viirya commented on a change in pull request #28560:
URL: https://github.com/apache/spark/pull/28560#discussion_r426343264
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/NestedColumnAliasing.scala
##########
@@ -68,10 +76,23 @@ object NestedColumnAliasing {
*/
def replaceChildrenWithAliases(
plan: LogicalPlan,
+ nestedFieldToAlias: Map[ExtractValue, Alias],
attrToAliases: Map[ExprId, Seq[Alias]]): LogicalPlan = {
plan.withNewChildren(plan.children.map { plan =>
Project(plan.output.flatMap(a => attrToAliases.getOrElse(a.exprId,
Seq(a))), plan)
- })
+ }).transformExpressions {
+ case f: ExtractValue if nestedFieldToAlias.contains(f) =>
+ nestedFieldToAlias(f).toAttribute
+ }
+ }
+
+ /**
+ * Returns true for those operators that we can prune nested column on it.
+ */
+ private def canPruneOn(plan: LogicalPlan) = plan match {
+ case _: Aggregate => true
+ case _: Expand => true
+ case _ => false
Review comment:
I think FlatMapGroupsInPandas is also supported.
As you see, we do `transformExpressions` to replace `ExtractValue` in
operator's expressions. So if the operator follows SparkSQL's operators, it
should be fine.
This patch adds the change needed for general nested column pruning for the
kind of operators which can prune nested column. Ideally we only need to add it
in `canPruneOn`, and add test for it.
Currently I think nested column pruning test cases are all in Scala, no
Python. I will think about how to add test.
----------------------------------------------------------------
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
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
