Examples with GenericInputDescriptor org.apache.samza.system.descriptors.GenericInputDescriptor used on opensource projects

Example 11 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class TestCouchbaseRemoteTableEndToEnd method testEndToEnd.

@Test
public void testEndToEnd() {
    Bucket inputBucket = cluster.openBucket(inputBucketName);
    inputBucket.upsert(ByteArrayDocument.create("Alice", "20".getBytes()));
    inputBucket.upsert(ByteArrayDocument.create("Bob", "30".getBytes()));
    inputBucket.upsert(ByteArrayDocument.create("Chris", "40".getBytes()));
    inputBucket.upsert(ByteArrayDocument.create("David", "50".getBytes()));
    inputBucket.close();
    List<String> users = Arrays.asList("Alice", "Bob", "Chris", "David");
    final StreamApplication app = appDesc -> {
        DelegatingSystemDescriptor inputSystemDescriptor = new DelegatingSystemDescriptor("test");
        GenericInputDescriptor<String> inputDescriptor = inputSystemDescriptor.getInputDescriptor("User", new NoOpSerde<>());
        CouchbaseTableReadFunction<String> readFunction = new CouchbaseTableReadFunction<>(inputBucketName, String.class, "couchbase://127.0.0.1").withBootstrapCarrierDirectPort(couchbaseMock.getCarrierPort(inputBucketName)).withBootstrapHttpDirectPort(couchbaseMock.getHttpPort()).withSerde(new StringSerde());
        CouchbaseTableWriteFunction<JsonObject> writeFunction = new CouchbaseTableWriteFunction<>(outputBucketName, JsonObject.class, "couchbase://127.0.0.1").withBootstrapCarrierDirectPort(couchbaseMock.getCarrierPort(outputBucketName)).withBootstrapHttpDirectPort(couchbaseMock.getHttpPort());
        RemoteTableDescriptor inputTableDesc = new RemoteTableDescriptor<String, String, Void>("input-table").withReadFunction(readFunction).withRateLimiterDisabled();
        Table<KV<String, String>> inputTable = appDesc.getTable(inputTableDesc);
        RemoteTableDescriptor outputTableDesc = new RemoteTableDescriptor<String, JsonObject, Object>("output-table").withReadFunction(new NoOpTableReadFunction<>()).withWriteFunction(writeFunction).withRateLimiterDisabled();
        Table<KV<String, JsonObject>> outputTable = appDesc.getTable(outputTableDesc);
        appDesc.getInputStream(inputDescriptor).map(k -> KV.of(k, k)).join(inputTable, new JoinFunction()).sendTo(outputTable);
    };
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor("test");
    InMemoryInputDescriptor<TestTableData.PageView> inputDescriptor = isd.getInputDescriptor("User", new NoOpSerde<>());
    TestRunner.of(app).addInputStream(inputDescriptor, users).run(Duration.ofSeconds(10));
    Bucket outputBucket = cluster.openBucket(outputBucketName);
    Assert.assertEquals("{\"name\":\"Alice\",\"age\":\"20\"}", outputBucket.get("Alice").content().toString());
    Assert.assertEquals("{\"name\":\"Bob\",\"age\":\"30\"}", outputBucket.get("Bob").content().toString());
    Assert.assertEquals("{\"name\":\"Chris\",\"age\":\"40\"}", outputBucket.get("Chris").content().toString());
    Assert.assertEquals("{\"name\":\"David\",\"age\":\"50\"}", outputBucket.get("David").content().toString());
    outputBucket.close();
}

Example 12 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class TestRemoteTableWithBatchEndToEnd method doTestStreamTableJoinRemoteTable.

private void doTestStreamTableJoinRemoteTable(String testName, boolean batchRead, boolean batchWrite) throws Exception {
    final InMemoryWriteFunction writer = new InMemoryWriteFunction(testName);
    BATCH_READS.put(testName, new AtomicInteger());
    BATCH_WRITES.put(testName, new AtomicInteger());
    WRITTEN_RECORDS.put(testName, new HashMap<>());
    int count = 16;
    int batchSize = 4;
    String profiles = Base64Serializer.serialize(generateProfiles(count));
    final RateLimiter readRateLimiter = mock(RateLimiter.class, withSettings().serializable());
    final RateLimiter writeRateLimiter = mock(RateLimiter.class, withSettings().serializable());
    final TableRateLimiter.CreditFunction creditFunction = (k, v, args) -> 1;
    final StreamApplication app = appDesc -> {
        RemoteTableDescriptor<Integer, Profile, Void> inputTableDesc = new RemoteTableDescriptor<>("profile-table-1");
        inputTableDesc.withReadFunction(InMemoryReadFunction.getInMemoryReadFunction(testName, profiles)).withRateLimiter(readRateLimiter, creditFunction, null);
        if (batchRead) {
            inputTableDesc.withBatchProvider(new CompactBatchProvider().withMaxBatchSize(batchSize).withMaxBatchDelay(Duration.ofHours(1)));
        }
        // dummy reader
        TableReadFunction readFn = new MyReadFunction();
        RemoteTableDescriptor<Integer, EnrichedPageView, EnrichedPageView> outputTableDesc = new RemoteTableDescriptor<>("enriched-page-view-table-1");
        outputTableDesc.withReadFunction(readFn).withWriteFunction(writer).withRateLimiter(writeRateLimiter, creditFunction, creditFunction);
        if (batchWrite) {
            outputTableDesc.withBatchProvider(new CompactBatchProvider().withMaxBatchSize(batchSize).withMaxBatchDelay(Duration.ofHours(1)));
        }
        Table outputTable = appDesc.getTable(outputTableDesc);
        Table<KV<Integer, Profile>> inputTable = appDesc.getTable(inputTableDesc);
        DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor("test");
        GenericInputDescriptor<PageView> isd = ksd.getInputDescriptor("PageView", new NoOpSerde<>());
        appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(inputTable, new PageViewToProfileJoinFunction()).map(m -> new KV<>(m.getMemberId(), m)).sendTo(outputTable);
    };
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor("test");
    InMemoryInputDescriptor<PageView> inputDescriptor = isd.getInputDescriptor("PageView", new NoOpSerde<>());
    TestRunner.of(app).addInputStream(inputDescriptor, Arrays.asList(generatePageViewsWithDistinctKeys(count))).addConfig("task.max.concurrency", String.valueOf(count)).addConfig("task.async.commit", String.valueOf(true)).run(Duration.ofSeconds(10));
    Assert.assertEquals(count, WRITTEN_RECORDS.get(testName).size());
    Assert.assertNotNull(WRITTEN_RECORDS.get(testName).get(0));
    if (batchWrite) {
        Assert.assertEquals(count / batchSize, BATCH_WRITES.get(testName).get());
    }
}

Example 13 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class WatermarkIntegrationTest method testWatermark.

@Test
public void testWatermark() throws Exception {
    Map<String, String> configs = new HashMap<>();
    configs.put(ApplicationConfig.APP_RUNNER_CLASS, MockLocalApplicationRunner.class.getName());
    configs.put("systems.test.samza.factory", TestSystemFactory.class.getName());
    configs.put("streams.PageView.samza.system", "test");
    configs.put("streams.PageView.partitionCount", String.valueOf(PARTITION_COUNT));
    configs.put(JobConfig.JOB_NAME, "test-watermark-job");
    configs.put(JobConfig.PROCESSOR_ID, "1");
    configs.put(JobCoordinatorConfig.JOB_COORDINATOR_FACTORY, PassthroughJobCoordinatorFactory.class.getName());
    configs.put(TaskConfig.GROUPER_FACTORY, SingleContainerGrouperFactory.class.getName());
    configs.put("systems.kafka.samza.factory", "org.apache.samza.system.kafka.KafkaSystemFactory");
    configs.put("systems.kafka.producer.bootstrap.servers", bootstrapUrl());
    configs.put("systems.kafka.consumer.zookeeper.connect", zkConnect());
    configs.put("systems.kafka.samza.key.serde", "int");
    configs.put("systems.kafka.samza.msg.serde", "json");
    configs.put("systems.kafka.default.stream.replication.factor", "1");
    configs.put("job.default.system", "kafka");
    configs.put("serializers.registry.int.class", IntegerSerdeFactory.class.getName());
    configs.put("serializers.registry.string.class", StringSerdeFactory.class.getName());
    configs.put("serializers.registry.json.class", PageViewJsonSerdeFactory.class.getName());
    List<PageView> received = new ArrayList<>();
    class TestStreamApp implements StreamApplication {

        @Override
        public void describe(StreamApplicationDescriptor appDescriptor) {
            DelegatingSystemDescriptor sd = new DelegatingSystemDescriptor("test");
            GenericInputDescriptor<KV<String, PageView>> isd = sd.getInputDescriptor("PageView", KVSerde.of(new NoOpSerde<>(), new NoOpSerde<>()));
            appDescriptor.getInputStream(isd).map(KV::getValue).partitionBy(pv -> pv.getMemberId(), pv -> pv, KVSerde.of(new NoOpSerde<>(), new NoOpSerde<>()), "p1").sink((m, collector, coordinator) -> {
                received.add(m.getValue());
            });
        }
    }
    Config config = new MapConfig(configs);
    final ApplicationRunner runner = ApplicationRunners.getApplicationRunner(new TestStreamApp(), config);
    executeRun(runner, config);
    // processors are only available when the app is running
    Map<String, StreamOperatorTask> tasks = getTaskOperationGraphs((MockLocalApplicationRunner) runner);
    runner.waitForFinish();
    // wait for the completion to ensure that all tasks are actually initialized and the OperatorImplGraph is initialized
    StreamOperatorTask task0 = tasks.get("Partition 0");
    OperatorImplGraph graph = TestStreamOperatorTask.getOperatorImplGraph(task0);
    OperatorImpl pb = getOperator(graph, OperatorSpec.OpCode.PARTITION_BY);
    assertEquals(TestOperatorImpl.getInputWatermark(pb), 4);
    assertEquals(TestOperatorImpl.getOutputWatermark(pb), 4);
    OperatorImpl sink = getOperator(graph, OperatorSpec.OpCode.SINK);
    assertEquals(TestOperatorImpl.getInputWatermark(sink), 3);
    assertEquals(TestOperatorImpl.getOutputWatermark(sink), 3);
    StreamOperatorTask task1 = tasks.get("Partition 1");
    graph = TestStreamOperatorTask.getOperatorImplGraph(task1);
    pb = getOperator(graph, OperatorSpec.OpCode.PARTITION_BY);
    assertEquals(TestOperatorImpl.getInputWatermark(pb), 3);
    assertEquals(TestOperatorImpl.getOutputWatermark(pb), 3);
    sink = getOperator(graph, OperatorSpec.OpCode.SINK);
    assertEquals(TestOperatorImpl.getInputWatermark(sink), 3);
    assertEquals(TestOperatorImpl.getOutputWatermark(sink), 3);
}

Example 14 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class TestRemoteTableEndToEnd method testSendToUpdatesWithoutUpdateOptions.

// Test will fail as we use sendTo with KV<K, UpdateMessage> stream without UpdateOptions
@Test(expected = SamzaException.class)
public void testSendToUpdatesWithoutUpdateOptions() throws Exception {
    // max member id for page views is 10
    final String profiles = Base64Serializer.serialize(generateProfiles(10));
    final RateLimiter readRateLimiter = mock(RateLimiter.class, withSettings().serializable());
    final TableRateLimiter.CreditFunction creditFunction = (k, v, args) -> 1;
    final StreamApplication app = appDesc -> {
        final RemoteTableDescriptor joinTableDesc = new RemoteTableDescriptor<Integer, TestTableData.Profile, Void>("profile-table-1").withReadFunction(InMemoryProfileReadFunction.getInMemoryReadFunction(profiles)).withRateLimiter(readRateLimiter, creditFunction, null);
        final RemoteTableDescriptor outputTableDesc = new RemoteTableDescriptor<Integer, EnrichedPageView, EnrichedPageView>("enriched-page-view-table-1").withReadFunction(new NoOpTableReadFunction<>()).withReadRateLimiterDisabled().withWriteFunction(new InMemoryEnrichedPageViewWriteFunction2("testUpdateWithoutUpdateOptions", false)).withWriteRateLimit(1000);
        final Table<KV<Integer, Profile>> outputTable = appDesc.getTable(outputTableDesc);
        final Table<KV<Integer, Profile>> joinTable = appDesc.getTable(joinTableDesc);
        final DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor("test");
        final GenericInputDescriptor<PageView> isd = ksd.getInputDescriptor("PageView", new NoOpSerde<>());
        appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(joinTable, new PageViewToProfileJoinFunction()).map(m -> new KV(m.getMemberId(), UpdateMessage.of(m, m))).sendTo(outputTable);
    };
    int numPageViews = 40;
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor("test");
    InMemoryInputDescriptor<PageView> inputDescriptor = isd.getInputDescriptor("PageView", new NoOpSerde<>());
    TestRunner.of(app).addInputStream(inputDescriptor, TestTableData.generatePartitionedPageViews(numPageViews, 4)).run(Duration.ofSeconds(10));
}

Example 15 with GenericInputDescriptor

use of org.apache.samza.system.descriptors.GenericInputDescriptor in project samza by apache.

the class TestRemoteTableEndToEnd method testSendToUpdatesFailureAfterPutDefault.

// Test fails with the following exception:
// org.apache.samza.SamzaException: Update after Put default failed with exception.
@Test(expected = SamzaException.class)
public void testSendToUpdatesFailureAfterPutDefault() throws Exception {
    // the test checks for failure when update after put default fails
    String testName = "testSendToUpdatesFailureAfterPutDefault";
    final String profiles = Base64Serializer.serialize(generateProfiles(30));
    final RateLimiter readRateLimiter = mock(RateLimiter.class, withSettings().serializable());
    final TableRateLimiter.CreditFunction creditFunction = (k, v, args) -> 1;
    final StreamApplication app = appDesc -> {
        final RemoteTableDescriptor joinTableDesc = new RemoteTableDescriptor<Integer, TestTableData.Profile, Void>("profile-table-1").withReadFunction(InMemoryProfileReadFunction.getInMemoryReadFunction(profiles)).withRateLimiter(readRateLimiter, creditFunction, null);
        final RemoteTableDescriptor outputTableDesc = new RemoteTableDescriptor<Integer, EnrichedPageView, EnrichedPageView>("enriched-page-view-table-1").withReadFunction(new NoOpTableReadFunction<>()).withReadRateLimiterDisabled().withWriteFunction(new InMemoryEnrichedPageViewWriteFunction2(testName, false, true)).withWriteRateLimit(1000);
        final Table<KV<Integer, Profile>> outputTable = appDesc.getTable(outputTableDesc);
        final Table<KV<Integer, Profile>> joinTable = appDesc.getTable(joinTableDesc);
        final DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor("test");
        final GenericInputDescriptor<PageView> isd = ksd.getInputDescriptor("PageView", new NoOpSerde<>());
        appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(joinTable, new PageViewToProfileJoinFunction()).map(m -> new KV(m.getMemberId(), UpdateMessage.of(m, m))).sendTo(outputTable, UpdateOptions.UPDATE_WITH_DEFAULTS);
    };
    int numPageViews = 15;
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor("test");
    InMemoryInputDescriptor<PageView> inputDescriptor = isd.getInputDescriptor("PageView", new NoOpSerde<>());
    Map<Integer, List<PageView>> integerListMap = TestTableData.generatePartitionedPageViews(numPageViews, 1);
    TestRunner.of(app).addInputStream(inputDescriptor, integerListMap).run(Duration.ofSeconds(10));
}