord

use of co.cask.cdap.api.data.format.StructuredRecord in project cdap by caskdata.

the class DataPipelineTest method testMultiSource.

private void testMultiSource(Engine engine) throws Exception {
    /*
     * source1 --|                 |--> sink1
     *           |--> transform1 --|
     * source2 --|                 |
     *                             |--> transform2 --> sink2
     *                                     ^
     *                                     |
     * source3 ----------------------------|
     */
    Schema schema = Schema.recordOf("testRecord", Schema.Field.of("name", Schema.of(Schema.Type.STRING)));
    String source1Name = String.format("msInput1-%s", engine);
    String source2Name = String.format("msInput2-%s", engine);
    String source3Name = String.format("msInput3-%s", engine);
    String sink1Name = String.format("msOutput1-%s", engine);
    String sink2Name = String.format("msOutput2-%s", engine);
    ETLBatchConfig etlConfig = ETLBatchConfig.builder("* * * * *").addStage(new ETLStage("source1", MockSource.getPlugin(source1Name, schema))).addStage(new ETLStage("source2", MockSource.getPlugin(source2Name, schema))).addStage(new ETLStage("source3", MockSource.getPlugin(source3Name, schema))).addStage(new ETLStage("transform1", IdentityTransform.getPlugin())).addStage(new ETLStage("transform2", IdentityTransform.getPlugin())).addStage(new ETLStage("sink1", MockSink.getPlugin(sink1Name))).addStage(new ETLStage("sink2", MockSink.getPlugin(sink2Name))).addConnection("source1", "transform1").addConnection("source2", "transform1").addConnection("transform1", "sink1").addConnection("transform1", "transform2").addConnection("transform2", "sink2").addConnection("source3", "transform2").setEngine(engine).build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("MultiSourceApp-" + engine);
    ApplicationManager appManager = deployApplication(appId.toId(), appRequest);
    // there should be only two programs - one workflow and one mapreduce/spark
    Assert.assertEquals(2, appManager.getInfo().getPrograms().size());
    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();
    // write one record to each source
    DataSetManager<Table> inputManager = getDataset(NamespaceId.DEFAULT.dataset(source1Name));
    MockSource.writeInput(inputManager, ImmutableList.of(recordSamuel));
    inputManager = getDataset(NamespaceId.DEFAULT.dataset(source2Name));
    MockSource.writeInput(inputManager, ImmutableList.of(recordBob));
    inputManager = getDataset(NamespaceId.DEFAULT.dataset(source3Name));
    MockSource.writeInput(inputManager, ImmutableList.of(recordJane));
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    // sink1 should get records from source1 and source2
    DataSetManager<Table> sinkManager = getDataset(sink1Name);
    Set<StructuredRecord> expected = ImmutableSet.of(recordSamuel, recordBob);
    Set<StructuredRecord> actual = Sets.newHashSet(MockSink.readOutput(sinkManager));
    Assert.assertEquals(expected, actual);
    // sink2 should get all records
    sinkManager = getDataset(sink2Name);
    expected = ImmutableSet.of(recordSamuel, recordBob, recordJane);
    actual = Sets.newHashSet(MockSink.readOutput(sinkManager));
    Assert.assertEquals(expected, actual);
    validateMetric(1, appId, "source1.records.out");
    validateMetric(1, appId, "source2.records.out");
    validateMetric(1, appId, "source3.records.out");
    validateMetric(2, appId, "transform1.records.in");
    validateMetric(2, appId, "transform1.records.out");
    validateMetric(3, appId, "transform2.records.in");
    validateMetric(3, appId, "transform2.records.out");
    validateMetric(2, appId, "sink1.records.in");
    validateMetric(3, appId, "sink2.records.in");
}

Example 4 with StructuredRecord

use of co.cask.cdap.api.data.format.StructuredRecord 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.toId(), 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 5 with StructuredRecord

use of co.cask.cdap.api.data.format.StructuredRecord in project cdap by caskdata.

the class DataPipelineTest method testExternalDatasetTracking.

private void testExternalDatasetTracking(Engine engine, boolean backwardsCompatible) throws Exception {
    String suffix = engine.name() + (backwardsCompatible ? "-bc" : "");
    // Define input/output datasets
    String expectedExternalDatasetInput = "fileInput-" + suffix;
    String expectedExternalDatasetOutput = "fileOutput-" + suffix;
    // Define input/output directories
    File inputDir = TMP_FOLDER.newFolder("input-" + suffix);
    String inputFile = "input-file1.txt";
    File outputDir = TMP_FOLDER.newFolder("output-" + suffix);
    File outputSubDir1 = new File(outputDir, "subdir1");
    File outputSubDir2 = new File(outputDir, "subdir2");
    if (!backwardsCompatible) {
        // Assert that there are no external datasets
        Assert.assertNull(getDataset(NamespaceId.DEFAULT.dataset(expectedExternalDatasetInput)).get());
        Assert.assertNull(getDataset(NamespaceId.DEFAULT.dataset(expectedExternalDatasetOutput)).get());
    }
    ETLBatchConfig.Builder builder = ETLBatchConfig.builder("* * * * *");
    ETLBatchConfig etlConfig = builder.setEngine(engine).addStage(new ETLStage("source", MockExternalSource.getPlugin(expectedExternalDatasetInput, inputDir.getAbsolutePath()))).addStage(new ETLStage("sink1", MockExternalSink.getPlugin(backwardsCompatible ? null : expectedExternalDatasetOutput, "dir1", outputSubDir1.getAbsolutePath()))).addStage(new ETLStage("sink2", MockExternalSink.getPlugin(backwardsCompatible ? null : expectedExternalDatasetOutput, "dir2", outputSubDir2.getAbsolutePath()))).addConnection("source", "sink1").addConnection("source", "sink2").build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, etlConfig);
    ApplicationId appId = NamespaceId.DEFAULT.app("ExternalDatasetApp-" + suffix);
    ApplicationManager appManager = deployApplication(appId.toId(), 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();
    ImmutableList<StructuredRecord> allInput = ImmutableList.of(recordSamuel, recordBob, recordJane);
    // Create input files
    MockExternalSource.writeInput(new File(inputDir, inputFile).getAbsolutePath(), allInput);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.start();
    workflowManager.waitForRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    List<RunRecord> history = workflowManager.getHistory();
    // there should be only one completed run
    Assert.assertEquals(1, history.size());
    Assert.assertEquals(ProgramRunStatus.COMPLETED, history.get(0).getStatus());
    // Assert output
    Assert.assertEquals(allInput, MockExternalSink.readOutput(outputSubDir1.getAbsolutePath()));
    Assert.assertEquals(allInput, MockExternalSink.readOutput(outputSubDir2.getAbsolutePath()));
    if (!backwardsCompatible) {
        // Assert that external datasets got created
        Assert.assertNotNull(getDataset(NamespaceId.DEFAULT.dataset(expectedExternalDatasetInput)).get());
        Assert.assertNotNull(getDataset(NamespaceId.DEFAULT.dataset(expectedExternalDatasetOutput)).get());
    }
}