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

Example 96 with ETLStage

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

the class ValidationUtils method validate.

/**
 * Validate plugin based on the {@link StageValidationRequest}
 *
 * @param validationRequest {@link StageValidationRequest} with plugin properties
 * @param pluginConfigurer  {@link PluginConfigurer} for using the plugin
 * @param macroFn           {@link Function} for evaluating macros
 * @return {@link StageValidationResponse} in json format
 */
public static StageValidationResponse validate(String namespace, StageValidationRequest validationRequest, PluginConfigurer pluginConfigurer, Function<Map<String, String>, Map<String, String>> macroFn, FeatureFlagsProvider featureFlagsProvider) {
    ETLStage stageConfig = validationRequest.getStage();
    ValidatingConfigurer validatingConfigurer = new ValidatingConfigurer(pluginConfigurer, featureFlagsProvider);
    // Batch or Streaming doesn't matter for a single stage.
    PipelineSpecGenerator<ETLBatchConfig, BatchPipelineSpec> pipelineSpecGenerator = new BatchPipelineSpecGenerator(namespace, validatingConfigurer, null, Collections.emptySet(), Collections.emptySet(), Engine.SPARK, featureFlagsProvider);
    DefaultStageConfigurer stageConfigurer = new DefaultStageConfigurer(stageConfig.getName());
    for (StageSchema stageSchema : validationRequest.getInputSchemas()) {
        stageConfigurer.addInputSchema(stageSchema.getStage(), stageSchema.getSchema());
        stageConfigurer.addInputStage(stageSchema.getStage());
    }
    DefaultPipelineConfigurer pipelineConfigurer = new DefaultPipelineConfigurer(validatingConfigurer, stageConfig.getName(), Engine.SPARK, stageConfigurer, featureFlagsProvider);
    // evaluate macros
    Map<String, String> evaluatedProperties = macroFn.apply(stageConfig.getPlugin().getProperties());
    ETLPlugin originalConfig = stageConfig.getPlugin();
    ETLPlugin evaluatedConfig = new ETLPlugin(originalConfig.getName(), originalConfig.getType(), evaluatedProperties, originalConfig.getArtifactConfig());
    try {
        StageSpec spec = pipelineSpecGenerator.configureStage(stageConfig.getName(), evaluatedConfig, pipelineConfigurer).build();
        return new StageValidationResponse(spec);
    } catch (ValidationException e) {
        return new StageValidationResponse(e.getFailures());
    }
}

Example 97 with ETLStage

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

the class AutoJoinerTest method testInnerBetweenCondition.

@Test
public void testInnerBetweenCondition() throws Exception {
    /*
         users ----------|
                         |--> join --> sink
         age_groups -----|

         joinOn: users.age > age_groups.lo and (users.age <= age_groups.hi or age_groups.hi is null)
     */
    Schema userSchema = Schema.recordOf("user", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("age", Schema.nullableOf(Schema.of(Schema.Type.INT))));
    Schema ageGroupSchema = Schema.recordOf("age_group", Schema.Field.of("name", Schema.of(Schema.Type.STRING)), Schema.Field.of("lo", Schema.of(Schema.Type.INT)), Schema.Field.of("hi", Schema.nullableOf(Schema.of(Schema.Type.INT))));
    Schema expectedSchema = Schema.recordOf("users.age_groups", Schema.Field.of("username", Schema.of(Schema.Type.STRING)), Schema.Field.of("age_group", Schema.of(Schema.Type.STRING)));
    String userInput = UUID.randomUUID().toString();
    String agesInput = UUID.randomUUID().toString();
    String output = UUID.randomUUID().toString();
    List<JoinField> select = new ArrayList<>();
    select.add(new JoinField("users", "name", "username"));
    select.add(new JoinField("age_groups", "name", "age_group"));
    JoinCondition.OnExpression condition = JoinCondition.onExpression().setExpression("users.age >= age_groups.lo and (users.age < age_groups.hi or age_groups.hi is null)").build();
    Map<String, String> joinerProperties = MockAutoJoiner.getProperties(Arrays.asList("users", "age_groups"), Collections.emptyList(), Arrays.asList("users", "age_groups"), Collections.emptyList(), select, false, null, condition);
    ETLBatchConfig config = ETLBatchConfig.builder().addStage(new ETLStage("users", MockSource.getPlugin(userInput, userSchema))).addStage(new ETLStage("age_groups", MockSource.getPlugin(agesInput, ageGroupSchema))).addStage(new ETLStage("join", new ETLPlugin(MockAutoJoiner.NAME, BatchJoiner.PLUGIN_TYPE, joinerProperties))).addStage(new ETLStage("sink", MockSink.getPlugin(output))).addConnection("users", "join").addConnection("age_groups", "join").addConnection("join", "sink").setEngine(Engine.SPARK).build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app(UUID.randomUUID().toString());
    ApplicationManager appManager = deployApplication(appId, appRequest);
    List<StructuredRecord> records = new ArrayList<>();
    records.add(StructuredRecord.builder(userSchema).set("name", "Alice").set("age", 35).build());
    records.add(StructuredRecord.builder(userSchema).set("name", "Bob").build());
    records.add(StructuredRecord.builder(userSchema).set("name", "Carl").set("age", 13).build());
    records.add(StructuredRecord.builder(userSchema).set("name", "Dave").set("age", 0).build());
    records.add(StructuredRecord.builder(userSchema).set("name", "Elaine").set("age", 68).build());
    records.add(StructuredRecord.builder(userSchema).set("name", "Fred").set("age", 4).build());
    DataSetManager<Table> inputManager = getDataset(userInput);
    MockSource.writeInput(inputManager, records);
    records.clear();
    records.add(StructuredRecord.builder(ageGroupSchema).set("name", "infant").set("lo", 0).set("hi", 2).build());
    records.add(StructuredRecord.builder(ageGroupSchema).set("name", "toddler").set("lo", 2).set("hi", 5).build());
    records.add(StructuredRecord.builder(ageGroupSchema).set("name", "child").set("lo", 5).set("hi", 13).build());
    records.add(StructuredRecord.builder(ageGroupSchema).set("name", "teen").set("lo", 13).set("hi", 20).build());
    records.add(StructuredRecord.builder(ageGroupSchema).set("name", "adult").set("lo", 20).set("hi", 65).build());
    records.add(StructuredRecord.builder(ageGroupSchema).set("name", "senior").set("lo", 65).build());
    inputManager = getDataset(agesInput);
    MockSource.writeInput(inputManager, records);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.startAndWaitForGoodRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    DataSetManager<Table> outputManager = getDataset(output);
    List<StructuredRecord> outputRecords = MockSink.readOutput(outputManager);
    Set<StructuredRecord> expected = new HashSet<>();
    expected.add(StructuredRecord.builder(expectedSchema).set("username", "Alice").set("age_group", "adult").build());
    expected.add(StructuredRecord.builder(expectedSchema).set("username", "Carl").set("age_group", "teen").build());
    expected.add(StructuredRecord.builder(expectedSchema).set("username", "Dave").set("age_group", "infant").build());
    expected.add(StructuredRecord.builder(expectedSchema).set("username", "Elaine").set("age_group", "senior").build());
    expected.add(StructuredRecord.builder(expectedSchema).set("username", "Fred").set("age_group", "toddler").build());
    Assert.assertEquals(expected, new HashSet<>(outputRecords));
    validateMetric(6, appId, "users.records.out");
    validateMetric(6, appId, "age_groups.records.out");
    validateMetric(12, appId, "join.records.in");
    validateMetric(expected.size(), appId, "join.records.out");
}

Example 98 with ETLStage

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

the class AutoJoinerTest method testSimpleAutoJoin.

private void testSimpleAutoJoin(List<String> required, List<String> broadcast, Set<StructuredRecord> expected, Engine engine) throws Exception {
    /*
         users ------|
                     |--> join --> sink
         purchases --|

         joinOn: users.region = purchases.region and users.user_id = purchases.user_id
     */
    String userInput = UUID.randomUUID().toString();
    String purchaseInput = UUID.randomUUID().toString();
    String output = UUID.randomUUID().toString();
    ETLBatchConfig config = ETLBatchConfig.builder().addStage(new ETLStage("users", MockSource.getPlugin(userInput, USER_SCHEMA))).addStage(new ETLStage("purchases", MockSource.getPlugin(purchaseInput, PURCHASE_SCHEMA))).addStage(new ETLStage("join", MockAutoJoiner.getPlugin(Arrays.asList("purchases", "users"), Arrays.asList("region", "user_id"), required, broadcast, Collections.emptyList(), true))).addStage(new ETLStage("sink", MockSink.getPlugin(output))).addConnection("users", "join").addConnection("purchases", "join").addConnection("join", "sink").setEngine(engine).build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app(UUID.randomUUID().toString());
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // write input data
    List<StructuredRecord> userData = Arrays.asList(USER_ALICE, USER_ALYCE, USER_BOB);
    DataSetManager<Table> inputManager = getDataset(userInput);
    MockSource.writeInput(inputManager, userData);
    List<StructuredRecord> purchaseData = new ArrayList<>();
    purchaseData.add(StructuredRecord.builder(PURCHASE_SCHEMA).set("region", "us").set("user_id", 0).set("purchase_id", 123).build());
    purchaseData.add(StructuredRecord.builder(PURCHASE_SCHEMA).set("region", "us").set("user_id", 2).set("purchase_id", 456).build());
    inputManager = getDataset(purchaseInput);
    MockSource.writeInput(inputManager, purchaseData);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    Map<String, String> args = Collections.singletonMap(MockAutoJoiner.PARTITIONS_ARGUMENT, "1");
    workflowManager.startAndWaitForGoodRun(args, ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    DataSetManager<Table> outputManager = getDataset(output);
    List<StructuredRecord> outputRecords = MockSink.readOutput(outputManager);
    Assert.assertEquals(expected, new HashSet<>(outputRecords));
    validateMetric(5, appId, "join.records.in");
    validateMetric(expected.size(), appId, "join.records.out");
    if (engine != Engine.SPARK) {
        // In SPARK number of partitions hint is ignored, so additional sinks are created
        validateMetric(1, appId, "sink." + MockSink.INITIALIZED_COUNT_METRIC);
    }
}

Example 99 with ETLStage

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

the class AutoJoinerTest method testLeftOuterComplexConditionBroadcast.

@Test
public void testLeftOuterComplexConditionBroadcast() throws Exception {
    /*
         sales ----------|
                         |--> join --> sink
         categories -----|

         joinOn:
           sales.price > 1000 and sales.date > 2020-01-01 and
           (sales.category <=> categories.id or (sales.category is null and sales.department = categories.department))
     */
    Schema salesSchema = Schema.recordOf("sale", Schema.Field.of("id", Schema.of(Schema.Type.INT)), Schema.Field.of("price", Schema.of(Schema.Type.DOUBLE)), Schema.Field.of("date", Schema.of(Schema.LogicalType.DATETIME)), Schema.Field.of("category", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("department", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    Schema categorySchema = Schema.recordOf("category", Schema.Field.of("id", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("department", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("flag", Schema.nullableOf(Schema.of(Schema.Type.BOOLEAN))));
    Schema expectedSchema = Schema.recordOf("sales.categories", Schema.Field.of("id", Schema.of(Schema.Type.INT)), Schema.Field.of("flag", Schema.nullableOf(Schema.of(Schema.Type.BOOLEAN))));
    String salesInput = UUID.randomUUID().toString();
    String categoriesInput = UUID.randomUUID().toString();
    String output = UUID.randomUUID().toString();
    List<JoinField> select = new ArrayList<>();
    select.add(new JoinField("sales", "id"));
    select.add(new JoinField("categories", "flag"));
    /*
           sales.price > 1000 and sales.date > 2020-01-01 and
           (sales.category <=> categories.id or (sales.category is null and sales.department = categories.department))
     */
    JoinCondition.OnExpression condition = JoinCondition.onExpression().addDatasetAlias("sales", "S").addDatasetAlias("categories", "C").setExpression("S.price > 1000 and S.date > '2020-01-01 00:00:00' and " + "(S.category = C.id or (S.category is null and S.department = C.department))").build();
    Map<String, String> joinerProperties = MockAutoJoiner.getProperties(Arrays.asList("sales", "categories"), Collections.emptyList(), Collections.singletonList("sales"), Collections.singletonList("categories"), select, false, null, condition);
    ETLBatchConfig config = ETLBatchConfig.builder().addStage(new ETLStage("sales", MockSource.getPlugin(salesInput, salesSchema))).addStage(new ETLStage("categories", MockSource.getPlugin(categoriesInput, categorySchema))).addStage(new ETLStage("join", new ETLPlugin(MockAutoJoiner.NAME, BatchJoiner.PLUGIN_TYPE, joinerProperties))).addStage(new ETLStage("sink", MockSink.getPlugin(output))).addConnection("sales", "join").addConnection("categories", "join").addConnection("join", "sink").setEngine(Engine.SPARK).build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app(UUID.randomUUID().toString());
    ApplicationManager appManager = deployApplication(appId, appRequest);
    List<StructuredRecord> records = new ArrayList<>();
    records.add(StructuredRecord.builder(salesSchema).set("id", 0).set("price", 123.45d).set("date", "2021-01-01 00:00:00").set("category", "electronics").set("department", "entertainment").build());
    records.add(StructuredRecord.builder(salesSchema).set("id", 1).set("price", 1000.01d).set("date", "2020-01-01 00:00:01").set("department", "home").build());
    records.add(StructuredRecord.builder(salesSchema).set("id", 2).set("price", 5000d).set("date", "2021-01-01 00:00:00").set("category", "furniture").build());
    records.add(StructuredRecord.builder(salesSchema).set("id", 3).set("price", 2000d).set("date", "2019-12-31 23:59:59").set("category", "furniture").build());
    records.add(StructuredRecord.builder(salesSchema).set("id", 4).set("price", 2000d).set("date", "2020-01-01 12:00:00").set("category", "tv").set("department", "entertainment").build());
    DataSetManager<Table> inputManager = getDataset(salesInput);
    MockSource.writeInput(inputManager, records);
    records.clear();
    records.add(StructuredRecord.builder(categorySchema).set("id", "electronics").set("department", "entertainment").set("flag", false).build());
    records.add(StructuredRecord.builder(categorySchema).set("id", "furniture").set("department", "home").set("flag", true).build());
    records.add(StructuredRecord.builder(categorySchema).set("id", "tv").set("department", "entertainment").set("flag", false).build());
    inputManager = getDataset(categoriesInput);
    MockSource.writeInput(inputManager, records);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.startAndWaitForGoodRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    DataSetManager<Table> outputManager = getDataset(output);
    List<StructuredRecord> outputRecords = MockSink.readOutput(outputManager);
    Set<StructuredRecord> expected = new HashSet<>();
    expected.add(StructuredRecord.builder(expectedSchema).set("id", 0).build());
    expected.add(StructuredRecord.builder(expectedSchema).set("id", 1).set("flag", true).build());
    expected.add(StructuredRecord.builder(expectedSchema).set("id", 2).set("flag", true).build());
    expected.add(StructuredRecord.builder(expectedSchema).set("id", 3).build());
    expected.add(StructuredRecord.builder(expectedSchema).set("id", 4).set("flag", false).build());
    Assert.assertEquals(expected, new HashSet<>(outputRecords));
}

Example 100 with ETLStage

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

the class AutoJoinerTest method testNullEquality.

private void testNullEquality(Engine engine, boolean nullIsEqual, Set<StructuredRecord> expected) throws Exception {
    Schema itemSchema = Schema.recordOf("item", Schema.Field.of("id", Schema.nullableOf(Schema.of(Schema.Type.INT))), Schema.Field.of("region", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("name", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    Schema attributeSchema = Schema.recordOf("attribute", Schema.Field.of("region", Schema.nullableOf(Schema.of(Schema.Type.STRING))), Schema.Field.of("id", Schema.nullableOf(Schema.of(Schema.Type.INT))), Schema.Field.of("attr", Schema.nullableOf(Schema.of(Schema.Type.STRING))));
    /*
         items -------|
                      |--> join --> sink
         attributes --|

         joinOn: items.region = attributes.region and items.id = attributes.id
     */
    String itemsInput = UUID.randomUUID().toString();
    String attributesInput = UUID.randomUUID().toString();
    String output = UUID.randomUUID().toString();
    ETLBatchConfig config = ETLBatchConfig.builder().addStage(new ETLStage("items", MockSource.getPlugin(itemsInput, itemSchema))).addStage(new ETLStage("attributes", MockSource.getPlugin(attributesInput, attributeSchema))).addStage(new ETLStage("join", MockAutoJoiner.getPlugin(Arrays.asList("items", "attributes"), Arrays.asList("region", "id"), Collections.singletonList("items"), Collections.emptyList(), Collections.emptyList(), nullIsEqual))).addStage(new ETLStage("sink", MockSink.getPlugin(output))).addConnection("items", "join").addConnection("attributes", "join").addConnection("join", "sink").setEngine(engine).build();
    AppRequest<ETLBatchConfig> appRequest = new AppRequest<>(APP_ARTIFACT, config);
    ApplicationId appId = NamespaceId.DEFAULT.app(UUID.randomUUID().toString());
    ApplicationManager appManager = deployApplication(appId, appRequest);
    // write input data
    List<StructuredRecord> itemData = new ArrayList<>();
    itemData.add(StructuredRecord.builder(itemSchema).set("region", "us").set("id", 0).set("name", "bacon").build());
    itemData.add(StructuredRecord.builder(itemSchema).set("id", 1).build());
    itemData.add(StructuredRecord.builder(itemSchema).set("region", "us").build());
    DataSetManager<Table> inputManager = getDataset(itemsInput);
    MockSource.writeInput(inputManager, itemData);
    List<StructuredRecord> attributesData = new ArrayList<>();
    attributesData.add(StructuredRecord.builder(attributeSchema).set("region", "us").set("id", 0).set("attr", "food").build());
    attributesData.add(StructuredRecord.builder(attributeSchema).set("id", 1).set("attr", "car").build());
    attributesData.add(StructuredRecord.builder(attributeSchema).set("region", "us").build());
    inputManager = getDataset(attributesInput);
    MockSource.writeInput(inputManager, attributesData);
    WorkflowManager workflowManager = appManager.getWorkflowManager(SmartWorkflow.NAME);
    workflowManager.startAndWaitForGoodRun(ProgramRunStatus.COMPLETED, 5, TimeUnit.MINUTES);
    DataSetManager<Table> outputManager = getDataset(output);
    List<StructuredRecord> outputRecords = MockSink.readOutput(outputManager);
    Assert.assertEquals(expected, new HashSet<>(outputRecords));
}