Merge pull request #10602 from xudong963/fix_mark_join

fix: incorrectly remove mark index

Author: xudong963

src/query/sql/src/planner/optimizer/rule/rewrite/filter_join/mark_join_to_semi_join.rs

@@ -32,11 +32,13 @@ pub fn convert_mark_to_semi_join(s_expr: &SExpr) -> Result<SExpr> {
     }
 
     let mark_index = join.marker_index.unwrap();
+    let mut find_mark_index = false;
 
     // remove mark index filter
     for (idx, predicate) in filter.predicates.iter().enumerate() {
         if let ScalarExpr::BoundColumnRef(col) = predicate {
             if col.column.index == mark_index {
+                find_mark_index = true;
                 filter.predicates.remove(idx);
                 break;
             }
@@ -51,6 +53,11 @@ pub fn convert_mark_to_semi_join(s_expr: &SExpr) -> Result<SExpr> {
         }
     }
 
+    if !find_mark_index {
+        // To be conservative, we do not convert
+        return Ok(s_expr.clone());
+    }
+
     join.join_type = match join.join_type {
         JoinType::LeftMark => JoinType::RightSemi,
         JoinType::RightMark => JoinType::LeftSemi,

tests/sqllogictests/suites/query/subquery.test

@@ -39,6 +39,19 @@ SELECT * FROM c WHERE EXISTS(SELECT * FROM o WHERE o.c_id=c.c_id) ORDER BY c_id
 4  TX
 6  FL
 
+query IT
+SELECT * FROM c, o WHERE c.c_id = o.c_id AND (EXISTS(SELECT * FROM o WHERE o.c_id=c.c_id) OR EXISTS(SELECT * FROM o where o.ship = c.bill)) ORDER BY c_id
+----
+1  CA   10 1  CA   
+1  CA   20 1  CA   
+1  CA   30 1  CA   
+2  TX   40 2  CA   
+2  TX   50 2  TX   
+2  TX   60 2  NULL 
+4  TX   70 4  WY
+4  TX   80 4  NULL 
+6  FL   90 6  WA   
+
 query IT
 SELECT * FROM c WHERE NOT EXISTS(SELECT * FROM o WHERE o.c_id=c.c_id) ORDER BY c_id
 ----