GROUP-BY prioritizes input columns in case of ambiguity (#9228)

* GROUP-BY prioritizes input columns in case of ambiguity * Update datafusion/sqllogictest/test_files/aggregate.slt Co-authored-by: Andrew Lamb <[email protected]> * Update datafusion/sqllogictest/test_files/aggregate.slt Co-authored-by: Andrew Lamb <[email protected]> --------- Co-authored-by: Andrew Lamb <[email protected]>
apache · Feb 16, 2024 · 40353fe · 40353fe
1 parent e4f4031
commit 40353fe
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 2 deletions.
diff --git a/datafusion/sql/src/select.rs b/datafusion/sql/src/select.rs
@@ -88,8 +88,11 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
 
         // having and group by clause may reference aliases defined in select projection
         let projected_plan = self.project(base_plan.clone(), select_exprs.clone())?;
-        let mut combined_schema = (**projected_plan.schema()).clone();
-        combined_schema.merge(base_plan.schema());
+        // Place the fields of the base plan at the front so that when there are references
+        // with the same name, the fields of the base plan will be searched first.
+        // See https://github.com/apache/arrow-datafusion/issues/9162
+        let mut combined_schema = base_plan.schema().as_ref().clone();
+        combined_schema.merge(projected_plan.schema());
 
         // this alias map is resolved and looked up in both having exprs and group by exprs
         let alias_map = extract_aliases(&select_exprs);

diff --git a/datafusion/sqllogictest/test_files/aggregate.slt b/datafusion/sqllogictest/test_files/aggregate.slt
@@ -3179,3 +3179,23 @@ NULL
 statement ok
 DROP TABLE t;
 
+
+# Test for the case when the column name is ambiguous
+statement ok
+CREATE TABLE t(a BIGINT) AS VALUES(1), (2), (3);
+
+# The column name referenced by GROUP-BY is ambiguous, prefer the column in base plan
+query I
+SELECT 0 as "t.a" FROM t GROUP BY t.a;
+----
+0
+0
+0
+
+# The column name referenced by HAVING is ambiguous, prefer the column in the base plan
+query I
+SELECT 0 AS "t.a" FROM t HAVING MAX(t.a) = 0;
+----
+
+statement ok
+DROP TABLE t;