fix(optimizer): avoid merging subquery with JOIN when outer query uses JOIN (#5999)

* fix(optimizer): prevent merging subqueries with joins when outer query uses a join

* add test

Author: geooo109

sqlglot/optimizer/merge_subqueries.py

@@ -201,6 +201,7 @@ def _is_recursive():
         and not outer_scope.pivots
         and not any(e.find(exp.AggFunc, exp.Select, exp.Explode) for e in inner_select.expressions)
         and not (leave_tables_isolated and len(outer_scope.selected_sources) > 1)
+        and not (isinstance(from_or_join, exp.Join) and inner_select.args.get("joins"))
         and not (
             isinstance(from_or_join, exp.Join)
             and inner_select.args.get("where")
@@ -283,13 +284,7 @@ def _merge_joins(outer_scope: Scope, inner_scope: Scope, from_or_join: FromOrJoi
 
     joins = inner_scope.expression.args.get("joins") or []
 
-    outer_is_left_join = isinstance(from_or_join, exp.Join) and from_or_join.side == "LEFT"
-
     for join in joins:
-        if outer_is_left_join and not join.method and join.kind in ("", "INNER"):
-            join.args.pop("kind", None)
-            join.set("side", "LEFT")
-
         new_joins.append(join)
         outer_scope.add_source(join.alias_or_name, inner_scope.sources[join.alias_or_name])
 

tests/fixtures/optimizer/merge_subqueries.sql

@@ -512,85 +512,30 @@ with cte as (
 select cte.mult from cte;
 SELECT x.a * x.b AS mult FROM x AS x;
 
-# title: replace INNER JOIN with LEFT JOIN when it exists in a subquery used as the RHS of a LEFT JOIN
-WITH 
-t0 AS (
-  SELECT 5 as id
-),
-t1 AS (
-  SELECT 1 AS id, 'US' AS cid
-),
-t2 AS (
-  SELECT 1 AS id, 'US' AS cid
-)
-SELECT
-  t0.id,
-  t3.cid AS cid
-FROM t0
-LEFT JOIN (
+# title: avoid merging subquery with JOIN
+WITH t0 AS (
   SELECT
-    t1.id,
-    t2.cid
-  FROM t1
-  INNER JOIN t2
-    ON t1.cid = t2.cid
-) AS t3
-  ON t0.id = t3.id;
-WITH t0 AS (SELECT 5 AS id), t1 AS (SELECT 1 AS id, 'US' AS cid), t2 AS (SELECT 1 AS id, 'US' AS cid) SELECT t0.id AS id, t2.cid AS cid FROM t0 AS t0 LEFT JOIN t1 AS t1 ON t0.id = t1.id LEFT JOIN t2 AS t2 ON t1.cid = t2.cid;
-
-# title: preserve INNER JOIN when it exists in a subquery used as the RHS of a RIGHT JOIN
-WITH 
-t0 AS (
-  SELECT 5 as id
-),
-t1 AS (
-  SELECT 1 AS id, 'US' AS cid
-),
-t2 AS (
-  SELECT 1 AS id, 'US' AS cid
+    5 AS id
+), t1 AS (
+  SELECT
+    1 AS id,
+    'US' AS cid
+), t2 AS (
+  SELECT
+    1 AS id,
+    'US' AS cid
 )
 SELECT
   t0.id,
   t3.cid AS cid
 FROM t0
-RIGHT JOIN (
+INNER JOIN (
   SELECT
     t1.id,
     t2.cid
   FROM t1
-  INNER JOIN t2
+  RIGHT JOIN t2
     ON t1.cid = t2.cid
 ) AS t3
   ON t0.id = t3.id;
-WITH t0 AS (SELECT 5 AS id), t1 AS (SELECT 1 AS id, 'US' AS cid), t2 AS (SELECT 1 AS id, 'US' AS cid) SELECT t0.id AS id, t2.cid AS cid FROM t0 AS t0 RIGHT JOIN t1 AS t1 ON t0.id = t1.id INNER JOIN t2 AS t2 ON t1.cid = t2.cid;
-
-# title: replace multiple INNER JOINS with LEFT JOINS when they exist in a subquery used as the RHS of a LEFT JOIN
-WITH
-t0 AS (
-  SELECT 5 as id
-),
-t1 AS (
-  SELECT 1 AS id, 'US' AS cid
-),
-t2 AS (
-  SELECT 1 AS id, 'US' AS cid
-),
-t3 AS (
-  SELECT 1 AS id, 'CA' AS cid
-)
-SELECT
-  t0.id,
-  t4.cid AS cid
-FROM t0
-LEFT JOIN (
-  SELECT
-    t1.id,
-    t3.cid
-  FROM t1
-  INNER JOIN t2
-    ON t1.cid = t2.cid
-  INNER JOIN t3
-    ON t2.id = t3.id
-) AS t4
-  ON t0.id = t4.id;
-WITH t0 AS (SELECT 5 AS id), t1 AS (SELECT 1 AS id, 'US' AS cid), t2 AS (SELECT 1 AS id, 'US' AS cid), t3 AS (SELECT 1 AS id, 'CA' AS cid) SELECT t0.id AS id, t3.cid AS cid FROM t0 AS t0 LEFT JOIN t1 AS t1 ON t0.id = t1.id LEFT JOIN t2 AS t2 ON t1.cid = t2.cid LEFT JOIN t3 AS t3 ON t2.id = t3.id;
+WITH t0 AS (SELECT 5 AS id), t1 AS (SELECT 1 AS id, 'US' AS cid), t2 AS (SELECT 1 AS id, 'US' AS cid) SELECT t0.id AS id, t3.cid AS cid FROM t0 AS t0 INNER JOIN (SELECT t1.id AS id, t2.cid AS cid FROM t1 AS t1 RIGHT JOIN t2 AS t2 ON t1.cid = t2.cid) AS t3 ON t0.id = t3.id;

tests/fixtures/optimizer/optimizer.sql

@@ -472,17 +472,22 @@ FROM (
       alias_1.id = alias_2.id
     )
 ) AS main_query;
+WITH "alias_2" AS (
+  SELECT
+    "company_table_2"."id" AS "id"
+  FROM "company_table" AS "company_table_2"
+  LEFT JOIN "unlocked" AS "unlocked"
+    ON "company_table_2"."id" = "unlocked"."company_id"
+  WHERE
+    CASE WHEN "unlocked"."company_id" IS NULL THEN 0 ELSE 1 END = FALSE
+    AND NOT "company_table_2"."id" IS NULL
+)
 SELECT
   "company_table"."id" AS "id",
   "company_table"."score" AS "score"
 FROM "company_table" AS "company_table"
-JOIN "company_table" AS "company_table_2"
-  ON "company_table"."id" = "company_table_2"."id"
-LEFT JOIN "unlocked" AS "unlocked"
-  ON "company_table_2"."id" = "unlocked"."company_id"
-WHERE
-  CASE WHEN "unlocked"."company_id" IS NULL THEN 0 ELSE 1 END = FALSE
-  AND NOT "company_table_2"."id" IS NULL;
+JOIN "alias_2" AS "alias_2"
+  ON "alias_2"."id" = "company_table"."id";
 
 # title: db.table alias clash
 # execute: false

tests/fixtures/optimizer/tpc-ds/tpc-ds.sql

@@ -198,24 +198,34 @@ WITH "wscs" AS (
     ON "date_dim"."d_date_sk" = "wscs"."sold_date_sk"
   GROUP BY
     "date_dim"."d_week_seq"
+), "z" AS (
+  SELECT
+    "wswscs"."d_week_seq" AS "d_week_seq2",
+    "wswscs"."sun_sales" AS "sun_sales2",
+    "wswscs"."mon_sales" AS "mon_sales2",
+    "wswscs"."tue_sales" AS "tue_sales2",
+    "wswscs"."wed_sales" AS "wed_sales2",
+    "wswscs"."thu_sales" AS "thu_sales2",
+    "wswscs"."fri_sales" AS "fri_sales2",
+    "wswscs"."sat_sales" AS "sat_sales2"
+  FROM "wswscs" AS "wswscs"
+  JOIN "date_dim" AS "date_dim"
+    ON "date_dim"."d_week_seq" = "wswscs"."d_week_seq" AND "date_dim"."d_year" = 1999
 )
 SELECT
   "wswscs"."d_week_seq" AS "d_week_seq1",
-  ROUND("wswscs"."sun_sales" / "wswscs_2"."sun_sales", 2) AS "_col_1",
-  ROUND("wswscs"."mon_sales" / "wswscs_2"."mon_sales", 2) AS "_col_2",
-  ROUND("wswscs"."tue_sales" / "wswscs_2"."tue_sales", 2) AS "_col_3",
-  ROUND("wswscs"."wed_sales" / "wswscs_2"."wed_sales", 2) AS "_col_4",
-  ROUND("wswscs"."thu_sales" / "wswscs_2"."thu_sales", 2) AS "_col_5",
-  ROUND("wswscs"."fri_sales" / "wswscs_2"."fri_sales", 2) AS "_col_6",
-  ROUND("wswscs"."sat_sales" / "wswscs_2"."sat_sales", 2) AS "_col_7"
+  ROUND("wswscs"."sun_sales" / "z"."sun_sales2", 2) AS "_col_1",
+  ROUND("wswscs"."mon_sales" / "z"."mon_sales2", 2) AS "_col_2",
+  ROUND("wswscs"."tue_sales" / "z"."tue_sales2", 2) AS "_col_3",
+  ROUND("wswscs"."wed_sales" / "z"."wed_sales2", 2) AS "_col_4",
+  ROUND("wswscs"."thu_sales" / "z"."thu_sales2", 2) AS "_col_5",
+  ROUND("wswscs"."fri_sales" / "z"."fri_sales2", 2) AS "_col_6",
+  ROUND("wswscs"."sat_sales" / "z"."sat_sales2", 2) AS "_col_7"
 FROM "wswscs" AS "wswscs"
 JOIN "date_dim" AS "date_dim"
   ON "date_dim"."d_week_seq" = "wswscs"."d_week_seq" AND "date_dim"."d_year" = 1998
-JOIN "wswscs" AS "wswscs_2"
-  ON "wswscs"."d_week_seq" = "wswscs_2"."d_week_seq" - 53
-JOIN "date_dim" AS "date_dim_2"
-  ON "date_dim_2"."d_week_seq" = "wswscs_2"."d_week_seq"
-  AND "date_dim_2"."d_year" = 1999
+JOIN "z" AS "z"
+  ON "wswscs"."d_week_seq" = "z"."d_week_seq2" - 53
 ORDER BY
   "d_week_seq1";