Examples with MessageStream org.apache.samza.operators.MessageStream used on opensource projects

Example 11 with MessageStream

use of org.apache.samza.operators.MessageStream in project samza by apache.

the class TumblingWindowApp method describe.

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
    JsonSerdeV2<PageView> inputSerde = new JsonSerdeV2<>(PageView.class);
    KVSerde<String, Integer> outputSerde = KVSerde.of(new StringSerde(), new IntegerSerde());
    KafkaSystemDescriptor ksd = new KafkaSystemDescriptor(SYSTEM);
    KafkaInputDescriptor<PageView> id = ksd.getInputDescriptor(INPUT_TOPIC, inputSerde);
    KafkaOutputDescriptor<KV<String, Integer>> od = ksd.getOutputDescriptor(OUTPUT_TOPIC, outputSerde);
    MessageStream<PageView> pageViews = appDescriptor.getInputStream(id);
    OutputStream<KV<String, Integer>> outputStream = appDescriptor.getOutputStream(od);
    pageViews.filter(m -> !FILTER_KEY.equals(m.getUserId())).window(Windows.keyedTumblingWindow(PageView::getUserId, Duration.ofSeconds(3), new StringSerde(), new JsonSerdeV2<>(PageView.class)), "tumblingWindow").map(m -> KV.of(m.getKey().getKey(), m.getMessage().size())).sendTo(outputStream);
}

Example 12 with MessageStream

use of org.apache.samza.operators.MessageStream in project samza by apache.

the class WindowExample method describe.

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
    KafkaSystemDescriptor trackingSystem = new KafkaSystemDescriptor("tracking");
    KafkaInputDescriptor<PageViewEvent> inputStreamDescriptor = trackingSystem.getInputDescriptor("pageViewEvent", new JsonSerdeV2<>(PageViewEvent.class));
    KafkaOutputDescriptor<Integer> outputStreamDescriptor = trackingSystem.getOutputDescriptor("pageViewEventPerMember", new IntegerSerde());
    SupplierFunction<Integer> initialValue = () -> 0;
    FoldLeftFunction<PageViewEvent, Integer> counter = (m, c) -> c == null ? 1 : c + 1;
    MessageStream<PageViewEvent> inputStream = appDescriptor.getInputStream(inputStreamDescriptor);
    OutputStream<Integer> outputStream = appDescriptor.getOutputStream(outputStreamDescriptor);
    // create a tumbling window that outputs the number of message collected every 10 minutes.
    // also emit early results if either the number of messages collected reaches 30000, or if no new messages arrive
    // for 1 minute.
    inputStream.window(Windows.tumblingWindow(Duration.ofMinutes(10), initialValue, counter, new IntegerSerde()).setLateTrigger(Triggers.any(Triggers.count(30000), Triggers.timeSinceLastMessage(Duration.ofMinutes(1)))), "window").map(WindowPane::getMessage).sendTo(outputStream);
}

Example 13 with MessageStream

use of org.apache.samza.operators.MessageStream in project samza by apache.

the class TestFilterTranslator method testTranslate.

@Test
public void testTranslate() throws IOException, ClassNotFoundException {
    // setup mock values to the constructor of FilterTranslator
    LogicalFilter mockFilter = PowerMockito.mock(LogicalFilter.class);
    Context mockContext = mock(Context.class);
    ContainerContext mockContainerContext = mock(ContainerContext.class);
    TranslatorContext mockTranslatorContext = mock(TranslatorContext.class);
    TestMetricsRegistryImpl metricsRegistry = new TestMetricsRegistryImpl();
    RelNode mockInput = mock(RelNode.class);
    when(mockFilter.getInput()).thenReturn(mockInput);
    when(mockInput.getId()).thenReturn(1);
    when(mockFilter.getId()).thenReturn(2);
    StreamApplicationDescriptorImpl mockGraph = mock(StreamApplicationDescriptorImpl.class);
    OperatorSpec<Object, SamzaSqlRelMessage> mockInputOp = mock(OperatorSpec.class);
    MessageStream<SamzaSqlRelMessage> mockStream = new MessageStreamImpl<>(mockGraph, mockInputOp);
    when(mockTranslatorContext.getMessageStream(eq(1))).thenReturn(mockStream);
    doAnswer(this.getRegisterMessageStreamAnswer()).when(mockTranslatorContext).registerMessageStream(eq(2), any(MessageStream.class));
    RexToJavaCompiler mockCompiler = mock(RexToJavaCompiler.class);
    when(mockTranslatorContext.getExpressionCompiler()).thenReturn(mockCompiler);
    Expression mockExpr = mock(Expression.class);
    when(mockCompiler.compile(any(), any())).thenReturn(mockExpr);
    when(mockContext.getContainerContext()).thenReturn(mockContainerContext);
    when(mockContainerContext.getContainerMetricsRegistry()).thenReturn(metricsRegistry);
    // Apply translate() method to verify that we are getting the correct filter operator constructed
    FilterTranslator filterTranslator = new FilterTranslator(1);
    filterTranslator.translate(mockFilter, LOGICAL_OP_ID, mockTranslatorContext);
    // make sure that context has been registered with LogicFilter and output message streams
    verify(mockTranslatorContext, times(1)).registerRelNode(2, mockFilter);
    verify(mockTranslatorContext, times(1)).registerMessageStream(2, this.getRegisteredMessageStream(2));
    when(mockTranslatorContext.getRelNode(2)).thenReturn(mockFilter);
    when(mockTranslatorContext.getMessageStream(2)).thenReturn(this.getRegisteredMessageStream(2));
    StreamOperatorSpec filterSpec = (StreamOperatorSpec) Whitebox.getInternalState(this.getRegisteredMessageStream(2), "operatorSpec");
    assertNotNull(filterSpec);
    assertEquals(filterSpec.getOpCode(), OperatorSpec.OpCode.FILTER);
    // Verify that the describe() method will establish the context for the filter function
    Map<Integer, TranslatorContext> mockContexts = new HashMap<>();
    mockContexts.put(1, mockTranslatorContext);
    when(mockContext.getApplicationTaskContext()).thenReturn(new SamzaSqlApplicationContext(mockContexts));
    filterSpec.getTransformFn().init(mockContext);
    FilterFunction filterFn = (FilterFunction) Whitebox.getInternalState(filterSpec, "filterFn");
    assertNotNull(filterFn);
    assertEquals(mockTranslatorContext, Whitebox.getInternalState(filterFn, "translatorContext"));
    assertEquals(mockFilter, Whitebox.getInternalState(filterFn, "filter"));
    assertEquals(mockExpr, Whitebox.getInternalState(filterFn, "expr"));
    // Verify MetricsRegistry works with Project
    assertEquals(1, metricsRegistry.getGauges().size());
    assertTrue(metricsRegistry.getGauges().get(LOGICAL_OP_ID).size() > 0);
    assertEquals(1, metricsRegistry.getCounters().size());
    assertEquals(3, metricsRegistry.getCounters().get(LOGICAL_OP_ID).size());
    assertEquals(0, metricsRegistry.getCounters().get(LOGICAL_OP_ID).get(0).getCount());
    assertEquals(0, metricsRegistry.getCounters().get(LOGICAL_OP_ID).get(1).getCount());
    // Calling filterFn.apply() to verify the filter function is correctly applied to the input message
    SamzaSqlRelMessage mockInputMsg = new SamzaSqlRelMessage(new ArrayList<>(), new ArrayList<>(), new SamzaSqlRelMsgMetadata(0L, 0L));
    SamzaSqlExecutionContext executionContext = mock(SamzaSqlExecutionContext.class);
    DataContext dataContext = mock(DataContext.class);
    when(mockTranslatorContext.getExecutionContext()).thenReturn(executionContext);
    when(mockTranslatorContext.getDataContext()).thenReturn(dataContext);
    Object[] result = new Object[1];
    doAnswer(invocation -> {
        Object[] retValue = invocation.getArgumentAt(4, Object[].class);
        retValue[0] = new Boolean(true);
        return null;
    }).when(mockExpr).execute(eq(executionContext), eq(mockContext), eq(dataContext), eq(mockInputMsg.getSamzaSqlRelRecord().getFieldValues().toArray()), eq(result));
    assertTrue(filterFn.apply(mockInputMsg));
    doAnswer(invocation -> {
        Object[] retValue = invocation.getArgumentAt(4, Object[].class);
        retValue[0] = new Boolean(false);
        return null;
    }).when(mockExpr).execute(eq(executionContext), eq(mockContext), eq(dataContext), eq(mockInputMsg.getSamzaSqlRelRecord().getFieldValues().toArray()), eq(result));
    assertFalse(filterFn.apply(mockInputMsg));
    // Verify filterFn.apply() updates the MetricsRegistry metrics
    assertEquals(2, metricsRegistry.getCounters().get(LOGICAL_OP_ID).get(0).getCount());
    assertEquals(1, metricsRegistry.getCounters().get(LOGICAL_OP_ID).get(1).getCount());
}

Example 14 with MessageStream

use of org.apache.samza.operators.MessageStream in project samza by apache.

the class TestJoinTranslator method testTranslateStreamToTableJoin.

private void testTranslateStreamToTableJoin(boolean isRemoteTable) throws IOException, ClassNotFoundException {
    // setup mock values to the constructor of JoinTranslator
    final String logicalOpId = "sql0_join3";
    final int queryId = 0;
    LogicalJoin mockJoin = PowerMockito.mock(LogicalJoin.class);
    TranslatorContext mockTranslatorContext = mock(TranslatorContext.class);
    RelNode mockLeftInput = PowerMockito.mock(LogicalTableScan.class);
    RelNode mockRightInput = mock(RelNode.class);
    List<RelNode> inputs = new ArrayList<>();
    inputs.add(mockLeftInput);
    inputs.add(mockRightInput);
    RelOptTable mockLeftTable = mock(RelOptTable.class);
    when(mockLeftInput.getTable()).thenReturn(mockLeftTable);
    List<String> qualifiedTableName = Arrays.asList("test", "LeftTable");
    when(mockLeftTable.getQualifiedName()).thenReturn(qualifiedTableName);
    when(mockLeftInput.getId()).thenReturn(1);
    when(mockRightInput.getId()).thenReturn(2);
    when(mockJoin.getId()).thenReturn(3);
    when(mockJoin.getInputs()).thenReturn(inputs);
    when(mockJoin.getLeft()).thenReturn(mockLeftInput);
    when(mockJoin.getRight()).thenReturn(mockRightInput);
    RexCall mockJoinCondition = mock(RexCall.class);
    when(mockJoinCondition.isAlwaysTrue()).thenReturn(false);
    when(mockJoinCondition.getKind()).thenReturn(SqlKind.EQUALS);
    when(mockJoin.getCondition()).thenReturn(mockJoinCondition);
    RexInputRef mockLeftConditionInput = mock(RexInputRef.class);
    RexInputRef mockRightConditionInput = mock(RexInputRef.class);
    when(mockLeftConditionInput.getIndex()).thenReturn(0);
    when(mockRightConditionInput.getIndex()).thenReturn(0);
    List<RexNode> condOperands = new ArrayList<>();
    condOperands.add(mockLeftConditionInput);
    condOperands.add(mockRightConditionInput);
    when(mockJoinCondition.getOperands()).thenReturn(condOperands);
    RelDataType mockLeftCondDataType = mock(RelDataType.class);
    RelDataType mockRightCondDataType = mock(RelDataType.class);
    when(mockLeftCondDataType.getSqlTypeName()).thenReturn(SqlTypeName.BOOLEAN);
    when(mockRightCondDataType.getSqlTypeName()).thenReturn(SqlTypeName.BOOLEAN);
    when(mockLeftConditionInput.getType()).thenReturn(mockLeftCondDataType);
    when(mockRightConditionInput.getType()).thenReturn(mockRightCondDataType);
    RelDataType mockLeftRowType = mock(RelDataType.class);
    // ?? why ??
    when(mockLeftRowType.getFieldCount()).thenReturn(0);
    when(mockLeftInput.getRowType()).thenReturn(mockLeftRowType);
    List<String> leftFieldNames = Collections.singletonList("test_table_field1");
    List<String> rightStreamFieldNames = Collections.singletonList("test_stream_field1");
    when(mockLeftRowType.getFieldNames()).thenReturn(leftFieldNames);
    RelDataType mockRightRowType = mock(RelDataType.class);
    when(mockRightInput.getRowType()).thenReturn(mockRightRowType);
    when(mockRightRowType.getFieldNames()).thenReturn(rightStreamFieldNames);
    StreamApplicationDescriptorImpl mockAppDesc = mock(StreamApplicationDescriptorImpl.class);
    OperatorSpec<Object, SamzaSqlRelMessage> mockLeftInputOp = mock(OperatorSpec.class);
    MessageStream<SamzaSqlRelMessage> mockLeftInputStream = new MessageStreamImpl<>(mockAppDesc, mockLeftInputOp);
    when(mockTranslatorContext.getMessageStream(eq(mockLeftInput.getId()))).thenReturn(mockLeftInputStream);
    OperatorSpec<Object, SamzaSqlRelMessage> mockRightInputOp = mock(OperatorSpec.class);
    MessageStream<SamzaSqlRelMessage> mockRightInputStream = new MessageStreamImpl<>(mockAppDesc, mockRightInputOp);
    when(mockTranslatorContext.getMessageStream(eq(mockRightInput.getId()))).thenReturn(mockRightInputStream);
    when(mockTranslatorContext.getStreamAppDescriptor()).thenReturn(mockAppDesc);
    InputOperatorSpec mockInputOp = mock(InputOperatorSpec.class);
    OutputStreamImpl mockOutputStream = mock(OutputStreamImpl.class);
    when(mockInputOp.isKeyed()).thenReturn(true);
    when(mockOutputStream.isKeyed()).thenReturn(true);
    doAnswer(this.getRegisterMessageStreamAnswer()).when(mockTranslatorContext).registerMessageStream(eq(3), any(MessageStream.class));
    RexToJavaCompiler mockCompiler = mock(RexToJavaCompiler.class);
    when(mockTranslatorContext.getExpressionCompiler()).thenReturn(mockCompiler);
    Expression mockExpr = mock(Expression.class);
    when(mockCompiler.compile(any(), any())).thenReturn(mockExpr);
    if (isRemoteTable) {
        doAnswer(this.getRegisteredTableAnswer()).when(mockAppDesc).getTable(any(RemoteTableDescriptor.class));
    } else {
        IntermediateMessageStreamImpl mockPartitionedStream = new IntermediateMessageStreamImpl(mockAppDesc, mockInputOp, mockOutputStream);
        when(mockAppDesc.getIntermediateStream(any(String.class), any(Serde.class), eq(false))).thenReturn(mockPartitionedStream);
        doAnswer(this.getRegisteredTableAnswer()).when(mockAppDesc).getTable(any(RocksDbTableDescriptor.class));
    }
    when(mockJoin.getJoinType()).thenReturn(JoinRelType.INNER);
    SamzaSqlExecutionContext mockExecutionContext = mock(SamzaSqlExecutionContext.class);
    when(mockTranslatorContext.getExecutionContext()).thenReturn(mockExecutionContext);
    SamzaSqlApplicationConfig mockAppConfig = mock(SamzaSqlApplicationConfig.class);
    when(mockExecutionContext.getSamzaSqlApplicationConfig()).thenReturn(mockAppConfig);
    Map<String, SqlIOConfig> ssConfigBySource = mock(HashMap.class);
    when(mockAppConfig.getInputSystemStreamConfigBySource()).thenReturn(ssConfigBySource);
    SqlIOConfig mockIOConfig = mock(SqlIOConfig.class);
    TableDescriptor mockTableDesc;
    if (isRemoteTable) {
        mockTableDesc = mock(RemoteTableDescriptor.class);
    } else {
        mockTableDesc = mock(RocksDbTableDescriptor.class);
    }
    when(ssConfigBySource.get(String.join(".", qualifiedTableName))).thenReturn(mockIOConfig);
    when(mockIOConfig.getTableDescriptor()).thenReturn(Optional.of(mockTableDesc));
    JoinTranslator joinTranslator = new JoinTranslator(logicalOpId, "", queryId);
    // Verify Metrics Works with Join
    Context mockContext = mock(Context.class);
    ContainerContext mockContainerContext = mock(ContainerContext.class);
    TestMetricsRegistryImpl testMetricsRegistryImpl = new TestMetricsRegistryImpl();
    when(mockContext.getContainerContext()).thenReturn(mockContainerContext);
    when(mockContainerContext.getContainerMetricsRegistry()).thenReturn(testMetricsRegistryImpl);
    TranslatorInputMetricsMapFunction inputMetricsMF = joinTranslator.getInputMetricsMF();
    assertNotNull(inputMetricsMF);
    inputMetricsMF.init(mockContext);
    TranslatorOutputMetricsMapFunction outputMetricsMF = joinTranslator.getOutputMetricsMF();
    assertNotNull(outputMetricsMF);
    outputMetricsMF.init(mockContext);
    assertEquals(1, testMetricsRegistryImpl.getCounters().size());
    assertEquals(2, testMetricsRegistryImpl.getCounters().get(logicalOpId).size());
    assertEquals(0, testMetricsRegistryImpl.getCounters().get(logicalOpId).get(0).getCount());
    assertEquals(0, testMetricsRegistryImpl.getCounters().get(logicalOpId).get(1).getCount());
    assertEquals(1, testMetricsRegistryImpl.getGauges().size());
    // Apply translate() method to verify that we are getting the correct map operator constructed
    joinTranslator.translate(mockJoin, mockTranslatorContext);
    // make sure that context has been registered with LogicFilter and output message streams
    verify(mockTranslatorContext, times(1)).registerMessageStream(3, this.getRegisteredMessageStream(3));
    when(mockTranslatorContext.getRelNode(3)).thenReturn(mockJoin);
    when(mockTranslatorContext.getMessageStream(3)).thenReturn(this.getRegisteredMessageStream(3));
    StreamTableJoinOperatorSpec joinSpec = (StreamTableJoinOperatorSpec) Whitebox.getInternalState(this.getRegisteredMessageStream(3), "operatorSpec");
    assertNotNull(joinSpec);
    assertEquals(joinSpec.getOpCode(), OperatorSpec.OpCode.JOIN);
    // Verify joinSpec has the corresponding setup
    StreamTableJoinFunction joinFn = joinSpec.getJoinFn();
    assertNotNull(joinFn);
    if (isRemoteTable) {
        assertTrue(joinFn instanceof SamzaSqlRemoteTableJoinFunction);
    } else {
        assertTrue(joinFn instanceof SamzaSqlLocalTableJoinFunction);
    }
    assertTrue(Whitebox.getInternalState(joinFn, "isTablePosOnRight").equals(false));
    assertEquals(Collections.singletonList(0), Whitebox.getInternalState(joinFn, "streamFieldIds"));
    assertEquals(leftFieldNames, Whitebox.getInternalState(joinFn, "tableFieldNames"));
    List<String> outputFieldNames = new ArrayList<>();
    outputFieldNames.addAll(leftFieldNames);
    outputFieldNames.addAll(rightStreamFieldNames);
    assertEquals(outputFieldNames, Whitebox.getInternalState(joinFn, "outFieldNames"));
}

Example 15 with MessageStream

use of org.apache.samza.operators.MessageStream in project samza by apache.

the class PageViewCounterExample method describe.

@Override
public void describe(StreamApplicationDescriptor appDescriptor) {
    KafkaSystemDescriptor trackingSystem = new KafkaSystemDescriptor("tracking");
    KafkaInputDescriptor<PageViewEvent> inputStreamDescriptor = trackingSystem.getInputDescriptor("pageViewEvent", new JsonSerdeV2<>(PageViewEvent.class));
    KafkaOutputDescriptor<KV<String, PageViewCount>> outputStreamDescriptor = trackingSystem.getOutputDescriptor("pageViewEventPerMember", KVSerde.of(new StringSerde(), new JsonSerdeV2<>(PageViewCount.class)));
    MessageStream<PageViewEvent> pageViewEvents = appDescriptor.getInputStream(inputStreamDescriptor);
    OutputStream<KV<String, PageViewCount>> pageViewEventPerMemberStream = appDescriptor.getOutputStream(outputStreamDescriptor);
    SupplierFunction<Integer> initialValue = () -> 0;
    FoldLeftFunction<PageViewEvent, Integer> foldLeftFn = (m, c) -> c + 1;
    pageViewEvents.window(Windows.keyedTumblingWindow(PageViewEvent::getMemberId, Duration.ofSeconds(10), initialValue, foldLeftFn, null, null).setEarlyTrigger(Triggers.repeat(Triggers.count(5))).setAccumulationMode(AccumulationMode.DISCARDING), "tumblingWindow").map(windowPane -> KV.of(windowPane.getKey().getKey(), buildPageViewCount(windowPane))).sendTo(pageViewEventPerMemberStream);
}