Github user dongjoon-hyun commented on a diff in the pull request:
https://github.com/apache/spark/pull/13906#discussion_r68838170
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
---
@@ -1053,6 +1055,49 @@ object PruneFilters extends Rule[LogicalPlan] with
PredicateHelper {
}
/**
+ * Collapse plans consisting all empty local relations generated by
[[PruneFilters]].
+ * Note that the ObjectProducer/Consumer and direct aggregations are the
exceptions.
+ * {{{
+ * SELECT a, b FROM t WHERE 1=0 GROUP BY a, b ORDER BY a, b ==> empty
result
+ * SELECT SUM(a) FROM t WHERE 1=0 GROUP BY a HAVING COUNT(*)>1 ORDER BY
a (Not optimized)
+ * }}}
+ */
+object CollapseEmptyPlan extends Rule[LogicalPlan] with PredicateHelper {
+ private def isEmptyLocalRelation(plan: LogicalPlan): Boolean =
+ plan.isInstanceOf[LocalRelation] &&
plan.asInstanceOf[LocalRelation].data.isEmpty
+
+ def apply(plan: LogicalPlan): LogicalPlan = plan transformUp {
+ case x if x.isInstanceOf[ObjectProducer] ||
x.isInstanceOf[ObjectConsumer] => x
+
+ // Case 1: If groupingExpressions contains all aggregation
expressions, the result is empty.
+ case a @ Aggregate(ge, ae, child) if isEmptyLocalRelation(child) &&
ae.forall(ge.contains(_)) =>
+ LocalRelation(a.output, data = Seq.empty)
+
+ // Case 2: General aggregations can generate non-empty results.
+ case a: Aggregate => a
+
+ // Case 3: The following plans having only empty relations return
empty results.
+ case p: LogicalPlan if p.children.nonEmpty &&
p.children.forall(isEmptyLocalRelation) =>
+ p match {
+ case _: Project | _: Generate | _: Filter | _: Sample | _: Join |
+ _: Sort | _: GlobalLimit | _: LocalLimit |
+ _: Distinct | _: Except | _: Union |
+ _: Repartition =>
+ LocalRelation(p.output, data = Seq.empty)
+ case _ => p
+ }
+
+ // Case 4: The following plans having at least one empty relation
return empty results.
+ case p: LogicalPlan if p.children.exists(isEmptyLocalRelation) =>
+ p match {
+ case Join(_, _, Inner, _) | _: Intersect =>
--- End diff --
Sure!
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
