anchovYu commented on code in PR #39040:
URL: https://github.com/apache/spark/pull/39040#discussion_r1051251048
##########
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/ResolveLateralColumnAliasReference.scala:
##########
@@ -129,6 +163,45 @@ object ResolveLateralColumnAliasReference extends
Rule[LogicalPlan] {
child = Project(innerProjectList.toSeq, child)
)
}
+
+ case agg @ Aggregate(groupingExpressions, aggregateExpressions, _) if
agg.resolved
+ &&
aggregateExpressions.exists(_.containsPattern(LATERAL_COLUMN_ALIAS_REFERENCE))
=>
+
+ val newAggExprs = collection.mutable.Set.empty[NamedExpression]
+ val expressionMap =
collection.mutable.LinkedHashMap.empty[Expression, NamedExpression]
+ val projectExprs = aggregateExpressions.map { exp =>
+ exp.transformDown {
+ case aggExpr: AggregateExpression =>
+ // Doesn't support referencing a lateral alias in aggregate
function
+ if (aggExpr.containsPattern(LATERAL_COLUMN_ALIAS_REFERENCE)) {
+ aggExpr.collectFirst {
+ case lcaRef: LateralColumnAliasReference =>
+ throw
QueryCompilationErrors.lateralColumnAliasInAggFuncUnsupportedError(
+ lcaRef.nameParts, aggExpr)
+ }
+ }
+ val ne = expressionMap.getOrElseUpdate(aggExpr.canonicalized,
assignAlias(aggExpr))
+ newAggExprs += ne
+ ne.toAttribute
+ case e if groupingExpressions.exists(_.semanticEquals(e)) =>
+ // TODO one concern here, is condition here be able to match
all grouping
Review Comment:
I surprisingly found out that this existing query can't analyze:
```
select 1 + dept + 10 from $testTable group by dept + 10
-- error: [MISSING_AGGREGATION] The non-aggregating expression "dept" is
based on columns which are not participating in the GROUP BY clause
```
Seems in our checkAnalysis, we don't canonicalize to compare the
expressions. It is structured as (1 + dept) + 10, and can't match the grouping
expression (dept + 10).
--
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]
