Examples with GeneratedRecordEqualiser org.apache.flink.table.runtime.generated.GeneratedRecordEqualiser used on opensource projects

Example 1 with GeneratedRecordEqualiser

use of org.apache.flink.table.runtime.generated.GeneratedRecordEqualiser in project flink by apache.

the class StreamExecGlobalGroupAggregate method translateToPlanInternal.

@SuppressWarnings("unchecked")
@Override
protected Transformation<RowData> translateToPlanInternal(PlannerBase planner, ExecNodeConfig config) {
    if (grouping.length > 0 && config.getStateRetentionTime() < 0) {
        LOG.warn("No state retention interval configured for a query which accumulates state. " + "Please provide a query configuration with valid retention interval to prevent excessive " + "state size. You may specify a retention time of 0 to not clean up the state.");
    }
    final ExecEdge inputEdge = getInputEdges().get(0);
    final Transformation<RowData> inputTransform = (Transformation<RowData>) inputEdge.translateToPlan(planner);
    final RowType inputRowType = (RowType) inputEdge.getOutputType();
    final AggregateInfoList localAggInfoList = AggregateUtil.transformToStreamAggregateInfoList(localAggInputRowType, JavaScalaConversionUtil.toScala(Arrays.asList(aggCalls)), aggCallNeedRetractions, needRetraction, JavaScalaConversionUtil.toScala(Optional.ofNullable(indexOfCountStar)), // isStateBackendDataViews
    false, // needDistinctInfo
    true);
    final AggregateInfoList globalAggInfoList = AggregateUtil.transformToStreamAggregateInfoList(localAggInputRowType, JavaScalaConversionUtil.toScala(Arrays.asList(aggCalls)), aggCallNeedRetractions, needRetraction, JavaScalaConversionUtil.toScala(Optional.ofNullable(indexOfCountStar)), // isStateBackendDataViews
    true, // needDistinctInfo
    true);
    final GeneratedAggsHandleFunction localAggsHandler = generateAggsHandler("LocalGroupAggsHandler", localAggInfoList, grouping.length, localAggInfoList.getAccTypes(), config, planner.getRelBuilder());
    final GeneratedAggsHandleFunction globalAggsHandler = generateAggsHandler("GlobalGroupAggsHandler", globalAggInfoList, // mergedAccOffset
    0, localAggInfoList.getAccTypes(), config, planner.getRelBuilder());
    final int indexOfCountStar = globalAggInfoList.getIndexOfCountStar();
    final LogicalType[] globalAccTypes = Arrays.stream(globalAggInfoList.getAccTypes()).map(LogicalTypeDataTypeConverter::fromDataTypeToLogicalType).toArray(LogicalType[]::new);
    final LogicalType[] globalAggValueTypes = Arrays.stream(globalAggInfoList.getActualValueTypes()).map(LogicalTypeDataTypeConverter::fromDataTypeToLogicalType).toArray(LogicalType[]::new);
    final GeneratedRecordEqualiser recordEqualiser = new EqualiserCodeGenerator(globalAggValueTypes).generateRecordEqualiser("GroupAggValueEqualiser");
    final OneInputStreamOperator<RowData, RowData> operator;
    final boolean isMiniBatchEnabled = config.get(ExecutionConfigOptions.TABLE_EXEC_MINIBATCH_ENABLED);
    if (isMiniBatchEnabled) {
        MiniBatchGlobalGroupAggFunction aggFunction = new MiniBatchGlobalGroupAggFunction(localAggsHandler, globalAggsHandler, recordEqualiser, globalAccTypes, indexOfCountStar, generateUpdateBefore, config.getStateRetentionTime());
        operator = new KeyedMapBundleOperator<>(aggFunction, AggregateUtil.createMiniBatchTrigger(config));
    } else {
        throw new TableException("Local-Global optimization is only worked in miniBatch mode");
    }
    // partitioned aggregation
    final OneInputTransformation<RowData, RowData> transform = ExecNodeUtil.createOneInputTransformation(inputTransform, createTransformationMeta(GLOBAL_GROUP_AGGREGATE_TRANSFORMATION, config), operator, InternalTypeInfo.of(getOutputType()), inputTransform.getParallelism());
    // set KeyType and Selector for state
    final RowDataKeySelector selector = KeySelectorUtil.getRowDataSelector(grouping, InternalTypeInfo.of(inputRowType));
    transform.setStateKeySelector(selector);
    transform.setStateKeyType(selector.getProducedType());
    return transform;
}

Example 2 with GeneratedRecordEqualiser

use of org.apache.flink.table.runtime.generated.GeneratedRecordEqualiser in project flink by apache.

the class StreamExecGroupAggregate method translateToPlanInternal.

@SuppressWarnings("unchecked")
@Override
protected Transformation<RowData> translateToPlanInternal(PlannerBase planner, ExecNodeConfig config) {
    if (grouping.length > 0 && config.getStateRetentionTime() < 0) {
        LOG.warn("No state retention interval configured for a query which accumulates state. " + "Please provide a query configuration with valid retention interval to prevent excessive " + "state size. You may specify a retention time of 0 to not clean up the state.");
    }
    final ExecEdge inputEdge = getInputEdges().get(0);
    final Transformation<RowData> inputTransform = (Transformation<RowData>) inputEdge.translateToPlan(planner);
    final RowType inputRowType = (RowType) inputEdge.getOutputType();
    final AggsHandlerCodeGenerator generator = new AggsHandlerCodeGenerator(new CodeGeneratorContext(config.getTableConfig()), planner.getRelBuilder(), JavaScalaConversionUtil.toScala(inputRowType.getChildren()), // TODO: but other operators do not copy this input field.....
    true).needAccumulate();
    if (needRetraction) {
        generator.needRetract();
    }
    final AggregateInfoList aggInfoList = AggregateUtil.transformToStreamAggregateInfoList(inputRowType, JavaScalaConversionUtil.toScala(Arrays.asList(aggCalls)), aggCallNeedRetractions, needRetraction, true, true);
    final GeneratedAggsHandleFunction aggsHandler = generator.generateAggsHandler("GroupAggsHandler", aggInfoList);
    final LogicalType[] accTypes = Arrays.stream(aggInfoList.getAccTypes()).map(LogicalTypeDataTypeConverter::fromDataTypeToLogicalType).toArray(LogicalType[]::new);
    final LogicalType[] aggValueTypes = Arrays.stream(aggInfoList.getActualValueTypes()).map(LogicalTypeDataTypeConverter::fromDataTypeToLogicalType).toArray(LogicalType[]::new);
    final GeneratedRecordEqualiser recordEqualiser = new EqualiserCodeGenerator(aggValueTypes).generateRecordEqualiser("GroupAggValueEqualiser");
    final int inputCountIndex = aggInfoList.getIndexOfCountStar();
    final boolean isMiniBatchEnabled = config.get(ExecutionConfigOptions.TABLE_EXEC_MINIBATCH_ENABLED);
    final OneInputStreamOperator<RowData, RowData> operator;
    if (isMiniBatchEnabled) {
        MiniBatchGroupAggFunction aggFunction = new MiniBatchGroupAggFunction(aggsHandler, recordEqualiser, accTypes, inputRowType, inputCountIndex, generateUpdateBefore, config.getStateRetentionTime());
        operator = new KeyedMapBundleOperator<>(aggFunction, AggregateUtil.createMiniBatchTrigger(config));
    } else {
        GroupAggFunction aggFunction = new GroupAggFunction(aggsHandler, recordEqualiser, accTypes, inputCountIndex, generateUpdateBefore, config.getStateRetentionTime());
        operator = new KeyedProcessOperator<>(aggFunction);
    }
    // partitioned aggregation
    final OneInputTransformation<RowData, RowData> transform = ExecNodeUtil.createOneInputTransformation(inputTransform, createTransformationMeta(GROUP_AGGREGATE_TRANSFORMATION, config), operator, InternalTypeInfo.of(getOutputType()), inputTransform.getParallelism());
    // set KeyType and Selector for state
    final RowDataKeySelector selector = KeySelectorUtil.getRowDataSelector(grouping, InternalTypeInfo.of(inputRowType));
    transform.setStateKeySelector(selector);
    transform.setStateKeyType(selector.getProducedType());
    return transform;
}

Example 3 with GeneratedRecordEqualiser

use of org.apache.flink.table.runtime.generated.GeneratedRecordEqualiser in project flink by apache.

the class StreamExecGroupWindowAggregate method translateToPlanInternal.

@SuppressWarnings("unchecked")
@Override
protected Transformation<RowData> translateToPlanInternal(PlannerBase planner, ExecNodeConfig config) {
    final boolean isCountWindow;
    if (window instanceof TumblingGroupWindow) {
        isCountWindow = hasRowIntervalType(((TumblingGroupWindow) window).size());
    } else if (window instanceof SlidingGroupWindow) {
        isCountWindow = hasRowIntervalType(((SlidingGroupWindow) window).size());
    } else {
        isCountWindow = false;
    }
    if (isCountWindow && grouping.length > 0 && config.getStateRetentionTime() < 0) {
        LOGGER.warn("No state retention interval configured for a query which accumulates state. " + "Please provide a query configuration with valid retention interval to prevent " + "excessive state size. You may specify a retention time of 0 to not clean up the state.");
    }
    final ExecEdge inputEdge = getInputEdges().get(0);
    final Transformation<RowData> inputTransform = (Transformation<RowData>) inputEdge.translateToPlan(planner);
    final RowType inputRowType = (RowType) inputEdge.getOutputType();
    final int inputTimeFieldIndex;
    if (isRowtimeAttribute(window.timeAttribute())) {
        inputTimeFieldIndex = timeFieldIndex(FlinkTypeFactory.INSTANCE().buildRelNodeRowType(inputRowType), planner.getRelBuilder(), window.timeAttribute());
        if (inputTimeFieldIndex < 0) {
            throw new TableException("Group window must defined on a time attribute, " + "but the time attribute can't be found.\n" + "This should never happen. Please file an issue.");
        }
    } else {
        inputTimeFieldIndex = -1;
    }
    final ZoneId shiftTimeZone = TimeWindowUtil.getShiftTimeZone(window.timeAttribute().getOutputDataType().getLogicalType(), config.getLocalTimeZone());
    final boolean[] aggCallNeedRetractions = new boolean[aggCalls.length];
    Arrays.fill(aggCallNeedRetractions, needRetraction);
    final AggregateInfoList aggInfoList = transformToStreamAggregateInfoList(inputRowType, JavaScalaConversionUtil.toScala(Arrays.asList(aggCalls)), aggCallNeedRetractions, needRetraction, // isStateBackendDataViews
    true, // needDistinctInfo
    true);
    final GeneratedClass<?> aggCodeGenerator = createAggsHandler(aggInfoList, config, planner.getRelBuilder(), inputRowType.getChildren(), shiftTimeZone);
    final LogicalType[] aggResultTypes = extractLogicalTypes(aggInfoList.getActualValueTypes());
    final LogicalType[] windowPropertyTypes = Arrays.stream(namedWindowProperties).map(p -> p.getProperty().getResultType()).toArray(LogicalType[]::new);
    final EqualiserCodeGenerator generator = new EqualiserCodeGenerator(ArrayUtils.addAll(aggResultTypes, windowPropertyTypes));
    final GeneratedRecordEqualiser equaliser = generator.generateRecordEqualiser("WindowValueEqualiser");
    final LogicalType[] aggValueTypes = extractLogicalTypes(aggInfoList.getActualValueTypes());
    final LogicalType[] accTypes = extractLogicalTypes(aggInfoList.getAccTypes());
    final int inputCountIndex = aggInfoList.getIndexOfCountStar();
    final WindowOperator<?, ?> operator = createWindowOperator(config, aggCodeGenerator, equaliser, accTypes, windowPropertyTypes, aggValueTypes, inputRowType.getChildren().toArray(new LogicalType[0]), inputTimeFieldIndex, shiftTimeZone, inputCountIndex);
    final OneInputTransformation<RowData, RowData> transform = ExecNodeUtil.createOneInputTransformation(inputTransform, createTransformationMeta(GROUP_WINDOW_AGGREGATE_TRANSFORMATION, config), operator, InternalTypeInfo.of(getOutputType()), inputTransform.getParallelism());
    // set KeyType and Selector for state
    final RowDataKeySelector selector = KeySelectorUtil.getRowDataSelector(grouping, InternalTypeInfo.of(inputRowType));
    transform.setStateKeySelector(selector);
    transform.setStateKeyType(selector.getProducedType());
    return transform;
}

Example 4 with GeneratedRecordEqualiser

use of org.apache.flink.table.runtime.generated.GeneratedRecordEqualiser in project flink by apache.

the class CommonExecSink method applyUpsertMaterialize.

private Transformation<RowData> applyUpsertMaterialize(Transformation<RowData> inputTransform, int[] primaryKeys, int sinkParallelism, ReadableConfig config, RowType physicalRowType) {
    GeneratedRecordEqualiser equaliser = new EqualiserCodeGenerator(physicalRowType).generateRecordEqualiser("SinkMaterializeEqualiser");
    SinkUpsertMaterializer operator = new SinkUpsertMaterializer(StateConfigUtil.createTtlConfig(config.get(ExecutionConfigOptions.IDLE_STATE_RETENTION).toMillis()), InternalSerializers.create(physicalRowType), equaliser);
    final String[] fieldNames = physicalRowType.getFieldNames().toArray(new String[0]);
    final List<String> pkFieldNames = Arrays.stream(primaryKeys).mapToObj(idx -> fieldNames[idx]).collect(Collectors.toList());
    OneInputTransformation<RowData, RowData> materializeTransform = ExecNodeUtil.createOneInputTransformation(inputTransform, createTransformationMeta(UPSERT_MATERIALIZE_TRANSFORMATION, String.format("SinkMaterializer(pk=[%s])", String.join(", ", pkFieldNames)), "SinkMaterializer", config), operator, inputTransform.getOutputType(), sinkParallelism);
    RowDataKeySelector keySelector = KeySelectorUtil.getRowDataSelector(primaryKeys, InternalTypeInfo.of(physicalRowType));
    materializeTransform.setStateKeySelector(keySelector);
    materializeTransform.setStateKeyType(keySelector.getProducedType());
    return materializeTransform;
}

Example 5 with GeneratedRecordEqualiser

use of org.apache.flink.table.runtime.generated.GeneratedRecordEqualiser in project flink by apache.

the class StreamExecChangelogNormalize method translateToPlanInternal.

@SuppressWarnings("unchecked")
@Override
protected Transformation<RowData> translateToPlanInternal(PlannerBase planner, ExecNodeConfig config) {
    final ExecEdge inputEdge = getInputEdges().get(0);
    final Transformation<RowData> inputTransform = (Transformation<RowData>) inputEdge.translateToPlan(planner);
    final InternalTypeInfo<RowData> rowTypeInfo = (InternalTypeInfo<RowData>) inputTransform.getOutputType();
    final OneInputStreamOperator<RowData, RowData> operator;
    final long stateIdleTime = config.getStateRetentionTime();
    final boolean isMiniBatchEnabled = config.get(ExecutionConfigOptions.TABLE_EXEC_MINIBATCH_ENABLED);
    GeneratedRecordEqualiser generatedEqualiser = new EqualiserCodeGenerator(rowTypeInfo.toRowType()).generateRecordEqualiser("DeduplicateRowEqualiser");
    if (isMiniBatchEnabled) {
        TypeSerializer<RowData> rowSerializer = rowTypeInfo.createSerializer(planner.getExecEnv().getConfig());
        ProcTimeMiniBatchDeduplicateKeepLastRowFunction processFunction = new ProcTimeMiniBatchDeduplicateKeepLastRowFunction(rowTypeInfo, rowSerializer, stateIdleTime, generateUpdateBefore, // generateInsert
        true, // inputInsertOnly
        false, generatedEqualiser);
        CountBundleTrigger<RowData> trigger = AggregateUtil.createMiniBatchTrigger(config);
        operator = new KeyedMapBundleOperator<>(processFunction, trigger);
    } else {
        ProcTimeDeduplicateKeepLastRowFunction processFunction = new ProcTimeDeduplicateKeepLastRowFunction(rowTypeInfo, stateIdleTime, generateUpdateBefore, // generateInsert
        true, // inputInsertOnly
        false, generatedEqualiser);
        operator = new KeyedProcessOperator<>(processFunction);
    }
    final OneInputTransformation<RowData, RowData> transform = ExecNodeUtil.createOneInputTransformation(inputTransform, createTransformationMeta(CHANGELOG_NORMALIZE_TRANSFORMATION, config), operator, rowTypeInfo, inputTransform.getParallelism());
    final RowDataKeySelector selector = KeySelectorUtil.getRowDataSelector(uniqueKeys, rowTypeInfo);
    transform.setStateKeySelector(selector);
    transform.setStateKeyType(selector.getProducedType());
    return transform;
}