use of org.apache.flink.streaming.api.datastream.MultipleConnectedStreams in project flink by apache.
the class StreamGraphGeneratorBatchExecutionTest method testInputSelectableMultiInputTransformation.
@Test
public void testInputSelectableMultiInputTransformation() {
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
DataStreamSource<Integer> elements1 = env.fromElements(1, 2);
DataStreamSource<Integer> elements2 = env.fromElements(1, 2);
DataStreamSource<Integer> elements3 = env.fromElements(1, 2);
MultipleInputOperatorFactory selectableOperator = new MultipleInputOperatorFactory(3, true);
KeyedMultipleInputTransformation<Integer> multipleInputTransformation = new KeyedMultipleInputTransformation<>("operator", selectableOperator, BasicTypeInfo.INT_TYPE_INFO, 1, BasicTypeInfo.INT_TYPE_INFO);
multipleInputTransformation.addInput(elements1.getTransformation(), e -> e);
multipleInputTransformation.addInput(elements2.getTransformation(), e -> e);
multipleInputTransformation.addInput(elements3.getTransformation(), e -> e);
DataStreamSink<Integer> sink = new MultipleConnectedStreams(env).transform(multipleInputTransformation).addSink(new DiscardingSink<>());
expectedException.expect(IllegalStateException.class);
expectedException.expectMessage("Batch state backend and sorting inputs are not supported in graphs with an InputSelectable operator.");
getStreamGraphInBatchMode(sink);
}
use of org.apache.flink.streaming.api.datastream.MultipleConnectedStreams in project flink by apache.
the class StreamGraphGeneratorBatchExecutionTest method testMultiInputTransformation.
@Test
public void testMultiInputTransformation() {
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
DataStreamSource<Integer> elements1 = env.fromElements(1, 2);
DataStreamSource<Integer> elements2 = env.fromElements(1, 2);
DataStreamSource<Integer> elements3 = env.fromElements(1, 2);
MultipleInputOperatorFactory selectableOperator = new MultipleInputOperatorFactory(3, false);
KeyedMultipleInputTransformation<Integer> multipleInputTransformation = new KeyedMultipleInputTransformation<>("operator", selectableOperator, BasicTypeInfo.INT_TYPE_INFO, 1, BasicTypeInfo.INT_TYPE_INFO);
multipleInputTransformation.addInput(elements1.getTransformation(), e -> e);
multipleInputTransformation.addInput(elements2.getTransformation(), e -> e);
multipleInputTransformation.addInput(elements3.getTransformation(), e -> e);
DataStreamSink<Integer> sink = new MultipleConnectedStreams(env).transform(multipleInputTransformation).addSink(new DiscardingSink<>());
StreamGraph graph = getStreamGraphInBatchMode(sink);
StreamNode operatorNode = graph.getStreamNode(multipleInputTransformation.getId());
assertThat(operatorNode.getInputRequirements().get(0), equalTo(StreamConfig.InputRequirement.SORTED));
assertThat(operatorNode.getInputRequirements().get(1), equalTo(StreamConfig.InputRequirement.SORTED));
assertThat(operatorNode.getOperatorFactory().getChainingStrategy(), equalTo(ChainingStrategy.HEAD));
assertThat(graph.getStateBackend(), instanceOf(BatchExecutionStateBackend.class));
// the provider is passed as a lambda therefore we cannot assert the class of the provider
assertThat(graph.getTimerServiceProvider(), notNullValue());
}
use of org.apache.flink.streaming.api.datastream.MultipleConnectedStreams in project flink by apache.
the class DataStreamBatchExecutionITCase method batchMixedKeyedAndNonKeyedMultiInputOperator.
@Test
public void batchMixedKeyedAndNonKeyedMultiInputOperator() throws Exception {
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setRuntimeMode(RuntimeExecutionMode.BATCH);
DataStream<Tuple2<String, Integer>> bc1Input = env.fromElements(Tuple2.of("bc3", 3), Tuple2.of("bc2", 2)).assignTimestampsAndWatermarks(WatermarkStrategy.<Tuple2<String, Integer>>forMonotonousTimestamps().withTimestampAssigner((in, ts) -> in.f1)).broadcast();
DataStream<Tuple2<String, Integer>> bc2Input = env.fromElements(Tuple2.of("bc1", 1)).assignTimestampsAndWatermarks(WatermarkStrategy.<Tuple2<String, Integer>>forMonotonousTimestamps().withTimestampAssigner((in, ts) -> in.f1)).broadcast();
DataStream<Tuple2<String, Integer>> regularInput = env.fromElements(Tuple2.of("regular1", 1), Tuple2.of("regular1", 2), Tuple2.of("regular1", 3), Tuple2.of("regular1", 4), Tuple2.of("regular2", 3), Tuple2.of("regular2", 5), Tuple2.of("regular1", 3)).assignTimestampsAndWatermarks(WatermarkStrategy.<Tuple2<String, Integer>>forMonotonousTimestamps().withTimestampAssigner((in, ts) -> in.f1)).keyBy(input -> input.f0);
KeyedMultipleInputTransformation<String> multipleInputTransformation = new KeyedMultipleInputTransformation<>("operator", mixedInputsOperatorFactory, BasicTypeInfo.STRING_TYPE_INFO, 1, BasicTypeInfo.STRING_TYPE_INFO);
multipleInputTransformation.addInput(regularInput.getTransformation(), input -> ((Tuple2<String, Integer>) input).f0);
multipleInputTransformation.addInput(bc1Input.getTransformation(), null);
multipleInputTransformation.addInput(bc2Input.getTransformation(), null);
DataStream<String> result = new MultipleConnectedStreams(env).transform(multipleInputTransformation);
try (CloseableIterator<String> resultIterator = result.executeAndCollect()) {
List<String> results = CollectionUtil.iteratorToList(resultIterator);
assertThat(results, equalTo(Arrays.asList("(regular1,1): [bc3, bc2, bc1]", "(regular1,2): [bc3, bc2, bc1]", "(regular1,3): [bc3, bc2, bc1]", "(regular1,3): [bc3, bc2, bc1]", "(regular1,4): [bc3, bc2, bc1]", "(regular2,3): [bc3, bc2, bc1]", "(regular2,5): [bc3, bc2, bc1]")));
}
}
use of org.apache.flink.streaming.api.datastream.MultipleConnectedStreams in project flink by apache.
the class SourceNAryInputChainingITCase method nAryInputStreamOperation.
private static DataStream<Long> nAryInputStreamOperation(final DataStream<?>... inputs) {
final StreamExecutionEnvironment env = inputs[0].getExecutionEnvironment();
// this is still clumsy due to the raw API
final MultipleInputTransformation<Long> transform = new MultipleInputTransformation<>("MultipleInputOperator", new NAryUnionOpFactory(inputs.length), Types.LONG, env.getParallelism());
for (DataStream<?> input : inputs) {
transform.addInput(input.getTransformation());
}
transform.setChainingStrategy(ChainingStrategy.HEAD_WITH_SOURCES);
env.addOperator(transform);
return new MultipleConnectedStreams(env).transform(transform);
}
use of org.apache.flink.streaming.api.datastream.MultipleConnectedStreams in project flink by apache.
the class MultipleInputITCase method testNonKeyed.
public void testNonKeyed(boolean withUnion) throws Exception {
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(1);
TestListResultSink<Long> resultSink = new TestListResultSink<>();
DataStream<Long> source1 = env.fromElements(1L, 10L);
DataStream<Long> source2 = env.fromElements(2L, 11L);
DataStream<String> source3 = env.fromElements("42", "44");
MultipleInputTransformation<Long> multipleInput = new MultipleInputTransformation<>("My Operator", new SumAllInputOperatorFactory(withUnion ? 2 : 3), BasicTypeInfo.LONG_TYPE_INFO, 1);
MultipleInputTransformation<Long> multipleInputTransformation;
if (withUnion) {
UnionTransformation<Long> union = new UnionTransformation<>(Arrays.asList(source1.getTransformation(), source2.getTransformation()));
multipleInputTransformation = multipleInput.addInput(union);
} else {
multipleInputTransformation = multipleInput.addInput(source1.getTransformation()).addInput(source2.getTransformation());
}
env.addOperator(multipleInputTransformation.addInput(source3.getTransformation()));
new MultipleConnectedStreams(env).transform(multipleInput).addSink(resultSink);
env.execute();
List<Long> result = resultSink.getResult();
Collections.sort(result);
long actualSum = result.get(result.size() - 1);
assertEquals(1 + 10 + 2 + 11 + 42 + 44, actualSum);
}
Aggregations