Examples with TopNQuery io.druid.query.topn.TopNQuery used on opensource projects

Example 1 with TopNQuery

use of io.druid.query.topn.TopNQuery in project druid by druid-io.

the class MultiValuedDimensionTest method testTopNWithDimFilterAndWithFilteredDimSpec.

@Test
public void testTopNWithDimFilterAndWithFilteredDimSpec() throws Exception {
    TopNQuery query = new TopNQueryBuilder().dataSource("xx").granularity(Granularities.ALL).dimension(new ListFilteredDimensionSpec(new DefaultDimensionSpec("tags", "tags"), ImmutableSet.of("t3"), null)).metric("count").intervals(QueryRunnerTestHelper.fullOnInterval).aggregators(Arrays.asList(new AggregatorFactory[] { new CountAggregatorFactory("count") })).threshold(5).filters(new SelectorDimFilter("tags", "t3", null)).build();
    QueryRunnerFactory factory = new TopNQueryRunnerFactory(TestQueryRunners.getPool(), new TopNQueryQueryToolChest(new TopNQueryConfig(), QueryRunnerTestHelper.NoopIntervalChunkingQueryRunnerDecorator()), QueryRunnerTestHelper.NOOP_QUERYWATCHER);
    QueryRunner<Result<TopNResultValue>> runner = QueryRunnerTestHelper.makeQueryRunner(factory, new QueryableIndexSegment("sid1", queryableIndex), null);
    Map<String, Object> context = Maps.newHashMap();
    Sequence<Result<TopNResultValue>> result = runner.run(query, context);
    List<Result<TopNResultValue>> expectedResults = Arrays.asList(new Result<TopNResultValue>(new DateTime("2011-01-12T00:00:00.000Z"), new TopNResultValue(Arrays.<Map<String, Object>>asList(ImmutableMap.<String, Object>of("tags", "t3", "count", 2L)))));
    TestHelper.assertExpectedObjects(expectedResults, Sequences.toList(result, new ArrayList<Result<TopNResultValue>>()), "");
}

Example 2 with TopNQuery

use of io.druid.query.topn.TopNQuery in project druid by druid-io.

the class AppendTest method testFilteredTopNSeries.

@Test
public void testFilteredTopNSeries() {
    List<Result<TopNResultValue>> expectedResults = Arrays.asList(new Result<TopNResultValue>(new DateTime("2011-01-12T00:00:00.000Z"), new TopNResultValue(Arrays.<Map<String, Object>>asList(ImmutableMap.<String, Object>builder().put("market", "spot").put("rows", 1L).put("index", 100.0D).put("addRowsIndexConstant", 102.0D).put("uniques", 0.0D).put("maxIndex", 100.0).put("minIndex", 100.0).build()))));
    TopNQuery query = makeFilteredTopNQuery();
    QueryRunner runner = TestQueryRunners.makeTopNQueryRunner(segment);
    HashMap<String, Object> context = new HashMap<String, Object>();
    TestHelper.assertExpectedResults(expectedResults, runner.run(query, context));
}

Example 3 with TopNQuery

use of io.druid.query.topn.TopNQuery in project druid by druid-io.

the class AppendTest method testFilteredTopNSeries2.

@Test
public void testFilteredTopNSeries2() {
    List<Result<TopNResultValue>> expectedResults = Arrays.asList(new Result<TopNResultValue>(new DateTime("2011-01-12T00:00:00.000Z"), new TopNResultValue(Lists.<Map<String, Object>>newArrayList())));
    TopNQuery query = makeFilteredTopNQuery();
    QueryRunner runner = TestQueryRunners.makeTopNQueryRunner(segment2);
    HashMap<String, Object> context = new HashMap<String, Object>();
    TestHelper.assertExpectedResults(expectedResults, runner.run(query, context));
}

Example 4 with TopNQuery

use of io.druid.query.topn.TopNQuery in project druid by druid-io.

the class QueryMaker method runQuery.

public Sequence<Object[]> runQuery(final DataSource dataSource, final RowSignature sourceRowSignature, final DruidQueryBuilder queryBuilder) {
    if (dataSource instanceof QueryDataSource) {
        final GroupByQuery outerQuery = queryBuilder.toGroupByQuery(dataSource, sourceRowSignature, plannerContext.getQueryContext());
        if (outerQuery == null) {
            // Bug in the planner rules. They shouldn't allow this to happen.
            throw new IllegalStateException("Can't use QueryDataSource without an outer groupBy query!");
        }
        return executeGroupBy(queryBuilder, outerQuery);
    }
    final TimeseriesQuery timeseriesQuery = queryBuilder.toTimeseriesQuery(dataSource, sourceRowSignature, plannerContext.getQueryContext());
    if (timeseriesQuery != null) {
        return executeTimeseries(queryBuilder, timeseriesQuery);
    }
    final TopNQuery topNQuery = queryBuilder.toTopNQuery(dataSource, sourceRowSignature, plannerContext.getQueryContext(), plannerContext.getPlannerConfig().getMaxTopNLimit(), plannerContext.getPlannerConfig().isUseApproximateTopN());
    if (topNQuery != null) {
        return executeTopN(queryBuilder, topNQuery);
    }
    final GroupByQuery groupByQuery = queryBuilder.toGroupByQuery(dataSource, sourceRowSignature, plannerContext.getQueryContext());
    if (groupByQuery != null) {
        return executeGroupBy(queryBuilder, groupByQuery);
    }
    final SelectQuery selectQuery = queryBuilder.toSelectQuery(dataSource, sourceRowSignature, plannerContext.getQueryContext());
    if (selectQuery != null) {
        return executeSelect(queryBuilder, selectQuery);
    }
    throw new IllegalStateException("WTF?! Cannot execute query even though we planned it?");
}

Example 5 with TopNQuery

use of io.druid.query.topn.TopNQuery in project druid by druid-io.

the class DruidQueryBuilder method toTopNQuery.

/**
   * Return this query as a TopN query, or null if this query is not compatible with TopN.
   *
   * @param dataSource         data source to query
   * @param sourceRowSignature row signature of the dataSource
   * @param context            query context
   * @param maxTopNLimit       maxTopNLimit from a PlannerConfig
   * @param useApproximateTopN from a PlannerConfig
   *
   * @return query or null
   */
public TopNQuery toTopNQuery(final DataSource dataSource, final RowSignature sourceRowSignature, final Map<String, Object> context, final int maxTopNLimit, final boolean useApproximateTopN) {
    // Must have GROUP BY one column, ORDER BY zero or one column, limit less than maxTopNLimit, and no HAVING.
    final boolean topNOk = grouping != null && grouping.getDimensions().size() == 1 && limitSpec != null && (limitSpec.getColumns().size() <= 1 && limitSpec.getLimit() <= maxTopNLimit) && having == null;
    if (!topNOk) {
        return null;
    }
    final DimensionSpec dimensionSpec = Iterables.getOnlyElement(grouping.getDimensions());
    final OrderByColumnSpec limitColumn;
    if (limitSpec.getColumns().isEmpty()) {
        limitColumn = new OrderByColumnSpec(dimensionSpec.getOutputName(), OrderByColumnSpec.Direction.ASCENDING, StringComparators.LEXICOGRAPHIC);
    } else {
        limitColumn = Iterables.getOnlyElement(limitSpec.getColumns());
    }
    final TopNMetricSpec topNMetricSpec;
    if (limitColumn.getDimension().equals(dimensionSpec.getOutputName())) {
        // DimensionTopNMetricSpec is exact; always return it even if allowApproximate is false.
        final DimensionTopNMetricSpec baseMetricSpec = new DimensionTopNMetricSpec(null, limitColumn.getDimensionComparator());
        topNMetricSpec = limitColumn.getDirection() == OrderByColumnSpec.Direction.ASCENDING ? baseMetricSpec : new InvertedTopNMetricSpec(baseMetricSpec);
    } else if (useApproximateTopN) {
        // ORDER BY metric
        final NumericTopNMetricSpec baseMetricSpec = new NumericTopNMetricSpec(limitColumn.getDimension());
        topNMetricSpec = limitColumn.getDirection() == OrderByColumnSpec.Direction.ASCENDING ? new InvertedTopNMetricSpec(baseMetricSpec) : baseMetricSpec;
    } else {
        return null;
    }
    final Filtration filtration = Filtration.create(filter).optimize(sourceRowSignature);
    return new TopNQuery(dataSource, VirtualColumns.EMPTY, Iterables.getOnlyElement(grouping.getDimensions()), topNMetricSpec, limitSpec.getLimit(), filtration.getQuerySegmentSpec(), filtration.getDimFilter(), Granularities.ALL, grouping.getAggregatorFactories(), grouping.getPostAggregators(), context);
}