From 2c73d2a948bdde798aaf0f87c18846281deb05fd Mon Sep 17 00:00:00 2001 From: Josh Rosen Date: Mon, 8 Jan 2018 16:04:03 +0800 Subject: [PATCH] [SPARK-22983] Don't push filters beneath aggregates with empty grouping expressions ## What changes were proposed in this pull request? The following SQL query should return zero rows, but in Spark it actually returns one row: ``` SELECT 1 from ( SELECT 1 AS z, MIN(a.x) FROM (select 1 as x) a WHERE false ) b where b.z != b.z ``` The problem stems from the `PushDownPredicate` rule: when this rule encounters a filter on top of an Aggregate operator, e.g. `Filter(Agg(...))`, it removes the original filter and adds a new filter onto Aggregate's child, e.g. `Agg(Filter(...))`. This is sometimes okay, but the case above is a counterexample: because there is no explicit `GROUP BY`, we are implicitly computing a global aggregate over the entire table so the original filter was not acting like a `HAVING` clause filtering the number of groups: if we push this filter then it fails to actually reduce the cardinality of the Aggregate output, leading to the wrong answer. In 2016 I fixed a similar problem involving invalid pushdowns of data-independent filters (filters which reference no columns of the filtered relation). There was additional discussion after my fix was merged which pointed out that my patch was an incomplete fix (see #15289), but it looks I must have either misunderstood the comment or forgot to follow up on the additional points raised there. This patch fixes the problem by choosing to never push down filters in cases where there are no grouping expressions. Since there are no grouping keys, the only columns are aggregate columns and we can't push filters defined over aggregate results, so this change won't cause us to miss out on any legitimate pushdown opportunities. ## How was this patch tested? New regression tests in `SQLQueryTestSuite` and `FilterPushdownSuite`. Author: Josh Rosen Closes #20180 from JoshRosen/SPARK-22983-dont-push-filters-beneath-aggs-with-empty-grouping-expressions. --- .../spark/sql/catalyst/optimizer/Optimizer.scala | 3 ++- .../catalyst/optimizer/FilterPushdownSuite.scala | 13 +++++++++++++ .../test/resources/sql-tests/inputs/group-by.sql | 9 +++++++++ .../resources/sql-tests/results/group-by.sql.out | 16 +++++++++++++++- 4 files changed, 39 insertions(+), 2 deletions(-) diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala index 0d4b02c6e7d8a..df0af8264a329 100644 --- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala +++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala @@ -795,7 +795,8 @@ object PushDownPredicate extends Rule[LogicalPlan] with PredicateHelper { project.copy(child = Filter(replaceAlias(condition, aliasMap), grandChild)) case filter @ Filter(condition, aggregate: Aggregate) - if aggregate.aggregateExpressions.forall(_.deterministic) => + if aggregate.aggregateExpressions.forall(_.deterministic) + && aggregate.groupingExpressions.nonEmpty => // Find all the aliased expressions in the aggregate list that don't include any actual // AggregateExpression, and create a map from the alias to the expression val aliasMap = AttributeMap(aggregate.aggregateExpressions.collect { diff --git a/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/FilterPushdownSuite.scala b/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/FilterPushdownSuite.scala index 85a5e979f6021..82a10254d846d 100644 --- a/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/FilterPushdownSuite.scala +++ b/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/FilterPushdownSuite.scala @@ -809,6 +809,19 @@ class FilterPushdownSuite extends PlanTest { comparePlans(optimized, correctAnswer) } + test("aggregate: don't push filters if the aggregate has no grouping expressions") { + val originalQuery = LocalRelation.apply(testRelation.output, Seq.empty) + .select('a, 'b) + .groupBy()(count(1)) + .where(false) + + val optimized = Optimize.execute(originalQuery.analyze) + + val correctAnswer = originalQuery.analyze + + comparePlans(optimized, correctAnswer) + } + test("broadcast hint") { val originalQuery = ResolvedHint(testRelation) .where('a === 2L && 'b + Rand(10).as("rnd") === 3) diff --git a/sql/core/src/test/resources/sql-tests/inputs/group-by.sql b/sql/core/src/test/resources/sql-tests/inputs/group-by.sql index 1e1384549a410..c5070b734d521 100644 --- a/sql/core/src/test/resources/sql-tests/inputs/group-by.sql +++ b/sql/core/src/test/resources/sql-tests/inputs/group-by.sql @@ -60,3 +60,12 @@ SELECT a, COUNT(1) FROM testData WHERE false GROUP BY a; -- Aggregate with empty input and empty GroupBy expressions. SELECT COUNT(1) FROM testData WHERE false; SELECT 1 FROM (SELECT COUNT(1) FROM testData WHERE false) t; + +-- Aggregate with empty GroupBy expressions and filter on top +SELECT 1 from ( + SELECT 1 AS z, + MIN(a.x) + FROM (select 1 as x) a + WHERE false +) b +where b.z != b.z diff --git a/sql/core/src/test/resources/sql-tests/results/group-by.sql.out b/sql/core/src/test/resources/sql-tests/results/group-by.sql.out index 986bb01c13fe4..c1abc6dff754b 100644 --- a/sql/core/src/test/resources/sql-tests/results/group-by.sql.out +++ b/sql/core/src/test/resources/sql-tests/results/group-by.sql.out @@ -1,5 +1,5 @@ -- Automatically generated by SQLQueryTestSuite --- Number of queries: 25 +-- Number of queries: 26 -- !query 0 @@ -227,3 +227,17 @@ SELECT 1 FROM (SELECT COUNT(1) FROM testData WHERE false) t struct<1:int> -- !query 24 output 1 + + +-- !query 25 +SELECT 1 from ( + SELECT 1 AS z, + MIN(a.x) + FROM (select 1 as x) a + WHERE false +) b +where b.z != b.z +-- !query 25 schema +struct<1:int> +-- !query 25 output +