maropu commented on a change in pull request #27058: [SPARK-30276][SQL] Support
Filter expression allows simultaneous use of DISTINCT
URL: https://github.com/apache/spark/pull/27058#discussion_r368809725
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/RewriteDistinctAggregates.scala
##########
@@ -148,24 +207,106 @@ object RewriteDistinctAggregates extends
Rule[LogicalPlan] {
val distinctAggs = exprs.flatMap { _.collect {
case ae: AggregateExpression if ae.isDistinct => ae
}}
- // We need at least two distinct aggregates for this rule because
aggregation
- // strategy can handle a single distinct group.
+ // This rule serves two purposes:
+ // One is to rewrite when there exists at least two distinct aggregates.
We need at least
+ // two distinct aggregates for this rule because aggregation strategy can
handle a single
+ // distinct group.
+ // Another is to expand distinct aggregates which exists filter clause so
that we can
+ // evaluate the filter locally.
// This check can produce false-positives, e.g., SUM(DISTINCT a) &
COUNT(DISTINCT a).
- distinctAggs.size > 1
+ distinctAggs.size >= 1 || distinctAggs.exists(_.filter.isDefined)
}
def apply(plan: LogicalPlan): LogicalPlan = plan transformUp {
- case a: Aggregate if mayNeedtoRewrite(a.aggregateExpressions) => rewrite(a)
+ case a: Aggregate if mayNeedtoRewrite(a.aggregateExpressions) =>
+ val expandAggregate = extractFiltersInDistinctAggregate(a)
+ rewriteDistinctAggregate(expandAggregate)
}
- def rewrite(a: Aggregate): Aggregate = {
+ private def extractFiltersInDistinctAggregate(a: Aggregate): Aggregate = {
+ val aggExpressions = collectAggregateExprs(a)
+ val (distinctAggExpressions, regularAggExpressions) =
aggExpressions.partition(_.isDistinct)
+ if (distinctAggExpressions.exists(_.filter.isDefined)) {
+ // Setup expand for the 'regular' aggregate expressions. Because we will
construct a new
Review comment:
Setup expand? It seems `Constructs a pair between old and new expressions
for regular aggregates`?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
