Merge pull request #12692 from taosdata/feature/3.0_wxy

fix: some problems of parser and planner

include/libs/function/functionMgt.h

@@ -142,6 +142,8 @@ void fmFuncMgtDestroy();
 
 int32_t fmGetFuncInfo(SFmGetFuncInfoParam* pParam, SFunctionNode* pFunc);
 
+bool fmIsBuiltinFunc(const char* pFunc);
+
 bool fmIsAggFunc(int32_t funcId);
 bool fmIsScalarFunc(int32_t funcId);
 bool fmIsNonstandardSQLFunc(int32_t funcId);

include/libs/nodes/cmdnodes.h

@@ -78,7 +78,7 @@ typedef struct SAlterDatabaseStmt {
 
 typedef struct STableOptions {
   ENodeType  type;
-  char       comment[TSDB_STB_COMMENT_LEN];
+  char       comment[TSDB_TB_COMMENT_LEN];
   int32_t    delay;
   float      filesFactor;
   SNodeList* pRollupFuncs;
@@ -90,7 +90,7 @@ typedef struct SColumnDefNode {
   ENodeType type;
   char      colName[TSDB_COL_NAME_LEN];
   SDataType dataType;
-  char      comments[TSDB_STB_COMMENT_LEN];
+  char      comments[TSDB_TB_COMMENT_LEN];
   bool      sma;
 } SColumnDefNode;
 

include/util/taoserror.h

@@ -647,6 +647,8 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_PAR_CANNOT_DROP_PRIMARY_KEY   TAOS_DEF_ERROR_CODE(0, 0x264A)
 #define TSDB_CODE_PAR_INVALID_MODIFY_COL        TAOS_DEF_ERROR_CODE(0, 0x264B)
 #define TSDB_CODE_PAR_INVALID_TBNAME            TAOS_DEF_ERROR_CODE(0, 0x264C)
+#define TSDB_CODE_PAR_INVALID_FUNCTION_NAME     TAOS_DEF_ERROR_CODE(0, 0x264D)
+#define TSDB_CODE_PAR_COMMENT_TOO_LONG          TAOS_DEF_ERROR_CODE(0, 0x264E)
 
 //planner
 #define TSDB_CODE_PLAN_INTERNAL_ERROR           TAOS_DEF_ERROR_CODE(0, 0x2700)

include/util/tdef.h

@@ -219,7 +219,7 @@ typedef enum ELogicConditionType {
 #define TSDB_MAX_ALLOWED_SQL_LEN (1 * 1024 * 1024u)  // sql length should be less than 1mb
 
 #define TSDB_APP_NAME_LEN   TSDB_UNI_LEN
-#define TSDB_STB_COMMENT_LEN 1024
+#define TSDB_TB_COMMENT_LEN 1025
 
 /**
  *  In some scenarios uint16_t (0~65535) is used to store the row len.

source/libs/function/src/functionMgt.c

@@ -100,6 +100,10 @@ int32_t fmGetFuncInfo(SFmGetFuncInfoParam* pParam, SFunctionNode* pFunc) {
   return getUdfInfo(pParam, pFunc);
 }
 
+bool fmIsBuiltinFunc(const char* pFunc) {
+  return NULL != taosHashGet(gFunMgtService.pFuncNameHashTable, pFunc, strlen(pFunc));
+}
+
 EFuncDataRequired fmFuncDataRequired(SFunctionNode* pFunc, STimeWindow* pTimeWindow) {
   if (fmIsUserDefinedFunc(pFunc->funcId) || pFunc->funcId < 0 || pFunc->funcId >= funcMgtBuiltinsNum) {
     return FUNC_DATA_REQUIRED_DATA_LOAD;

source/libs/nodes/src/nodesCodeFuncs.c

@@ -946,6 +946,23 @@ static int32_t jsonToLogicSubplan(const SJson* pJson, void* pObj) {
   return code;
 }
 
+static const char* jkLogicPlanSubplans = "Subplans";
+
+static int32_t logicPlanToJson(const void* pObj, SJson* pJson) {
+  const SQueryLogicPlan* pNode = (const SQueryLogicPlan*)pObj;
+  return tjsonAddObject(pJson, jkLogicPlanSubplans, nodeToJson, nodesListGetNode(pNode->pTopSubplans, 0));
+}
+
+static int32_t jsonToLogicPlan(const SJson* pJson, void* pObj) {
+  SQueryLogicPlan* pNode = (SQueryLogicPlan*)pObj;
+  SNode*           pChild = NULL;
+  int32_t          code = jsonToNodeObject(pJson, jkLogicPlanSubplans, &pChild);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = nodesListMakeStrictAppend(&pNode->pTopSubplans, pChild);
+  }
+  return code;
+}
+
 static const char* jkJoinLogicPlanJoinType = "JoinType";
 static const char* jkJoinLogicPlanOnConditions = "OnConditions";
 
@@ -3029,7 +3046,7 @@ static int32_t specificNodeToJson(const void* pObj, SJson* pJson) {
     case QUERY_NODE_LOGIC_SUBPLAN:
       return logicSubplanToJson(pObj, pJson);
     case QUERY_NODE_LOGIC_PLAN:
-      break;
+      return logicPlanToJson(pObj, pJson);
     case QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN:
       return physiTagScanNodeToJson(pObj, pJson);
     case QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN:
@@ -3126,6 +3143,8 @@ static int32_t jsonToSpecificNode(const SJson* pJson, void* pObj) {
       return jsonToLogicPartitionNode(pJson, pObj);
     case QUERY_NODE_LOGIC_SUBPLAN:
       return jsonToLogicSubplan(pJson, pObj);
+    case QUERY_NODE_LOGIC_PLAN:
+      return jsonToLogicPlan(pJson, pObj);
     case QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN:
       return jsonToPhysiTagScanNode(pJson, pObj);
     case QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN:

source/libs/parser/src/parAstCreater.c

@@ -196,6 +196,15 @@ static bool checkIndexName(SAstCreateContext* pCxt, SToken* pIndexName) {
   return true;
 }
 
+static bool checkComment(SAstCreateContext* pCxt, const SToken* pCommentToken, bool demand) {
+  if (NULL == pCommentToken) {
+    pCxt->errCode = demand ? TSDB_CODE_PAR_SYNTAX_ERROR : TSDB_CODE_SUCCESS;
+  } else if (pCommentToken->n >= (TSDB_TB_COMMENT_LEN + 2)) {
+    pCxt->errCode = generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_COMMENT_TOO_LONG);
+  }
+  return TSDB_CODE_SUCCESS == pCxt->errCode;
+}
+
 SNode* createRawExprNode(SAstCreateContext* pCxt, const SToken* pToken, SNode* pNode) {
   SRawExprNode* target = (SRawExprNode*)nodesMakeNode(QUERY_NODE_RAW_EXPR);
   CHECK_OUT_OF_MEM(target);
@@ -823,8 +832,10 @@ SNode* createAlterTableOptions(SAstCreateContext* pCxt) {
 SNode* setTableOption(SAstCreateContext* pCxt, SNode* pOptions, ETableOptionType type, void* pVal) {
   switch (type) {
     case TABLE_OPTION_COMMENT:
+      if (checkComment(pCxt, (SToken*)pVal, true)) {
         copyStringFormStringToken((SToken*)pVal, ((STableOptions*)pOptions)->comment,
                                   sizeof(((STableOptions*)pOptions)->comment));
+      }
       break;
     case TABLE_OPTION_DELAY:
       ((STableOptions*)pOptions)->delay = taosStr2Int32(((SToken*)pVal)->z, NULL, 10);
@@ -848,7 +859,7 @@ SNode* setTableOption(SAstCreateContext* pCxt, SNode* pOptions, ETableOptionType
 }
 
 SNode* createColumnDefNode(SAstCreateContext* pCxt, SToken* pColName, SDataType dataType, const SToken* pComment) {
-  if (!checkColumnName(pCxt, pColName)) {
+  if (!checkColumnName(pCxt, pColName) || !checkComment(pCxt, pComment, false)) {
     return NULL;
   }
   SColumnDefNode* pCol = (SColumnDefNode*)nodesMakeNode(QUERY_NODE_COLUMN_DEF);

source/libs/parser/src/parTranslater.c

@@ -272,6 +272,10 @@ static bool isTimelineFunc(const SNode* pNode) {
   return (QUERY_NODE_FUNCTION == nodeType(pNode) && fmIsTimelineFunc(((SFunctionNode*)pNode)->funcId));
 }
 
+static bool isScanPseudoColumnFunc(const SNode* pNode) {
+  return (QUERY_NODE_FUNCTION == nodeType(pNode) && fmIsScanPseudoColumnFunc(((SFunctionNode*)pNode)->funcId));
+}
+
 static bool isDistinctOrderBy(STranslateContext* pCxt) {
   return (SQL_CLAUSE_ORDER_BY == pCxt->currClause && pCxt->pCurrStmt->isDistinct);
 }
@@ -892,7 +896,7 @@ static EDealRes doCheckExprForGroupBy(SNode** pNode, void* pContext) {
       return DEAL_RES_IGNORE_CHILD;
     }
   }
-  if (QUERY_NODE_COLUMN == nodeType(*pNode)) {
+  if (isScanPseudoColumnFunc(*pNode) || QUERY_NODE_COLUMN == nodeType(*pNode)) {
     if (pCxt->selectFuncNum > 1) {
       return generateDealNodeErrMsg(pCxt->pTranslateCxt, getGroupByErrorCode(pCxt->pTranslateCxt));
     } else {
@@ -930,7 +934,7 @@ static EDealRes rewriteColsToSelectValFuncImpl(SNode** pNode, void* pContext) {
   if (isAggFunc(*pNode)) {
     return DEAL_RES_IGNORE_CHILD;
   }
-  if (QUERY_NODE_COLUMN == nodeType(*pNode)) {
+  if (isScanPseudoColumnFunc(*pNode) || QUERY_NODE_COLUMN == nodeType(*pNode)) {
     return rewriteColToSelectValFunc((STranslateContext*)pContext, NULL, pNode);
   }
   return DEAL_RES_CONTINUE;
@@ -958,7 +962,7 @@ static EDealRes doCheckAggColCoexist(SNode* pNode, void* pContext) {
     pCxt->existAggFunc = true;
     return DEAL_RES_IGNORE_CHILD;
   }
-  if (QUERY_NODE_COLUMN == nodeType(pNode)) {
+  if (isScanPseudoColumnFunc(pNode) || QUERY_NODE_COLUMN == nodeType(pNode)) {
     pCxt->existCol = true;
   }
   return DEAL_RES_CONTINUE;
@@ -3261,6 +3265,9 @@ static int32_t readFromFile(char* pName, int32_t* len, char** buf) {
 }
 
 static int32_t translateCreateFunction(STranslateContext* pCxt, SCreateFunctionStmt* pStmt) {
+  if (fmIsBuiltinFunc(pStmt->funcName)) {
+    return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_FUNCTION_NAME);
+  }
   SCreateFuncReq req = {0};
   strcpy(req.name, pStmt->funcName);
   req.igExists = pStmt->ignoreExists;

source/libs/parser/src/parUtil.c

@@ -160,6 +160,10 @@ static char* getSyntaxErrFormat(int32_t errCode) {
       return "Only binary/nchar column length could be modified";
     case TSDB_CODE_PAR_INVALID_TBNAME:
       return "Invalid tbname pseudo column";
+    case TSDB_CODE_PAR_INVALID_FUNCTION_NAME:
+      return "Invalid function name";
+    case TSDB_CODE_PAR_COMMENT_TOO_LONG:
+      return "Comment too long";
     case TSDB_CODE_OUT_OF_MEMORY:
       return "Out of memory";
     default:

source/libs/planner/src/planSpliter.c

@@ -24,6 +24,7 @@
 #define SPLIT_FLAG_TEST_MASK(val, mask) (((val) & (mask)) != 0)
 
 typedef struct SSplitContext {
+  int32_t queryId;
   int32_t groupId;
   bool    split;
 } SSplitContext;
@@ -62,6 +63,7 @@ static SLogicSubplan* splCreateScanSubplan(SSplitContext* pCxt, SScanLogicNode*
   if (NULL == pSubplan) {
     return NULL;
   }
+  pSubplan->id.queryId = pCxt->queryId;
   pSubplan->id.groupId = pCxt->groupId;
   pSubplan->subplanType = SUBPLAN_TYPE_SCAN;
   pSubplan->pNode = (SLogicNode*)nodesCloneNode(pScan);
@@ -242,6 +244,7 @@ static SLogicSubplan* unionCreateSubplan(SSplitContext* pCxt, SLogicNode* pNode)
   if (NULL == pSubplan) {
     return NULL;
   }
+  pSubplan->id.queryId = pCxt->queryId;
   pSubplan->id.groupId = pCxt->groupId;
   pSubplan->subplanType = SUBPLAN_TYPE_SCAN;
   pSubplan->pNode = pNode;
@@ -406,7 +409,7 @@ static const SSplitRule splitRuleSet[] = {{.pName = "SuperTableScan", .splitFunc
 static const int32_t splitRuleNum = (sizeof(splitRuleSet) / sizeof(SSplitRule));
 
 static int32_t applySplitRule(SLogicSubplan* pSubplan) {
-  SSplitContext cxt = {.groupId = pSubplan->id.groupId + 1, .split = false};
+  SSplitContext cxt = {.queryId = pSubplan->id.queryId, .groupId = pSubplan->id.groupId + 1, .split = false};
   do {
     cxt.split = false;
     for (int32_t i = 0; i < splitRuleNum; ++i) {

source/libs/planner/test/planSetOpTest.cpp

@@ -32,6 +32,15 @@ TEST_F(PlanSetOpTest, unionAllSubquery) {
   run("SELECT * FROM (SELECT c1, c2 FROM t1 UNION ALL SELECT c1, c2 FROM t1)");
 }
 
+TEST_F(PlanSetOpTest, unionAllWithSubquery) {
+  useDb("root", "test");
+
+  // child table
+  run("SELECT ts FROM (SELECT ts FROM st1s1) UNION ALL SELECT ts FROM (SELECT ts FROM st1s2)");
+  // super table
+  run("SELECT ts FROM (SELECT ts FROM st1) UNION ALL SELECT ts FROM (SELECT ts FROM st1)");
+}
+
 TEST_F(PlanSetOpTest, union) {
   useDb("root", "test");
 

source/libs/planner/test/planTestUtil.cpp

@@ -322,6 +322,7 @@ class PlannerTestBaseImpl {
   }
 
   void setPlanContext(SQuery* pQuery, SPlanContext* pCxt) {
+    pCxt->queryId = 1;
     if (QUERY_NODE_CREATE_TOPIC_STMT == nodeType(pQuery->pRoot)) {
       pCxt->pAstRoot = ((SCreateTopicStmt*)pQuery->pRoot)->pQuery;
       pCxt->topicQuery = true;