sigmod commented on code in PR #35864:
URL: https://github.com/apache/spark/pull/35864#discussion_r842205189
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/NestedColumnAliasing.scala:
##########
@@ -312,6 +312,32 @@ object NestedColumnAliasing {
}
}
+object GeneratorUnrequiredChildrenPruning {
+ def unapply(plan: LogicalPlan): Option[LogicalPlan] = plan match {
+ case p@Project(_, g: Generate) =>
+ val requiredAttrs = p.references ++ g.generator.references
+ var pruned = false
+ val newChild = if (!g.child.outputSet.subsetOf(requiredAttrs)) {
+ pruned = true
+ Project(g.child.output.filter(requiredAttrs.contains), g.child)
+ } else {
+ g.child
+ }
+ val unrequired = g.generator.references -- p.references
+ val unrequiredIndices = newChild.output.zipWithIndex.filter(t =>
unrequired.contains(t._1))
+ .map(_._2)
+ if (unrequiredIndices.toSet != g.unrequiredChildIndex.toSet) {
+ pruned = true
+ }
+ if (pruned) {
+ Some(p.copy(child = g.copy(child = newChild, unrequiredChildIndex =
unrequiredIndices)))
+ } else {
+ None
+ }
Review Comment:
Is it possible to not replicate the logic of `prunedChild`?
`prunedChild` is used by every other branches. In case the logic in
`prunedChild` evolve in the future, this call site will not get the benefit?
>> I don't think g.producedAttributes should be removed from the
requiredAttrs list.
iiuc, `g.producedAttributes` shouldn't exist in `g.child.output`? If so,
why do we need to include them in `requiredAttrs`?
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
