Examples with TypeSerializer org.apache.flink.api.common.typeutils.TypeSerializer used on opensource projects

Example 11 with TypeSerializer

use of org.apache.flink.api.common.typeutils.TypeSerializer in project flink by apache.

the class AllWindowedStream method apply.

/**
	 * Applies the given window function to each window. The window function is called for each
	 * evaluation of the window for each key individually. The output of the window function is
	 * interpreted as a regular non-windowed stream.
	 *
	 * <p>
	 * Arriving data is incrementally aggregated using the given reducer.
	 *
	 * @param reduceFunction The reduce function that is used for incremental aggregation.
	 * @param function The window function.
	 * @param resultType Type information for the result type of the window function
	 * @return The data stream that is the result of applying the window function to the window.
	 *
	 * @deprecated Use {@link #reduce(ReduceFunction, AllWindowFunction, TypeInformation)} instead.
	 */
@Deprecated
public <R> SingleOutputStreamOperator<R> apply(ReduceFunction<T> reduceFunction, AllWindowFunction<T, R, W> function, TypeInformation<R> resultType) {
    if (reduceFunction instanceof RichFunction) {
        throw new UnsupportedOperationException("ReduceFunction of apply can not be a RichFunction.");
    }
    //clean the closures
    function = input.getExecutionEnvironment().clean(function);
    reduceFunction = input.getExecutionEnvironment().clean(reduceFunction);
    String callLocation = Utils.getCallLocationName();
    String udfName = "AllWindowedStream." + callLocation;
    String opName;
    KeySelector<T, Byte> keySel = input.getKeySelector();
    OneInputStreamOperator<T, R> operator;
    if (evictor != null) {
        @SuppressWarnings({ "unchecked", "rawtypes" }) TypeSerializer<StreamRecord<T>> streamRecordSerializer = (TypeSerializer<StreamRecord<T>>) new StreamElementSerializer(input.getType().createSerializer(getExecutionEnvironment().getConfig()));
        ListStateDescriptor<StreamRecord<T>> stateDesc = new ListStateDescriptor<>("window-contents", streamRecordSerializer);
        opName = "TriggerWindow(" + windowAssigner + ", " + stateDesc + ", " + trigger + ", " + evictor + ", " + udfName + ")";
        operator = new EvictingWindowOperator<>(windowAssigner, windowAssigner.getWindowSerializer(getExecutionEnvironment().getConfig()), keySel, input.getKeyType().createSerializer(getExecutionEnvironment().getConfig()), stateDesc, new InternalIterableAllWindowFunction<>(new ReduceApplyAllWindowFunction<>(reduceFunction, function)), trigger, evictor, allowedLateness, lateDataOutputTag);
    } else {
        ReducingStateDescriptor<T> stateDesc = new ReducingStateDescriptor<>("window-contents", reduceFunction, input.getType().createSerializer(getExecutionEnvironment().getConfig()));
        opName = "TriggerWindow(" + windowAssigner + ", " + stateDesc + ", " + trigger + ", " + udfName + ")";
        operator = new WindowOperator<>(windowAssigner, windowAssigner.getWindowSerializer(getExecutionEnvironment().getConfig()), keySel, input.getKeyType().createSerializer(getExecutionEnvironment().getConfig()), stateDesc, new InternalSingleValueAllWindowFunction<>(function), trigger, allowedLateness, lateDataOutputTag);
    }
    return input.transform(opName, resultType, operator).forceNonParallel();
}

Example 12 with TypeSerializer

use of org.apache.flink.api.common.typeutils.TypeSerializer in project flink by apache.

the class WindowedStream method aggregate.

/**
	 * Applies the given window function to each window. The window function is called for each
	 * evaluation of the window for each key individually. The output of the window function is
	 * interpreted as a regular non-windowed stream.
	 *
	 * <p>Arriving data is incrementally aggregated using the given aggregate function. This means
	 * that the window function typically has only a single value to process when called.
	 *
	 * @param aggregateFunction The aggregation function that is used for incremental aggregation.
	 * @param windowFunction The window function.
	 * @param accumulatorType Type information for the internal accumulator type of the aggregation function
	 * @param resultType Type information for the result type of the window function
	 *
	 * @return The data stream that is the result of applying the window function to the window.
	 *
	 * @param <ACC> The type of the AggregateFunction's accumulator
	 * @param <V> The type of AggregateFunction's result, and the WindowFunction's input
	 * @param <R> The type of the elements in the resulting stream, equal to the
	 *            WindowFunction's result type
	 */
@PublicEvolving
public <ACC, V, R> SingleOutputStreamOperator<R> aggregate(AggregateFunction<T, ACC, V> aggregateFunction, ProcessWindowFunction<V, R, K, W> windowFunction, TypeInformation<ACC> accumulatorType, TypeInformation<V> aggregateResultType, TypeInformation<R> resultType) {
    checkNotNull(aggregateFunction, "aggregateFunction");
    checkNotNull(windowFunction, "windowFunction");
    checkNotNull(accumulatorType, "accumulatorType");
    checkNotNull(aggregateResultType, "aggregateResultType");
    checkNotNull(resultType, "resultType");
    if (aggregateFunction instanceof RichFunction) {
        throw new UnsupportedOperationException("This aggregate function cannot be a RichFunction.");
    }
    //clean the closures
    windowFunction = input.getExecutionEnvironment().clean(windowFunction);
    aggregateFunction = input.getExecutionEnvironment().clean(aggregateFunction);
    String callLocation = Utils.getCallLocationName();
    String udfName = "WindowedStream." + callLocation;
    String opName;
    KeySelector<T, K> keySel = input.getKeySelector();
    OneInputStreamOperator<T, R> operator;
    if (evictor != null) {
        @SuppressWarnings({ "unchecked", "rawtypes" }) TypeSerializer<StreamRecord<T>> streamRecordSerializer = (TypeSerializer<StreamRecord<T>>) new StreamElementSerializer(input.getType().createSerializer(getExecutionEnvironment().getConfig()));
        ListStateDescriptor<StreamRecord<T>> stateDesc = new ListStateDescriptor<>("window-contents", streamRecordSerializer);
        opName = "TriggerWindow(" + windowAssigner + ", " + stateDesc + ", " + trigger + ", " + evictor + ", " + udfName + ")";
        operator = new EvictingWindowOperator<>(windowAssigner, windowAssigner.getWindowSerializer(getExecutionEnvironment().getConfig()), keySel, input.getKeyType().createSerializer(getExecutionEnvironment().getConfig()), stateDesc, new InternalAggregateProcessWindowFunction<>(aggregateFunction, windowFunction), trigger, evictor, allowedLateness, lateDataOutputTag);
    } else {
        AggregatingStateDescriptor<T, ACC, V> stateDesc = new AggregatingStateDescriptor<>("window-contents", aggregateFunction, accumulatorType.createSerializer(getExecutionEnvironment().getConfig()));
        opName = "TriggerWindow(" + windowAssigner + ", " + stateDesc + ", " + trigger + ", " + udfName + ")";
        operator = new WindowOperator<>(windowAssigner, windowAssigner.getWindowSerializer(getExecutionEnvironment().getConfig()), keySel, input.getKeyType().createSerializer(getExecutionEnvironment().getConfig()), stateDesc, new InternalSingleValueProcessWindowFunction<>(windowFunction), trigger, allowedLateness, lateDataOutputTag);
    }
    return input.transform(opName, resultType, operator);
}

Example 13 with TypeSerializer

use of org.apache.flink.api.common.typeutils.TypeSerializer in project flink by apache.

the class WindowedStream method fold.

/**
	 * Applies the given window function to each window. The window function is called for each
	 * evaluation of the window for each key individually. The output of the window function is
	 * interpreted as a regular non-windowed stream.
	 *
	 * <p>
	 * Arriving data is incrementally aggregated using the given fold function.
	 *
	 * @param initialValue The initial value of the fold.
	 * @param foldFunction The fold function that is used for incremental aggregation.
	 * @param function The window function.
	 * @param foldAccumulatorType Type information for the result type of the fold function
	 * @param resultType Type information for the result type of the window function
	 * @return The data stream that is the result of applying the window function to the window.
	 */
@PublicEvolving
public <ACC, R> SingleOutputStreamOperator<R> fold(ACC initialValue, FoldFunction<T, ACC> foldFunction, WindowFunction<ACC, R, K, W> function, TypeInformation<ACC> foldAccumulatorType, TypeInformation<R> resultType) {
    if (foldFunction instanceof RichFunction) {
        throw new UnsupportedOperationException("FoldFunction of fold can not be a RichFunction.");
    }
    if (windowAssigner instanceof MergingWindowAssigner) {
        throw new UnsupportedOperationException("Fold cannot be used with a merging WindowAssigner.");
    }
    if (windowAssigner instanceof BaseAlignedWindowAssigner) {
        throw new UnsupportedOperationException("Fold cannot be used with a " + windowAssigner.getClass().getSimpleName() + " assigner.");
    }
    //clean the closures
    function = input.getExecutionEnvironment().clean(function);
    foldFunction = input.getExecutionEnvironment().clean(foldFunction);
    String callLocation = Utils.getCallLocationName();
    String udfName = "WindowedStream." + callLocation;
    String opName;
    KeySelector<T, K> keySel = input.getKeySelector();
    OneInputStreamOperator<T, R> operator;
    if (evictor != null) {
        @SuppressWarnings({ "unchecked", "rawtypes" }) TypeSerializer<StreamRecord<T>> streamRecordSerializer = (TypeSerializer<StreamRecord<T>>) new StreamElementSerializer(input.getType().createSerializer(getExecutionEnvironment().getConfig()));
        ListStateDescriptor<StreamRecord<T>> stateDesc = new ListStateDescriptor<>("window-contents", streamRecordSerializer);
        opName = "TriggerWindow(" + windowAssigner + ", " + stateDesc + ", " + trigger + ", " + evictor + ", " + udfName + ")";
        operator = new EvictingWindowOperator<>(windowAssigner, windowAssigner.getWindowSerializer(getExecutionEnvironment().getConfig()), keySel, input.getKeyType().createSerializer(getExecutionEnvironment().getConfig()), stateDesc, new InternalIterableWindowFunction<>(new FoldApplyWindowFunction<>(initialValue, foldFunction, function, foldAccumulatorType)), trigger, evictor, allowedLateness, lateDataOutputTag);
    } else {
        FoldingStateDescriptor<T, ACC> stateDesc = new FoldingStateDescriptor<>("window-contents", initialValue, foldFunction, foldAccumulatorType.createSerializer(getExecutionEnvironment().getConfig()));
        opName = "TriggerWindow(" + windowAssigner + ", " + stateDesc + ", " + trigger + ", " + udfName + ")";
        operator = new WindowOperator<>(windowAssigner, windowAssigner.getWindowSerializer(getExecutionEnvironment().getConfig()), keySel, input.getKeyType().createSerializer(getExecutionEnvironment().getConfig()), stateDesc, new InternalSingleValueWindowFunction<>(function), trigger, allowedLateness, lateDataOutputTag);
    }
    return input.transform(opName, resultType, operator);
}

Example 14 with TypeSerializer

use of org.apache.flink.api.common.typeutils.TypeSerializer in project flink by apache.

the class EvictingWindowOperatorTest method testCountTrigger.

@Test
@SuppressWarnings("unchecked")
public void testCountTrigger() throws Exception {
    final int WINDOW_SIZE = 4;
    final int WINDOW_SLIDE = 2;
    TypeInformation<Tuple2<String, Integer>> inputType = TypeInfoParser.parse("Tuple2<String, Integer>");
    @SuppressWarnings({ "unchecked", "rawtypes" }) TypeSerializer<StreamRecord<Tuple2<String, Integer>>> streamRecordSerializer = (TypeSerializer<StreamRecord<Tuple2<String, Integer>>>) new StreamElementSerializer(inputType.createSerializer(new ExecutionConfig()));
    ListStateDescriptor<StreamRecord<Tuple2<String, Integer>>> stateDesc = new ListStateDescriptor<>("window-contents", streamRecordSerializer);
    EvictingWindowOperator<String, Tuple2<String, Integer>, Tuple2<String, Integer>, GlobalWindow> operator = new EvictingWindowOperator<>(GlobalWindows.create(), new GlobalWindow.Serializer(), new TupleKeySelector(), BasicTypeInfo.STRING_TYPE_INFO.createSerializer(new ExecutionConfig()), stateDesc, new InternalIterableWindowFunction<>(new ReduceIterableWindowFunction<String, GlobalWindow, Tuple2<String, Integer>>(new SumReducer())), CountTrigger.of(WINDOW_SLIDE), CountEvictor.of(WINDOW_SIZE), 0, null);
    OneInputStreamOperatorTestHarness<Tuple2<String, Integer>, Tuple2<String, Integer>> testHarness = new KeyedOneInputStreamOperatorTestHarness<>(operator, new TupleKeySelector(), BasicTypeInfo.STRING_TYPE_INFO);
    long initialTime = 0L;
    ConcurrentLinkedQueue<Object> expectedOutput = new ConcurrentLinkedQueue<>();
    testHarness.open();
    // The global window actually ignores these timestamps...
    // add elements out-of-order
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 3000));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 3999));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key1", 1), initialTime + 20));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key1", 1), initialTime));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key1", 1), initialTime + 999));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 1998));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 1999));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 1000));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key2", 2), Long.MAX_VALUE));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key2", 4), Long.MAX_VALUE));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key1", 2), Long.MAX_VALUE));
    TestHarnessUtil.assertOutputEqualsSorted("Output was not correct.", expectedOutput, testHarness.getOutput(), new ResultSortComparator());
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key1", 1), initialTime + 10999));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 1000));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key1", 4), Long.MAX_VALUE));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key2", 4), Long.MAX_VALUE));
    TestHarnessUtil.assertOutputEqualsSorted("Output was not correct.", expectedOutput, testHarness.getOutput(), new ResultSortComparator());
    testHarness.close();
}

Example 15 with TypeSerializer

use of org.apache.flink.api.common.typeutils.TypeSerializer in project flink by apache.

the class EvictingWindowOperatorTest method testCountTriggerWithApply.

@Test
@SuppressWarnings("unchecked")
public void testCountTriggerWithApply() throws Exception {
    AtomicInteger closeCalled = new AtomicInteger(0);
    final int WINDOW_SIZE = 4;
    final int WINDOW_SLIDE = 2;
    TypeInformation<Tuple2<String, Integer>> inputType = TypeInfoParser.parse("Tuple2<String, Integer>");
    @SuppressWarnings({ "unchecked", "rawtypes" }) TypeSerializer<StreamRecord<Tuple2<String, Integer>>> streamRecordSerializer = (TypeSerializer<StreamRecord<Tuple2<String, Integer>>>) new StreamElementSerializer(inputType.createSerializer(new ExecutionConfig()));
    ListStateDescriptor<StreamRecord<Tuple2<String, Integer>>> stateDesc = new ListStateDescriptor<>("window-contents", streamRecordSerializer);
    EvictingWindowOperator<String, Tuple2<String, Integer>, Tuple2<String, Integer>, GlobalWindow> operator = new EvictingWindowOperator<>(GlobalWindows.create(), new GlobalWindow.Serializer(), new TupleKeySelector(), BasicTypeInfo.STRING_TYPE_INFO.createSerializer(new ExecutionConfig()), stateDesc, new InternalIterableWindowFunction<>(new RichSumReducer<GlobalWindow>(closeCalled)), CountTrigger.of(WINDOW_SLIDE), CountEvictor.of(WINDOW_SIZE), 0, null);
    OneInputStreamOperatorTestHarness<Tuple2<String, Integer>, Tuple2<String, Integer>> testHarness = new KeyedOneInputStreamOperatorTestHarness<>(operator, new TupleKeySelector(), BasicTypeInfo.STRING_TYPE_INFO);
    long initialTime = 0L;
    ConcurrentLinkedQueue<Object> expectedOutput = new ConcurrentLinkedQueue<>();
    testHarness.open();
    // The global window actually ignores these timestamps...
    // add elements out-of-order
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 3000));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 3999));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key1", 1), initialTime + 20));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key1", 1), initialTime));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key1", 1), initialTime + 999));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 1998));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 1999));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 1000));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key2", 2), Long.MAX_VALUE));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key2", 4), Long.MAX_VALUE));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key1", 2), Long.MAX_VALUE));
    TestHarnessUtil.assertOutputEqualsSorted("Output was not correct.", expectedOutput, testHarness.getOutput(), new ResultSortComparator());
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key1", 1), initialTime + 10999));
    testHarness.processElement(new StreamRecord<>(new Tuple2<>("key2", 1), initialTime + 1000));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key1", 4), Long.MAX_VALUE));
    expectedOutput.add(new StreamRecord<>(new Tuple2<>("key2", 4), Long.MAX_VALUE));
    TestHarnessUtil.assertOutputEqualsSorted("Output was not correct.", expectedOutput, testHarness.getOutput(), new ResultSortComparator());
    testHarness.close();
    Assert.assertEquals("Close was not called.", 1, closeCalled.get());
}