fix: group by two same columns (#84)

Author: gandronchik

datafusion/core/src/logical_plan/builder.rs

@@ -868,11 +868,18 @@ impl LogicalPlanBuilder {
     /// value of the `group_expr`;
     pub fn aggregate(
         &self,
-        group_expr: impl IntoIterator<Item = impl Into<Expr>>,
-        aggr_expr: impl IntoIterator<Item = impl Into<Expr>>,
+        group_exprs: impl IntoIterator<Item = impl Into<Expr>>,
+        aggr_exprs: impl IntoIterator<Item = impl Into<Expr>>,
     ) -> Result<Self> {
-        let group_expr = normalize_cols(group_expr, &self.plan)?;
-        let aggr_expr = normalize_cols(aggr_expr, &self.plan)?;
+        let mut group_expr: Vec<Expr> = vec![];
+        normalize_cols(group_exprs, &self.plan)?
+            .iter()
+            .for_each(|e| {
+                if !group_expr.contains(e) {
+                    group_expr.push(e.clone());
+                }
+            });
+        let aggr_expr = normalize_cols(aggr_exprs, &self.plan)?;
         let all_expr = group_expr.iter().chain(aggr_expr.iter());
         validate_unique_names("Aggregations", all_expr.clone(), self.plan.schema())?;
         let aggr_schema = DFSchema::new_with_metadata(

datafusion/core/src/sql/planner.rs

@@ -3279,6 +3279,17 @@ mod tests {
         quick_test(sql, expected);
     }
 
+    #[test]
+    fn select_aggregate_with_group_by_with_two_same_exprs() {
+        let sql = "SELECT first_name as name, first_name as nickname, 1 as first_num, 1 as second_num
+                   FROM person
+                   GROUP BY 1, 2, 3, 4";
+        let expected = "Projection: #person.first_name AS name, #person.first_name AS nickname, #Int64(1) AS first_num, #Int64(1) AS second_num\
+                        \n  Aggregate: groupBy=[[#person.first_name, Int64(1)]], aggr=[[]]\
+                        \n    TableScan: person projection=None";
+        quick_test(sql, expected);
+    }
+
     #[test]
     fn select_binary_expr() {
         let sql = "SELECT age + salary from person";