fix: problem of parser and planner

include/libs/function/functionMgt.h

@@ -154,6 +154,7 @@ bool fmIsWindowClauseFunc(int32_t funcId);
 bool fmIsSpecialDataRequiredFunc(int32_t funcId);
 bool fmIsDynamicScanOptimizedFunc(int32_t funcId);
 bool fmIsMultiResFunc(int32_t funcId);
+bool fmIsRepeatScanFunc(int32_t funcId);
 bool fmIsUserDefinedFunc(int32_t funcId);
 
 typedef enum EFuncDataRequired {
@@ -170,7 +171,7 @@ int32_t fmGetScalarFuncExecFuncs(int32_t funcId, SScalarFuncExecFuncs* pFpSet);
 int32_t fmGetUdafExecFuncs(int32_t funcId, SFuncExecFuncs* pFpSet);
 int32_t fmSetInvertFunc(int32_t funcId, SFuncExecFuncs* pFpSet);
 int32_t fmSetNormalFunc(int32_t funcId, SFuncExecFuncs* pFpSet);
-bool fmIsInvertible(int32_t funcId);
+bool    fmIsInvertible(int32_t funcId);
 
 #ifdef __cplusplus
 }

include/libs/nodes/nodes.h

@@ -206,6 +206,7 @@ typedef enum ENodeType {
   QUERY_NODE_PHYSICAL_PLAN_EXCHANGE,
   QUERY_NODE_PHYSICAL_PLAN_SORT,
   QUERY_NODE_PHYSICAL_PLAN_INTERVAL,
+  QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL,
   QUERY_NODE_PHYSICAL_PLAN_FILL,
   QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW,
   QUERY_NODE_PHYSICAL_PLAN_STATE_WINDOW,

include/libs/nodes/plannodes.h

@@ -274,6 +274,8 @@ typedef struct SIntervalPhysiNode {
   int8_t           slidingUnit;
 } SIntervalPhysiNode;
 
+typedef SIntervalPhysiNode SStreamIntervalPhysiNode;
+
 typedef struct SFillPhysiNode {
   SPhysiNode  node;
   EFillMode   mode;

include/libs/nodes/querynodes.h

@@ -233,6 +233,7 @@ typedef struct SSelectStmt {
   uint8_t     precision;
   bool        isEmptyResult;
   bool        hasAggFuncs;
+  bool        hasRepeatScanFuncs;
   bool        isTimeOrderQuery;
 } SSelectStmt;
 

source/libs/function/inc/functionMgtInt.h

@@ -40,6 +40,7 @@ extern "C" {
 #define FUNC_MGT_MULTI_RES_FUNC         FUNC_MGT_FUNC_CLASSIFICATION_MASK(11)
 #define FUNC_MGT_SCAN_PC_FUNC           FUNC_MGT_FUNC_CLASSIFICATION_MASK(12)
 #define FUNC_MGT_SELECT_FUNC            FUNC_MGT_FUNC_CLASSIFICATION_MASK(13)
+#define FUNC_MGT_REPEAT_SCAN_FUNC       FUNC_MGT_FUNC_CLASSIFICATION_MASK(14)
 
 #define FUNC_MGT_TEST_MASK(val, mask) (((val) & (mask)) != 0)
 

source/libs/function/src/builtins.c

@@ -241,7 +241,7 @@ static int32_t translateHistogram(SFunctionNode* pFunc, char* pErrBuf, int32_t l
     return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
   }
 
-  pFunc->node.resType = (SDataType) { .bytes = 512, .type = TSDB_DATA_TYPE_BINARY };
+  pFunc->node.resType = (SDataType){.bytes = 512, .type = TSDB_DATA_TYPE_BINARY};
   return TSDB_CODE_SUCCESS;
 }
 
@@ -273,7 +273,8 @@ static int32_t translateDiff(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
   }
 
   SExprNode* p1 = (SExprNode*)nodesListGetNode(pFunc->pParameterList, 0);
-  if (!IS_NUMERIC_TYPE(p1->resType.type)) {
+  if (!IS_SIGNED_NUMERIC_TYPE(p1->resType.type) && !IS_FLOAT_TYPE(p1->resType.type) &&
+      TSDB_DATA_TYPE_BOOL != p1->resType.type) {
     return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
   }
   pFunc->node.resType = p1->resType;
@@ -549,7 +550,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "percentile",
     .type = FUNCTION_TYPE_PERCENTILE,
-    .classification = FUNC_MGT_AGG_FUNC,
+    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_REPEAT_SCAN_FUNC,
     .translateFunc = translatePercentile,
     .getEnvFunc   = getPercentileFuncEnv,
     .initFunc     = percentileFunctionSetup,

source/libs/function/src/functionMgt.c

@@ -169,6 +169,8 @@ bool fmIsDynamicScanOptimizedFunc(int32_t funcId) {
 
 bool fmIsMultiResFunc(int32_t funcId) { return isSpecificClassifyFunc(funcId, FUNC_MGT_MULTI_RES_FUNC); }
 
+bool fmIsRepeatScanFunc(int32_t funcId) { return isSpecificClassifyFunc(funcId, FUNC_MGT_REPEAT_SCAN_FUNC); }
+
 bool fmIsUserDefinedFunc(int32_t funcId) { return funcId > FUNC_UDF_ID_START; }
 
 void fmFuncMgtDestroy() {
@@ -197,15 +199,14 @@ int32_t fmSetNormalFunc(int32_t funcId, SFuncExecFuncs* pFpSet) {
 bool fmIsInvertible(int32_t funcId) {
   bool res = false;
   switch (funcMgtBuiltins[funcId].type) {
-  case FUNCTION_TYPE_COUNT:
-  case FUNCTION_TYPE_SUM:
-  case FUNCTION_TYPE_STDDEV:
-  case FUNCTION_TYPE_AVG:
-    res = true;
-    break;
-  default:
-    break;
+    case FUNCTION_TYPE_COUNT:
+    case FUNCTION_TYPE_SUM:
+    case FUNCTION_TYPE_STDDEV:
+    case FUNCTION_TYPE_AVG:
+      res = true;
+      break;
+    default:
+      break;
   }
   return res;
 }
-

source/libs/nodes/src/nodesCodeFuncs.c

@@ -222,6 +222,8 @@ const char* nodesNodeName(ENodeType type) {
       return "PhysiSort";
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
       return "PhysiInterval";
+    case QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL:
+      return "PhysiStreamInterval";
     case QUERY_NODE_PHYSICAL_PLAN_FILL:
       return "PhysiFill";
     case QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW:
@@ -2893,6 +2895,7 @@ static int32_t specificNodeToJson(const void* pObj, SJson* pJson) {
     case QUERY_NODE_PHYSICAL_PLAN_SORT:
       return physiSortNodeToJson(pObj, pJson);
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
+    case QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL:
       return physiIntervalNodeToJson(pObj, pJson);
     case QUERY_NODE_PHYSICAL_PLAN_FILL:
       return physiFillNodeToJson(pObj, pJson);
@@ -2983,6 +2986,7 @@ static int32_t jsonToSpecificNode(const SJson* pJson, void* pObj) {
     case QUERY_NODE_PHYSICAL_PLAN_SORT:
       return jsonToPhysiSortNode(pJson, pObj);
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
+    case QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL:
       return jsonToPhysiIntervalNode(pJson, pObj);
     case QUERY_NODE_PHYSICAL_PLAN_FILL:
       return jsonToPhysiFillNode(pJson, pObj);
@@ -3099,6 +3103,7 @@ int32_t nodesStringToNode(const char* pStr, SNode** pNode) {
     return TSDB_CODE_FAILED;
   }
   int32_t code = makeNodeByJson(pJson, pNode);
+  tjsonDelete(pJson);
   if (TSDB_CODE_SUCCESS != code) {
     nodesDestroyNode(*pNode);
     *pNode = NULL;

source/libs/nodes/src/nodesTraverseFuncs.c

@@ -513,6 +513,7 @@ static EDealRes dispatchPhysiPlan(SNode* pNode, ETraversalOrder order, FNodeWalk
       break;
     }
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
+    case QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL:
       res = walkWindowPhysi((SWinodwPhysiNode*)pNode, order, walker, pContext);
       break;
     case QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW:

source/libs/nodes/src/nodesUtilFuncs.c

@@ -252,6 +252,8 @@ SNodeptr nodesMakeNode(ENodeType type) {
       return makeNode(type, sizeof(SSortPhysiNode));
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
       return makeNode(type, sizeof(SIntervalPhysiNode));
+    case QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL:
+      return makeNode(type, sizeof(SStreamIntervalPhysiNode));
     case QUERY_NODE_PHYSICAL_PLAN_FILL:
       return makeNode(type, sizeof(SFillPhysiNode));
     case QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW:
@@ -644,6 +646,7 @@ void nodesDestroyNode(SNodeptr pNode) {
       break;
     }
     case QUERY_NODE_PHYSICAL_PLAN_INTERVAL:
+    case QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL:
       destroyWinodwPhysiNode((SWinodwPhysiNode*)pNode);
       break;
     case QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW:

source/libs/parser/src/parAstParser.c

@@ -64,8 +64,8 @@ int32_t parse(SParseContext* pParseCxt, SQuery** pQuery) {
         goto abort_parse;
       }
       default:
-        Parse(pParser, t0.type, t0, &cxt);
         // ParseTrace(stdout, "");
+        Parse(pParser, t0.type, t0, &cxt);
         if (TSDB_CODE_SUCCESS != cxt.errCode) {
           goto abort_parse;
         }

source/libs/parser/src/parTokenizer.c

@@ -590,6 +590,8 @@ uint32_t tGetToken(const char* z, uint32_t* tokenId) {
       if (seg == 4) {  // ip address
         *tokenId = TK_NK_IPTOKEN;
         return i;
+      } else if (seg > 2) {
+        break;
       }
 
       if ((z[i] == 'e' || z[i] == 'E') &&

source/libs/parser/src/parTranslater.c

@@ -699,6 +699,10 @@ static EDealRes translateFunction(STranslateContext* pCxt, SFunctionNode* pFunc)
     if (isCountStar(pFunc)) {
       pCxt->errCode = rewriteCountStar(pCxt, pFunc);
     }
+
+    if (fmIsRepeatScanFunc(pFunc->funcId)) {
+      pCxt->pCurrStmt->hasRepeatScanFuncs = true;
+    }
   }
   return TSDB_CODE_SUCCESS == pCxt->errCode ? DEAL_RES_CONTINUE : DEAL_RES_ERROR;
 }
@@ -2255,8 +2259,8 @@ static int32_t checkTableColsSchema(STranslateContext* pCxt, SHashObj* pHash, SN
       code = generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_DUPLICATED_COLUMN);
     }
     if (TSDB_CODE_SUCCESS == code) {
-      if ((TSDB_DATA_TYPE_VARCHAR == pCol->dataType.type && pCol->dataType.bytes > TSDB_MAX_BINARY_LEN) ||
-          (TSDB_DATA_TYPE_NCHAR == pCol->dataType.type && pCol->dataType.bytes > TSDB_MAX_NCHAR_LEN)) {
+      if ((TSDB_DATA_TYPE_VARCHAR == pCol->dataType.type && calcTypeBytes(pCol->dataType) > TSDB_MAX_BINARY_LEN) ||
+          (TSDB_DATA_TYPE_NCHAR == pCol->dataType.type && calcTypeBytes(pCol->dataType) > TSDB_MAX_NCHAR_LEN)) {
         code = code = generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_VAR_COLUMN_LEN);
       }
     }

source/libs/parser/test/parSelectTest.cpp

@@ -37,13 +37,15 @@ TEST_F(ParserSelectTest, basic) {
 
 TEST_F(ParserSelectTest, constant) {
   useDb("root", "test");
-
+#if 0
   run("SELECT 123, 20.4, 'abc', \"wxy\", timestamp '2022-02-09 17:30:20', true, false, 10s FROM t1");
 
   run("SELECT 1234567890123456789012345678901234567890, 20.1234567890123456789012345678901234567890, 'abc', \"wxy\", "
       "timestamp '2022-02-09 17:30:20', true, false, 15s FROM t1");
 
   run("SELECT 123 + 45 FROM t1 WHERE 2 - 1");
+#endif
+  run("SELECT 1.1.1 FROM t1");
 }
 
 TEST_F(ParserSelectTest, expression) {

source/libs/planner/src/planLogicCreater.c

@@ -54,6 +54,9 @@ static EDealRes doRewriteExpr(SNode** pNode, void* pContext) {
           pCol->node.resType = pToBeRewrittenExpr->resType;
           strcpy(pCol->node.aliasName, pToBeRewrittenExpr->aliasName);
           strcpy(pCol->colName, ((SExprNode*)pExpr)->aliasName);
+          if (QUERY_NODE_FUNCTION == nodeType(pExpr) && FUNCTION_TYPE_WSTARTTS == ((SFunctionNode*)pExpr)->funcType) {
+            pCol->colId = PRIMARYKEY_TIMESTAMP_COL_ID;
+          }
           nodesDestroyNode(*pNode);
           *pNode = (SNode*)pCol;
           return DEAL_RES_IGNORE_CHILD;
@@ -253,7 +256,7 @@ static int32_t createScanLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSelect
 
   TSWAP(pScan->pMeta, pRealTable->pMeta);
   TSWAP(pScan->pVgroupList, pRealTable->pVgroupList);
-  pScan->scanSeq[0] = 1;
+  pScan->scanSeq[0] = pSelect->hasRepeatScanFuncs ? 2 : 1;
   pScan->scanSeq[1] = 0;
   pScan->scanRange = TSWINDOW_INITIALIZER;
   pScan->tableName.type = TSDB_TABLE_NAME_T;

source/libs/planner/src/planOptimizer.c

@@ -754,10 +754,7 @@ static int32_t opkDoOptimized(SOptimizeContext* pCxt, SSortLogicNode* pSort, SNo
   EOrder order = opkGetPrimaryKeyOrder(pSort);
   if (ORDER_DESC == order) {
     SNode* pScan = NULL;
-    FOREACH(pScan, pScanNodes) {
-      ((SScanLogicNode*)pScan)->scanSeq[0] = 0;
-      ((SScanLogicNode*)pScan)->scanSeq[1] = 1;
-    }
+    FOREACH(pScan, pScanNodes) { TSWAP(((SScanLogicNode*)pScan)->scanSeq[0], ((SScanLogicNode*)pScan)->scanSeq[1]); }
   }
 
   if (NULL == pSort->node.pParent) {

source/libs/planner/src/planPhysiCreater.c

@@ -869,7 +869,8 @@ static int32_t createWindowPhysiNodeFinalize(SPhysiPlanContext* pCxt, SNodeList*
 static int32_t createIntervalPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren,
                                        SWindowLogicNode* pWindowLogicNode, SPhysiNode** pPhyNode) {
   SIntervalPhysiNode* pInterval = (SIntervalPhysiNode*)makePhysiNode(
-      pCxt, getPrecision(pChildren), (SLogicNode*)pWindowLogicNode, QUERY_NODE_PHYSICAL_PLAN_INTERVAL);
+      pCxt, getPrecision(pChildren), (SLogicNode*)pWindowLogicNode,
+      (pCxt->pPlanCxt->streamQuery ? QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL : QUERY_NODE_PHYSICAL_PLAN_INTERVAL));
   if (NULL == pInterval) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }

source/libs/planner/test/planBasicTest.cpp

@@ -48,4 +48,6 @@ TEST_F(PlanBasicTest, func) {
   useDb("root", "test");
 
   run("SELECT DIFF(c1) FROM t1");
+
+  run("SELECT PERCENTILE(c1, 60) FROM t1");
 }

source/libs/planner/test/planOptimizeTest.cpp

@@ -37,4 +37,6 @@ TEST_F(PlanOptimizeTest, orderByPrimaryKey) {
   run("SELECT * FROM t1 ORDER BY ts DESC");
   run("SELECT c1 FROM t1 ORDER BY ts");
   run("SELECT c1 FROM t1 ORDER BY ts DESC");
+
+  run("SELECT COUNT(*) FROM t1 INTERVAL(10S) ORDER BY _WSTARTTS DESC");
 }

tests/script/sh/exec.sh

@@ -101,8 +101,8 @@ if [ "$EXEC_OPTON" = "start" ]; then
   if [ "$VALGRIND_OPTION" = "true" ]; then 
     TT=`date +%s`
     #mkdir ${LOG_DIR}/${TT}
-    echo "nohup valgrind --log-file=${LOG_DIR}/valgrind-taosd-${NODE_NAME}-${TT}.log --tool=memcheck --leak-check=full --show-reachable=no  --track-origins=yes --show-leak-kinds=all  -v  --workaround-gcc296-bugs=yes   $EXE_DIR/taosd -c $CFG_DIR > /dev/null 2>&1 &"
-    nohup valgrind --log-file=${LOG_DIR}/valgrind-taosd-${NODE_NAME}-${TT}.log --tool=memcheck --leak-check=full --show-reachable=no  --track-origins=yes --show-leak-kinds=all  -v  --workaround-gcc296-bugs=yes   $EXE_DIR/taosd -c $CFG_DIR > /dev/null 2>&1 &   
+    echo "nohup valgrind --log-file=${LOG_DIR}/valgrind-taosd-${NODE_NAME}-${TT}.log --tool=memcheck --leak-check=full --show-reachable=no  --track-origins=yes --show-leak-kinds=all --num-callers=20 -v -v  --workaround-gcc296-bugs=yes   $EXE_DIR/taosd -c $CFG_DIR > /dev/null 2>&1 &"
+    nohup valgrind --log-file=${LOG_DIR}/valgrind-taosd-${NODE_NAME}-${TT}.log --tool=memcheck --leak-check=full --show-reachable=no  --track-origins=yes --show-leak-kinds=all --num-callers=20 -v -v  --workaround-gcc296-bugs=yes   $EXE_DIR/taosd -c $CFG_DIR > /dev/null 2>&1 &   
   else
     echo "nohup $EXE_DIR/taosd -c $CFG_DIR > /dev/null 2>&1 &"
     nohup $EXE_DIR/taosd -c $CFG_DIR > /dev/null 2>&1 & 

tests/script/test.sh

@@ -131,8 +131,8 @@ if [ -n "$FILE_NAME" ]; then
       FLAG="-v"    
     fi 
   
-    echo valgrind --tool=memcheck --leak-check=full --show-reachable=no  --track-origins=yes --child-silent-after-fork=yes --show-leak-kinds=all  -v  --workaround-gcc296-bugs=yes  --log-file=${LOG_DIR}/valgrind-tsim.log $PROGRAM -c $CFG_DIR -f $FILE_NAME $FLAG
-    valgrind --tool=memcheck --leak-check=full --show-reachable=no  --track-origins=yes --child-silent-after-fork=yes --show-leak-kinds=all  -v  --workaround-gcc296-bugs=yes  --log-file=${LOG_DIR}/valgrind-tsim.log $PROGRAM -c $CFG_DIR -f $FILE_NAME $FLAG
+    echo valgrind --tool=memcheck --leak-check=full --show-reachable=no  --track-origins=yes --child-silent-after-fork=yes --show-leak-kinds=all --num-callers=20 -v  --workaround-gcc296-bugs=yes  --log-file=${LOG_DIR}/valgrind-tsim.log $PROGRAM -c $CFG_DIR -f $FILE_NAME $FLAG
+    valgrind --tool=memcheck --leak-check=full --show-reachable=no  --track-origins=yes --child-silent-after-fork=yes --show-leak-kinds=all --num-callers=20 -v  --workaround-gcc296-bugs=yes  --log-file=${LOG_DIR}/valgrind-tsim.log $PROGRAM -c $CFG_DIR -f $FILE_NAME $FLAG
   else
     if [[ $MULTIPROCESS -eq 1 ]];then
       echo "ExcuteCmd(multiprocess):" $PROGRAM -m -c $CFG_DIR -f $FILE_NAME