Examples with RemoteTableDescriptor org.apache.samza.table.descriptors.RemoteTableDescriptor used on opensource projects

Example 11 with RemoteTableDescriptor

use of org.apache.samza.table.descriptors.RemoteTableDescriptor in project samza by apache.

the class TestRemoteTableEndToEnd method testSendToWithDefaultsAndUpdateOnly.

// Test fails with the following exception:
// org.apache.samza.SamzaException: Put default failed for update as the UpdateOptions was set to UPDATE_ONLY.
// Please use UpdateOptions.UPDATE_WITH_DEFAULTS instead.
@Test(expected = SamzaException.class)
public void testSendToWithDefaultsAndUpdateOnly() throws Exception {
    String testName = "testSendToWithDefaultsAndUpdateOnly";
    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)).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<>());
        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_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));
}

Example 12 with RemoteTableDescriptor

use of org.apache.samza.table.descriptors.RemoteTableDescriptor in project samza by apache.

the class ScanTranslator method translate.

// ScanMapFunction
void translate(final TableScan tableScan, final String queryLogicalId, final String logicalOpId, final TranslatorContext context, Map<String, DelegatingSystemDescriptor> systemDescriptors, Map<String, MessageStream<SamzaSqlInputMessage>> inputMsgStreams) {
    StreamApplicationDescriptor streamAppDesc = context.getStreamAppDescriptor();
    List<String> tableNameParts = tableScan.getTable().getQualifiedName();
    String sourceName = SqlIOConfig.getSourceFromSourceParts(tableNameParts);
    Validate.isTrue(relMsgConverters.containsKey(sourceName), String.format("Unknown source %s", sourceName));
    SqlIOConfig sqlIOConfig = systemStreamConfig.get(sourceName);
    final String systemName = sqlIOConfig.getSystemName();
    final String streamId = sqlIOConfig.getStreamId();
    final String source = sqlIOConfig.getSource();
    final boolean isRemoteTable = sqlIOConfig.getTableDescriptor().isPresent() && (sqlIOConfig.getTableDescriptor().get() instanceof RemoteTableDescriptor || sqlIOConfig.getTableDescriptor().get() instanceof CachingTableDescriptor);
    // descriptor to load the local table.
    if (isRemoteTable) {
        return;
    }
    // set the wrapper input transformer (SamzaSqlInputTransformer) in system descriptor
    DelegatingSystemDescriptor systemDescriptor = systemDescriptors.get(systemName);
    if (systemDescriptor == null) {
        systemDescriptor = new DelegatingSystemDescriptor(systemName, new SamzaSqlInputTransformer());
        systemDescriptors.put(systemName, systemDescriptor);
    } else {
        /* in SamzaSQL, there should be no systemDescriptor setup by user, so this branch happens only
       * in case of Fan-OUT (i.e., same input stream used in multiple sql statements), or when same input
       * used twice in same sql statement (e.g., select ... from input as i1, input as i2 ...), o.w., throw error */
        if (systemDescriptor.getTransformer().isPresent()) {
            InputTransformer existingTransformer = systemDescriptor.getTransformer().get();
            if (!(existingTransformer instanceof SamzaSqlInputTransformer)) {
                throw new SamzaException("SamzaSQL Exception: existing transformer for " + systemName + " is not SamzaSqlInputTransformer");
            }
        }
    }
    InputDescriptor inputDescriptor = systemDescriptor.getInputDescriptor(streamId, new NoOpSerde<>());
    if (!inputMsgStreams.containsKey(source)) {
        MessageStream<SamzaSqlInputMessage> inputMsgStream = streamAppDesc.getInputStream(inputDescriptor);
        inputMsgStreams.put(source, inputMsgStream.map(new SystemMessageMapperFunction(source, queryId)));
    }
    MessageStream<SamzaSqlRelMessage> samzaSqlRelMessageStream = inputMsgStreams.get(source).filter(new FilterSystemMessageFunction(sourceName, queryId)).map(new ScanMapFunction(sourceName, queryId, queryLogicalId, logicalOpId));
    context.registerMessageStream(tableScan.getId(), samzaSqlRelMessageStream);
}

Example 13 with RemoteTableDescriptor

use of org.apache.samza.table.descriptors.RemoteTableDescriptor 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());
}

Example 14 with RemoteTableDescriptor

use of org.apache.samza.table.descriptors.RemoteTableDescriptor 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("-")));
}

Example 15 with RemoteTableDescriptor

use of org.apache.samza.table.descriptors.RemoteTableDescriptor 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());
    }
}