chore: try_eliminate_in_subquery skip if left has join or child expr is nullable

Signed-off-by: Kould <[email protected]>

Author: KKould

src/query/sql/src/planner/optimizer/decorrelate/subquery_rewriter.rs

@@ -570,6 +570,10 @@ impl SubqueryRewriter {
         else {
             return Ok(None);
         };
+        // `xx in null` != `where xx = null`
+        if left_column.column.data_type.is_nullable() {
+            return Ok(None);
+        }
         let (Some(left_table_index), Some(right_table_index)) = (
             left_column.column.table_index,
             right_expr_binding.table_index,
@@ -583,6 +587,11 @@ impl SubqueryRewriter {
         ) else {
             return Ok(None);
         };
+        // when left is a join, filtering becomes complicated. If in subquery is eliminated,
+        // the filter will be pushed down to Scan before join, resulting in inaccurate data results.
+        if left.has_join() {
+            return Ok(None);
+        }
         if left_source_binding != right_source_binding {
             return Ok(None);
         }
@@ -592,10 +601,9 @@ impl SubqueryRewriter {
             let left_columns = guard.columns_by_table_index(left_table_index);
             let right_columns = guard.columns_by_table_index(right_table_index);
             let left_table = guard.table(left_table_index);
+            let right_table = guard.table(right_table_index);
             // filter table function
-            if left_table.database() == "system"
-                || guard.table(right_table_index).database() == "system"
-            {
+            if left_table.database() == "system" || right_table.database() == "system" {
                 return Ok(None);
             }
             let left_source_table_index = guard

src/query/sql/src/planner/optimizer/s_expr.rs

@@ -426,6 +426,13 @@ impl SExpr {
         }
         self.children.iter().any(|child| child.has_merge_exchange())
     }
+    #[recursive::recursive]
+    pub fn has_join(&self) -> bool {
+        if let RelOperator::Join(_) = self.plan.as_ref() {
+            return true;
+        }
+        self.children.iter().any(|child| child.has_join())
+    }
 }
 
 fn find_subquery(rel_op: &RelOperator) -> bool {

tests/sqllogictests/suites/mode/standalone/explain/explain.test

@@ -1330,10 +1330,10 @@ statement ok
 drop table t2;
 
 statement ok
-create OR REPLACE table t1(a int, b int, c varchar(20));
+create OR REPLACE table t1(a int not null, b int, c varchar(20));
 
 statement ok
-create OR REPLACE table t2(a int, b int, c varchar(20));
+create OR REPLACE table t2(a int not null, b int, c varchar(20));
 
 # eliminate subquery
 query T
@@ -1350,7 +1350,7 @@ Filter
     ├── read size: 0
     ├── partitions total: 0
     ├── partitions scanned: 0
-    ├── push downs: [filters: [and_filters(t1.c (#2) = 'D', t1.b (#1) = 7)], limit: NONE]
+    ├── push downs: [filters: [and_filters(t1.b (#1) = 7, t1.c (#2) = 'D')], limit: NONE]
     └── estimated rows: 0.00
 
 

tests/sqllogictests/suites/query/subquery.test

@@ -944,10 +944,10 @@ WHERE t1.i1 = (
 1
 
 statement ok
-create or replace table t1(a int, b int, c varchar(20));
+create or replace table t1(a int not null, b int, c varchar(20));
 
 statement ok
-create or replace table t2(a int, b int, c varchar(20));
+create or replace table t2(a int not null, b int, c varchar(20));
 
 statement ok
 insert into t1 (a, b, c) values (1, 7, 'D'), (2, 7, 'D'), (3, 7, 'D'), (4, 8, 'D'), (5, 7, 'E'), (6, 8, 'E');