Examples with DoubleSumAggregatorFactory io.druid.query.aggregation.DoubleSumAggregatorFactory used on opensource projects

Example 31 with DoubleSumAggregatorFactory

use of io.druid.query.aggregation.DoubleSumAggregatorFactory in project druid by druid-io.

the class DataSchemaTest method testDefaultExclusions.

@Test
public void testDefaultExclusions() throws Exception {
    Map<String, Object> parser = jsonMapper.convertValue(new StringInputRowParser(new JSONParseSpec(new TimestampSpec("time", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("dimB", "dimA")), null, null), null, null), null), new TypeReference<Map<String, Object>>() {
    });
    DataSchema schema = new DataSchema("test", parser, new AggregatorFactory[] { new DoubleSumAggregatorFactory("metric1", "col1"), new DoubleSumAggregatorFactory("metric2", "col2") }, new ArbitraryGranularitySpec(Granularities.DAY, ImmutableList.of(Interval.parse("2014/2015"))), jsonMapper);
    Assert.assertEquals(ImmutableSet.of("time", "col1", "col2", "metric1", "metric2"), schema.getParser().getParseSpec().getDimensionsSpec().getDimensionExclusions());
}

Example 32 with DoubleSumAggregatorFactory

use of io.druid.query.aggregation.DoubleSumAggregatorFactory in project druid by druid-io.

the class TopNTypeInterfaceBenchmark method setupQueries.

private void setupQueries() {
    // queries for the basic schema
    Map<String, TopNQueryBuilder> basicQueries = new LinkedHashMap<>();
    BenchmarkSchemaInfo basicSchema = BenchmarkSchemas.SCHEMA_MAP.get("basic");
    {
        // basic.A
        QuerySegmentSpec intervalSpec = new MultipleIntervalSegmentSpec(Arrays.asList(basicSchema.getDataInterval()));
        List<AggregatorFactory> queryAggs = new ArrayList<>();
        queryAggs.add(new LongSumAggregatorFactory("sumLongSequential", "sumLongSequential"));
        queryAggs.add(new LongMaxAggregatorFactory("maxLongUniform", "maxLongUniform"));
        queryAggs.add(new DoubleSumAggregatorFactory("sumFloatNormal", "sumFloatNormal"));
        queryAggs.add(new DoubleMinAggregatorFactory("minFloatZipf", "minFloatZipf"));
        queryAggs.add(new HyperUniquesAggregatorFactory("hyperUniquesMet", "hyper"));
        // Use an IdentityExtractionFn to force usage of DimExtractionTopNAlgorithm
        TopNQueryBuilder queryBuilderString = new TopNQueryBuilder().dataSource("blah").granularity(Granularities.ALL).dimension(new ExtractionDimensionSpec("dimSequential", "dimSequential", IdentityExtractionFn.getInstance())).metric("sumFloatNormal").intervals(intervalSpec).aggregators(queryAggs);
        // DimExtractionTopNAlgorithm is always used for numeric columns
        TopNQueryBuilder queryBuilderLong = new TopNQueryBuilder().dataSource("blah").granularity(Granularities.ALL).dimension("metLongUniform").metric("sumFloatNormal").intervals(intervalSpec).aggregators(queryAggs);
        TopNQueryBuilder queryBuilderFloat = new TopNQueryBuilder().dataSource("blah").granularity(Granularities.ALL).dimension("metFloatNormal").metric("sumFloatNormal").intervals(intervalSpec).aggregators(queryAggs);
        basicQueries.put("string", queryBuilderString);
        basicQueries.put("long", queryBuilderLong);
        basicQueries.put("float", queryBuilderFloat);
    }
    {
        // basic.numericSort
        QuerySegmentSpec intervalSpec = new MultipleIntervalSegmentSpec(Arrays.asList(basicSchema.getDataInterval()));
        List<AggregatorFactory> queryAggs = new ArrayList<>();
        queryAggs.add(new LongSumAggregatorFactory("sumLongSequential", "sumLongSequential"));
        TopNQueryBuilder queryBuilderA = new TopNQueryBuilder().dataSource("blah").granularity(Granularities.ALL).dimension("dimUniform").metric(new DimensionTopNMetricSpec(null, StringComparators.NUMERIC)).intervals(intervalSpec).aggregators(queryAggs);
        basicQueries.put("numericSort", queryBuilderA);
    }
    {
        // basic.alphanumericSort
        QuerySegmentSpec intervalSpec = new MultipleIntervalSegmentSpec(Arrays.asList(basicSchema.getDataInterval()));
        List<AggregatorFactory> queryAggs = new ArrayList<>();
        queryAggs.add(new LongSumAggregatorFactory("sumLongSequential", "sumLongSequential"));
        TopNQueryBuilder queryBuilderA = new TopNQueryBuilder().dataSource("blah").granularity(Granularities.ALL).dimension("dimUniform").metric(new DimensionTopNMetricSpec(null, StringComparators.ALPHANUMERIC)).intervals(intervalSpec).aggregators(queryAggs);
        basicQueries.put("alphanumericSort", queryBuilderA);
    }
    SCHEMA_QUERY_MAP.put("basic", basicQueries);
}

Example 33 with DoubleSumAggregatorFactory

use of io.druid.query.aggregation.DoubleSumAggregatorFactory in project druid by druid-io.

the class IngestSegmentFirehoseFactoryTest method constructorFeeder.

@Parameterized.Parameters(name = "{1}")
public static Collection<Object[]> constructorFeeder() throws IOException {
    final IndexSpec indexSpec = new IndexSpec();
    final HeapMemoryTaskStorage ts = new HeapMemoryTaskStorage(new TaskStorageConfig(null) {
    });
    final IncrementalIndexSchema schema = new IncrementalIndexSchema.Builder().withQueryGranularity(Granularities.NONE).withMinTimestamp(JodaUtils.MIN_INSTANT).withDimensionsSpec(ROW_PARSER).withMetrics(new AggregatorFactory[] { new LongSumAggregatorFactory(METRIC_LONG_NAME, DIM_LONG_NAME), new DoubleSumAggregatorFactory(METRIC_FLOAT_NAME, DIM_FLOAT_NAME) }).build();
    final OnheapIncrementalIndex index = new OnheapIncrementalIndex(schema, true, MAX_ROWS * MAX_SHARD_NUMBER);
    for (Integer i = 0; i < MAX_ROWS; ++i) {
        index.add(ROW_PARSER.parse(buildRow(i.longValue())));
    }
    if (!persistDir.mkdirs() && !persistDir.exists()) {
        throw new IOException(String.format("Could not create directory at [%s]", persistDir.getAbsolutePath()));
    }
    INDEX_MERGER.persist(index, persistDir, indexSpec);
    final TaskLockbox tl = new TaskLockbox(ts);
    final IndexerSQLMetadataStorageCoordinator mdc = new IndexerSQLMetadataStorageCoordinator(null, null, null) {

        private final Set<DataSegment> published = Sets.newHashSet();

        private final Set<DataSegment> nuked = Sets.newHashSet();

        @Override
        public List<DataSegment> getUsedSegmentsForInterval(String dataSource, Interval interval) throws IOException {
            return ImmutableList.copyOf(segmentSet);
        }

        @Override
        public List<DataSegment> getUsedSegmentsForIntervals(String dataSource, List<Interval> interval) throws IOException {
            return ImmutableList.copyOf(segmentSet);
        }

        @Override
        public List<DataSegment> getUnusedSegmentsForInterval(String dataSource, Interval interval) {
            return ImmutableList.of();
        }

        @Override
        public Set<DataSegment> announceHistoricalSegments(Set<DataSegment> segments) {
            Set<DataSegment> added = Sets.newHashSet();
            for (final DataSegment segment : segments) {
                if (published.add(segment)) {
                    added.add(segment);
                }
            }
            return ImmutableSet.copyOf(added);
        }

        @Override
        public void deleteSegments(Set<DataSegment> segments) {
            nuked.addAll(segments);
        }
    };
    final LocalTaskActionClientFactory tac = new LocalTaskActionClientFactory(ts, new TaskActionToolbox(tl, mdc, newMockEmitter(), EasyMock.createMock(SupervisorManager.class)));
    SegmentHandoffNotifierFactory notifierFactory = EasyMock.createNiceMock(SegmentHandoffNotifierFactory.class);
    EasyMock.replay(notifierFactory);
    final TaskToolboxFactory taskToolboxFactory = new TaskToolboxFactory(new TaskConfig(tmpDir.getAbsolutePath(), null, null, 50000, null, false, null, null), tac, newMockEmitter(), new DataSegmentPusher() {

        @Deprecated
        @Override
        public String getPathForHadoop(String dataSource) {
            return getPathForHadoop();
        }

        @Override
        public String getPathForHadoop() {
            throw new UnsupportedOperationException();
        }

        @Override
        public DataSegment push(File file, DataSegment segment) throws IOException {
            return segment;
        }
    }, new DataSegmentKiller() {

        @Override
        public void kill(DataSegment segments) throws SegmentLoadingException {
        }

        @Override
        public void killAll() throws IOException {
            throw new UnsupportedOperationException("not implemented");
        }
    }, new DataSegmentMover() {

        @Override
        public DataSegment move(DataSegment dataSegment, Map<String, Object> targetLoadSpec) throws SegmentLoadingException {
            return dataSegment;
        }
    }, new DataSegmentArchiver() {

        @Override
        public DataSegment archive(DataSegment segment) throws SegmentLoadingException {
            return segment;
        }

        @Override
        public DataSegment restore(DataSegment segment) throws SegmentLoadingException {
            return segment;
        }
    }, // segment announcer
    null, notifierFactory, // query runner factory conglomerate corporation unionized collective
    null, // query executor service
    null, // monitor scheduler
    null, new SegmentLoaderFactory(new SegmentLoaderLocalCacheManager(null, new SegmentLoaderConfig() {

        @Override
        public List<StorageLocationConfig> getLocations() {
            return Lists.newArrayList();
        }
    }, MAPPER)), MAPPER, INDEX_MERGER, INDEX_IO, null, null, INDEX_MERGER_V9);
    Collection<Object[]> values = new LinkedList<>();
    for (InputRowParser parser : Arrays.<InputRowParser>asList(ROW_PARSER, new MapInputRowParser(new JSONParseSpec(new TimestampSpec(TIME_COLUMN, "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.<String>of()), ImmutableList.of(DIM_FLOAT_NAME, DIM_LONG_NAME), ImmutableList.<SpatialDimensionSchema>of()), null, null)))) {
        for (List<String> dim_names : Arrays.<List<String>>asList(null, ImmutableList.of(DIM_NAME))) {
            for (List<String> metric_names : Arrays.<List<String>>asList(null, ImmutableList.of(METRIC_LONG_NAME, METRIC_FLOAT_NAME))) {
                values.add(new Object[] { new IngestSegmentFirehoseFactory(DATA_SOURCE_NAME, FOREVER, new SelectorDimFilter(DIM_NAME, DIM_VALUE, null), dim_names, metric_names, Guice.createInjector(new Module() {

                    @Override
                    public void configure(Binder binder) {
                        binder.bind(TaskToolboxFactory.class).toInstance(taskToolboxFactory);
                    }
                }), INDEX_IO), String.format("DimNames[%s]MetricNames[%s]ParserDimNames[%s]", dim_names == null ? "null" : "dims", metric_names == null ? "null" : "metrics", parser == ROW_PARSER ? "dims" : "null"), parser });
            }
        }
    }
    return values;
}

Example 34 with DoubleSumAggregatorFactory

use of io.druid.query.aggregation.DoubleSumAggregatorFactory in project druid by druid-io.

the class TaskLifecycleTest method testResumeTasks.

@Test
public void testResumeTasks() throws Exception {
    final Task indexTask = new IndexTask(null, null, new IndexTask.IndexIngestionSpec(new DataSchema("foo", null, new AggregatorFactory[] { new DoubleSumAggregatorFactory("met", "met") }, new UniformGranularitySpec(Granularities.DAY, null, ImmutableList.of(new Interval("2010-01-01/P2D"))), mapper), new IndexTask.IndexIOConfig(new MockFirehoseFactory(false), false, null), new IndexTask.IndexTuningConfig(10000, 10, null, null, indexSpec, null, false, null, null)), null, MAPPER);
    final long startTime = System.currentTimeMillis();
    // manually insert the task into TaskStorage, waiting for TaskQueue to sync from storage
    taskQueue.start();
    taskStorage.insert(indexTask, TaskStatus.running(indexTask.getId()));
    while (tsqa.getStatus(indexTask.getId()).get().isRunnable()) {
        if (System.currentTimeMillis() > startTime + 10 * 1000) {
            throw new ISE("Where did the task go?!: %s", indexTask.getId());
        }
        Thread.sleep(100);
    }
    final TaskStatus status = taskStorage.getStatus(indexTask.getId()).get();
    final List<DataSegment> publishedSegments = byIntervalOrdering.sortedCopy(mdc.getPublished());
    final List<DataSegment> loggedSegments = byIntervalOrdering.sortedCopy(tsqa.getInsertedSegments(indexTask.getId()));
    Assert.assertEquals("statusCode", TaskStatus.Status.SUCCESS, status.getStatusCode());
    Assert.assertEquals("segments logged vs published", loggedSegments, publishedSegments);
    Assert.assertEquals("num segments published", 2, mdc.getPublished().size());
    Assert.assertEquals("num segments nuked", 0, mdc.getNuked().size());
    Assert.assertEquals("segment1 datasource", "foo", publishedSegments.get(0).getDataSource());
    Assert.assertEquals("segment1 interval", new Interval("2010-01-01/P1D"), publishedSegments.get(0).getInterval());
    Assert.assertEquals("segment1 dimensions", ImmutableList.of("dim1", "dim2"), publishedSegments.get(0).getDimensions());
    Assert.assertEquals("segment1 metrics", ImmutableList.of("met"), publishedSegments.get(0).getMetrics());
    Assert.assertEquals("segment2 datasource", "foo", publishedSegments.get(1).getDataSource());
    Assert.assertEquals("segment2 interval", new Interval("2010-01-02/P1D"), publishedSegments.get(1).getInterval());
    Assert.assertEquals("segment2 dimensions", ImmutableList.of("dim1", "dim2"), publishedSegments.get(1).getDimensions());
    Assert.assertEquals("segment2 metrics", ImmutableList.of("met"), publishedSegments.get(1).getMetrics());
}

Example 35 with DoubleSumAggregatorFactory

use of io.druid.query.aggregation.DoubleSumAggregatorFactory in project druid by druid-io.

the class OnheapIncrementalIndexBenchmark method testConcurrentAddRead.

@Ignore
@Test
@BenchmarkOptions(callgc = true, clock = Clock.REAL_TIME, warmupRounds = 10, benchmarkRounds = 20)
public void testConcurrentAddRead() throws InterruptedException, ExecutionException, NoSuchMethodException, IllegalAccessException, InvocationTargetException, InstantiationException {
    final int taskCount = 30;
    final int concurrentThreads = 3;
    final int elementsPerThread = 1 << 15;
    final OnheapIncrementalIndex incrementalIndex = this.incrementalIndex.getConstructor(Long.TYPE, Granularity.class, AggregatorFactory[].class, Integer.TYPE).newInstance(0, Granularities.NONE, factories, elementsPerThread * taskCount);
    final ArrayList<AggregatorFactory> queryAggregatorFactories = new ArrayList<>(dimensionCount + 1);
    queryAggregatorFactories.add(new CountAggregatorFactory("rows"));
    for (int i = 0; i < dimensionCount; ++i) {
        queryAggregatorFactories.add(new LongSumAggregatorFactory(String.format("sumResult%s", i), String.format("sumResult%s", i)));
        queryAggregatorFactories.add(new DoubleSumAggregatorFactory(String.format("doubleSumResult%s", i), String.format("doubleSumResult%s", i)));
    }
    final ListeningExecutorService indexExecutor = MoreExecutors.listeningDecorator(Executors.newFixedThreadPool(concurrentThreads, new ThreadFactoryBuilder().setDaemon(false).setNameFormat("index-executor-%d").setPriority(Thread.MIN_PRIORITY).build()));
    final ListeningExecutorService queryExecutor = MoreExecutors.listeningDecorator(Executors.newFixedThreadPool(concurrentThreads, new ThreadFactoryBuilder().setDaemon(false).setNameFormat("query-executor-%d").build()));
    final long timestamp = System.currentTimeMillis();
    final Interval queryInterval = new Interval("1900-01-01T00:00:00Z/2900-01-01T00:00:00Z");
    final List<ListenableFuture<?>> indexFutures = new LinkedList<>();
    final List<ListenableFuture<?>> queryFutures = new LinkedList<>();
    final Segment incrementalIndexSegment = new IncrementalIndexSegment(incrementalIndex, null);
    final QueryRunnerFactory factory = new TimeseriesQueryRunnerFactory(new TimeseriesQueryQueryToolChest(QueryRunnerTestHelper.NoopIntervalChunkingQueryRunnerDecorator()), new TimeseriesQueryEngine(), QueryRunnerTestHelper.NOOP_QUERYWATCHER);
    final AtomicInteger currentlyRunning = new AtomicInteger(0);
    final AtomicBoolean concurrentlyRan = new AtomicBoolean(false);
    final AtomicBoolean someoneRan = new AtomicBoolean(false);
    for (int j = 0; j < taskCount; j++) {
        indexFutures.add(indexExecutor.submit(new Runnable() {

            @Override
            public void run() {
                currentlyRunning.incrementAndGet();
                try {
                    for (int i = 0; i < elementsPerThread; i++) {
                        incrementalIndex.add(getLongRow(timestamp + i, 1, dimensionCount));
                    }
                } catch (IndexSizeExceededException e) {
                    throw Throwables.propagate(e);
                }
                currentlyRunning.decrementAndGet();
                someoneRan.set(true);
            }
        }));
        queryFutures.add(queryExecutor.submit(new Runnable() {

            @Override
            public void run() {
                QueryRunner<Result<TimeseriesResultValue>> runner = new FinalizeResultsQueryRunner<Result<TimeseriesResultValue>>(factory.createRunner(incrementalIndexSegment), factory.getToolchest());
                TimeseriesQuery query = Druids.newTimeseriesQueryBuilder().dataSource("xxx").granularity(Granularities.ALL).intervals(ImmutableList.of(queryInterval)).aggregators(queryAggregatorFactories).build();
                Map<String, Object> context = new HashMap<String, Object>();
                for (Result<TimeseriesResultValue> result : Sequences.toList(runner.run(query, context), new LinkedList<Result<TimeseriesResultValue>>())) {
                    if (someoneRan.get()) {
                        Assert.assertTrue(result.getValue().getDoubleMetric("doubleSumResult0") > 0);
                    }
                }
                if (currentlyRunning.get() > 0) {
                    concurrentlyRan.set(true);
                }
            }
        }));
    }
    List<ListenableFuture<?>> allFutures = new ArrayList<>(queryFutures.size() + indexFutures.size());
    allFutures.addAll(queryFutures);
    allFutures.addAll(indexFutures);
    Futures.allAsList(allFutures).get();
    //Assert.assertTrue("Did not hit concurrency, please try again", concurrentlyRan.get());
    queryExecutor.shutdown();
    indexExecutor.shutdown();
    QueryRunner<Result<TimeseriesResultValue>> runner = new FinalizeResultsQueryRunner<Result<TimeseriesResultValue>>(factory.createRunner(incrementalIndexSegment), factory.getToolchest());
    TimeseriesQuery query = Druids.newTimeseriesQueryBuilder().dataSource("xxx").granularity(Granularities.ALL).intervals(ImmutableList.of(queryInterval)).aggregators(queryAggregatorFactories).build();
    Map<String, Object> context = new HashMap<String, Object>();
    List<Result<TimeseriesResultValue>> results = Sequences.toList(runner.run(query, context), new LinkedList<Result<TimeseriesResultValue>>());
    final int expectedVal = elementsPerThread * taskCount;
    for (Result<TimeseriesResultValue> result : results) {
        Assert.assertEquals(elementsPerThread, result.getValue().getLongMetric("rows").intValue());
        for (int i = 0; i < dimensionCount; ++i) {
            Assert.assertEquals(String.format("Failed long sum on dimension %d", i), expectedVal, result.getValue().getLongMetric(String.format("sumResult%s", i)).intValue());
            Assert.assertEquals(String.format("Failed double sum on dimension %d", i), expectedVal, result.getValue().getDoubleMetric(String.format("doubleSumResult%s", i)).intValue());
        }
    }
}