Examples with FilteredAggregatorFactory org.apache.druid.query.aggregation.FilteredAggregatorFactory used on opensource projects

Example 56 with FilteredAggregatorFactory

use of org.apache.druid.query.aggregation.FilteredAggregatorFactory in project druid by druid-io.

the class CalciteQueryTest method testMinMaxAvgDailyCountWithLimit.

@Test
public void testMinMaxAvgDailyCountWithLimit() throws Exception {
    // Cannot vectorize due to virtual columns.
    cannotVectorize();
    testQuery("SELECT * FROM (" + "  SELECT max(cnt), min(cnt), avg(cnt), TIME_EXTRACT(max(t), 'EPOCH') last_time, count(1) num_days FROM (\n" + "      SELECT TIME_FLOOR(__time, 'P1D') AS t, count(1) cnt\n" + "      FROM \"foo\"\n" + "      GROUP BY 1\n" + "  )" + ") LIMIT 1\n", ImmutableList.of(GroupByQuery.builder().setDataSource(new QueryDataSource(Druids.newTimeseriesQueryBuilder().dataSource(CalciteTests.DATASOURCE1).granularity(new PeriodGranularity(Period.days(1), null, DateTimeZone.UTC)).intervals(querySegmentSpec(Filtration.eternity())).aggregators(new CountAggregatorFactory("a0")).context(getTimeseriesContextWithFloorTime(TIMESERIES_CONTEXT_BY_GRAN, "d0")).build())).setInterval(querySegmentSpec(Filtration.eternity())).setGranularity(Granularities.ALL).setAggregatorSpecs(useDefault ? aggregators(new LongMaxAggregatorFactory("_a0", "a0"), new LongMinAggregatorFactory("_a1", "a0"), new LongSumAggregatorFactory("_a2:sum", "a0"), new CountAggregatorFactory("_a2:count"), new LongMaxAggregatorFactory("_a3", "d0"), new CountAggregatorFactory("_a4")) : aggregators(new LongMaxAggregatorFactory("_a0", "a0"), new LongMinAggregatorFactory("_a1", "a0"), new LongSumAggregatorFactory("_a2:sum", "a0"), new FilteredAggregatorFactory(new CountAggregatorFactory("_a2:count"), not(selector("a0", null, null))), new LongMaxAggregatorFactory("_a3", "d0"), new CountAggregatorFactory("_a4"))).setPostAggregatorSpecs(ImmutableList.of(new ArithmeticPostAggregator("_a2", "quotient", ImmutableList.of(new FieldAccessPostAggregator(null, "_a2:sum"), new FieldAccessPostAggregator(null, "_a2:count"))), expressionPostAgg("p0", "timestamp_extract(\"_a3\",'EPOCH','UTC')"))).setContext(QUERY_CONTEXT_DEFAULT).build()), ImmutableList.of(new Object[] { 1L, 1L, 1L, 978480000L, 6L }));
}

Example 57 with FilteredAggregatorFactory

use of org.apache.druid.query.aggregation.FilteredAggregatorFactory in project druid by druid-io.

the class CalciteQueryTest method testExactCountDistinctWithFilter.

@Test
public void testExactCountDistinctWithFilter() throws Exception {
    final String sqlQuery = "SELECT COUNT(DISTINCT foo.dim1) FILTER(WHERE foo.cnt = 1), SUM(foo.cnt) FROM druid.foo";
    // to a bug in the Calcite's rule (AggregateExpandDistinctAggregatesRule)
    try {
        testQuery(PLANNER_CONFIG_NO_HLL.withOverrides(ImmutableMap.of(PlannerConfig.CTX_KEY_USE_GROUPING_SET_FOR_EXACT_DISTINCT, "false")), // Enable exact count distinct
        sqlQuery, CalciteTests.REGULAR_USER_AUTH_RESULT, ImmutableList.of(), ImmutableList.of());
        Assert.fail("query execution should fail");
    } catch (RuntimeException e) {
        Assert.assertTrue(e.getMessage().contains("Error while applying rule AggregateExpandDistinctAggregatesRule"));
    }
    requireMergeBuffers(3);
    testQuery(PLANNER_CONFIG_NO_HLL.withOverrides(ImmutableMap.of(PlannerConfig.CTX_KEY_USE_GROUPING_SET_FOR_EXACT_DISTINCT, "true")), sqlQuery, CalciteTests.REGULAR_USER_AUTH_RESULT, ImmutableList.of(GroupByQuery.builder().setDataSource(new QueryDataSource(GroupByQuery.builder().setDataSource(CalciteTests.DATASOURCE1).setInterval(querySegmentSpec(Filtration.eternity())).setGranularity(Granularities.ALL).setVirtualColumns(expressionVirtualColumn("v0", NullHandling.replaceWithDefault() ? "(\"cnt\" == 1)" : "((\"cnt\" == 1) > 0)", ColumnType.LONG)).setDimensions(dimensions(new DefaultDimensionSpec("dim1", "d0", ColumnType.STRING), new DefaultDimensionSpec("v0", "d1", ColumnType.LONG))).setAggregatorSpecs(aggregators(new LongSumAggregatorFactory("a0", "cnt"), new GroupingAggregatorFactory("a1", Arrays.asList("dim1", "v0")))).setSubtotalsSpec(ImmutableList.of(ImmutableList.of("d0", "d1"), ImmutableList.of())).build())).setInterval(querySegmentSpec(Filtration.eternity())).setGranularity(Granularities.ALL).setAggregatorSpecs(aggregators(new FilteredAggregatorFactory(new CountAggregatorFactory("_a0"), and(not(selector("d0", null, null)), selector("a1", "0", null))), new FilteredAggregatorFactory(new LongMinAggregatorFactory("_a1", "a0"), selector("a1", "3", null)))).setContext(QUERY_CONTEXT_DEFAULT).build()), ImmutableList.of(new Object[] { NullHandling.replaceWithDefault() ? 5L : 6L, 6L }));
}

Example 58 with FilteredAggregatorFactory

use of org.apache.druid.query.aggregation.FilteredAggregatorFactory in project druid by druid-io.

the class StringSqlAggregator method toDruidAggregation.

@Nullable
@Override
public Aggregation toDruidAggregation(PlannerContext plannerContext, RowSignature rowSignature, VirtualColumnRegistry virtualColumnRegistry, RexBuilder rexBuilder, String name, AggregateCall aggregateCall, Project project, List<Aggregation> existingAggregations, boolean finalizeAggregations) {
    final List<DruidExpression> arguments = aggregateCall.getArgList().stream().map(i -> Expressions.fromFieldAccess(rowSignature, project, i)).map(rexNode -> Expressions.toDruidExpression(plannerContext, rowSignature, rexNode)).collect(Collectors.toList());
    if (arguments.stream().anyMatch(Objects::isNull)) {
        return null;
    }
    RexNode separatorNode = Expressions.fromFieldAccess(rowSignature, project, aggregateCall.getArgList().get(1));
    if (!separatorNode.isA(SqlKind.LITERAL)) {
        // separator must be a literal
        return null;
    }
    String separator = RexLiteral.stringValue(separatorNode);
    if (separator == null) {
        // separator must not be null
        return null;
    }
    Integer maxSizeBytes = null;
    if (arguments.size() > 2) {
        RexNode maxBytes = Expressions.fromFieldAccess(rowSignature, project, aggregateCall.getArgList().get(2));
        if (!maxBytes.isA(SqlKind.LITERAL)) {
            // maxBytes must be a literal
            return null;
        }
        maxSizeBytes = ((Number) RexLiteral.value(maxBytes)).intValue();
    }
    final DruidExpression arg = arguments.get(0);
    final ExprMacroTable macroTable = plannerContext.getExprMacroTable();
    final String initialvalue = "[]";
    final ColumnType elementType = ColumnType.STRING;
    final String fieldName;
    if (arg.isDirectColumnAccess()) {
        fieldName = arg.getDirectColumn();
    } else {
        fieldName = virtualColumnRegistry.getOrCreateVirtualColumnForExpression(arg, elementType);
    }
    final String finalizer = StringUtils.format("if(array_length(o) == 0, null, array_to_string(o, '%s'))", separator);
    final NotDimFilter dimFilter = new NotDimFilter(new SelectorDimFilter(fieldName, null, null));
    if (aggregateCall.isDistinct()) {
        return Aggregation.create(// string_agg ignores nulls
        new FilteredAggregatorFactory(new ExpressionLambdaAggregatorFactory(name, ImmutableSet.of(fieldName), null, initialvalue, null, true, false, false, StringUtils.format("array_set_add(\"__acc\", \"%s\")", fieldName), StringUtils.format("array_set_add_all(\"__acc\", \"%s\")", name), null, finalizer, maxSizeBytes != null ? new HumanReadableBytes(maxSizeBytes) : null, macroTable), dimFilter));
    } else {
        return Aggregation.create(// string_agg ignores nulls
        new FilteredAggregatorFactory(new ExpressionLambdaAggregatorFactory(name, ImmutableSet.of(fieldName), null, initialvalue, null, true, false, false, StringUtils.format("array_append(\"__acc\", \"%s\")", fieldName), StringUtils.format("array_concat(\"__acc\", \"%s\")", name), null, finalizer, maxSizeBytes != null ? new HumanReadableBytes(maxSizeBytes) : null, macroTable), dimFilter));
    }
}

Example 59 with FilteredAggregatorFactory

use of org.apache.druid.query.aggregation.FilteredAggregatorFactory in project druid by druid-io.

the class BitwiseSqlAggregator method toDruidAggregation.

@Nullable
@Override
public Aggregation toDruidAggregation(PlannerContext plannerContext, RowSignature rowSignature, VirtualColumnRegistry virtualColumnRegistry, RexBuilder rexBuilder, String name, AggregateCall aggregateCall, Project project, List<Aggregation> existingAggregations, boolean finalizeAggregations) {
    final List<DruidExpression> arguments = aggregateCall.getArgList().stream().map(i -> Expressions.fromFieldAccess(rowSignature, project, i)).map(rexNode -> Expressions.toDruidExpression(plannerContext, rowSignature, rexNode)).collect(Collectors.toList());
    if (arguments.stream().anyMatch(Objects::isNull)) {
        return null;
    }
    final DruidExpression arg = arguments.get(0);
    final ExprMacroTable macroTable = plannerContext.getExprMacroTable();
    final String fieldName;
    if (arg.isDirectColumnAccess()) {
        fieldName = arg.getDirectColumn();
    } else {
        fieldName = virtualColumnRegistry.getOrCreateVirtualColumnForExpression(arg, ColumnType.LONG);
    }
    return Aggregation.create(new FilteredAggregatorFactory(new ExpressionLambdaAggregatorFactory(name, ImmutableSet.of(fieldName), null, "0", null, null, false, false, StringUtils.format("%s(\"__acc\", \"%s\")", op.getDruidFunction(), fieldName), null, null, null, null, macroTable), new NotDimFilter(new SelectorDimFilter(fieldName, null, null))));
}

Example 60 with FilteredAggregatorFactory

use of org.apache.druid.query.aggregation.FilteredAggregatorFactory in project druid by druid-io.

the class CalciteJoinQueryTest method testSemiAndAntiJoinSimultaneouslyUsingWhereInSubquery.

@Test
@Parameters(source = QueryContextForJoinProvider.class)
public void testSemiAndAntiJoinSimultaneouslyUsingWhereInSubquery(Map<String, Object> queryContext) throws Exception {
    cannotVectorize();
    testQuery("SELECT dim1, COUNT(*) FROM foo\n" + "WHERE dim1 IN ('abc', 'def')\n" + "AND __time IN (SELECT MAX(__time) FROM foo)\n" + "AND __time NOT IN (SELECT MIN(__time) FROM foo)\n" + "GROUP BY 1", queryContext, ImmutableList.of(GroupByQuery.builder().setDataSource(join(join(join(new TableDataSource(CalciteTests.DATASOURCE1), new QueryDataSource(Druids.newTimeseriesQueryBuilder().dataSource(CalciteTests.DATASOURCE1).intervals(querySegmentSpec(Filtration.eternity())).granularity(Granularities.ALL).aggregators(new LongMaxAggregatorFactory("a0", "__time")).context(QUERY_CONTEXT_DEFAULT).build()), "j0.", "(\"__time\" == \"j0.a0\")", JoinType.INNER), new QueryDataSource(GroupByQuery.builder().setDataSource(new QueryDataSource(Druids.newTimeseriesQueryBuilder().dataSource(CalciteTests.DATASOURCE1).intervals(querySegmentSpec(Filtration.eternity())).granularity(Granularities.ALL).aggregators(new LongMinAggregatorFactory("a0", "__time")).context(QUERY_CONTEXT_DEFAULT).build())).setInterval(querySegmentSpec(Filtration.eternity())).setGranularity(Granularities.ALL).setAggregatorSpecs(new CountAggregatorFactory("_a0"), NullHandling.sqlCompatible() ? new FilteredAggregatorFactory(new CountAggregatorFactory("_a1"), not(selector("a0", null, null))) : new CountAggregatorFactory("_a1")).setContext(QUERY_CONTEXT_DEFAULT).build()), "_j0.", "1", JoinType.INNER), new QueryDataSource(Druids.newTimeseriesQueryBuilder().dataSource(CalciteTests.DATASOURCE1).intervals(querySegmentSpec(Filtration.eternity())).granularity(Granularities.ALL).aggregators(new LongMinAggregatorFactory("a0", "__time")).postAggregators(expressionPostAgg("p0", "1")).context(QUERY_CONTEXT_DEFAULT).build()), "__j0.", "(\"__time\" == \"__j0.a0\")", JoinType.LEFT)).setInterval(querySegmentSpec(Filtration.eternity())).setGranularity(Granularities.ALL).setDimFilter(and(in("dim1", ImmutableList.of("abc", "def"), null), or(selector("_j0._a0", "0", null), and(selector("__j0.p0", null, null), expressionFilter("(\"_j0._a1\" >= \"_j0._a0\")"))))).setDimensions(dimensions(new DefaultDimensionSpec("dim1", "d0", ColumnType.STRING))).setAggregatorSpecs(aggregators(new CountAggregatorFactory("a0"))).setContext(queryContext).build()), ImmutableList.of(new Object[] { "abc", 1L }));
}