Search in sources :

Example 11 with InputOperatorSpec

use of org.apache.samza.operators.spec.InputOperatorSpec in project samza by apache.

the class OperatorSpecGraphAnalyzer method getCustomGetNextOpSpecs.

/**
 * Creates a function that retrieves the next {@link OperatorSpec}s of any given {@link OperatorSpec} in the specified
 * {@code operatorSpecGraph}.
 *
 * Calling the returned function with any {@link SendToTableOperatorSpec} will return a collection of all
 * {@link StreamTableJoinOperatorSpec}s that reference the same table as the specified
 * {@link SendToTableOperatorSpec}, as if they were actually connected.
 */
private static Function<OperatorSpec, Iterable<OperatorSpec>> getCustomGetNextOpSpecs(Iterable<InputOperatorSpec> inputOpSpecs) {
    // Traverse operatorSpecGraph to create mapping between every SendToTableOperatorSpec and all
    // StreamTableJoinOperatorSpecs referencing the same table.
    TableJoinVisitor tableJoinVisitor = new TableJoinVisitor();
    for (InputOperatorSpec inputOpSpec : inputOpSpecs) {
        traverse(inputOpSpec, tableJoinVisitor, opSpec -> opSpec.getRegisteredOperatorSpecs());
    }
    Multimap<SendToTableOperatorSpec, StreamTableJoinOperatorSpec> sendToTableOpSpecToStreamTableJoinOpSpecs = tableJoinVisitor.getSendToTableOpSpecToStreamTableJoinOpSpecs();
    return operatorSpec -> {
        // For all other types of operator specs, return the next registered operator specs.
        if (operatorSpec instanceof SendToTableOperatorSpec) {
            SendToTableOperatorSpec sendToTableOperatorSpec = (SendToTableOperatorSpec) operatorSpec;
            return Collections.unmodifiableCollection(sendToTableOpSpecToStreamTableJoinOpSpecs.get(sendToTableOperatorSpec));
        }
        return operatorSpec.getRegisteredOperatorSpecs();
    };
}
Also used : Arrays(java.util.Arrays) Collection(java.util.Collection) Set(java.util.Set) JoinOperatorSpec(org.apache.samza.operators.spec.JoinOperatorSpec) Multimap(com.google.common.collect.Multimap) Function(java.util.function.Function) SendToTableOperatorSpec(org.apache.samza.operators.spec.SendToTableOperatorSpec) Multimaps(com.google.common.collect.Multimaps) HashSet(java.util.HashSet) Consumer(java.util.function.Consumer) HashMultimap(com.google.common.collect.HashMultimap) OperatorSpec(org.apache.samza.operators.spec.OperatorSpec) StreamTableJoinOperatorSpec(org.apache.samza.operators.spec.StreamTableJoinOperatorSpec) Collections(java.util.Collections) InputOperatorSpec(org.apache.samza.operators.spec.InputOperatorSpec) InputOperatorSpec(org.apache.samza.operators.spec.InputOperatorSpec) StreamTableJoinOperatorSpec(org.apache.samza.operators.spec.StreamTableJoinOperatorSpec) SendToTableOperatorSpec(org.apache.samza.operators.spec.SendToTableOperatorSpec)

Example 12 with InputOperatorSpec

use of org.apache.samza.operators.spec.InputOperatorSpec in project samza by apache.

the class TestInputOperatorImpl method testWithUnkeyedInput.

@Test
public void testWithUnkeyedInput() {
    InputOperatorImpl inputOperator = new InputOperatorImpl(new InputOperatorSpec("stream-id", null, null, null, false, "input-op-id"));
    IncomingMessageEnvelope ime = new IncomingMessageEnvelope(mock(SystemStreamPartition.class), "123", "key", "msg");
    Collection<Object> results = inputOperator.handleMessage(ime, mock(MessageCollector.class), mock(TaskCoordinator.class));
    Object result = results.iterator().next();
    assertEquals("msg", result);
}
Also used : InputOperatorSpec(org.apache.samza.operators.spec.InputOperatorSpec) IncomingMessageEnvelope(org.apache.samza.system.IncomingMessageEnvelope) MessageCollector(org.apache.samza.task.MessageCollector) TaskCoordinator(org.apache.samza.task.TaskCoordinator) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Test(org.junit.Test)

Example 13 with InputOperatorSpec

use of org.apache.samza.operators.spec.InputOperatorSpec 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"));
}
Also used : Serde(org.apache.samza.serializers.Serde) IntermediateMessageStreamImpl(org.apache.samza.operators.stream.IntermediateMessageStreamImpl) MessageStreamImpl(org.apache.samza.operators.MessageStreamImpl) OutputStreamImpl(org.apache.samza.operators.spec.OutputStreamImpl) ArrayList(java.util.ArrayList) RelDataType(org.apache.calcite.rel.type.RelDataType) RexCall(org.apache.calcite.rex.RexCall) ContainerContext(org.apache.samza.context.ContainerContext) StreamApplicationDescriptorImpl(org.apache.samza.application.descriptors.StreamApplicationDescriptorImpl) RocksDbTableDescriptor(org.apache.samza.storage.kv.descriptors.RocksDbTableDescriptor) MessageStream(org.apache.samza.operators.MessageStream) StreamTableJoinFunction(org.apache.samza.operators.functions.StreamTableJoinFunction) SqlIOConfig(org.apache.samza.sql.interfaces.SqlIOConfig) SamzaSqlExecutionContext(org.apache.samza.sql.data.SamzaSqlExecutionContext) Context(org.apache.samza.context.Context) ContainerContext(org.apache.samza.context.ContainerContext) InputOperatorSpec(org.apache.samza.operators.spec.InputOperatorSpec) SamzaSqlApplicationConfig(org.apache.samza.sql.runner.SamzaSqlApplicationConfig) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) IntermediateMessageStreamImpl(org.apache.samza.operators.stream.IntermediateMessageStreamImpl) RexToJavaCompiler(org.apache.samza.sql.data.RexToJavaCompiler) TableDescriptor(org.apache.samza.table.descriptors.TableDescriptor) RocksDbTableDescriptor(org.apache.samza.storage.kv.descriptors.RocksDbTableDescriptor) RemoteTableDescriptor(org.apache.samza.table.descriptors.RemoteTableDescriptor) TestMetricsRegistryImpl(org.apache.samza.sql.util.TestMetricsRegistryImpl) RelNode(org.apache.calcite.rel.RelNode) Expression(org.apache.samza.sql.data.Expression) LogicalJoin(org.apache.calcite.rel.logical.LogicalJoin) RexInputRef(org.apache.calcite.rex.RexInputRef) SamzaSqlExecutionContext(org.apache.samza.sql.data.SamzaSqlExecutionContext) RelOptTable(org.apache.calcite.plan.RelOptTable) StreamTableJoinOperatorSpec(org.apache.samza.operators.spec.StreamTableJoinOperatorSpec) SamzaSqlRelMessage(org.apache.samza.sql.data.SamzaSqlRelMessage) RexNode(org.apache.calcite.rex.RexNode)

Example 14 with InputOperatorSpec

use of org.apache.samza.operators.spec.InputOperatorSpec in project samza by apache.

the class TestInputOperatorImpl method testWithInputTransformer.

@Test
public void testWithInputTransformer() {
    InputOperatorSpec inputOpSpec = new InputOperatorSpec("stream-id", null, null, IncomingMessageEnvelope::getOffset, true, "input-op-id");
    InputOperatorImpl inputOperator = new InputOperatorImpl(inputOpSpec);
    IncomingMessageEnvelope ime = new IncomingMessageEnvelope(mock(SystemStreamPartition.class), "123", "key", "msg");
    Collection<Object> results = inputOperator.handleMessage(ime, mock(MessageCollector.class), mock(TaskCoordinator.class));
    Object result = results.iterator().next();
    assertEquals("123", result);
}
Also used : InputOperatorSpec(org.apache.samza.operators.spec.InputOperatorSpec) IncomingMessageEnvelope(org.apache.samza.system.IncomingMessageEnvelope) MessageCollector(org.apache.samza.task.MessageCollector) TaskCoordinator(org.apache.samza.task.TaskCoordinator) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Test(org.junit.Test)

Example 15 with InputOperatorSpec

use of org.apache.samza.operators.spec.InputOperatorSpec in project samza by apache.

the class TestInputOperatorImpl method testWithFilteringInputTransformer.

@Test
public void testWithFilteringInputTransformer() {
    InputOperatorSpec inputOpSpec = new InputOperatorSpec("stream-id", null, null, (ime) -> null, true, "input-op-id");
    InputOperatorImpl inputOperator = new InputOperatorImpl(inputOpSpec);
    IncomingMessageEnvelope ime = new IncomingMessageEnvelope(mock(SystemStreamPartition.class), "123", "key", "msg");
    Collection<Object> results = inputOperator.handleMessage(ime, mock(MessageCollector.class), mock(TaskCoordinator.class));
    assertTrue("Transformer doesn't return any record. Expected an empty collection", results.isEmpty());
}
Also used : InputOperatorSpec(org.apache.samza.operators.spec.InputOperatorSpec) IncomingMessageEnvelope(org.apache.samza.system.IncomingMessageEnvelope) MessageCollector(org.apache.samza.task.MessageCollector) TaskCoordinator(org.apache.samza.task.TaskCoordinator) SystemStreamPartition(org.apache.samza.system.SystemStreamPartition) Test(org.junit.Test)

Aggregations

InputOperatorSpec (org.apache.samza.operators.spec.InputOperatorSpec)23 Test (org.junit.Test)13 Serde (org.apache.samza.serializers.Serde)11 KVSerde (org.apache.samza.serializers.KVSerde)10 OperatorSpec (org.apache.samza.operators.spec.OperatorSpec)9 NoOpSerde (org.apache.samza.serializers.NoOpSerde)9 IntegerSerde (org.apache.samza.serializers.IntegerSerde)8 GenericSystemDescriptor (org.apache.samza.system.descriptors.GenericSystemDescriptor)7 GenericInputDescriptor (org.apache.samza.system.descriptors.GenericInputDescriptor)6 ArrayList (java.util.ArrayList)5 OutputStreamImpl (org.apache.samza.operators.spec.OutputStreamImpl)5 StreamTableJoinOperatorSpec (org.apache.samza.operators.spec.StreamTableJoinOperatorSpec)5 IntermediateMessageStreamImpl (org.apache.samza.operators.stream.IntermediateMessageStreamImpl)5 Config (org.apache.samza.config.Config)4 HashSet (java.util.HashSet)3 ApplicationConfig (org.apache.samza.config.ApplicationConfig)3 MapConfig (org.apache.samza.config.MapConfig)3 TestMessageEnvelope (org.apache.samza.operators.data.TestMessageEnvelope)3 IncomingMessageEnvelope (org.apache.samza.system.IncomingMessageEnvelope)3 SystemStreamPartition (org.apache.samza.system.SystemStreamPartition)3