From fd34087b8cb42935f5cf1974a9f5f1c5acb84c18 Mon Sep 17 00:00:00 2001
From: wangjiaming0909 <wangjiaming0909@sina.com>
Date: Mon, 27 May 2024 16:20:13 +0800
Subject: [PATCH] fix select from union all caused crash

---
 include/libs/nodes/plannodes.h             |  1 +
 source/libs/planner/src/planLogicCreater.c |  1 +
 source/libs/planner/src/planOptimizer.c    | 30 +++++++++++++++++-----
 tests/system-test/2-query/union1.py        | 13 ++++++++++
 4 files changed, 38 insertions(+), 7 deletions(-)

diff --git a/include/libs/nodes/plannodes.h b/include/libs/nodes/plannodes.h
index 18bc24d612..a691433ee6 100644
--- a/include/libs/nodes/plannodes.h
+++ b/include/libs/nodes/plannodes.h
@@ -183,6 +183,7 @@ typedef struct SProjectLogicNode {
   char       stmtName[TSDB_TABLE_NAME_LEN];
   bool       ignoreGroupId;
   bool       inputIgnoreGroup;
+  bool       isSetOpProj;
 } SProjectLogicNode;
 
 typedef struct SIndefRowsFuncLogicNode {
diff --git a/source/libs/planner/src/planLogicCreater.c b/source/libs/planner/src/planLogicCreater.c
index 23b8baf031..acc29997a2 100644
--- a/source/libs/planner/src/planLogicCreater.c
+++ b/source/libs/planner/src/planLogicCreater.c
@@ -1587,6 +1587,7 @@ static int32_t createSetOpProjectLogicNode(SLogicPlanContext* pCxt, SSetOperator
     TSWAP(pProject->node.pLimit, pSetOperator->pLimit);
   }
   pProject->ignoreGroupId = true;
+  pProject->isSetOpProj = true;
 
   int32_t code = TSDB_CODE_SUCCESS;
 
diff --git a/source/libs/planner/src/planOptimizer.c b/source/libs/planner/src/planOptimizer.c
index a2a0343316..7d10c02529 100644
--- a/source/libs/planner/src/planOptimizer.c
+++ b/source/libs/planner/src/planOptimizer.c
@@ -3284,18 +3284,34 @@ static int32_t eliminateProjOptimizeImpl(SOptimizeContext* pCxt, SLogicSubplan*
   SNodeList*  pNewChildTargets = nodesMakeList();
 
   if (NULL == pProjectNode->node.pParent) {
-    SNode* pProjection = NULL;
-    FOREACH(pProjection, pProjectNode->pProjections) {
-      SNode* pChildTarget = NULL;
-      FOREACH(pChildTarget, pChild->pTargets) {
-        if (0 == strcmp(((SColumnNode*)pProjection)->colName, ((SColumnNode*)pChildTarget)->colName)) {
-          nodesListAppend(pNewChildTargets, nodesCloneNode(pChildTarget));
+    SNode *pProjection = NULL, *pChildTarget = NULL;
+    bool needOrderMatch = QUERY_NODE_LOGIC_PLAN_PROJECT == nodeType(pChild) && ((SProjectLogicNode*)pChild)->isSetOpProj;
+    bool orderMatch = true;
+    if (needOrderMatch) {
+      // For sql: select ... from (select ... union all select ...);
+      // When eliminating the outer proj (the outer select), we have to make sure that the outer proj projections and
+      // union all project targets have same columns in the same order. See detail in TD-30188
+      FORBOTH(pProjection, pProjectNode->pProjections, pChildTarget, pChild->pTargets) {
+        if (!pProjection) break;
+        if (0 != strcmp(((SColumnNode*)pProjection)->colName, ((SColumnNode*)pChildTarget)->colName)) {
+          orderMatch = false;
           break;
         }
+        nodesListAppend(pNewChildTargets, nodesCloneNode(pChildTarget));
+      }
+    } else {
+      FOREACH(pProjection, pProjectNode->pProjections) {
+        FOREACH(pChildTarget, pChild->pTargets) {
+          if (0 == strcmp(((SColumnNode*)pProjection)->colName, ((SColumnNode*)pChildTarget)->colName)) {
+            nodesListAppend(pNewChildTargets, nodesCloneNode(pChildTarget));
+            break;
+          }
+        }
       }
     }
 
-    if (eliminateProjOptCanChildConditionUseChildTargets(pChild, pNewChildTargets)) {
+    if (eliminateProjOptCanChildConditionUseChildTargets(pChild, pNewChildTargets) &&
+        (!needOrderMatch || (needOrderMatch && orderMatch))) {
       nodesDestroyList(pChild->pTargets);
       pChild->pTargets = pNewChildTargets;
     } else {
diff --git a/tests/system-test/2-query/union1.py b/tests/system-test/2-query/union1.py
index 8db5ce01f3..853dfe9582 100644
--- a/tests/system-test/2-query/union1.py
+++ b/tests/system-test/2-query/union1.py
@@ -240,9 +240,22 @@ class TDTestCase:
         tdSql.error( " '' union all select c1 from ct1 " )
         # tdSql.error( "select c1 from ct1 union select c1 from ct2 union select c1 from ct4 ")
 
+    def test_select_from_union_all(self):
+        tdSql.query('select c8, ts from ((select ts, c8,c1 from stb1 order by c1) union all select ts, c8, c1 from stb1 limit 15)')
+        tdSql.checkRows(15)
+        tdSql.query('select c8, ts from ((select ts, c8,c1 from stb1 order by c1) union all (select ts, c8, c1 from stb1 order by c8 limit 10) limit 15)')
+        tdSql.checkRows(15)
+        tdSql.query('select ts, c1 from ((select ts, c8,c1 from stb1 order by c1) union all (select ts, c8, c1 from stb1 order by c8 limit 10) limit 15)')
+        tdSql.checkRows(15)
+        tdSql.query('select ts, c1, c8 from ((select ts, c8,c1 from stb1 order by c1) union all (select ts, c8, c1 from stb1 order by c8 limit 10) limit 15)')
+        tdSql.checkRows(15)
+        tdSql.query('select ts, c8, c1, 123 from ((select ts, c8,c1 from stb1 order by c1) union all (select ts, c8, c1 from stb1 order by c8 limit 10) limit 15)')
+        tdSql.checkRows(15)
+
     def all_test(self):
         self.__test_error()
         self.union_check()
+        self.test_select_from_union_all()
 
 
     def __create_tb(self):