Search in sources :

Example 6 with InMemoryInputDescriptor

use of org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor in project samza by apache.

the class TestRemoteTableWithBatchEndToEnd method doTestStreamTableJoinRemoteTablePartialUpdates.

private void doTestStreamTableJoinRemoteTablePartialUpdates(String testName, boolean isCompactBatch) 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);
        // dummy reader
        TableReadFunction<Integer, EnrichedPageView> 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 (isCompactBatch) {
            outputTableDesc.withBatchProvider(new CompactBatchProvider<Integer, EnrichedPageView, EnrichedPageView>().withMaxBatchSize(batchSize).withMaxBatchDelay(Duration.ofHours(1)));
        } else {
            outputTableDesc.withBatchProvider(new CompleteBatchProvider<Integer, EnrichedPageView, EnrichedPageView>().withMaxBatchSize(batchSize).withMaxBatchDelay(Duration.ofHours(1)));
        }
        Table<KV<Integer, EnrichedPageView>> table = 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(), UpdateMessage.of(m, m))).sendTo(table, UpdateOptions.UPDATE_WITH_DEFAULTS);
    };
    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));
    Assert.assertEquals(count / batchSize, BATCH_WRITES.get(testName).get());
}
Also used : Arrays(java.util.Arrays) UpdateMessage(org.apache.samza.operators.UpdateMessage) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) ObjectInputStream(java.io.ObjectInputStream) HashMap(java.util.HashMap) CompletableFuture(java.util.concurrent.CompletableFuture) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) Function(java.util.function.Function) TableReadFunction(org.apache.samza.table.remote.TableReadFunction) Base64Serializer(org.apache.samza.test.util.Base64Serializer) InMemoryInputDescriptor(org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) TestTableData(org.apache.samza.test.table.TestTableData) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) Duration(java.time.Duration) Map(java.util.Map) TableWriteFunction(org.apache.samza.table.remote.TableWriteFunction) KV(org.apache.samza.operators.KV) NoOpSerde(org.apache.samza.serializers.NoOpSerde) Table(org.apache.samza.table.Table) Collection(java.util.Collection) BaseTableFunction(org.apache.samza.table.remote.BaseTableFunction) IOException(java.io.IOException) Test(org.junit.Test) Collectors(java.util.stream.Collectors) SamzaException(org.apache.samza.SamzaException) TestRunner(org.apache.samza.test.framework.TestRunner) Mockito(org.mockito.Mockito) Entry(org.apache.samza.storage.kv.Entry) RateLimiter(org.apache.samza.util.RateLimiter) UpdateOptions(org.apache.samza.operators.UpdateOptions) CompactBatchProvider(org.apache.samza.table.batching.CompactBatchProvider) StreamApplication(org.apache.samza.application.StreamApplication) Assert(org.junit.Assert) CompleteBatchProvider(org.apache.samza.table.batching.CompleteBatchProvider) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) TableReadFunction(org.apache.samza.table.remote.TableReadFunction) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) CompleteBatchProvider(org.apache.samza.table.batching.CompleteBatchProvider) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) Table(org.apache.samza.table.Table) StreamApplication(org.apache.samza.application.StreamApplication) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) RateLimiter(org.apache.samza.util.RateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) NoOpSerde(org.apache.samza.serializers.NoOpSerde) CompactBatchProvider(org.apache.samza.table.batching.CompactBatchProvider)

Example 7 with InMemoryInputDescriptor

use of org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor in project samza by apache.

the class TestRemoteTableEndToEnd method doTestStreamTableJoinRemoteTable.

private void doTestStreamTableJoinRemoteTable(boolean withCache, boolean defaultCache, boolean withUpdate, String testName) throws Exception {
    WRITTEN_RECORDS.put(testName, new ArrayList<>());
    // 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 InMemoryEnrichedPageViewWriteFunction(testName)).withWriteRateLimit(1000);
        final Table<KV<Integer, Profile>> outputTable = withCache ? getCachingTable(outputTableDesc, defaultCache, appDesc) : appDesc.getTable(outputTableDesc);
        final Table<KV<Integer, Profile>> joinTable = withCache ? getCachingTable(joinTableDesc, defaultCache, appDesc) : appDesc.getTable(joinTableDesc);
        final DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor("test");
        final GenericInputDescriptor<PageView> isd = ksd.getInputDescriptor("PageView", new NoOpSerde<>());
        if (withUpdate) {
            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);
        } else {
            appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(joinTable, new PageViewToProfileJoinFunction()).map(m -> KV.of(m.getMemberId(), 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));
    Assert.assertEquals(numPageViews, WRITTEN_RECORDS.get(testName).size());
    Assert.assertNotNull(WRITTEN_RECORDS.get(testName).get(0));
    WRITTEN_RECORDS.get(testName).forEach(epv -> Assert.assertFalse(epv.company.contains("-")));
}
Also used : GuavaCacheTableDescriptor(org.apache.samza.table.descriptors.GuavaCacheTableDescriptor) Arrays(java.util.Arrays) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) TableDescriptor(org.apache.samza.table.descriptors.TableDescriptor) ObjectInputStream(java.io.ObjectInputStream) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) RemoteTable(org.apache.samza.table.remote.RemoteTable) CachingTableDescriptor(org.apache.samza.table.descriptors.CachingTableDescriptor) Counter(org.apache.samza.metrics.Counter) InMemoryInputDescriptor(org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) Duration(java.time.Duration) Map(java.util.Map) NoOpTableReadFunction(org.apache.samza.table.remote.NoOpTableReadFunction) MapConfig(org.apache.samza.config.MapConfig) KV(org.apache.samza.operators.KV) NoOpSerde(org.apache.samza.serializers.NoOpSerde) Table(org.apache.samza.table.Table) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) MetricsRegistry(org.apache.samza.metrics.MetricsRegistry) Collectors(java.util.stream.Collectors) Executors(java.util.concurrent.Executors) Context(org.apache.samza.context.Context) RecordNotFoundException(org.apache.samza.table.RecordNotFoundException) TestRunner(org.apache.samza.test.framework.TestRunner) Matchers.any(org.mockito.Matchers.any) List(java.util.List) StreamApplicationDescriptor(org.apache.samza.application.descriptors.StreamApplicationDescriptor) TestTableData.generateProfiles(org.apache.samza.test.table.TestTableData.generateProfiles) CacheBuilder(com.google.common.cache.CacheBuilder) PageView(org.apache.samza.test.table.TestTableData.PageView) StreamApplication(org.apache.samza.application.StreamApplication) UpdateMessage(org.apache.samza.operators.UpdateMessage) HashMap(java.util.HashMap) CompletableFuture(java.util.concurrent.CompletableFuture) Profile(org.apache.samza.test.table.TestTableData.Profile) Function(java.util.function.Function) TableReadFunction(org.apache.samza.table.remote.TableReadFunction) Matchers.anyString(org.mockito.Matchers.anyString) ArrayList(java.util.ArrayList) Base64Serializer(org.apache.samza.test.util.Base64Serializer) MockContext(org.apache.samza.context.MockContext) TableWriteFunction(org.apache.samza.table.remote.TableWriteFunction) ExpectedException(org.junit.rules.ExpectedException) Timer(org.apache.samza.metrics.Timer) BaseTableFunction(org.apache.samza.table.remote.BaseTableFunction) IOException(java.io.IOException) Test(org.junit.Test) SamzaException(org.apache.samza.SamzaException) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) Rule(org.junit.Rule) RateLimiter(org.apache.samza.util.RateLimiter) UpdateOptions(org.apache.samza.operators.UpdateOptions) Assert(org.junit.Assert) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) Matchers.anyString(org.mockito.Matchers.anyString) Profile(org.apache.samza.test.table.TestTableData.Profile) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) PageView(org.apache.samza.test.table.TestTableData.PageView) RemoteTable(org.apache.samza.table.remote.RemoteTable) Table(org.apache.samza.table.Table) StreamApplication(org.apache.samza.application.StreamApplication) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) KV(org.apache.samza.operators.KV) RateLimiter(org.apache.samza.util.RateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) NoOpSerde(org.apache.samza.serializers.NoOpSerde) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView)

Example 8 with InMemoryInputDescriptor

use of org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor in project samza by apache.

the class TestRemoteTableEndToEnd method doTestStreamTableJoinRemoteTableWithFirstTimeUpdates.

private void doTestStreamTableJoinRemoteTableWithFirstTimeUpdates(String testName, boolean withDefaults, boolean failUpdatesAlways) throws IOException {
    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, failUpdatesAlways)).withWriteRateLimit(1000);
        // counters to count puts and updates
        COUNTERS.put(testName + "-put", new AtomicInteger());
        COUNTERS.put(testName + "-update", new AtomicInteger());
        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<>());
        if (withDefaults) {
            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);
        } else {
            appDesc.getInputStream(isd).map(pv -> new KV<>(pv.getMemberId(), pv)).join(joinTable, new PageViewToProfileJoinFunction()).map(m -> new KV(m.getMemberId(), UpdateMessage.of(m))).sendTo(outputTable, UpdateOptions.UPDATE_ONLY);
        }
    };
    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));
    if (withDefaults) {
        Assert.assertEquals(10, COUNTERS.get(testName + "-put").intValue());
        Assert.assertEquals(15, COUNTERS.get(testName + "-update").intValue());
    }
}
Also used : GuavaCacheTableDescriptor(org.apache.samza.table.descriptors.GuavaCacheTableDescriptor) Arrays(java.util.Arrays) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) TableDescriptor(org.apache.samza.table.descriptors.TableDescriptor) ObjectInputStream(java.io.ObjectInputStream) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) RemoteTable(org.apache.samza.table.remote.RemoteTable) CachingTableDescriptor(org.apache.samza.table.descriptors.CachingTableDescriptor) Counter(org.apache.samza.metrics.Counter) InMemoryInputDescriptor(org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) Duration(java.time.Duration) Map(java.util.Map) NoOpTableReadFunction(org.apache.samza.table.remote.NoOpTableReadFunction) MapConfig(org.apache.samza.config.MapConfig) KV(org.apache.samza.operators.KV) NoOpSerde(org.apache.samza.serializers.NoOpSerde) Table(org.apache.samza.table.Table) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) MetricsRegistry(org.apache.samza.metrics.MetricsRegistry) Collectors(java.util.stream.Collectors) Executors(java.util.concurrent.Executors) Context(org.apache.samza.context.Context) RecordNotFoundException(org.apache.samza.table.RecordNotFoundException) TestRunner(org.apache.samza.test.framework.TestRunner) Matchers.any(org.mockito.Matchers.any) List(java.util.List) StreamApplicationDescriptor(org.apache.samza.application.descriptors.StreamApplicationDescriptor) TestTableData.generateProfiles(org.apache.samza.test.table.TestTableData.generateProfiles) CacheBuilder(com.google.common.cache.CacheBuilder) PageView(org.apache.samza.test.table.TestTableData.PageView) StreamApplication(org.apache.samza.application.StreamApplication) UpdateMessage(org.apache.samza.operators.UpdateMessage) HashMap(java.util.HashMap) CompletableFuture(java.util.concurrent.CompletableFuture) Profile(org.apache.samza.test.table.TestTableData.Profile) Function(java.util.function.Function) TableReadFunction(org.apache.samza.table.remote.TableReadFunction) Matchers.anyString(org.mockito.Matchers.anyString) ArrayList(java.util.ArrayList) Base64Serializer(org.apache.samza.test.util.Base64Serializer) MockContext(org.apache.samza.context.MockContext) TableWriteFunction(org.apache.samza.table.remote.TableWriteFunction) ExpectedException(org.junit.rules.ExpectedException) Timer(org.apache.samza.metrics.Timer) BaseTableFunction(org.apache.samza.table.remote.BaseTableFunction) IOException(java.io.IOException) Test(org.junit.Test) SamzaException(org.apache.samza.SamzaException) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) Rule(org.junit.Rule) RateLimiter(org.apache.samza.util.RateLimiter) UpdateOptions(org.apache.samza.operators.UpdateOptions) Assert(org.junit.Assert) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) Matchers.anyString(org.mockito.Matchers.anyString) Profile(org.apache.samza.test.table.TestTableData.Profile) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) List(java.util.List) ArrayList(java.util.ArrayList) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) PageView(org.apache.samza.test.table.TestTableData.PageView) RemoteTable(org.apache.samza.table.remote.RemoteTable) Table(org.apache.samza.table.Table) StreamApplication(org.apache.samza.application.StreamApplication) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) KV(org.apache.samza.operators.KV) RateLimiter(org.apache.samza.util.RateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) TableRateLimiter(org.apache.samza.table.remote.TableRateLimiter) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) NoOpSerde(org.apache.samza.serializers.NoOpSerde) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView)

Example 9 with InMemoryInputDescriptor

use of org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor in project samza by apache.

the class TestLocalTableEndToEnd method testSendTo.

@Test
public void testSendTo() {
    MyMapFunction mapFn = new MyMapFunction();
    StreamApplication app = appDesc -> {
        Table<KV<Integer, Profile>> table = appDesc.getTable(new InMemoryTableDescriptor<>("t1", KVSerde.of(new IntegerSerde(), new ProfileJsonSerde())));
        DelegatingSystemDescriptor ksd = new DelegatingSystemDescriptor(SYSTEM_NAME);
        GenericInputDescriptor<Profile> isd = ksd.getInputDescriptor(PROFILE_STREAM, new NoOpSerde<>());
        appDesc.getInputStream(isd).map(mapFn).sendTo(table);
    };
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor(SYSTEM_NAME);
    InMemoryInputDescriptor<Profile> profileStreamDesc = isd.getInputDescriptor(PROFILE_STREAM, new NoOpSerde<>());
    int numProfilesPerPartition = 10;
    int numInputPartitions = 4;
    Map<Integer, List<Profile>> inputProfiles = TestTableData.generatePartitionedProfiles(numProfilesPerPartition * numInputPartitions, numInputPartitions);
    TestRunner.of(app).addInputStream(profileStreamDesc, inputProfiles).run(Duration.ofSeconds(10));
    for (int i = 0; i < numInputPartitions; i++) {
        MyMapFunction mapFnCopy = MyMapFunction.getMapFunctionByTask(String.format("Partition %d", i));
        assertEquals(numProfilesPerPartition, mapFnCopy.received.size());
        mapFnCopy.received.forEach(p -> assertNotNull(mapFnCopy.table.get(p.getMemberId())));
    }
}
Also used : PageViewJsonSerdeFactory(org.apache.samza.test.table.TestTableData.PageViewJsonSerdeFactory) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) JobConfig(org.apache.samza.config.JobConfig) HashMap(java.util.HashMap) SingleContainerGrouperFactory(org.apache.samza.container.grouper.task.SingleContainerGrouperFactory) Profile(org.apache.samza.test.table.TestTableData.Profile) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) ArrayList(java.util.ArrayList) MapFunction(org.apache.samza.operators.functions.MapFunction) InMemoryInputDescriptor(org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) Duration(java.time.Duration) Map(java.util.Map) ArraySystemFactory(org.apache.samza.test.util.ArraySystemFactory) LinkedList(java.util.LinkedList) KV(org.apache.samza.operators.KV) IntegerSerde(org.apache.samza.serializers.IntegerSerde) NoOpSerde(org.apache.samza.serializers.NoOpSerde) ProfileJsonSerde(org.apache.samza.test.table.TestTableData.ProfileJsonSerde) InMemoryTableDescriptor(org.apache.samza.storage.kv.inmemory.descriptors.InMemoryTableDescriptor) MessageStream(org.apache.samza.operators.MessageStream) Table(org.apache.samza.table.Table) TaskConfig(org.apache.samza.config.TaskConfig) Assert.assertNotNull(org.junit.Assert.assertNotNull) EnrichedPageView(org.apache.samza.test.table.TestTableData.EnrichedPageView) Test(org.junit.Test) PageViewJsonSerde(org.apache.samza.test.table.TestTableData.PageViewJsonSerde) PassthroughJobCoordinatorFactory(org.apache.samza.standalone.PassthroughJobCoordinatorFactory) Context(org.apache.samza.context.Context) TestRunner(org.apache.samza.test.framework.TestRunner) List(java.util.List) ReadWriteUpdateTable(org.apache.samza.table.ReadWriteUpdateTable) StreamApplicationDescriptor(org.apache.samza.application.descriptors.StreamApplicationDescriptor) JobCoordinatorConfig(org.apache.samza.config.JobCoordinatorConfig) KVSerde(org.apache.samza.serializers.KVSerde) PageView(org.apache.samza.test.table.TestTableData.PageView) StreamApplication(org.apache.samza.application.StreamApplication) Assert.assertEquals(org.junit.Assert.assertEquals) Table(org.apache.samza.table.Table) ReadWriteUpdateTable(org.apache.samza.table.ReadWriteUpdateTable) ProfileJsonSerde(org.apache.samza.test.table.TestTableData.ProfileJsonSerde) StreamApplication(org.apache.samza.application.StreamApplication) Profile(org.apache.samza.test.table.TestTableData.Profile) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) IntegerSerde(org.apache.samza.serializers.IntegerSerde) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) NoOpSerde(org.apache.samza.serializers.NoOpSerde) InMemoryTableDescriptor(org.apache.samza.storage.kv.inmemory.descriptors.InMemoryTableDescriptor) ArrayList(java.util.ArrayList) LinkedList(java.util.LinkedList) List(java.util.List) Test(org.junit.Test)

Example 10 with InMemoryInputDescriptor

use of org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor in project samza by apache.

the class EndOfStreamIntegrationTest method testPipeline.

@Test
public void testPipeline() {
    class PipelineApplication 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(PageView::getMemberId, pv -> pv, KVSerde.of(new IntegerSerde(), new TestTableData.PageViewJsonSerde()), "p1").sink((m, collector, coordinator) -> {
                RECEIVED.add(m.getValue());
            });
        }
    }
    int numPageViews = 40;
    InMemorySystemDescriptor isd = new InMemorySystemDescriptor("test");
    InMemoryInputDescriptor<TestTableData.PageView> inputDescriptor = isd.getInputDescriptor("PageView", new NoOpSerde<>());
    TestRunner.of(new PipelineApplication()).addInputStream(inputDescriptor, TestTableData.generatePartitionedPageViews(numPageViews, 4)).run(Duration.ofSeconds(10));
    assertEquals(RECEIVED.size(), numPageViews);
}
Also used : InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) Test(org.junit.Test) GenericInputDescriptor(org.apache.samza.system.descriptors.GenericInputDescriptor) ArrayList(java.util.ArrayList) TestRunner(org.apache.samza.test.framework.TestRunner) List(java.util.List) InMemoryInputDescriptor(org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor) StreamApplicationDescriptor(org.apache.samza.application.descriptors.StreamApplicationDescriptor) TestTableData(org.apache.samza.test.table.TestTableData) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) Duration(java.time.Duration) KVSerde(org.apache.samza.serializers.KVSerde) PageView(org.apache.samza.test.table.TestTableData.PageView) StreamApplication(org.apache.samza.application.StreamApplication) KV(org.apache.samza.operators.KV) IntegerSerde(org.apache.samza.serializers.IntegerSerde) NoOpSerde(org.apache.samza.serializers.NoOpSerde) Assert.assertEquals(org.junit.Assert.assertEquals) PageView(org.apache.samza.test.table.TestTableData.PageView) StreamApplication(org.apache.samza.application.StreamApplication) KV(org.apache.samza.operators.KV) InMemorySystemDescriptor(org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor) IntegerSerde(org.apache.samza.serializers.IntegerSerde) StreamApplicationDescriptor(org.apache.samza.application.descriptors.StreamApplicationDescriptor) DelegatingSystemDescriptor(org.apache.samza.system.descriptors.DelegatingSystemDescriptor) NoOpSerde(org.apache.samza.serializers.NoOpSerde) TestTableData(org.apache.samza.test.table.TestTableData) Test(org.junit.Test)

Aggregations

Duration (java.time.Duration)11 StreamApplication (org.apache.samza.application.StreamApplication)11 KV (org.apache.samza.operators.KV)11 NoOpSerde (org.apache.samza.serializers.NoOpSerde)11 InMemoryInputDescriptor (org.apache.samza.test.framework.system.descriptors.InMemoryInputDescriptor)11 InMemorySystemDescriptor (org.apache.samza.test.framework.system.descriptors.InMemorySystemDescriptor)11 Test (org.junit.Test)11 DelegatingSystemDescriptor (org.apache.samza.system.descriptors.DelegatingSystemDescriptor)10 GenericInputDescriptor (org.apache.samza.system.descriptors.GenericInputDescriptor)10 Table (org.apache.samza.table.Table)10 TestRunner (org.apache.samza.test.framework.TestRunner)10 ArrayList (java.util.ArrayList)9 Arrays (java.util.Arrays)9 List (java.util.List)9 Assert (org.junit.Assert)9 HashMap (java.util.HashMap)8 Map (java.util.Map)8 Collectors (java.util.stream.Collectors)8 SamzaException (org.apache.samza.SamzaException)8 StreamApplicationDescriptor (org.apache.samza.application.descriptors.StreamApplicationDescriptor)8