[SPARK-41985][SQL][FOLLOWUP] Remove alias in GROUP BY only when the e…

…xpr is resolved

### What changes were proposed in this pull request?

This is a followup of apache#39508 to fix a regression. We should not remove aliases from grouping expressions if they are not resolved, as the alias may be necessary for resolution, such as `CreateNamedStruct`.

### Why are the changes needed?

fix a regression

### Does this PR introduce _any_ user-facing change?

no

### How was this patch tested?

new test

Closes apache#39867 from cloud-fan/column.

Lead-authored-by: Wenchen Fan <wenchen@databricks.com>
Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
Signed-off-by: Max Gekk <max.gekk@gmail.com>

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/ResolveReferencesInAggregate.scala

@@ -96,7 +96,13 @@ object ResolveReferencesInAggregate extends SQLConfHelper
       // can't find the grouping expressions via `semanticEquals` and the analysis will fail.
       // Example rules: ResolveGroupingAnalytics (See SPARK-31670 for more details) and
       // ResolveLateralColumnAliasReference.
-      groupingExpressions = resolvedGroupExprs.map(trimAliases),
+      groupingExpressions = resolvedGroupExprs.map { e =>
+        // Only trim the alias if the expression is resolved, as the alias may be needed to resolve
+        // the expression, such as `NamePlaceHolder` in `CreateNamedStruct`.
+        // Note: this rule will be invoked even if the Aggregate is fully resolved. So alias in
+        //       GROUP BY will be removed eventually, by following iterations.
+        if (e.resolved) trimAliases(e) else e
+      },
       aggregateExpressions = resolvedAggExprsWithOuter)
   }
 

sql/core/src/test/resources/sql-tests/inputs/group-by.sql

@@ -34,6 +34,9 @@ SELECT a + b, COUNT(b) FROM testData GROUP BY a + b;
 SELECT a + 2, COUNT(b) FROM testData GROUP BY a + 1;
 SELECT a + 1 + 1, COUNT(b) FROM testData GROUP BY a + 1;
 
+-- struct() in group by
+SELECT count(1) FROM testData GROUP BY struct(a + 0.1 AS aa);
+
 -- Aggregate with nulls.
 SELECT SKEWNESS(a), KURTOSIS(a), MIN(a), MAX(a), AVG(a), VARIANCE(a), STDDEV(a), SUM(a), COUNT(a)
 FROM testData;

sql/core/src/test/resources/sql-tests/results/group-by.sql.out

@@ -145,6 +145,17 @@ struct<((a + 1) + 1):int,count(b):bigint>
 NULL	1
 
 
+-- !query
+SELECT count(1) FROM testData GROUP BY struct(a + 0.1 AS aa)
+-- !query schema
+struct<count(1):bigint>
+-- !query output
+2
+2
+2
+3
+
+
 -- !query
 SELECT SKEWNESS(a), KURTOSIS(a), MIN(a), MAX(a), AVG(a), VARIANCE(a), STDDEV(a), SUM(a), COUNT(a)
 FROM testData