use of org.apache.samza.operators.functions.SupplierFunction in project samza by apache.
the class TestMessageStreamImpl method testWindowWithRelaxedTypes.
@Test
public void testWindowWithRelaxedTypes() throws Exception {
StreamApplicationDescriptorImpl mockGraph = mock(StreamApplicationDescriptorImpl.class);
OperatorSpec mockOpSpec = mock(OperatorSpec.class);
MessageStream<TestInputMessageEnvelope> inputStream = new MessageStreamImpl<>(mockGraph, mockOpSpec);
MapFunction<TestMessageEnvelope, String> keyExtractor = m -> m.getKey();
FoldLeftFunction<TestMessageEnvelope, Integer> aggregator = (m, c) -> c + 1;
SupplierFunction<Integer> initialValue = () -> 0;
// should compile since TestMessageEnvelope (input for functions) is base class of TestInputMessageEnvelope (M)
Window<TestInputMessageEnvelope, String, Integer> window = Windows.keyedTumblingWindow(keyExtractor, Duration.ofHours(1), initialValue, aggregator, null, mock(Serde.class));
MessageStream<WindowPane<String, Integer>> windowedStream = inputStream.window(window, "w1");
ArgumentCaptor<OperatorSpec> registeredOpCaptor = ArgumentCaptor.forClass(OperatorSpec.class);
verify(mockOpSpec).registerNextOperatorSpec(registeredOpCaptor.capture());
OperatorSpec<?, TestMessageEnvelope> registeredOpSpec = registeredOpCaptor.getValue();
assertTrue(registeredOpSpec instanceof WindowOperatorSpec);
assertEquals(OpCode.WINDOW, registeredOpSpec.getOpCode());
assertEquals(window, ((WindowOperatorSpec) registeredOpSpec).getWindow());
}
use of org.apache.samza.operators.functions.SupplierFunction in project samza by apache.
the class LogicalAggregateTranslator method translate.
void translate(final LogicalAggregate aggregate, final TranslatorContext context) {
validateAggregateFunctions(aggregate);
MessageStream<SamzaSqlRelMessage> inputStream = context.getMessageStream(aggregate.getInput().getId());
// At this point, the assumption is that only count function is supported.
SupplierFunction<Long> initialValue = () -> (long) 0;
FoldLeftFunction<SamzaSqlRelMessage, Long> foldCountFn = (m, c) -> c + 1;
final ArrayList<String> aggFieldNames = getAggFieldNames(aggregate);
MessageStream<SamzaSqlRelMessage> outputStream = inputStream.map(new TranslatorInputMetricsMapFunction(logicalOpId)).window(Windows.keyedTumblingWindow(m -> m, Duration.ofMillis(context.getExecutionContext().getSamzaSqlApplicationConfig().getWindowDurationMs()), initialValue, foldCountFn, new SamzaSqlRelMessageSerdeFactory.SamzaSqlRelMessageSerde(), new LongSerde()).setAccumulationMode(AccumulationMode.DISCARDING), changeLogStorePrefix + "_tumblingWindow_" + logicalOpId).map(windowPane -> {
List<String> fieldNames = windowPane.getKey().getKey().getSamzaSqlRelRecord().getFieldNames();
List<Object> fieldValues = windowPane.getKey().getKey().getSamzaSqlRelRecord().getFieldValues();
fieldNames.add(aggFieldNames.get(0));
fieldValues.add(windowPane.getMessage());
return new SamzaSqlRelMessage(fieldNames, fieldValues, new SamzaSqlRelMsgMetadata(0L, 0L));
});
context.registerMessageStream(aggregate.getId(), outputStream);
outputStream.map(new TranslatorOutputMetricsMapFunction(logicalOpId));
}
use of org.apache.samza.operators.functions.SupplierFunction 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);
}
use of org.apache.samza.operators.functions.SupplierFunction 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);
}
Aggregations