Search in sources :

Example 21 with Metadata

use of com.yahoo.bullet.pubsub.Metadata in project bullet-storm by yahoo.

the class JoinBoltTest method testDistribution.

@Test
public void testDistribution() {
    BulletConfig bulletConfig = DistributionTest.makeConfiguration(10, 128);
    Distribution distribution = DistributionTest.makeDistribution(bulletConfig, makeAttributes(Distribution.Type.PMF, 3), "field", 10, null);
    IntStream.range(0, 50).mapToObj(i -> RecordBox.get().add("field", i).getRecord()).forEach(distribution::consume);
    byte[] first = distribution.getData();
    distribution = DistributionTest.makeDistribution(bulletConfig, makeAttributes(Distribution.Type.PMF, 3), "field", 10, null);
    IntStream.range(50, 101).mapToObj(i -> RecordBox.get().add("field", i).getRecord()).forEach(distribution::consume);
    byte[] second = distribution.getData();
    bolt = new DonableJoinBolt(config, 2, true);
    setup(bolt);
    Tuple query = TupleUtils.makeIDTuple(TupleClassifier.Type.QUERY_TUPLE, "42", makeAggregationQuery(DISTRIBUTION, 10, Distribution.Type.PMF, "field", null, null, null, null, 3), EMPTY);
    bolt.execute(query);
    sendRawByteTuplesTo(bolt, "42", asList(first, second));
    BulletRecord expectedA = RecordBox.get().add(RANGE_FIELD, NEGATIVE_INFINITY_START + SEPARATOR + 0.0 + END_EXCLUSIVE).add(COUNT_FIELD, 0.0).add(PROBABILITY_FIELD, 0.0).getRecord();
    BulletRecord expectedB = RecordBox.get().add(RANGE_FIELD, START_INCLUSIVE + 0.0 + SEPARATOR + 50.0 + END_EXCLUSIVE).add(COUNT_FIELD, 50.0).add(PROBABILITY_FIELD, 50.0 / 101).getRecord();
    BulletRecord expectedC = RecordBox.get().add(RANGE_FIELD, START_INCLUSIVE + 50.0 + SEPARATOR + 100.0 + END_EXCLUSIVE).add(COUNT_FIELD, 50.0).add(PROBABILITY_FIELD, 50.0 / 101).getRecord();
    BulletRecord expectedD = RecordBox.get().add(RANGE_FIELD, START_INCLUSIVE + 100.0 + SEPARATOR + POSITIVE_INFINITY_END).add(COUNT_FIELD, 1.0).add(PROBABILITY_FIELD, 1.0 / 101).getRecord();
    List<BulletRecord> results = asList(expectedA, expectedB, expectedC, expectedD);
    Tuple expected = TupleUtils.makeTuple(TupleClassifier.Type.RESULT_TUPLE, "42", Clip.of(results).asJSON(), COMPLETED);
    Tuple tick = TupleUtils.makeTuple(TupleClassifier.Type.TICK_TUPLE);
    bolt.execute(tick);
    for (int i = 0; i < BulletStormConfig.DEFAULT_JOIN_BOLT_QUERY_TICK_TIMEOUT - 1; ++i) {
        bolt.execute(tick);
        Assert.assertFalse(wasResultEmittedTo(TopologyConstants.RESULT_STREAM, expected));
    }
    bolt.execute(tick);
    Assert.assertTrue(wasResultEmittedTo(TopologyConstants.RESULT_STREAM, expected));
    Tuple metadata = TupleUtils.makeTuple(TupleClassifier.Type.FEEDBACK_TUPLE, "42", new Metadata(Metadata.Signal.COMPLETE, null));
    Assert.assertTrue(wasMetadataEmittedTo(TopologyConstants.FEEDBACK_STREAM, metadata));
    Assert.assertEquals(collector.getAllEmittedTo(TopologyConstants.RESULT_STREAM).count(), 1);
    Assert.assertEquals(collector.getAllEmittedTo(TopologyConstants.FEEDBACK_STREAM).count(), 1);
}
Also used : JsonObject(com.google.gson.JsonObject) GROUP(com.yahoo.bullet.parsing.Aggregation.Type.GROUP) COUNT(com.yahoo.bullet.aggregations.grouping.GroupOperation.GroupOperationType.COUNT) TopK(com.yahoo.bullet.aggregations.TopK) BulletError(com.yahoo.bullet.common.BulletError) Concept(com.yahoo.bullet.result.Meta.Concept) ParsingError(com.yahoo.bullet.parsing.ParsingError) Test(org.testng.annotations.Test) RecordBox(com.yahoo.bullet.result.RecordBox) ErrorType(com.yahoo.sketches.frequencies.ErrorType) EQUALS(com.yahoo.bullet.parsing.Clause.Operation.EQUALS) PROBABILITY_FIELD(com.yahoo.bullet.aggregations.sketches.QuantileSketch.PROBABILITY_FIELD) Collections.singletonList(java.util.Collections.singletonList) CustomCollector(com.yahoo.bullet.storm.testing.CustomCollector) Pair(org.apache.commons.lang3.tuple.Pair) Arrays.asList(java.util.Arrays.asList) Map(java.util.Map) Mockito.doAnswer(org.mockito.Mockito.doAnswer) Metadata(com.yahoo.bullet.pubsub.Metadata) Mockito.doReturn(org.mockito.Mockito.doReturn) GroupData(com.yahoo.bullet.aggregations.grouping.GroupData) QueryUtils.makeAggregationQuery(com.yahoo.bullet.parsing.QueryUtils.makeAggregationQuery) GroupBy(com.yahoo.bullet.aggregations.GroupBy) ComponentUtils(com.yahoo.bullet.storm.testing.ComponentUtils) BulletRecord(com.yahoo.bullet.record.BulletRecord) AggregationUtils(com.yahoo.bullet.parsing.AggregationUtils) BeforeMethod(org.testng.annotations.BeforeMethod) GroupByTest(com.yahoo.bullet.aggregations.GroupByTest) Fields(org.apache.storm.tuple.Fields) CustomTopologyContext(com.yahoo.bullet.storm.testing.CustomTopologyContext) SlidingRecord(com.yahoo.bullet.windowing.SlidingRecord) DistributionTest(com.yahoo.bullet.aggregations.DistributionTest) Serializable(java.io.Serializable) RateLimitError(com.yahoo.bullet.querying.RateLimitError) GroupOperation(com.yahoo.bullet.aggregations.grouping.GroupOperation) Querier(com.yahoo.bullet.querying.Querier) JsonArray(com.google.gson.JsonArray) List(java.util.List) Distribution(com.yahoo.bullet.aggregations.Distribution) NEGATIVE_INFINITY_START(com.yahoo.bullet.aggregations.sketches.QuantileSketch.NEGATIVE_INFINITY_START) BulletConfig(com.yahoo.bullet.common.BulletConfig) AdditionalAnswers.returnsElementsOf(org.mockito.AdditionalAnswers.returnsElementsOf) Window(com.yahoo.bullet.parsing.Window) IRichBolt(org.apache.storm.topology.IRichBolt) END_EXCLUSIVE(com.yahoo.bullet.aggregations.sketches.QuantileSketch.END_EXCLUSIVE) TupleUtils(com.yahoo.bullet.storm.testing.TupleUtils) IntStream(java.util.stream.IntStream) TopKTest(com.yahoo.bullet.aggregations.TopKTest) Setter(lombok.Setter) TestHelpers.assertJSONEquals(com.yahoo.bullet.storm.testing.TestHelpers.assertJSONEquals) SerializerDeserializer(com.yahoo.bullet.common.SerializerDeserializer) HashMap(java.util.HashMap) JsonParser(com.google.gson.JsonParser) Mockito.spy(org.mockito.Mockito.spy) TestHelpers.getListBytes(com.yahoo.bullet.storm.testing.TestHelpers.getListBytes) Clip(com.yahoo.bullet.result.Clip) SUM(com.yahoo.bullet.aggregations.grouping.GroupOperation.GroupOperationType.SUM) SEPARATOR(com.yahoo.bullet.aggregations.sketches.QuantileSketch.SEPARATOR) ArrayList(java.util.ArrayList) HashSet(java.util.HashSet) Tuple(org.apache.storm.tuple.Tuple) Assert(org.testng.Assert) CountDistinct(com.yahoo.bullet.aggregations.CountDistinct) AggregationUtils.makeAttributes(com.yahoo.bullet.parsing.AggregationUtils.makeAttributes) Collections.singletonMap(java.util.Collections.singletonMap) TOP_K(com.yahoo.bullet.parsing.Aggregation.Type.TOP_K) Aggregation(com.yahoo.bullet.parsing.Aggregation) Meta(com.yahoo.bullet.result.Meta) COUNT_FIELD(com.yahoo.bullet.aggregations.sketches.QuantileSketch.COUNT_FIELD) POSITIVE_INFINITY_END(com.yahoo.bullet.aggregations.sketches.QuantileSketch.POSITIVE_INFINITY_END) COUNT_DISTINCT(com.yahoo.bullet.parsing.Aggregation.Type.COUNT_DISTINCT) CustomOutputFieldsDeclarer(com.yahoo.bullet.storm.testing.CustomOutputFieldsDeclarer) START_INCLUSIVE(com.yahoo.bullet.aggregations.sketches.QuantileSketch.START_INCLUSIVE) CountDistinctTest(com.yahoo.bullet.aggregations.CountDistinctTest) RAW(com.yahoo.bullet.parsing.Aggregation.Type.RAW) DISTRIBUTION(com.yahoo.bullet.parsing.Aggregation.Type.DISTRIBUTION) QueryUtils.makeGroupFilterQuery(com.yahoo.bullet.parsing.QueryUtils.makeGroupFilterQuery) RANGE_FIELD(com.yahoo.bullet.aggregations.sketches.QuantileSketch.RANGE_FIELD) BulletRecord(com.yahoo.bullet.record.BulletRecord) Distribution(com.yahoo.bullet.aggregations.Distribution) Metadata(com.yahoo.bullet.pubsub.Metadata) BulletConfig(com.yahoo.bullet.common.BulletConfig) Tuple(org.apache.storm.tuple.Tuple) Test(org.testng.annotations.Test) GroupByTest(com.yahoo.bullet.aggregations.GroupByTest) DistributionTest(com.yahoo.bullet.aggregations.DistributionTest) TopKTest(com.yahoo.bullet.aggregations.TopKTest) CountDistinctTest(com.yahoo.bullet.aggregations.CountDistinctTest)

Example 22 with Metadata

use of com.yahoo.bullet.pubsub.Metadata in project bullet-storm by yahoo.

the class JoinBoltTest method testQueryNotDoneButIsTimeBased.

@Test
public void testQueryNotDoneButIsTimeBased() {
    // This bolt will be done after combining RAW_MAX_SIZE - 1 times and is a shouldBuffer, so it is buffered.
    bolt = new DonableJoinBolt(config, RAW_MAX_SIZE - 1, true);
    setup(bolt);
    Tuple query = TupleUtils.makeIDTuple(TupleClassifier.Type.QUERY_TUPLE, "42", "{}", EMPTY);
    bolt.execute(query);
    List<BulletRecord> sent = sendRawRecordTuplesTo(bolt, "42", RAW_MAX_SIZE - 1);
    Tuple tick = TupleUtils.makeTuple(TupleClassifier.Type.TICK_TUPLE);
    // Should make isDone true but query is a time based window so it gets buffered.
    bolt.execute(tick);
    Tuple expected = TupleUtils.makeTuple(TupleClassifier.Type.RESULT_TUPLE, "42", Clip.of(sent).asJSON(), COMPLETED);
    // We need to tick the default query tickout to make the query emit
    for (int i = 0; i < BulletStormConfig.DEFAULT_JOIN_BOLT_QUERY_TICK_TIMEOUT - 1; ++i) {
        bolt.execute(tick);
        Assert.assertFalse(wasResultEmittedTo(TopologyConstants.RESULT_STREAM, expected));
    }
    // Should emit on the last tick
    bolt.execute(tick);
    Assert.assertTrue(wasResultEmittedTo(TopologyConstants.RESULT_STREAM, expected));
    Tuple metadata = TupleUtils.makeTuple(TupleClassifier.Type.FEEDBACK_TUPLE, "42", new Metadata(Metadata.Signal.COMPLETE, null));
    Assert.assertTrue(wasMetadataEmittedTo(TopologyConstants.FEEDBACK_STREAM, metadata));
    Assert.assertEquals(collector.getAllEmittedTo(TopologyConstants.RESULT_STREAM).count(), 1);
    Assert.assertEquals(collector.getAllEmittedTo(TopologyConstants.FEEDBACK_STREAM).count(), 1);
}
Also used : BulletRecord(com.yahoo.bullet.record.BulletRecord) Metadata(com.yahoo.bullet.pubsub.Metadata) Tuple(org.apache.storm.tuple.Tuple) Test(org.testng.annotations.Test) GroupByTest(com.yahoo.bullet.aggregations.GroupByTest) DistributionTest(com.yahoo.bullet.aggregations.DistributionTest) TopKTest(com.yahoo.bullet.aggregations.TopKTest) CountDistinctTest(com.yahoo.bullet.aggregations.CountDistinctTest)

Example 23 with Metadata

use of com.yahoo.bullet.pubsub.Metadata in project bullet-storm by yahoo.

the class JoinBoltTest method testRawQueryDoneButNotTimedOutWithExcessRecords.

@Test
public void testRawQueryDoneButNotTimedOutWithExcessRecords() {
    Tuple query = TupleUtils.makeIDTuple(TupleClassifier.Type.QUERY_TUPLE, "42", makeAggregationQuery(RAW, 5), EMPTY);
    bolt.execute(query);
    // This will send 2 batches of 3 records each (total of 6 records).
    List<BulletRecord> sent = sendRawRecordTuplesTo(bolt, "42", 5, 3);
    List<BulletRecord> actualSent = sent.subList(0, 5);
    Tuple expected = TupleUtils.makeTuple(TupleClassifier.Type.RESULT_TUPLE, "42", Clip.of(actualSent).asJSON(), COMPLETED);
    Assert.assertTrue(wasResultEmittedTo(TopologyConstants.RESULT_STREAM, expected));
    Tuple metadata = TupleUtils.makeTuple(TupleClassifier.Type.FEEDBACK_TUPLE, "42", new Metadata(Metadata.Signal.COMPLETE, null));
    Assert.assertTrue(wasMetadataEmittedTo(TopologyConstants.FEEDBACK_STREAM, metadata));
    Assert.assertEquals(collector.getAllEmittedTo(TopologyConstants.RESULT_STREAM).count(), 1);
    Assert.assertEquals(collector.getAllEmittedTo(TopologyConstants.FEEDBACK_STREAM).count(), 1);
}
Also used : BulletRecord(com.yahoo.bullet.record.BulletRecord) Metadata(com.yahoo.bullet.pubsub.Metadata) Tuple(org.apache.storm.tuple.Tuple) Test(org.testng.annotations.Test) GroupByTest(com.yahoo.bullet.aggregations.GroupByTest) DistributionTest(com.yahoo.bullet.aggregations.DistributionTest) TopKTest(com.yahoo.bullet.aggregations.TopKTest) CountDistinctTest(com.yahoo.bullet.aggregations.CountDistinctTest)

Example 24 with Metadata

use of com.yahoo.bullet.pubsub.Metadata in project bullet-storm by yahoo.

the class QuerySpoutTest method testSignalOnlyMessagesAreSentOnTheMetadataStream.

@Test
public void testSignalOnlyMessagesAreSentOnTheMetadataStream() {
    // Add messages to be received from subscriber
    PubSubMessage messageA = new PubSubMessage("42", Metadata.Signal.KILL);
    PubSubMessage messageB = new PubSubMessage("43", Metadata.Signal.COMPLETE);
    PubSubMessage messageC = new PubSubMessage("44", null, new Metadata());
    subscriber.addMessages(messageA, messageB, messageC);
    Assert.assertEquals(subscriber.getReceived().size(), 0);
    Assert.assertEquals(emitter.getEmitted().size(), 0);
    spout.nextTuple();
    spout.nextTuple();
    spout.nextTuple();
    Assert.assertEquals(subscriber.getReceived().size(), 3);
    Assert.assertEquals(subscriber.getReceived().get(0), messageA);
    Assert.assertEquals(subscriber.getReceived().get(1), messageB);
    Assert.assertEquals(subscriber.getReceived().get(2), messageC);
    Assert.assertEquals(emitter.getEmitted().size(), 3);
    Tuple emittedFirst = TupleUtils.makeTuple(TupleClassifier.Type.METADATA_TUPLE, messageA.getId(), messageA.getMetadata());
    Tuple emittedSecond = TupleUtils.makeTuple(TupleClassifier.Type.METADATA_TUPLE, messageB.getId(), messageB.getMetadata());
    Tuple emittedThird = TupleUtils.makeTuple(TupleClassifier.Type.METADATA_TUPLE, messageC.getId(), messageC.getMetadata());
    Assert.assertTrue(emitter.wasTupleEmittedTo(emittedFirst, TopologyConstants.METADATA_STREAM));
    Assert.assertTrue(emitter.wasTupleEmittedTo(emittedSecond, TopologyConstants.METADATA_STREAM));
    Assert.assertTrue(emitter.wasTupleEmittedTo(emittedThird, TopologyConstants.METADATA_STREAM));
    Assert.assertTrue(emitter.wasNthEmitted(emittedFirst, 1));
    Assert.assertTrue(emitter.wasNthEmitted(emittedSecond, 2));
    Assert.assertTrue(emitter.wasNthEmitted(emittedThird, 3));
}
Also used : PubSubMessage(com.yahoo.bullet.pubsub.PubSubMessage) Metadata(com.yahoo.bullet.pubsub.Metadata) Tuple(org.apache.storm.tuple.Tuple) Test(org.testng.annotations.Test)

Example 25 with Metadata

use of com.yahoo.bullet.pubsub.Metadata in project bullet-storm by yahoo.

the class QuerySpoutTest method testNextTupleMessagesAreReceivedAndTupleIsEmitted.

@Test
public void testNextTupleMessagesAreReceivedAndTupleIsEmitted() {
    // Add messages to be received from subscriber
    PubSubMessage messageA = new PubSubMessage("42", "This is a PubSubMessage", new Metadata());
    PubSubMessage messageB = new PubSubMessage("43", "This is also a PubSubMessage", new Metadata());
    subscriber.addMessages(messageA, messageB);
    Assert.assertEquals(subscriber.getReceived().size(), 0);
    Assert.assertEquals(emitter.getEmitted().size(), 0);
    // subscriber.receive() -> messageA
    spout.nextTuple();
    Assert.assertEquals(subscriber.getReceived().size(), 1);
    Assert.assertEquals(subscriber.getReceived().get(0), messageA);
    Tuple emittedFirst = TupleUtils.makeTuple(TupleClassifier.Type.QUERY_TUPLE, messageA.getId(), messageA.getContent(), messageA.getMetadata());
    Assert.assertEquals(emitter.getEmitted().size(), 1);
    Assert.assertTrue(emitter.wasNthEmitted(emittedFirst, 1));
    // subscriber.receive() -> messageB
    spout.nextTuple();
    Assert.assertEquals(subscriber.getReceived().size(), 2);
    Assert.assertEquals(subscriber.getReceived().get(0), messageA);
    Assert.assertEquals(subscriber.getReceived().get(1), messageB);
    Tuple emittedSecond = TupleUtils.makeTuple(TupleClassifier.Type.QUERY_TUPLE, messageB.getId(), messageB.getContent(), messageB.getMetadata());
    Assert.assertEquals(emitter.getEmitted().size(), 2);
    Assert.assertTrue(emitter.wasNthEmitted(emittedFirst, 1));
    Assert.assertTrue(emitter.wasNthEmitted(emittedSecond, 2));
}
Also used : PubSubMessage(com.yahoo.bullet.pubsub.PubSubMessage) Metadata(com.yahoo.bullet.pubsub.Metadata) Tuple(org.apache.storm.tuple.Tuple) Test(org.testng.annotations.Test)

Aggregations

Metadata (com.yahoo.bullet.pubsub.Metadata)45 Test (org.testng.annotations.Test)35 Tuple (org.apache.storm.tuple.Tuple)31 CountDistinctTest (com.yahoo.bullet.aggregations.CountDistinctTest)25 DistributionTest (com.yahoo.bullet.aggregations.DistributionTest)25 TopKTest (com.yahoo.bullet.aggregations.TopKTest)25 BulletRecord (com.yahoo.bullet.record.BulletRecord)25 GroupByTest (com.yahoo.bullet.aggregations.GroupByTest)22 PubSubMessage (com.yahoo.bullet.pubsub.PubSubMessage)11 GroupOperation (com.yahoo.bullet.aggregations.grouping.GroupOperation)7 RateLimitError (com.yahoo.bullet.querying.RateLimitError)7 ArrayList (java.util.ArrayList)6 BulletConfig (com.yahoo.bullet.common.BulletConfig)5 Meta (com.yahoo.bullet.result.Meta)5 CountDistinct (com.yahoo.bullet.aggregations.CountDistinct)4 Distribution (com.yahoo.bullet.aggregations.Distribution)4 TopK (com.yahoo.bullet.aggregations.TopK)4 GroupData (com.yahoo.bullet.aggregations.grouping.GroupData)4 COUNT (com.yahoo.bullet.aggregations.grouping.GroupOperation.GroupOperationType.COUNT)4 COUNT_FIELD (com.yahoo.bullet.aggregations.sketches.QuantileSketch.COUNT_FIELD)4