Merge pull request #7013 from leiysky/fix-prune-columns

fix(planner): Fix prune projection

query/src/sql/optimizer/heuristic/prune_columns.rs

@@ -116,7 +116,13 @@ impl ColumnPruner {
                 ))
             }
             RelOperator::Project(p) => {
-                let used: ColumnSet = p.columns.intersection(&required).cloned().collect();
+                let mut used: ColumnSet = p.columns.intersection(&required).cloned().collect();
+                if used.is_empty() {
+                    // Keep at least one column for project.
+                    used.insert(*p.columns.iter().sorted().take(1).next().ok_or_else(|| {
+                        ErrorCode::LogicalError("Invalid Project without output column")
+                    })?);
+                }
                 Ok(SExpr::create_unary(
                     RelOperator::Project(Project {
                         columns: used.clone(),

query/tests/it/sql/optimizer/heuristic/prune_columns.rs

@@ -60,6 +60,11 @@ pub async fn test_heuristic_optimizer_prune_columns() -> Result<()> {
             query: "select name from system.functions order by example".to_string(),
             rules: vec![],
         },
+        Suite {
+            comment: "# Prune unused columns with cross join".to_string(),
+            query: "select t.number from numbers(10) t where exists(select * from numbers(10))".to_string(),
+            rules: vec![],
+        },
     ];
 
     run_suites(ctx, &mut file, &suites, run_test).await

query/tests/it/sql/optimizer/heuristic/testdata/prune_columns.test

@@ -88,3 +88,20 @@ Project: [name (#0)]
             Scan: default.system.functions
 
 
+# Prune unused columns with cross join
+select t.number from numbers(10) t where exists(select * from numbers(10))
+----
+Project: [number (#0)]
+    EvalScalar: [t.number (#0)]
+        Filter: [subquery_3 (#3)]
+            CrossJoin
+                Scan: default.system.numbers
+                Project: [subquery (#3)]
+                    EvalScalar: [count(*) (#2) = 1]
+                        Aggregate(Initial): group items: [], aggregate functions: [count(*)]
+                            Limit: [1], Offset: [0]
+                                Project: [number (#1)]
+                                    EvalScalar: [numbers.number (#1)]
+                                        Scan: default.system.numbers
+
+

query/tests/it/sql/optimizer/heuristic/testdata/subquery.test

@@ -36,7 +36,7 @@ select t.number from numbers(1) as t where exists (select * from numbers(1) wher
 Project: [number (#0)]
     CrossJoin
         Scan: default.system.numbers
-        Project: []
+        Project: [subquery (#3)]
             Filter: [subquery_3 (#3)]
                 EvalScalar: [count(*) (#2) = 1]
                     Aggregate(Final): group items: [], aggregate functions: [count(*)]

tests/logictest/suites/base/15_query/subquery.test

@@ -615,6 +615,16 @@ ORDER BY c.c_id
 5  0
 6  0
 
+statement query I
+select * from numbers(5) where exists(select * from numbers(1)) order by number;
+
+----
+0
+1
+2
+3
+4
+
 
 statement ok
 drop table o;