[GitHub] [spark] maropu commented on a change in pull request #26918: [SPARK-30279][SQL] Support 32 or more grouping attributes for GROUPING_ID

[GitBox <gi...@apache.org>]

maropu commented on a change in pull request #26918: [SPARK-30279][SQL] Support 32 or more grouping attributes for GROUPING_ID 
URL: https://github.com/apache/spark/pull/26918#discussion_r359188374
 
 

 ##########
 File path: sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/Analyzer.scala
 ##########
 @@ -563,15 +570,44 @@ class Analyzer(
       }
     }
 
+    private object EmptyGroupingIDExtractor {
+
+      private def hasEmptyGroupingID(aggExprs: Seq[Expression]): Boolean = {
+        aggExprs.exists { p => p.collectFirst { case GroupingID(Nil) => true }.isDefined }
+      }
+
+      def unapply(p: LogicalPlan): Option[Seq[Expression]] = p match {
+        case Aggregate(Seq(Cube(groupByExprs)), aggExprs, _) if hasEmptyGroupingID(aggExprs) =>
+          Some(constructGroupBy(cubeExprs(groupByExprs)))
+        case Aggregate(Seq(Rollup(groupByExprs)), aggExprs, _) if hasEmptyGroupingID(aggExprs) =>
+          Some(constructGroupBy(rollupExprs(groupByExprs)))
+        case x: GroupingSets if hasEmptyGroupingID(x.aggregations) =>
+          Some(constructGroupBy(x.selectedGroupByExprs))
+        case Filter(cond, child) if hasEmptyGroupingID(cond :: Nil) =>
+          Some(findGroupingExprs(child))
+        case Sort(order, _, child) if hasEmptyGroupingID(order) =>
+          Some(findGroupingExprs(child))
+        case _ =>
+          None
+      }
+    }
+
     // This require transformUp to replace grouping()/grouping_id() in resolved Filter/Sort
     def apply(plan: LogicalPlan): LogicalPlan = plan resolveOperatorsUp {
       case a if !a.childrenResolved => a // be sure all of the children are resolved.
 
+      // If a plan has unresolved grouping IDs, we resolve them first
+      case p @ EmptyGroupingIDExtractor(groupByExprs) =>
+        p.transformExpressions {
+          // In case that `GroupingID` has an empty group-by expressions
+          case gid: GroupingID if !gid.resolved => GroupingID(groupByExprs)
+        }
 
 Review comment:
   As you said, I think now we don't need this handling for string grouping IDs. Rather, we need to use `bigint` for grouping IDs intead of `int`.

----------------------------------------------------------------
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:
users@infra.apache.org


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org