Examples with GroupByQuery io.druid.query.groupby.GroupByQuery used on opensource projects

Example 21 with GroupByQuery

use of io.druid.query.groupby.GroupByQuery in project druid by druid-io.

the class VarianceGroupByQueryTest method testGroupBy.

@Test
public void testGroupBy() {
    GroupByQuery query = GroupByQuery.builder().setDataSource(QueryRunnerTestHelper.dataSource).setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird).setDimensions(Lists.<DimensionSpec>newArrayList(new DefaultDimensionSpec("quality", "alias"))).setAggregatorSpecs(Arrays.asList(VarianceTestHelper.rowsCount, VarianceTestHelper.indexVarianceAggr, new LongSumAggregatorFactory("idx", "index"))).setPostAggregatorSpecs(Arrays.<PostAggregator>asList(VarianceTestHelper.stddevOfIndexPostAggr)).setGranularity(QueryRunnerTestHelper.dayGran).build();
    VarianceTestHelper.RowBuilder builder = new VarianceTestHelper.RowBuilder(new String[] { "alias", "rows", "idx", "index_stddev", "index_var" });
    List<Row> expectedResults = builder.add("2011-04-01", "automotive", 1L, 135L, 0d, 0d).add("2011-04-01", "business", 1L, 118L, 0d, 0d).add("2011-04-01", "entertainment", 1L, 158L, 0d, 0d).add("2011-04-01", "health", 1L, 120L, 0d, 0d).add("2011-04-01", "mezzanine", 3L, 2870L, 737.0179286322613d, 543195.4271253889d).add("2011-04-01", "news", 1L, 121L, 0d, 0d).add("2011-04-01", "premium", 3L, 2900L, 726.6322593583996d, 527994.4403402924d).add("2011-04-01", "technology", 1L, 78L, 0d, 0d).add("2011-04-01", "travel", 1L, 119L, 0d, 0d).add("2011-04-02", "automotive", 1L, 147L, 0d, 0d).add("2011-04-02", "business", 1L, 112L, 0d, 0d).add("2011-04-02", "entertainment", 1L, 166L, 0d, 0d).add("2011-04-02", "health", 1L, 113L, 0d, 0d).add("2011-04-02", "mezzanine", 3L, 2447L, 611.3420766546617d, 373739.13468843425d).add("2011-04-02", "news", 1L, 114L, 0d, 0d).add("2011-04-02", "premium", 3L, 2505L, 621.3898134843073d, 386125.30030206224d).add("2011-04-02", "technology", 1L, 97L, 0d, 0d).add("2011-04-02", "travel", 1L, 126L, 0d, 0d).build();
    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, runner, query);
    TestHelper.assertExpectedObjects(expectedResults, results, "");
}

Example 22 with GroupByQuery

use of io.druid.query.groupby.GroupByQuery in project druid by druid-io.

the class DataSourceTest method testQueryDataSource.

@Test
public void testQueryDataSource() throws IOException {
    GroupByQuery query = GroupByQuery.builder().setDataSource(QueryRunnerTestHelper.dataSource).setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird).setDimensions(Lists.<DimensionSpec>newArrayList(new DefaultDimensionSpec("quality", "alias"))).setAggregatorSpecs(Arrays.<AggregatorFactory>asList(QueryRunnerTestHelper.rowsCount, new LongSumAggregatorFactory("idx", "index"))).setGranularity(QueryRunnerTestHelper.dayGran).build();
    String dataSourceJSON = "{\"type\":\"query\", \"query\":" + jsonMapper.writeValueAsString(query) + "}";
    DataSource dataSource = jsonMapper.readValue(dataSourceJSON, DataSource.class);
    Assert.assertEquals(new QueryDataSource(query), dataSource);
}

Example 23 with GroupByQuery

use of io.druid.query.groupby.GroupByQuery in project druid by druid-io.

the class MultiValuedDimensionTest method testGroupByWithDimFilterAndWithFilteredDimSpec.

@Test
public void testGroupByWithDimFilterAndWithFilteredDimSpec() throws Exception {
    GroupByQuery query = GroupByQuery.builder().setDataSource("xx").setQuerySegmentSpec(new LegacySegmentSpec("1970/3000")).setGranularity(Granularities.ALL).setDimensions(Lists.<DimensionSpec>newArrayList(new RegexFilteredDimensionSpec(new DefaultDimensionSpec("tags", "tags"), "t3"))).setAggregatorSpecs(Arrays.asList(new AggregatorFactory[] { new CountAggregatorFactory("count") })).setDimFilter(new SelectorDimFilter("tags", "t3", null)).build();
    Sequence<Row> result = helper.runQueryOnSegmentsObjs(ImmutableList.<Segment>of(new QueryableIndexSegment("sid1", queryableIndex), new IncrementalIndexSegment(incrementalIndex, "sid2")), query);
    List<Row> expectedResults = Arrays.asList(GroupByQueryRunnerTestHelper.createExpectedRow("1970-01-01T00:00:00.000Z", "tags", "t3", "count", 4L));
    TestHelper.assertExpectedObjects(expectedResults, Sequences.toList(result, new ArrayList<Row>()), "");
}

Example 24 with GroupByQuery

use of io.druid.query.groupby.GroupByQuery in project druid by druid-io.

the class DistinctCountGroupByQueryTest method testGroupByWithDistinctCountAgg.

@Test
public void testGroupByWithDistinctCountAgg() throws Exception {
    final GroupByQueryConfig config = new GroupByQueryConfig();
    config.setMaxIntermediateRows(10000);
    final GroupByQueryRunnerFactory factory = GroupByQueryRunnerTest.makeQueryRunnerFactory(config);
    IncrementalIndex index = new OnheapIncrementalIndex(0, Granularities.SECOND, new AggregatorFactory[] { new CountAggregatorFactory("cnt") }, 1000);
    String visitor_id = "visitor_id";
    String client_type = "client_type";
    long timestamp = System.currentTimeMillis();
    index.add(new MapBasedInputRow(timestamp, Lists.newArrayList(visitor_id, client_type), ImmutableMap.<String, Object>of(visitor_id, "0", client_type, "iphone")));
    index.add(new MapBasedInputRow(timestamp + 1, Lists.newArrayList(visitor_id, client_type), ImmutableMap.<String, Object>of(visitor_id, "1", client_type, "iphone")));
    index.add(new MapBasedInputRow(timestamp + 2, Lists.newArrayList(visitor_id, client_type), ImmutableMap.<String, Object>of(visitor_id, "2", client_type, "android")));
    GroupByQuery query = new GroupByQuery.Builder().setDataSource(QueryRunnerTestHelper.dataSource).setGranularity(QueryRunnerTestHelper.allGran).setDimensions(Arrays.<DimensionSpec>asList(new DefaultDimensionSpec(client_type, client_type))).setInterval(QueryRunnerTestHelper.fullOnInterval).setLimitSpec(new DefaultLimitSpec(Lists.newArrayList(new OrderByColumnSpec(client_type, OrderByColumnSpec.Direction.DESCENDING)), 10)).setAggregatorSpecs(Lists.newArrayList(QueryRunnerTestHelper.rowsCount, new DistinctCountAggregatorFactory("UV", visitor_id, null))).build();
    final Segment incrementalIndexSegment = new IncrementalIndexSegment(index, null);
    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, factory.createRunner(incrementalIndexSegment), query);
    List<Row> expectedResults = Arrays.asList(GroupByQueryRunnerTestHelper.createExpectedRow("1970-01-01T00:00:00.000Z", client_type, "iphone", "UV", 2L, "rows", 2L), GroupByQueryRunnerTestHelper.createExpectedRow("1970-01-01T00:00:00.000Z", client_type, "android", "UV", 1L, "rows", 1L));
    TestHelper.assertExpectedObjects(expectedResults, results, "distinct-count");
}

Example 25 with GroupByQuery

use of io.druid.query.groupby.GroupByQuery in project druid by druid-io.

the class GroupByMergingQueryRunnerV2 method run.

@Override
public Sequence<Row> run(final Query queryParam, final Map responseContext) {
    final GroupByQuery query = (GroupByQuery) queryParam;
    final GroupByQueryConfig querySpecificConfig = config.withOverrides(query);
    // CTX_KEY_MERGE_RUNNERS_USING_CHAINED_EXECUTION is here because realtime servers use nested mergeRunners calls
    // (one for the entire query and one for each sink). We only want the outer call to actually do merging with a
    // merge buffer, otherwise the query will allocate too many merge buffers. This is potentially sub-optimal as it
    // will involve materializing the results for each sink before starting to feed them into the outer merge buffer.
    // I'm not sure of a better way to do this without tweaking how realtime servers do queries.
    final boolean forceChainedExecution = query.getContextBoolean(CTX_KEY_MERGE_RUNNERS_USING_CHAINED_EXECUTION, false);
    final GroupByQuery queryForRunners = query.withOverriddenContext(ImmutableMap.<String, Object>of(CTX_KEY_MERGE_RUNNERS_USING_CHAINED_EXECUTION, true));
    if (BaseQuery.getContextBySegment(query, false) || forceChainedExecution) {
        return new ChainedExecutionQueryRunner(exec, queryWatcher, queryables).run(query, responseContext);
    }
    final boolean isSingleThreaded = querySpecificConfig.isSingleThreaded();
    final AggregatorFactory[] combiningAggregatorFactories = new AggregatorFactory[query.getAggregatorSpecs().size()];
    for (int i = 0; i < query.getAggregatorSpecs().size(); i++) {
        combiningAggregatorFactories[i] = query.getAggregatorSpecs().get(i).getCombiningFactory();
    }
    final File temporaryStorageDirectory = new File(processingTmpDir, String.format("druid-groupBy-%s_%s", UUID.randomUUID(), query.getId()));
    final int priority = BaseQuery.getContextPriority(query, 0);
    // Figure out timeoutAt time now, so we can apply the timeout to both the mergeBufferPool.take and the actual
    // query processing together.
    final Number queryTimeout = query.getContextValue(QueryContextKeys.TIMEOUT, null);
    final long timeoutAt = queryTimeout == null ? JodaUtils.MAX_INSTANT : System.currentTimeMillis() + queryTimeout.longValue();
    return new BaseSequence<>(new BaseSequence.IteratorMaker<Row, CloseableGrouperIterator<RowBasedKey, Row>>() {

        @Override
        public CloseableGrouperIterator<RowBasedKey, Row> make() {
            final List<ReferenceCountingResourceHolder> resources = Lists.newArrayList();
            try {
                final LimitedTemporaryStorage temporaryStorage = new LimitedTemporaryStorage(temporaryStorageDirectory, querySpecificConfig.getMaxOnDiskStorage());
                final ReferenceCountingResourceHolder<LimitedTemporaryStorage> temporaryStorageHolder = ReferenceCountingResourceHolder.fromCloseable(temporaryStorage);
                resources.add(temporaryStorageHolder);
                final ReferenceCountingResourceHolder<ByteBuffer> mergeBufferHolder;
                try {
                    // This will potentially block if there are no merge buffers left in the pool.
                    final long timeout = timeoutAt - System.currentTimeMillis();
                    if (timeout <= 0 || (mergeBufferHolder = mergeBufferPool.take(timeout)) == null) {
                        throw new TimeoutException();
                    }
                    resources.add(mergeBufferHolder);
                } catch (Exception e) {
                    throw new QueryInterruptedException(e);
                }
                Pair<Grouper<RowBasedKey>, Accumulator<Grouper<RowBasedKey>, Row>> pair = RowBasedGrouperHelper.createGrouperAccumulatorPair(query, false, null, config, Suppliers.ofInstance(mergeBufferHolder.get()), concurrencyHint, temporaryStorage, spillMapper, combiningAggregatorFactories);
                final Grouper<RowBasedKey> grouper = pair.lhs;
                final Accumulator<Grouper<RowBasedKey>, Row> accumulator = pair.rhs;
                grouper.init();
                final ReferenceCountingResourceHolder<Grouper<RowBasedKey>> grouperHolder = ReferenceCountingResourceHolder.fromCloseable(grouper);
                resources.add(grouperHolder);
                ListenableFuture<List<Boolean>> futures = Futures.allAsList(Lists.newArrayList(Iterables.transform(queryables, new Function<QueryRunner<Row>, ListenableFuture<Boolean>>() {

                    @Override
                    public ListenableFuture<Boolean> apply(final QueryRunner<Row> input) {
                        if (input == null) {
                            throw new ISE("Null queryRunner! Looks to be some segment unmapping action happening");
                        }
                        ListenableFuture<Boolean> future = exec.submit(new AbstractPrioritizedCallable<Boolean>(priority) {

                            @Override
                            public Boolean call() throws Exception {
                                try (Releaser bufferReleaser = mergeBufferHolder.increment();
                                    Releaser grouperReleaser = grouperHolder.increment()) {
                                    final Object retVal = input.run(queryForRunners, responseContext).accumulate(grouper, accumulator);
                                    // Return true if OK, false if resources were exhausted.
                                    return retVal == grouper;
                                } catch (QueryInterruptedException e) {
                                    throw e;
                                } catch (Exception e) {
                                    log.error(e, "Exception with one of the sequences!");
                                    throw Throwables.propagate(e);
                                }
                            }
                        });
                        if (isSingleThreaded) {
                            waitForFutureCompletion(query, Futures.allAsList(ImmutableList.of(future)), timeoutAt - System.currentTimeMillis());
                        }
                        return future;
                    }
                })));
                if (!isSingleThreaded) {
                    waitForFutureCompletion(query, futures, timeoutAt - System.currentTimeMillis());
                }
                return RowBasedGrouperHelper.makeGrouperIterator(grouper, query, new Closeable() {

                    @Override
                    public void close() throws IOException {
                        for (Closeable closeable : Lists.reverse(resources)) {
                            CloseQuietly.close(closeable);
                        }
                    }
                });
            } catch (Throwable e) {
                // Exception caught while setting up the iterator; release resources.
                for (Closeable closeable : Lists.reverse(resources)) {
                    CloseQuietly.close(closeable);
                }
                throw e;
            }
        }

        @Override
        public void cleanup(CloseableGrouperIterator<RowBasedKey, Row> iterFromMake) {
            iterFromMake.close();
        }
    });
}