From 32577d41966059b7a66ba3ce5a00e5c2a270319c Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Thu, 14 Apr 2022 18:24:21 +0800
Subject: [PATCH 01/15] refactor: add json tag function for sql parser

---
 include/common/tdataformat.h           |   3 +-
 include/common/ttypes.h                |   2 +-
 include/util/taoserror.h               |   1 +
 include/util/tdef.h                    |   8 ++
 include/util/tjson.h                   |   1 +
 include/util/tutil.h                   |   2 -
 source/common/src/ttypes.c             |   3 +-
 source/common/src/tvariant.c           |   2 +-
 source/libs/parser/inc/parUtil.h       |   1 +
 source/libs/parser/src/parAstCreater.c |   1 +
 source/libs/parser/src/parInsert.c     |  54 ++++----
 source/libs/parser/src/parTranslater.c |  24 +++-
 source/libs/parser/src/parUtil.c       | 169 ++++++++++++++++++++++++-
 source/libs/scalar/src/filter.c        |   2 +-
 source/util/src/tjson.c                |   1 +
 source/util/src/tutil.c                |  59 ---------
 tools/shell/src/shellEngine.c          |  71 +++--------
 17 files changed, 252 insertions(+), 152 deletions(-)

diff --git a/include/common/tdataformat.h b/include/common/tdataformat.h
index 8a87a1bd2e..2dd887ae74 100644
--- a/include/common/tdataformat.h
+++ b/include/common/tdataformat.h
@@ -608,7 +608,7 @@ void    tdDestroyKVRowBuilder(SKVRowBuilder *pBuilder);
 void    tdResetKVRowBuilder(SKVRowBuilder *pBuilder);
 SKVRow  tdGetKVRowFromBuilder(SKVRowBuilder *pBuilder);
 
-static FORCE_INLINE int32_t tdAddColToKVRow(SKVRowBuilder *pBuilder, col_id_t colId, int8_t type, const void *value) {
+static FORCE_INLINE int32_t tdAddColToKVRow(SKVRowBuilder *pBuilder, col_id_t colId, const void *value, int32_t tlen) {
   if (pBuilder->nCols >= pBuilder->tCols) {
     pBuilder->tCols *= 2;
     SColIdx *pColIdx = (SColIdx *)taosMemoryRealloc((void *)(pBuilder->pColIdx), sizeof(SColIdx) * pBuilder->tCols);
@@ -621,7 +621,6 @@ static FORCE_INLINE int32_t tdAddColToKVRow(SKVRowBuilder *pBuilder, col_id_t co
 
   pBuilder->nCols++;
 
-  int32_t tlen = IS_VAR_DATA_TYPE(type) ? varDataTLen(value) : TYPE_BYTES[type];
   if (tlen > pBuilder->alloc - pBuilder->size) {
     while (tlen > pBuilder->alloc - pBuilder->size) {
       pBuilder->alloc *= 2;
diff --git a/include/common/ttypes.h b/include/common/ttypes.h
index 032cb44122..1112021124 100644
--- a/include/common/ttypes.h
+++ b/include/common/ttypes.h
@@ -252,7 +252,7 @@ typedef struct tDataTypeDescriptor {
                      int16_t *maxindex, int16_t *numofnull);
 } tDataTypeDescriptor;
 
-extern tDataTypeDescriptor tDataTypes[15];
+extern tDataTypeDescriptor tDataTypes[TSDB_DATA_TYPE_MAX];
 
 bool isValidDataType(int32_t type);
 
diff --git a/include/util/taoserror.h b/include/util/taoserror.h
index 60535e2f49..a11bb5162d 100644
--- a/include/util/taoserror.h
+++ b/include/util/taoserror.h
@@ -599,6 +599,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_PAR_GROUPBY_WINDOW_COEXIST    TAOS_DEF_ERROR_CODE(0, 0x2624)
 #define TSDB_CODE_PAR_INVALID_OPTION_UNIT       TAOS_DEF_ERROR_CODE(0, 0x2625)
 #define TSDB_CODE_PAR_INVALID_KEEP_UNIT         TAOS_DEF_ERROR_CODE(0, 0x2626)
+#define TSDB_CODE_PAR_ONLY_ONE_JSON_TAG         TAOS_DEF_ERROR_CODE(0, 0x2627)
 
 //planner
 #define TSDB_CODE_PLAN_INTERNAL_ERROR              TAOS_DEF_ERROR_CODE(0, 0x2700)
diff --git a/include/util/tdef.h b/include/util/tdef.h
index f276a1a812..0289bb0df7 100644
--- a/include/util/tdef.h
+++ b/include/util/tdef.h
@@ -254,6 +254,14 @@ typedef enum ELogicConditionType {
 #define TSDB_MAX_TAGS           128
 #define TSDB_MAX_TAG_CONDITIONS 1024
 
+#define TSDB_MAX_JSON_TAG_LEN   16384
+#define TSDB_JSON_PLACEHOLDER      0x7F
+#define TSDB_JSON_null             0x00
+#define TSDB_JSON_KEY_NULL         0x00
+#define TSDB_JSON_NOT_NULL         0x01
+#define TSDB_JSON_NULL             0x00
+
+
 #define TSDB_AUTH_LEN          16
 #define TSDB_PASSWORD_LEN      32
 #define TSDB_USET_PASSWORD_LEN 129
diff --git a/include/util/tjson.h b/include/util/tjson.h
index 6b2221f704..9f347e5896 100644
--- a/include/util/tjson.h
+++ b/include/util/tjson.h
@@ -76,6 +76,7 @@ char* tjsonToString(const SJson* pJson);
 char* tjsonToUnformattedString(const SJson* pJson);
 
 SJson* tjsonParse(const char* pStr);
+bool   tjsonIsObject(const SJson* pJson);
 
 #ifdef __cplusplus
 }
diff --git a/include/util/tutil.h b/include/util/tutil.h
index e0f92be76a..444a893a88 100644
--- a/include/util/tutil.h
+++ b/include/util/tutil.h
@@ -26,8 +26,6 @@ extern "C" {
 #endif
 
 int32_t strdequote(char *src);
-int32_t strndequote(char *dst, const char *z, int32_t len);
-int32_t strRmquote(char *z, int32_t len);
 size_t  strtrim(char *src);
 char   *strnchr(const char *haystack, char needle, int32_t len, bool skipquote);
 char  **strsplit(char *src, const char *delim, int32_t *num);
diff --git a/source/common/src/ttypes.c b/source/common/src/ttypes.c
index ef75adeb5d..6af4777537 100644
--- a/source/common/src/ttypes.c
+++ b/source/common/src/ttypes.c
@@ -365,7 +365,7 @@ static void getStatics_nchr(const void *pData, int32_t numOfRow, int64_t *min, i
   *maxIndex = 0;
 }
 
-tDataTypeDescriptor tDataTypes[15] = {
+tDataTypeDescriptor tDataTypes[TSDB_DATA_TYPE_MAX] = {
     {TSDB_DATA_TYPE_NULL, 6, 1, "NOTYPE", 0, 0, NULL, NULL, NULL},
     {TSDB_DATA_TYPE_BOOL, 4, CHAR_BYTES, "BOOL", false, true, tsCompressBool, tsDecompressBool, getStatics_bool},
     {TSDB_DATA_TYPE_TINYINT, 7, CHAR_BYTES, "TINYINT", INT8_MIN, INT8_MAX, tsCompressTinyint, tsDecompressTinyint,
@@ -388,6 +388,7 @@ tDataTypeDescriptor tDataTypes[15] = {
     {TSDB_DATA_TYPE_UINT, 12, INT_BYTES, "INT UNSIGNED", 0, UINT32_MAX, tsCompressInt, tsDecompressInt, getStatics_u32},
     {TSDB_DATA_TYPE_UBIGINT, 15, LONG_BYTES, "BIGINT UNSIGNED", 0, UINT64_MAX, tsCompressBigint, tsDecompressBigint,
      getStatics_u64},
+    {TSDB_DATA_TYPE_JSON, 4, TSDB_MAX_JSON_TAG_LEN, "JSON", 0, 0, tsCompressString, tsDecompressString, getStatics_nchr},
 };
 
 char tTokenTypeSwitcher[13] = {
diff --git a/source/common/src/tvariant.c b/source/common/src/tvariant.c
index a0503e43a1..a546d9bb6b 100644
--- a/source/common/src/tvariant.c
+++ b/source/common/src/tvariant.c
@@ -118,7 +118,7 @@ void taosVariantCreate(SVariant *pVar, const char *z, int32_t n, int32_t type) {
     }
     case TSDB_DATA_TYPE_BINARY: {
       pVar->pz = strndup(z, n);
-      pVar->nLen = strRmquote(pVar->pz, n);
+      //pVar->nLen = strRmquote(pVar->pz, n);
       break;
     }
     case TSDB_DATA_TYPE_TIMESTAMP: {
diff --git a/source/libs/parser/inc/parUtil.h b/source/libs/parser/inc/parUtil.h
index 742ab303d3..460a089802 100644
--- a/source/libs/parser/inc/parUtil.h
+++ b/source/libs/parser/inc/parUtil.h
@@ -47,6 +47,7 @@ SSchema *getTableTagSchema(const STableMeta* pTableMeta);
 int32_t  getNumOfColumns(const STableMeta* pTableMeta);
 int32_t  getNumOfTags(const STableMeta* pTableMeta);
 STableComInfo getTableInfo(const STableMeta* pTableMeta);
+int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* errMsg, int16_t startColId);
 
 int32_t trimString(const char* src, int32_t len, char* dst, int32_t dlen);
 
diff --git a/source/libs/parser/src/parAstCreater.c b/source/libs/parser/src/parAstCreater.c
index a14053ada7..0dcf04ecc0 100644
--- a/source/libs/parser/src/parAstCreater.c
+++ b/source/libs/parser/src/parAstCreater.c
@@ -47,6 +47,7 @@ void initAstCreateContext(SParseContext* pParseCxt, SAstCreateContext* pCxt) {
 }
 
 static void trimEscape(SToken* pName) {
+  // todo need to deal with `ioo``ii` -> ioo`ii
   if (NULL != pName && pName->n > 1 && '`' == pName->z[0]) {
     pName->z += 1;
     pName->n -= 2;
diff --git a/source/libs/parser/src/parInsert.c b/source/libs/parser/src/parInsert.c
index 435ab317e6..c9a0f4114e 100644
--- a/source/libs/parser/src/parInsert.c
+++ b/source/libs/parser/src/parInsert.c
@@ -377,35 +377,15 @@ static FORCE_INLINE int32_t checkAndTrimValue(SToken* pToken, uint32_t type, cha
     return buildSyntaxErrMsg(pMsgBuf, "invalid data or symbol", pToken->z);
   }
 
-  if (IS_NUMERIC_TYPE(type) && pToken->n == 0) {
-    return buildSyntaxErrMsg(pMsgBuf, "invalid numeric data", pToken->z);
-  }
-
   // Remove quotation marks
   if (TK_NK_STRING == pToken->type) {
     if (pToken->n >= TSDB_MAX_BYTES_PER_ROW) {
       return buildSyntaxErrMsg(pMsgBuf, "too long string", pToken->z);
     }
 
-    // delete escape character: \\, \', \"
-    char delim = pToken->z[0];
-    int32_t cnt = 0;
-    int32_t j = 0;
-    for (uint32_t k = 1; k < pToken->n - 1; ++k) {
-      if (pToken->z[k] == '\\' || (pToken->z[k] == delim && pToken->z[k + 1] == delim)) {
-        tmpTokenBuf[j] = pToken->z[k + 1];
-        cnt++;
-        j++;
-        k++;
-        continue;
-      }
-      tmpTokenBuf[j] = pToken->z[k];
-      j++;
-    }
-
-    tmpTokenBuf[j] = 0;
+    int32_t len = trimString(pToken->z, pToken->n, tmpTokenBuf, TSDB_MAX_BYTES_PER_ROW);
     pToken->z = tmpTokenBuf;
-    pToken->n -= 2 + cnt;
+    pToken->n = len;
   }
 
   return TSDB_CODE_SUCCESS;
@@ -582,6 +562,13 @@ static int32_t parseValueToken(char** end, SToken* pToken, SSchema* pSchema, int
       return func(pMsgBuf, pToken->z, pToken->n, param);
     }
 
+    case TSDB_DATA_TYPE_JSON: {
+      if(pToken->n > (TSDB_MAX_JSON_TAG_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE){
+        return buildSyntaxErrMsg(pMsgBuf, "json string too long than 4095", pToken->z);
+      }
+      return func(pMsgBuf, pToken->z, pToken->n, param);
+    }
+
     case TSDB_DATA_TYPE_TIMESTAMP: {
       int64_t tmpVal;
       if (parseTime(end, pToken, timePrec, &tmpVal, pMsgBuf) != TSDB_CODE_SUCCESS) {
@@ -702,8 +689,10 @@ static int32_t parseBoundColumns(SInsertParseContext* pCxt, SParsedDataColInfo*
     qsort(pColIdx, pColList->numOfBound, sizeof(SBoundIdxInfo), boundIdxCompar);
   }
 
-  memset(&pColList->boundColumns[pColList->numOfBound], 0,
-         sizeof(col_id_t) * (pColList->numOfCols - pColList->numOfBound));
+  if(pColList->numOfCols > pColList->numOfBound){
+    memset(&pColList->boundColumns[pColList->numOfBound], 0,
+           sizeof(col_id_t) * (pColList->numOfCols - pColList->numOfBound));
+  }
 
   return TSDB_CODE_SUCCESS;
 }
@@ -720,22 +709,31 @@ static int32_t KvRowAppend(SMsgBuf* pMsgBuf, const void *value, int32_t len, voi
   int8_t  type = pa->schema->type;
   int16_t colId = pa->schema->colId;
 
+  if(TSDB_DATA_TYPE_JSON == type){
+    return parseJsontoTagData(value, pa->builder, pMsgBuf, colId);
+  }
+
+  if (value == NULL) {  // it is a null data
+    // tdAppendColValToRow(rb, pa->schema->colId, pa->schema->type, TD_VTYPE_NULL, value, false, pa->toffset, pa->colIdx);
+    return TSDB_CODE_SUCCESS;
+  }
+
   if (TSDB_DATA_TYPE_BINARY == type) {
     STR_WITH_SIZE_TO_VARSTR(pa->buf, value, len);
-    tdAddColToKVRow(pa->builder, colId, type, pa->buf);
+    tdAddColToKVRow(pa->builder, colId, pa->buf, varDataTLen(pa->buf));
   } else if (TSDB_DATA_TYPE_NCHAR == type) {
     // if the converted output len is over than pColumnModel->bytes, return error: 'Argument list too long'
     int32_t output = 0;
     if (!taosMbsToUcs4(value, len, (TdUcs4*)varDataVal(pa->buf), pa->schema->bytes - VARSTR_HEADER_SIZE, &output)) {
       char buf[512] = {0};
       snprintf(buf, tListLen(buf), "%s", strerror(errno));
-      return buildSyntaxErrMsg(pMsgBuf, buf, value);;
+      return buildSyntaxErrMsg(pMsgBuf, buf, value);
     }
 
     varDataSetLen(pa->buf, output);
-    tdAddColToKVRow(pa->builder, colId, type, pa->buf);
+    tdAddColToKVRow(pa->builder, colId, varDataTLen(pa->buf));
   } else {
-    tdAddColToKVRow(pa->builder, colId, type, value);
+    tdAddColToKVRow(pa->builder, colId, value, TYPE_BYTES[type]);
   }
 
   return TSDB_CODE_SUCCESS;
diff --git a/source/libs/parser/src/parTranslater.c b/source/libs/parser/src/parTranslater.c
index b3947b872e..bd461c7622 100644
--- a/source/libs/parser/src/parTranslater.c
+++ b/source/libs/parser/src/parTranslater.c
@@ -1542,6 +1542,17 @@ static int32_t checkTableSmaOption(STranslateContext* pCxt, SCreateTableStmt* pS
   return TSDB_CODE_SUCCESS;
 }
 
+static int32_t checkTableTags(STranslateContext* pCxt, SCreateTableStmt* pStmt) {
+  SNode* pNode;
+  FOREACH(pNode, pStmt->pTags) {
+    SColumnDefNode* pCol = (SColumnDefNode*)pNode;
+    if(pCol->dataType.type == TSDB_DATA_TYPE_JSON && LIST_LENGTH(pStmt->pTags) > 1){
+      return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_ONLY_ONE_JSON_TAG);
+    }
+  }
+  return TSDB_CODE_SUCCESS;
+}
+
 static int32_t checkTableRollupOption(STranslateContext* pCxt, SNodeList* pFuncs) {
   if (NULL == pFuncs) {
     return TSDB_CODE_SUCCESS;
@@ -1577,6 +1588,9 @@ static int32_t checkCreateTable(STranslateContext* pCxt, SCreateTableStmt* pStmt
   if (TSDB_CODE_SUCCESS == code) {
     code = checkRangeOption(pCxt, "delay", pStmt->pOptions->pDelay, TSDB_MIN_DB_DELAY, TSDB_MAX_DB_DELAY);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = checkTableTags(pCxt, pStmt);
+  }
   return code;
 }
 
@@ -2893,11 +2907,19 @@ static int32_t addValToKVRow(STranslateContext* pCxt, SValueNode* pVal, const SS
   }
   SVariant var;
   valueNodeToVariant(pVal, &var);
+  if(pSchema->type == TSDB_DATA_TYPE_JSON){
+    if(var.nLen > (TSDB_MAX_JSON_TAG_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE){
+      return buildSyntaxErrMsg(&pCxt->msgBuf, "json string too long than 4095", var.pz);
+    }
+    return parseJsontoTagData(var.pz, pBuilder, &pCxt->msgBuf, pSchema->colId);
+  }
+
   char    tagVal[TSDB_MAX_TAGS_LEN] = {0};
   int32_t code = taosVariantDump(&var, tagVal, pSchema->type, true);
   if (TSDB_CODE_SUCCESS == code) {
-    tdAddColToKVRow(pBuilder, pSchema->colId, pSchema->type, tagVal);
+    tdAddColToKVRow(pBuilder, pSchema->colId, tagVal, IS_VAR_DATA_TYPE(pSchema->type) ? varDataTLen(tagVal) : TYPE_BYTES[pSchema->type]);
   }
+
   return code;
 }
 
diff --git a/source/libs/parser/src/parUtil.c b/source/libs/parser/src/parUtil.c
index 85d574d9e7..4ace62a2f1 100644
--- a/source/libs/parser/src/parUtil.c
+++ b/source/libs/parser/src/parUtil.c
@@ -14,6 +14,7 @@
  */
 
 #include "parUtil.h"
+#include "cJSON.h"
 
 static char* getSyntaxErrFormat(int32_t errCode) {
   switch (errCode) {
@@ -95,6 +96,8 @@ static char* getSyntaxErrFormat(int32_t errCode) {
       return "Invalid option %s unit: %c, only m, h, d allowed";
     case TSDB_CODE_PAR_INVALID_KEEP_UNIT:
       return "Invalid option keep unit: %c, %c, %c, only m, h, d allowed";
+    case TSDB_CODE_PAR_ONLY_ONE_JSON_TAG:
+      return "Only one tag if there is a json tag";
     case TSDB_CODE_OUT_OF_MEMORY:
       return "Out of memory";
     default:
@@ -195,24 +198,180 @@ STableComInfo getTableInfo(const STableMeta* pTableMeta) {
 }
 
 int32_t trimString(const char* src, int32_t len, char* dst, int32_t dlen) {
-  // delete escape character: \\, \', \"
+  if (len <=0 || dlen <= 0) return 0;
+
   char delim = src[0];
-  int32_t cnt = 0;
   int32_t j = 0;
   for (uint32_t k = 1; k < len - 1; ++k) {
     if (j >= dlen) {
-      break;
+      dst[j - 1] = '\0';
+      return j;
     }
-    if (src[k] == '\\' || (src[k] == delim && src[k + 1] == delim)) {
+    if (src[k] == delim && src[k + 1] == delim) {   // deal with "", ''
       dst[j] = src[k + 1];
-      cnt++;
       j++;
       k++;
       continue;
     }
+
+    if (src[k] == '\\') {   // deal with escape character
+      if(src[k+1] == 'n'){
+        dst[j] = '\n';
+      }else if(src[k+1] == 'r'){
+        dst[j] = '\r';
+      }else if(src[k+1] == 't'){
+        dst[j] = '\t';
+      }else if(src[k+1] == '\\'){
+        dst[j] = '\\';
+      }else if(src[k+1] == '\''){
+        dst[j] = '\'';
+      }else if(src[k+1] == '"'){
+        dst[j] = '"';
+      }else if(src[k+1] == '%' || src[k+1] == '_'){
+        dst[j++] = src[k];
+        dst[j] = src[k+1];
+      }else{
+        dst[j] = src[k+1];
+      }
+      j++;
+      k++;
+      continue;
+    }
+
     dst[j] = src[k];
     j++;
   }
   dst[j] = '\0';
   return j;
 }
+
+static bool isValidateTag(char *input) {
+  if (!input) return false;
+  for (size_t i = 0; i < strlen(input); ++i) {
+    if (isprint(input[i]) == 0) return false;
+  }
+  return true;
+}
+
+int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* pMsgBuf, int16_t startColId){
+  // set json NULL data
+  uint8_t jsonKeyNULL = TSDB_JSON_KEY_NULL;
+  uint8_t jsonNULL = TSDB_JSON_NULL;
+  int jsonIndex = startColId + 1;
+  tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonKeyNULL, CHAR_BYTES);   // add json null type
+  if (!json || strtrim(json) == 0 || strncasecmp(json, "null", 4) == 0){
+    tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonNULL, CHAR_BYTES);   // add json null value
+    return TSDB_CODE_SUCCESS;
+  }
+
+  // set json real data
+  cJSON *root = cJSON_Parse(json);
+  if (root == NULL){
+    return buildSyntaxErrMsg(pMsgBuf, "json parse error", json);
+  }
+
+  int size = cJSON_GetArraySize(root);
+  if(!cJSON_IsObject(root)){
+    return buildSyntaxErrMsg(pMsgBuf, "json error invalide value", json);
+  }
+
+  int retCode = 0;
+  SHashObj* keyHash = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, false);
+  for(int i = 0; i < size; i++) {
+    cJSON* item = cJSON_GetArrayItem(root, i);
+    if (!item) {
+      qError("json inner error:%d", i);
+      return buildSyntaxErrMsg(pMsgBuf, "json inner error", json);
+      goto end;
+    }
+
+    char *jsonKey = item->string;
+    if(!isValidateTag(jsonKey)){
+      retCode =  buildSyntaxErrMsg(pMsgBuf, "json key not validate", jsonKey);
+      goto end;
+    }
+//    if(strlen(jsonKey) > TSDB_MAX_JSON_KEY_LEN){
+//      tscError("json key too long error");
+//      retCode =  tscSQLSyntaxErrMsg(errMsg, "json key too long, more than 256", NULL);
+//      goto end;
+//    }
+    size_t keyLen = strlen(jsonKey);
+    if(keyLen == 0 || taosHashGet(keyHash, jsonKey, keyLen) != NULL){
+      continue;
+    }
+
+    if(taosHashGetSize(keyHash) == 0){
+      uint8_t jsonNotNULL = TSDB_JSON_NOT_NULL;
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonNotNULL, CHAR_BYTES);   // add json type
+    }
+    // json key encode by binary
+    void *tagKey = taosMemoryCalloc(keyLen + VARSTR_HEADER_SIZE, 1);
+    if(!tagKey) {
+      retCode = TSDB_CODE_TSC_OUT_OF_MEMORY;
+      goto end;
+    }
+    strncpy(varDataVal(tagKey), jsonKey, keyLen);
+    taosHashPut(keyHash, jsonKey, keyLen, &keyLen, CHAR_BYTES);  // add key to hash to remove dumplicate, value is useless
+
+    varDataSetLen(tagKey, keyLen);
+    tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagKey, varDataTLen(tagKey));   // add json key
+    taosMemoryFree(tagKey);
+
+    if(item->type == cJSON_String){     // add json value  format: type|data
+      char *jsonValue = item->valuestring;
+      int32_t valLen = (int32_t)strlen(jsonValue);
+      char *tagVal = taosMemoryCalloc(valLen * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE + CHAR_BYTES, 1);
+      if(!tagVal) {
+        retCode = TSDB_CODE_TSC_OUT_OF_MEMORY;
+        goto end;
+      } ;
+      tagVal[0] = TSDB_DATA_TYPE_NCHAR;
+      char* tagData = POINTER_SHIFT(tagVal, CHAR_BYTES);
+      if (valLen > 0 && !taosMbsToUcs4(jsonValue, valLen, (TdUcs4*)varDataVal(tagData),
+                                                  (int32_t)(valLen * TSDB_NCHAR_SIZE), &valLen)) {
+        qError("charset:%s to %s. val:%s, errno:%s, convert failed.", DEFAULT_UNICODE_ENCODEC, tsCharset, jsonValue, strerror(errno));
+        retCode = buildSyntaxErrMsg(pMsgBuf, "charset convert json error", jsonValue);
+        taosMemoryFree(tagVal);
+        goto end;
+      }
+
+      varDataSetLen(tagData, valLen);
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagVal, CHAR_BYTES + varDataTLen(tagData));
+      taosMemoryFree(tagVal);
+    }else if(item->type == cJSON_Number){
+      if(!isfinite(item->valuedouble)){
+        qError("json value is invalidate");
+        retCode =  buildSyntaxErrMsg(pMsgBuf, "json value number is illegal", json);
+        goto end;
+      }
+      char tagVal[LONG_BYTES + CHAR_BYTES] = {0};
+      tagVal[0] = (item->valuedouble - (int64_t)(item->valuedouble) == 0) ? TSDB_DATA_TYPE_BIGINT
+                                                                        : TSDB_DATA_TYPE_DOUBLE;
+      char* tagData = POINTER_SHIFT(tagVal,CHAR_BYTES);
+      if(tagVal[0]== TSDB_DATA_TYPE_DOUBLE) *((double *)tagData) = item->valuedouble;
+      else if(tagVal[0] == TSDB_DATA_TYPE_BIGINT) *((int64_t *)tagData) = item->valueint;
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagVal, LONG_BYTES + CHAR_BYTES);
+    }else if(item->type == cJSON_True || item->type == cJSON_False){
+      char tagVal[CHAR_BYTES + CHAR_BYTES] = {0};
+      tagVal[0] = TSDB_DATA_TYPE_BOOL;
+      tagVal[1] = (char)(item->valueint);
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagVal, CHAR_BYTES + CHAR_BYTES);
+    }else if(item->type == cJSON_NULL){
+      char tagVal[CHAR_BYTES] = {TSDB_DATA_TYPE_NULL};
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagVal, CHAR_BYTES);
+    }
+    else{
+      retCode = buildSyntaxErrMsg(pMsgBuf, "invalidate json value", json);
+      goto end;
+    }
+  }
+
+  if(taosHashGetSize(keyHash) == 0){  // set json NULL true
+    tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonNULL, CHAR_BYTES);
+  }
+
+end:
+  taosHashCleanup(keyHash);
+  cJSON_Delete(root);
+  return retCode;
+}
\ No newline at end of file
diff --git a/source/libs/scalar/src/filter.c b/source/libs/scalar/src/filter.c
index 191143de12..711ece3b01 100644
--- a/source/libs/scalar/src/filter.c
+++ b/source/libs/scalar/src/filter.c
@@ -57,7 +57,7 @@ OptrStr gOptrStr[] = {
   {OP_TYPE_IS_NOT_UNKNOWN,                 "not unknown"},
 
   // json operator
-  {OP_TYPE_JSON_GET_VALUE,                 "json get"},
+  {OP_TYPE_JSON_GET_VALUE,                 "->"},
   {OP_TYPE_JSON_CONTAINS,                  "json contains"}
 };
 
diff --git a/source/util/src/tjson.c b/source/util/src/tjson.c
index 93a843fee8..8df404f85f 100644
--- a/source/util/src/tjson.c
+++ b/source/util/src/tjson.c
@@ -276,3 +276,4 @@ int32_t tjsonToArray(const SJson* pJson, const char* pName, FToObject func, void
 }
 
 SJson* tjsonParse(const char* pStr) { return cJSON_Parse(pStr); }
+bool tjsonIsObject(const SJson* pJson) { return cJSON_IsObject(pJson); }
\ No newline at end of file
diff --git a/source/util/src/tutil.c b/source/util/src/tutil.c
index 8133e4d237..adb6a37ba7 100644
--- a/source/util/src/tutil.c
+++ b/source/util/src/tutil.c
@@ -47,65 +47,6 @@ int32_t strdequote(char *z) {
   return j + 1;  // only one quote, do nothing
 }
 
-int32_t strRmquote(char *z, int32_t len) {
-  // delete escape character: \\, \', \"
-  char delim = z[0];
-  if (delim != '\'' && delim != '\"') {
-    return len;
-  }
-
-  int32_t cnt = 0;
-  int32_t j = 0;
-  for (uint32_t k = 1; k < len - 1; ++k) {
-    if (z[k] == '\\' || (z[k] == delim && z[k + 1] == delim)) {
-      if (z[k] == '\\' && z[k + 1] == '_') {
-        // match '_' self
-      } else {
-        z[j] = z[k + 1];
-        cnt++;
-        j++;
-        k++;
-        continue;
-      }
-    }
-
-    z[j] = z[k];
-    j++;
-  }
-
-  z[j] = 0;
-
-  return len - 2 - cnt;
-}
-
-int32_t strndequote(char *dst, const char *z, int32_t len) {
-  assert(dst != NULL);
-  if (z == NULL || len == 0) {
-    return 0;
-  }
-
-  int32_t quote = z[0];
-  int32_t i = 1, j = 0;
-
-  while (z[i] != 0) {
-    if (z[i] == quote) {
-      if (z[i + 1] == quote) {
-        dst[j++] = (char)quote;
-        i++;
-      } else {
-        dst[j++] = 0;
-        return (j - 1);
-      }
-    } else {
-      dst[j++] = z[i];
-    }
-
-    i++;
-  }
-
-  return j + 1;  // only one quote, do nothing
-}
-
 size_t strtrim(char *z) {
   int32_t i = 0;
   int32_t j = 0;
diff --git a/tools/shell/src/shellEngine.c b/tools/shell/src/shellEngine.c
index b89d517ad3..5a96a2a87d 100644
--- a/tools/shell/src/shellEngine.c
+++ b/tools/shell/src/shellEngine.c
@@ -226,63 +226,27 @@ int32_t shellRunCommand(TAOS *con, char *command) {
     }
   }
 
-  bool esc = false;
-  char quote = 0, *cmd = command, *p = command;
+  char quote = 0, *cmd = command;
   for (char c = *command++; c != 0; c = *command++) {
-    if (esc) {
-      switch (c) {
-        case 'n':
-          c = '\n';
-          break;
-        case 'r':
-          c = '\r';
-          break;
-        case 't':
-          c = '\t';
-          break;
-        case 'G':
-          *p++ = '\\';
-          break;
-        case '\'':
-        case '"':
-          if (quote) {
-            *p++ = '\\';
-          }
-          break;
-      }
-      *p++ = c;
-      esc = false;
+    if (c == '\\' && (*command == '\'' || *command == '"' || *command == '`')) {
+      command ++;
       continue;
     }
 
-    if (c == '\\') {
-      if (quote != 0 && (*command == '_' || *command == '\\')) {
-        // DO nothing
-      } else {
-        esc = true;
-        continue;
-      }
-    }
-
     if (quote == c) {
       quote = 0;
-    } else if (quote == 0 && (c == '\'' || c == '"')) {
+    } else if (quote == 0 && (c == '\'' || c == '"' || c == '`')) {
       quote = c;
-    }
-
-    *p++ = c;
-    if (c == ';' && quote == 0) {
-      c = *p;
-      *p = 0;
+    } else if (c == ';' && quote == 0) {
+      c = *command;
+      *command = 0;
       if (shellRunSingleCommand(con, cmd) < 0) {
         return -1;
       }
-      *p = c;
-      p = cmd;
+      *command = c;
+      cmd = command;
     }
   }
-
-  *p = 0;
   return shellRunSingleCommand(con, cmd);
 }
 
@@ -573,19 +537,23 @@ static void shellPrintNChar(const char *str, int length, int width) {
   while (pos < length) {
     TdWchar wc;
     int     bytes = taosMbToWchar(&wc, str + pos, MB_CUR_MAX);
-    if (bytes == 0) {
+    if (bytes <= 0) {
       break;
     }
-    pos += bytes;
-    if (pos > length) {
+    if (pos + bytes > length) {
       break;
     }
-
+    int w = 0;
 #ifdef WINDOWS
-    int w = bytes;
+    w = bytes;
 #else
-    int w = taosWcharWidth(wc);
+    if(*(str + pos) == '\t' || *(str + pos) == '\n' || *(str + pos) == '\r'){
+      w = bytes;
+    }else{
+      w = taosWcharWidth(wc);
+    }
 #endif
+    pos += bytes;
     if (w <= 0) {
       continue;
     }
@@ -679,6 +647,7 @@ static void printField(const char *val, TAOS_FIELD *field, int width, int32_t le
       break;
     case TSDB_DATA_TYPE_BINARY:
     case TSDB_DATA_TYPE_NCHAR:
+    case TSDB_DATA_TYPE_JSON:
       shellPrintNChar(val, length, width);
       break;
     case TSDB_DATA_TYPE_TIMESTAMP:

From 807252356147847cb4b3018e2aee04a651e128ac Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Thu, 14 Apr 2022 19:01:11 +0800
Subject: [PATCH 02/15] refactor: add json tag function for sql parser

---
 source/libs/parser/src/parInsert.c | 2 +-
 source/libs/parser/src/parUtil.c   | 7 ++++---
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/source/libs/parser/src/parInsert.c b/source/libs/parser/src/parInsert.c
index 9bc81d8fba..7d2d928aa9 100644
--- a/source/libs/parser/src/parInsert.c
+++ b/source/libs/parser/src/parInsert.c
@@ -738,7 +738,7 @@ static int32_t KvRowAppend(SMsgBuf* pMsgBuf, const void *value, int32_t len, voi
     }
 
     varDataSetLen(pa->buf, output);
-    tdAddColToKVRow(pa->builder, colId, varDataTLen(pa->buf));
+    tdAddColToKVRow(pa->builder, colId, pa->buf, varDataTLen(pa->buf));
   } else {
     tdAddColToKVRow(pa->builder, colId, value, TYPE_BYTES[type]);
   }
diff --git a/source/libs/parser/src/parUtil.c b/source/libs/parser/src/parUtil.c
index 4ace62a2f1..a423643e42 100644
--- a/source/libs/parser/src/parUtil.c
+++ b/source/libs/parser/src/parUtil.c
@@ -242,6 +242,7 @@ int32_t trimString(const char* src, int32_t len, char* dst, int32_t dlen) {
     j++;
   }
   dst[j] = '\0';
+  strtrim(dst);
   return j;
 }
 
@@ -259,7 +260,7 @@ int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* p
   uint8_t jsonNULL = TSDB_JSON_NULL;
   int jsonIndex = startColId + 1;
   tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonKeyNULL, CHAR_BYTES);   // add json null type
-  if (!json || strtrim(json) == 0 || strncasecmp(json, "null", 4) == 0){
+  if (!json || strcasecmp(json, TSDB_DATA_NULL_STR_L) == 0){
     tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonNULL, CHAR_BYTES);   // add json null value
     return TSDB_CODE_SUCCESS;
   }
@@ -281,13 +282,13 @@ int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* p
     cJSON* item = cJSON_GetArrayItem(root, i);
     if (!item) {
       qError("json inner error:%d", i);
-      return buildSyntaxErrMsg(pMsgBuf, "json inner error", json);
+      retCode = buildSyntaxErrMsg(pMsgBuf, "json inner error", json);
       goto end;
     }
 
     char *jsonKey = item->string;
     if(!isValidateTag(jsonKey)){
-      retCode =  buildSyntaxErrMsg(pMsgBuf, "json key not validate", jsonKey);
+      retCode = buildSyntaxErrMsg(pMsgBuf, "json key not validate", jsonKey);
       goto end;
     }
 //    if(strlen(jsonKey) > TSDB_MAX_JSON_KEY_LEN){

From b7bd654c773306a781045ed7f5b21705d71cdb97 Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Sat, 16 Apr 2022 21:24:07 +0800
Subject: [PATCH 03/15] refactor: add compare logic for json value

---
 include/common/ttypes.h                |   2 +-
 include/util/tcompare.h                |   2 +
 include/util/tdef.h                    |   1 -
 source/libs/nodes/src/nodesUtilFuncs.c |   2 +-
 source/libs/parser/src/parTranslater.c |  38 ++--
 source/libs/parser/src/parUtil.c       |  76 ++++---
 source/libs/scalar/inc/sclvector.h     |   4 +
 source/libs/scalar/src/filter.c        |   9 +-
 source/libs/scalar/src/scalar.c        |  77 ++++---
 source/libs/scalar/src/sclvector.c     | 283 +++++++++++++++++++++----
 source/util/src/tcompare.c             |   5 +
 11 files changed, 358 insertions(+), 141 deletions(-)

diff --git a/include/common/ttypes.h b/include/common/ttypes.h
index 19455db258..7c87235c39 100644
--- a/include/common/ttypes.h
+++ b/include/common/ttypes.h
@@ -49,7 +49,7 @@ typedef struct {
 #define varDataCopy(dst, v)    memcpy((dst), (void *)(v), varDataTLen(v))
 #define varDataLenByData(v)    (*(VarDataLenT *)(((char *)(v)) - VARSTR_HEADER_SIZE))
 #define varDataSetLen(v, _len) (((VarDataLenT *)(v))[0] = (VarDataLenT)(_len))
-#define IS_VAR_DATA_TYPE(t)    (((t) == TSDB_DATA_TYPE_VARCHAR) || ((t) == TSDB_DATA_TYPE_NCHAR))
+#define IS_VAR_DATA_TYPE(t)    (((t) == TSDB_DATA_TYPE_VARCHAR) || ((t) == TSDB_DATA_TYPE_NCHAR) || ((t) == TSDB_DATA_TYPE_JSON))
 
 #define varDataNetLen(v)  (htons(((VarDataLenT *)(v))[0]))
 #define varDataNetTLen(v) (sizeof(VarDataLenT) + varDataNetLen(v))
diff --git a/include/util/tcompare.h b/include/util/tcompare.h
index cc9e8ae464..ed07ae1c9c 100644
--- a/include/util/tcompare.h
+++ b/include/util/tcompare.h
@@ -105,6 +105,8 @@ int32_t compareStrPatternNotMatch(const void *pLeft, const void *pRight);
 int32_t compareWStrPatternMatch(const void *pLeft, const void *pRight);
 int32_t compareWStrPatternNotMatch(const void *pLeft, const void *pRight);
 
+int32_t compareJsonContainsKey(const void *pLeft, const void *pRight);
+
 __compar_fn_t getComparFunc(int32_t type, int32_t optr);
 __compar_fn_t getKeyComparFunc(int32_t keyType, int32_t order);
 int32_t       doCompare(const char *a, const char *b, int32_t type, size_t size);
diff --git a/include/util/tdef.h b/include/util/tdef.h
index caf01f5a22..eacc3aa08e 100644
--- a/include/util/tdef.h
+++ b/include/util/tdef.h
@@ -274,7 +274,6 @@ typedef enum ELogicConditionType {
 #define TSDB_MAX_JSON_TAG_LEN   16384
 #define TSDB_JSON_PLACEHOLDER      0x7F
 #define TSDB_JSON_null             0x00
-#define TSDB_JSON_KEY_NULL         0x00
 #define TSDB_JSON_NOT_NULL         0x01
 #define TSDB_JSON_NULL             0x00
 
diff --git a/source/libs/nodes/src/nodesUtilFuncs.c b/source/libs/nodes/src/nodesUtilFuncs.c
index 758f9d5d6f..1154894050 100644
--- a/source/libs/nodes/src/nodesUtilFuncs.c
+++ b/source/libs/nodes/src/nodesUtilFuncs.c
@@ -1006,6 +1006,7 @@ bool nodesIsComparisonOp(const SOperatorNode* pOp) {
     case OP_TYPE_NOT_LIKE:
     case OP_TYPE_MATCH:
     case OP_TYPE_NMATCH:
+    case OP_TYPE_JSON_CONTAINS:
     case OP_TYPE_IS_NULL:
     case OP_TYPE_IS_NOT_NULL:
     case OP_TYPE_IS_TRUE:
@@ -1024,7 +1025,6 @@ bool nodesIsComparisonOp(const SOperatorNode* pOp) {
 bool nodesIsJsonOp(const SOperatorNode* pOp) {
   switch (pOp->opType) {
     case OP_TYPE_JSON_GET_VALUE:
-    case OP_TYPE_JSON_CONTAINS:
       return true;
     default:
       break;
diff --git a/source/libs/parser/src/parTranslater.c b/source/libs/parser/src/parTranslater.c
index e0708e8cea..4a20d929b5 100644
--- a/source/libs/parser/src/parTranslater.c
+++ b/source/libs/parser/src/parTranslater.c
@@ -469,14 +469,18 @@ static EDealRes translateOperator(STranslateContext* pCxt, SOperatorNode* pOp) {
     pOp->node.resType.type = TSDB_DATA_TYPE_DOUBLE;
     pOp->node.resType.bytes = tDataTypes[TSDB_DATA_TYPE_DOUBLE].bytes;
   } else if (nodesIsComparisonOp(pOp)) {
-    if (TSDB_DATA_TYPE_JSON == ldt.type || TSDB_DATA_TYPE_BLOB == ldt.type || TSDB_DATA_TYPE_JSON == rdt.type ||
+    if (TSDB_DATA_TYPE_BLOB == ldt.type || TSDB_DATA_TYPE_JSON == rdt.type ||
         TSDB_DATA_TYPE_BLOB == rdt.type) {
       return generateDealNodeErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, ((SExprNode*)(pOp->pRight))->aliasName);
     }
     pOp->node.resType.type = TSDB_DATA_TYPE_BOOL;
     pOp->node.resType.bytes = tDataTypes[TSDB_DATA_TYPE_BOOL].bytes;
-  } else {
-    // todo json operator
+  } else if (nodesIsJsonOp(pOp)){
+    if (TSDB_DATA_TYPE_JSON != ldt.type || TSDB_DATA_TYPE_BINARY != rdt.type) {
+      return generateDealNodeErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, ((SExprNode*)(pOp->pRight))->aliasName);
+    }
+    pOp->node.resType.type = TSDB_DATA_TYPE_JSON;
+    pOp->node.resType.bytes = tDataTypes[TSDB_DATA_TYPE_JSON].bytes;
   }
   return DEAL_RES_CONTINUE;
 }
@@ -2814,25 +2818,25 @@ static void addCreateTbReqIntoVgroup(int32_t acctId, SHashObj* pVgroupHashmap, c
 
 static int32_t addValToKVRow(STranslateContext* pCxt, SValueNode* pVal, const SSchema* pSchema,
                              SKVRowBuilder* pBuilder) {
+  if(pSchema->type == TSDB_DATA_TYPE_JSON){
+    if(pVal->literal && strlen(pVal->literal) > (TSDB_MAX_JSON_TAG_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE){
+      return buildSyntaxErrMsg(&pCxt->msgBuf, "json string too long than 4095", pVal->literal);
+    }
+
+    return parseJsontoTagData(pVal->literal, pBuilder, &pCxt->msgBuf, pSchema->colId);
+  }
+
   if (DEAL_RES_ERROR == translateValue(pCxt, pVal)) {
     return pCxt->errCode;
   }
-  SVariant var;
-  valueNodeToVariant(pVal, &var);
-  if(pSchema->type == TSDB_DATA_TYPE_JSON){
-    if(var.nLen > (TSDB_MAX_JSON_TAG_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE){
-      return buildSyntaxErrMsg(&pCxt->msgBuf, "json string too long than 4095", var.pz);
-    }
-    return parseJsontoTagData(var.pz, pBuilder, &pCxt->msgBuf, pSchema->colId);
+
+  if(pVal->node.resType.type == TSDB_DATA_TYPE_NULL){
+    // todo
+  }else{
+    tdAddColToKVRow(pBuilder, pSchema->colId, &(pVal->datum.p), IS_VAR_DATA_TYPE(pSchema->type) ? varDataTLen(pVal->datum.p) : TYPE_BYTES[pSchema->type]);
   }
 
-  char    tagVal[TSDB_MAX_TAGS_LEN] = {0};
-  int32_t code = taosVariantDump(&var, tagVal, pSchema->type, true);
-  if (TSDB_CODE_SUCCESS == code) {
-    tdAddColToKVRow(pBuilder, pSchema->colId, tagVal, IS_VAR_DATA_TYPE(pSchema->type) ? varDataTLen(tagVal) : TYPE_BYTES[pSchema->type]);
-  }
-
-  return code;
+  return TSDB_CODE_SUCCESS;
 }
 
 static int32_t buildKVRowForBindTags(STranslateContext* pCxt, SCreateSubTableClause* pStmt, STableMeta* pSuperTableMeta,
diff --git a/source/libs/parser/src/parUtil.c b/source/libs/parser/src/parUtil.c
index 968b799816..2bb867e8d4 100644
--- a/source/libs/parser/src/parUtil.c
+++ b/source/libs/parser/src/parUtil.c
@@ -252,12 +252,10 @@ static bool isValidateTag(char *input) {
 
 int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* pMsgBuf, int16_t startColId){
   // set json NULL data
-  uint8_t jsonKeyNULL = TSDB_JSON_KEY_NULL;
   uint8_t jsonNULL = TSDB_JSON_NULL;
   int jsonIndex = startColId + 1;
-  tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonKeyNULL, CHAR_BYTES);   // add json null type
   if (!json || strcasecmp(json, TSDB_DATA_NULL_STR_L) == 0){
-    tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonNULL, CHAR_BYTES);   // add json null value
+    tdAddColToKVRow(kvRowBuilder, jsonIndex, &jsonNULL, CHAR_BYTES);
     return TSDB_CODE_SUCCESS;
   }
 
@@ -273,6 +271,7 @@ int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* p
   }
 
   int retCode = 0;
+  char *tagKV = NULL;
   SHashObj* keyHash = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, false);
   for(int i = 0; i < size; i++) {
     cJSON* item = cJSON_GetArrayItem(root, i);
@@ -296,66 +295,64 @@ int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* p
     if(keyLen == 0 || taosHashGet(keyHash, jsonKey, keyLen) != NULL){
       continue;
     }
-
+    // key: keyLen + VARSTR_HEADER_SIZE, value type: CHAR_BYTES, value reserved: LONG_BYTES
+    tagKV = taosMemoryCalloc(keyLen + VARSTR_HEADER_SIZE + CHAR_BYTES + LONG_BYTES, 1);
+    if(!tagKV) {
+      retCode = TSDB_CODE_TSC_OUT_OF_MEMORY;
+      goto end;
+    }
+    strncpy(varDataVal(tagKV), jsonKey, keyLen);
+    varDataSetLen(tagKV, keyLen);
     if(taosHashGetSize(keyHash) == 0){
       uint8_t jsonNotNULL = TSDB_JSON_NOT_NULL;
       tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonNotNULL, CHAR_BYTES);   // add json type
     }
-    // json key encode by binary
-    void *tagKey = taosMemoryCalloc(keyLen + VARSTR_HEADER_SIZE, 1);
-    if(!tagKey) {
-      retCode = TSDB_CODE_TSC_OUT_OF_MEMORY;
-      goto end;
-    }
-    strncpy(varDataVal(tagKey), jsonKey, keyLen);
     taosHashPut(keyHash, jsonKey, keyLen, &keyLen, CHAR_BYTES);  // add key to hash to remove dumplicate, value is useless
 
-    varDataSetLen(tagKey, keyLen);
-    tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagKey, varDataTLen(tagKey));   // add json key
-    taosMemoryFree(tagKey);
-
     if(item->type == cJSON_String){     // add json value  format: type|data
       char *jsonValue = item->valuestring;
       int32_t valLen = (int32_t)strlen(jsonValue);
-      char *tagVal = taosMemoryCalloc(valLen * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE + CHAR_BYTES, 1);
-      if(!tagVal) {
+      int32_t totalLen = keyLen + VARSTR_HEADER_SIZE + valLen * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE + CHAR_BYTES;
+      char *tmp = taosMemoryRealloc(tagKV, totalLen);
+      if(!tmp) {
         retCode = TSDB_CODE_TSC_OUT_OF_MEMORY;
         goto end;
-      } ;
-      tagVal[0] = TSDB_DATA_TYPE_NCHAR;
-      char* tagData = POINTER_SHIFT(tagVal, CHAR_BYTES);
-      if (valLen > 0 && !taosMbsToUcs4(jsonValue, valLen, (TdUcs4*)varDataVal(tagData),
+      }
+      tagKV = tmp;
+      char* valueType = POINTER_SHIFT(tagKV, keyLen + VARSTR_HEADER_SIZE);
+      char* valueData = POINTER_SHIFT(tagKV, keyLen + VARSTR_HEADER_SIZE + CHAR_BYTES);
+      *valueType = TSDB_DATA_TYPE_NCHAR;
+      if (valLen > 0 && !taosMbsToUcs4(jsonValue, valLen, (TdUcs4*)varDataVal(valueData),
                                                   (int32_t)(valLen * TSDB_NCHAR_SIZE), &valLen)) {
         qError("charset:%s to %s. val:%s, errno:%s, convert failed.", DEFAULT_UNICODE_ENCODEC, tsCharset, jsonValue, strerror(errno));
         retCode = buildSyntaxErrMsg(pMsgBuf, "charset convert json error", jsonValue);
-        taosMemoryFree(tagVal);
         goto end;
       }
 
-      varDataSetLen(tagData, valLen);
-      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagVal, CHAR_BYTES + varDataTLen(tagData));
-      taosMemoryFree(tagVal);
+      varDataSetLen(valueData, valLen);
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagKV, totalLen);
     }else if(item->type == cJSON_Number){
       if(!isfinite(item->valuedouble)){
         qError("json value is invalidate");
         retCode =  buildSyntaxErrMsg(pMsgBuf, "json value number is illegal", json);
         goto end;
       }
-      char tagVal[LONG_BYTES + CHAR_BYTES] = {0};
-      tagVal[0] = (item->valuedouble - (int64_t)(item->valuedouble) == 0) ? TSDB_DATA_TYPE_BIGINT
-                                                                        : TSDB_DATA_TYPE_DOUBLE;
-      char* tagData = POINTER_SHIFT(tagVal,CHAR_BYTES);
-      if(tagVal[0]== TSDB_DATA_TYPE_DOUBLE) *((double *)tagData) = item->valuedouble;
-      else if(tagVal[0] == TSDB_DATA_TYPE_BIGINT) *((int64_t *)tagData) = item->valueint;
-      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagVal, LONG_BYTES + CHAR_BYTES);
+      char* valueType = POINTER_SHIFT(tagKV, keyLen + VARSTR_HEADER_SIZE);
+      char* valueData = POINTER_SHIFT(tagKV, keyLen + VARSTR_HEADER_SIZE + CHAR_BYTES);
+      *valueType = (item->valuedouble - (int64_t)(item->valuedouble) == 0) ? TSDB_DATA_TYPE_BIGINT : TSDB_DATA_TYPE_DOUBLE;
+      if(*valueType== TSDB_DATA_TYPE_DOUBLE) *((double *)valueData) = item->valuedouble;
+      else if(*valueType == TSDB_DATA_TYPE_BIGINT) *((int64_t *)valueData) = item->valueint;
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagKV, keyLen + VARSTR_HEADER_SIZE + CHAR_BYTES +LONG_BYTES);
     }else if(item->type == cJSON_True || item->type == cJSON_False){
-      char tagVal[CHAR_BYTES + CHAR_BYTES] = {0};
-      tagVal[0] = TSDB_DATA_TYPE_BOOL;
-      tagVal[1] = (char)(item->valueint);
-      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagVal, CHAR_BYTES + CHAR_BYTES);
+      char* valueType = POINTER_SHIFT(tagKV, keyLen + VARSTR_HEADER_SIZE);
+      char* valueData = POINTER_SHIFT(tagKV, keyLen + VARSTR_HEADER_SIZE + CHAR_BYTES);
+      *valueType = TSDB_DATA_TYPE_BOOL;
+      *valueData = (char)(item->valueint);
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagKV, keyLen + VARSTR_HEADER_SIZE + CHAR_BYTES + CHAR_BYTES);
     }else if(item->type == cJSON_NULL){
-      char tagVal[CHAR_BYTES] = {TSDB_DATA_TYPE_NULL};
-      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagVal, CHAR_BYTES);
+      char* valueType = POINTER_SHIFT(tagKV, keyLen + VARSTR_HEADER_SIZE);
+      *valueType = TSDB_DATA_TYPE_NULL;
+      tdAddColToKVRow(kvRowBuilder, jsonIndex++, tagKV, keyLen + VARSTR_HEADER_SIZE + CHAR_BYTES);
     }
     else{
       retCode = buildSyntaxErrMsg(pMsgBuf, "invalidate json value", json);
@@ -364,10 +361,11 @@ int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* p
   }
 
   if(taosHashGetSize(keyHash) == 0){  // set json NULL true
-    tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonNULL, CHAR_BYTES);
+    tdAddColToKVRow(kvRowBuilder, jsonIndex, &jsonNULL, CHAR_BYTES);
   }
 
 end:
+  taosMemoryFree(tagKV);
   taosHashCleanup(keyHash);
   cJSON_Delete(root);
   return retCode;
diff --git a/source/libs/scalar/inc/sclvector.h b/source/libs/scalar/inc/sclvector.h
index e51115e9c0..47205b7fba 100644
--- a/source/libs/scalar/inc/sclvector.h
+++ b/source/libs/scalar/inc/sclvector.h
@@ -53,6 +53,8 @@ static FORCE_INLINE double getVectorDoubleValue_DOUBLE(void *src, int32_t index)
   return (double)*((double *)src + index);
 }
 
+double getVectorDoubleValue_JSON(void *src, int32_t index);
+
 static FORCE_INLINE _getDoubleValue_fn_t getVectorDoubleValueFn(int32_t srcType) {
   _getDoubleValue_fn_t p = NULL;
   if (srcType == TSDB_DATA_TYPE_TINYINT) {
@@ -77,6 +79,8 @@ static FORCE_INLINE _getDoubleValue_fn_t getVectorDoubleValueFn(int32_t srcType)
     p = getVectorDoubleValue_DOUBLE;
   } else if (srcType == TSDB_DATA_TYPE_TIMESTAMP) {
     p = getVectorDoubleValue_BIGINT;
+  } else if (srcType == TSDB_DATA_TYPE_JSON) {
+    p = getVectorDoubleValue_JSON;
   } else {
     assert(0);
   }
diff --git a/source/libs/scalar/src/filter.c b/source/libs/scalar/src/filter.c
index 711ece3b01..ea249005b4 100644
--- a/source/libs/scalar/src/filter.c
+++ b/source/libs/scalar/src/filter.c
@@ -170,7 +170,7 @@ __compar_fn_t gDataCompare[] = {compareInt32Val, compareInt8Val, compareInt16Val
   compareLenPrefixedWStr, compareUint8Val, compareUint16Val, compareUint32Val, compareUint64Val,
   setChkInBytes1, setChkInBytes2, setChkInBytes4, setChkInBytes8, compareStrRegexCompMatch, 
   compareStrRegexCompNMatch, setChkNotInBytes1, setChkNotInBytes2, setChkNotInBytes4, setChkNotInBytes8,
-  compareChkNotInString, compareStrPatternNotMatch, compareWStrPatternNotMatch
+  compareChkNotInString, compareStrPatternNotMatch, compareWStrPatternNotMatch, compareJsonContainsKey
 };
 
 int8_t filterGetCompFuncIdx(int32_t type, int32_t optr) {
@@ -221,7 +221,12 @@ int8_t filterGetCompFuncIdx(int32_t type, int32_t optr) {
         assert(0);
     }
   }
-  
+
+  if (optr == OP_TYPE_JSON_CONTAINS && type == TSDB_DATA_TYPE_JSON) {
+    return 28;
+  }
+
+
   switch (type) {
     case TSDB_DATA_TYPE_BOOL:
     case TSDB_DATA_TYPE_TINYINT:   comparFn = 1;   break;
diff --git a/source/libs/scalar/src/scalar.c b/source/libs/scalar/src/scalar.c
index 116dfdd5d5..371b011c40 100644
--- a/source/libs/scalar/src/scalar.c
+++ b/source/libs/scalar/src/scalar.c
@@ -600,44 +600,39 @@ EDealRes sclWalkOperator(SNode* pNode, SScalarCtx *ctx) {
   return DEAL_RES_CONTINUE;
 }
 
-EDealRes sclWalkTarget(SNode* pNode, SScalarCtx *ctx) {
-  STargetNode *target = (STargetNode *)pNode;
-  
-  if (target->dataBlockId >= taosArrayGetSize(ctx->pBlockList)) {
-    sclError("target tupleId is too big, tupleId:%d, dataBlockNum:%d", target->dataBlockId, (int32_t)taosArrayGetSize(ctx->pBlockList));
-    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
-    return DEAL_RES_ERROR;
-  }
-
-  SSDataBlock *block = *(SSDataBlock **)taosArrayGet(ctx->pBlockList, target->dataBlockId);
-  if (target->slotId >= taosArrayGetSize(block->pDataBlock)) {
-    sclError("target slot not exist, dataBlockId:%d, slotId:%d, dataBlockNum:%d", target->dataBlockId, target->slotId, (int32_t)taosArrayGetSize(block->pDataBlock));
-    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
-    return DEAL_RES_ERROR;
-  }
-
-  SColumnInfoData *col = taosArrayGet(block->pDataBlock, target->slotId);
-  
-  SScalarParam *res = (SScalarParam *)taosHashGet(ctx->pRes, (void *)&target->pExpr, POINTER_BYTES);
-  if (NULL == res) {
-    sclError("no valid res in hash, node:%p, type:%d", target->pExpr, nodeType(target->pExpr));
-    ctx->code = TSDB_CODE_QRY_APP_ERROR;
-    return DEAL_RES_ERROR;
-  }
-  
-  for (int32_t i = 0; i < res->numOfRows; ++i) {
-    if (colDataIsNull(res->columnData, res->numOfRows, i, NULL)) {
-      colDataAppend(col, i, NULL, true);
-    } else {
-      char *p = colDataGetData(res->columnData, i);
-      colDataAppend(col, i, p, false);
-    }
-  }
-
-  sclFreeParam(res);
-  taosHashRemove(ctx->pRes, (void *)&target->pExpr, POINTER_BYTES);
-  return DEAL_RES_CONTINUE;
-}
+//EDealRes sclWalkTarget(SNode* pNode, SScalarCtx *ctx) {
+//  STargetNode *target = (STargetNode *)pNode;
+//
+//  if (target->dataBlockId >= taosArrayGetSize(ctx->pBlockList)) {
+//    sclError("target tupleId is too big, tupleId:%d, dataBlockNum:%d", target->dataBlockId, (int32_t)taosArrayGetSize(ctx->pBlockList));
+//    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
+//    return DEAL_RES_ERROR;
+//  }
+//
+//  SSDataBlock *block = *(SSDataBlock **)taosArrayGet(ctx->pBlockList, target->dataBlockId);
+//  if (target->slotId >= taosArrayGetSize(block->pDataBlock)) {
+//    sclError("target slot not exist, dataBlockId:%d, slotId:%d, dataBlockNum:%d", target->dataBlockId, target->slotId, (int32_t)taosArrayGetSize(block->pDataBlock));
+//    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
+//    return DEAL_RES_ERROR;
+//  }
+//
+//  // if block->pDataBlock is not enough, there are problems if target->slotId bigger than the size of block->pDataBlock,
+//  SColumnInfoData *col = taosArrayGet(block->pDataBlock, target->slotId);
+//
+//  SScalarParam *res = (SScalarParam *)taosHashGet(ctx->pRes, (void *)&target->pExpr, POINTER_BYTES);
+//  if (NULL == res) {
+//    sclError("no valid res in hash, node:%p, type:%d", target->pExpr, nodeType(target->pExpr));
+//    ctx->code = TSDB_CODE_QRY_APP_ERROR;
+//    return DEAL_RES_ERROR;
+//  }
+//
+//  colDataAssign(col, res->columnData, res->numOfRows);
+//  block->info.rows = res->numOfRows;
+//
+//  sclFreeParam(res);
+//  taosHashRemove(ctx->pRes, (void *)&target->pExpr, POINTER_BYTES);
+//  return DEAL_RES_CONTINUE;
+//}
 
 EDealRes sclCalcWalker(SNode* pNode, void* pContext) {
   if (QUERY_NODE_VALUE == nodeType(pNode) || QUERY_NODE_NODE_LIST == nodeType(pNode) || QUERY_NODE_COLUMN == nodeType(pNode)) {
@@ -657,9 +652,9 @@ EDealRes sclCalcWalker(SNode* pNode, void* pContext) {
     return sclWalkOperator(pNode, ctx);
   }
 
-  if (QUERY_NODE_TARGET == nodeType(pNode)) {
-    return sclWalkTarget(pNode, ctx);
-  }
+//  if (QUERY_NODE_TARGET == nodeType(pNode)) {
+//    return sclWalkTarget(pNode, ctx);
+//  }
 
   sclError("invalid node type for scalar calculating, type:%d", nodeType(pNode));
   ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c
index 05456790a5..11dc60dab3 100644
--- a/source/libs/scalar/src/sclvector.c
+++ b/source/libs/scalar/src/sclvector.c
@@ -177,24 +177,24 @@ static FORCE_INLINE void varToBool(char *buf, SScalarParam* pOut, int32_t rowInd
   colDataAppendInt8(pOut->columnData, rowIndex, (int8_t*) &v);
 }
 
-static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIndex) {
-  int32_t len = 0;
-  int32_t inputLen = varDataLen(buf);
-
-  char* t = taosMemoryCalloc(1,(inputLen + 1) * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE);
-  /*int32_t resLen = */taosMbsToUcs4(varDataVal(buf), inputLen, (TdUcs4*) varDataVal(t), pOut->columnData->info.bytes, &len);
-  varDataSetLen(t, len);
-
-  colDataAppend(pOut->columnData, rowIndex, t, false);
-  taosMemoryFree(t);
-}
+//static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIndex) {
+//  int32_t len = 0;
+//  int32_t inputLen = varDataLen(buf);
+//
+//  char* t = taosMemoryCalloc(1,(inputLen + 1) * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE);
+//  /*int32_t resLen = */taosMbsToUcs4(varDataVal(buf), inputLen, (TdUcs4*) varDataVal(t), pOut->columnData->info.bytes, &len);
+//  varDataSetLen(t, len);
+//
+//  colDataAppend(pOut->columnData, rowIndex, t, false);
+//  taosMemoryFree(t);
+//}
 
 //TODO opt performance, tmp is not needed.
 int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, int32_t inType, int32_t outType) {
   int32_t bufSize = pIn->columnData->info.bytes;
   char *tmp = taosMemoryMalloc(bufSize + VARSTR_HEADER_SIZE);
 
-  bool vton = false;
+//  bool vton = false;
 
   _bufConverteFunc func = NULL;
   if (TSDB_DATA_TYPE_BOOL == outType) {
@@ -205,9 +205,9 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
     func = varToUnsigned;
   } else if (IS_FLOAT_TYPE(outType)) {
     func = varToFloat;
-  } else if (outType == TSDB_DATA_TYPE_NCHAR) {
-    func = varToNchar;
-    vton = true;
+//  } else if (outType == TSDB_DATA_TYPE_NCHAR) {   // can not be nchar or binary
+//    func = varToNchar;
+//    vton = true;
   } else {
     sclError("invalid convert outType:%d", outType);
     return TSDB_CODE_QRY_APP_ERROR;
@@ -221,9 +221,9 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
     }
 
     char* data = colDataGetData(pIn->columnData, i);
-    if (vton) {
-      memcpy(tmp, data, varDataTLen(data));
-    } else {
+//    if (vton) {
+//      memcpy(tmp, data, varDataTLen(data));
+//    } else {
       if (TSDB_DATA_TYPE_VARCHAR == inType) {
         memcpy(tmp, varDataVal(data), varDataLen(data));
         tmp[varDataLen(data)] = 0;
@@ -239,7 +239,7 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
 
         tmp[len] = 0;
       }
-    }
+//    }
     
     (*func)(tmp, pOut, i);
   }
@@ -248,6 +248,135 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
   return TSDB_CODE_SUCCESS;
 }
 
+void convertNumberToNumber(const void *inData, void *outData, int8_t inType, int8_t outType){
+  switch (outType) {
+    case TSDB_DATA_TYPE_BOOL: {
+      GET_TYPED_DATA(*((bool *)outData), bool, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_TINYINT: {
+      GET_TYPED_DATA(*((int8_t *)outData), int8_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_SMALLINT: {
+      GET_TYPED_DATA(*((int16_t *)outData), int16_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_INT: {
+      GET_TYPED_DATA(*((int32_t *)outData), int32_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_BIGINT:
+    case TSDB_DATA_TYPE_TIMESTAMP: {
+      GET_TYPED_DATA(*((int64_t *)outData), int64_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_UTINYINT: {
+      GET_TYPED_DATA(*((uint8_t *)outData), uint8_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_USMALLINT: {
+      GET_TYPED_DATA(*((uint16_t *)outData), uint16_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_UINT: {
+      GET_TYPED_DATA(*((uint32_t *)outData), uint32_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_UBIGINT: {
+      GET_TYPED_DATA(*((uint64_t *)outData), uint64_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_FLOAT: {
+      GET_TYPED_DATA(*((float *)outData), float, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_DOUBLE: {
+      GET_TYPED_DATA(*((double *)outData), double, inType, inData);
+      break;
+    }
+    default:{
+      ASSERT(0);
+    }
+  }
+}
+
+void convertStringToDouble(const void *inData, void *outData, int8_t inType, int8_t outType){
+  char *tmp = taosMemoryMalloc(varDataTLen(inData));
+  if (inType == TSDB_DATA_TYPE_NCHAR) {
+    int len = taosUcs4ToMbs((TdUcs4 *)varDataVal(inData), varDataLen(inData), tmp);
+    if (len < 0) {
+      sclError("castConvert taosUcs4ToMbs error 1");
+    }
+
+    tmp[len] = 0;
+  } else {
+    memcpy(tmp, varDataVal(inData), varDataLen(inData));
+    tmp[varDataLen(inData)] = 0;
+  }
+
+  ASSERT(outType == TSDB_DATA_TYPE_DOUBLE);
+  double value = strtod(tmp, NULL);
+
+  *((double *)outData) = value;
+  taosMemoryFreeClear(tmp);
+}
+
+double getVectorDoubleValue_JSON(void *src, int32_t index){
+  ASSERT(!colDataIsNull_s(((SColumnInfoData*)src), index));
+  char *data = colDataGetData((SColumnInfoData*)src, index);
+  double out = 0;
+  if(*data == TSDB_DATA_TYPE_NCHAR) {
+    convertStringToDouble(data+CHAR_BYTES, &out, *data, TSDB_DATA_TYPE_DOUBLE);
+  } {
+    convertNumberToNumber(data+CHAR_BYTES, &out, *data, TSDB_DATA_TYPE_DOUBLE);
+  }
+  return out;
+}
+
+void convertJsonValue(__compar_fn_t *fp, int32_t optr, int8_t typeLeft, int8_t typeRight, char **pLeftData, char **pRightData, void *pLeftOut, void *pRightOut){
+  if(optr == OP_TYPE_JSON_CONTAINS) {
+    return;
+  }
+
+  if(typeLeft != TSDB_DATA_TYPE_JSON && typeRight != TSDB_DATA_TYPE_JSON){
+    return;
+  }
+
+  if(typeLeft == TSDB_DATA_TYPE_JSON){
+    typeLeft = **pLeftData;
+    (*pLeftData) ++;
+  }
+  if(typeRight == TSDB_DATA_TYPE_JSON){
+    typeRight = **pRightData;
+    (*pRightData) ++;
+  }
+  int8_t type = vectorGetConvertType(typeLeft, typeRight);
+
+  if(type == 0) {
+    *fp = filterGetCompFunc(typeLeft, optr);
+    return;
+  }
+
+  *fp = filterGetCompFunc(type, optr);
+
+  if(typeLeft == TSDB_DATA_TYPE_NCHAR) {
+    convertStringToDouble(*pLeftData, pLeftOut, typeLeft, type);
+    *pLeftData = pLeftOut;
+  } else if(typeLeft != type) {
+    convertNumberToNumber(*pLeftData, pLeftOut, typeLeft, type);
+    *pLeftData = pLeftOut;
+  }
+
+  if(IS_VAR_DATA_TYPE(typeRight)) {
+    convertStringToDouble(*pRightData, pRightOut, typeRight, type);
+    *pRightData = pRightOut;
+  } else if(typeRight != type) {
+    convertNumberToNumber(*pRightData, pRightOut, typeRight, type);
+    *pRightData = pRightOut;
+  }
+}
+
 // TODO opt performance
 int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut) {
   SColumnInfoData* pInputCol  = pIn->columnData;
@@ -522,7 +651,7 @@ enum {
 static int32_t doConvertHelper(SScalarParam* pDest, int32_t* convert, const SScalarParam* pParam, int32_t type) {
   SColumnInfoData* pCol = pParam->columnData;
 
-  if (IS_VAR_DATA_TYPE(pCol->info.type)) {
+  if (IS_VAR_DATA_TYPE(pCol->info.type) || pCol->info.type != TSDB_DATA_TYPE_JSON) {
     pDest->numOfRows = pParam->numOfRows;
 
     SDataType t = {.type = type, .bytes = tDataTypes[type].bytes};
@@ -556,7 +685,8 @@ static void vectorMathAddHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRig
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
-      *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) + getVectorDoubleValueFnRight(pRightCol->pData, 0);
+      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
+                + getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0);
     }
     pOutputCol->hasNull = pLeftCol->hasNull;
     if (pOutputCol->hasNull) {
@@ -587,6 +717,36 @@ static void doReleaseVec(SColumnInfoData* pCol, int32_t type) {
   }
 }
 
+char *getJsonValue(char *json, char *key){    //todo
+  return NULL;
+}
+
+void vectorJsonArrow(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
+  SColumnInfoData *pOutputCol = pOut->columnData;
+
+  int32_t i = ((_ord) == TSDB_ORDER_ASC)? 0 : TMAX(pLeft->numOfRows, pRight->numOfRows) - 1;
+  int32_t step = ((_ord) == TSDB_ORDER_ASC)? 1 : -1;
+
+  pOut->numOfRows = TMAX(pLeft->numOfRows, pRight->numOfRows);
+
+  char *pRightData = colDataGetData(pRight->columnData, 0);
+  for (; i >= 0 && i < pLeft->numOfRows; i += step) {
+    if (pLeft->columnData->varmeta.offset[i] == -1) {
+      pOutputCol->varmeta.offset[i] = -1;
+      pOutputCol->hasNull = true;
+      continue;
+    }
+    char *pLeftData = colDataGetData(pLeft->columnData, i);
+    char *value = getJsonValue(pLeftData, pRightData);
+    if (!value || *value == TSDB_DATA_TYPE_NULL) {
+      pOutputCol->varmeta.offset[i] = -1;
+      pOutputCol->hasNull = true;
+      continue;
+    }
+    colDataAppend(pOutputCol, i, pLeftData, false);
+  }
+}
+
 void vectorMathAdd(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
   SColumnInfoData *pOutputCol = pOut->columnData;
 
@@ -605,7 +765,12 @@ void vectorMathAdd(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
   double *output = (double *)pOutputCol->pData;
   if (pLeft->numOfRows == pRight->numOfRows) {
     for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-      *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) + getVectorDoubleValueFnRight(pRightCol->pData, i);
+      if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
+                + getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -637,7 +802,8 @@ static void vectorMathSubHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRig
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
-      *output = (getVectorDoubleValueFnLeft(pLeftCol->pData, i) - getVectorDoubleValueFnRight(pRightCol->pData, 0)) * factor;
+      *output = (getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
+                 - getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0)) * factor;
     }
     pOutputCol->hasNull = pLeftCol->hasNull;
     if (pOutputCol->hasNull) {
@@ -664,7 +830,12 @@ void vectorMathSub(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
   double *output = (double *)pOutputCol->pData;
   if (pLeft->numOfRows == pRight->numOfRows) {
     for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-      *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) - getVectorDoubleValueFnRight(pRightCol->pData, i);
+      if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
+                - getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -696,7 +867,8 @@ static void vectorMathMultiplyHelper(SColumnInfoData* pLeftCol, SColumnInfoData*
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
-      *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) * getVectorDoubleValueFnRight(pRightCol->pData, 0);
+      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
+                * getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0);
     }
     pOutputCol->hasNull = pLeftCol->hasNull;
     if (pOutputCol->hasNull) {
@@ -722,7 +894,12 @@ void vectorMathMultiply(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
   double *output = (double *)pOutputCol->pData;
   if (pLeft->numOfRows == pRight->numOfRows) {
     for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-      *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) * getVectorDoubleValueFnRight(pRightCol->pData, i);
+      if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
+                * getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -760,7 +937,12 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
   double *output = (double *)pOutputCol->pData;
   if (pLeft->numOfRows == pRight->numOfRows) {  // check for the 0 value
     for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-      *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) / getVectorDoubleValueFnRight(pRightCol->pData, i);
+      if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
+                 /getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -776,7 +958,8 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
       colDataAppendNNULL(pOutputCol, 0, pRight->numOfRows);
     } else {
       for (; i >= 0 && i < pRight->numOfRows; i += step, output += 1) {
-        *output = getVectorDoubleValueFnLeft(pLeftCol->pData, 0) / getVectorDoubleValueFnRight(pRightCol->pData, i);
+        *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), 0)
+                  / getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
       }
       pOutputCol->hasNull = pRightCol->hasNull;
       if (pOutputCol->hasNull) {
@@ -788,7 +971,8 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
       colDataAppendNNULL(pOutputCol, 0, pLeft->numOfRows);
     } else {
       for (; i >= 0 && i < pLeft->numOfRows; i += step, output += 1) {
-        *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) / getVectorDoubleValueFnRight(pRightCol->pData, 0);
+        *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
+                  / getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0);
       }
       pOutputCol->hasNull = pLeftCol->hasNull;
       if (pOutputCol->hasNull) {
@@ -825,8 +1009,8 @@ void vectorMathRemainder(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
         continue;
       }
 
-      double lx = getVectorDoubleValueFnLeft(pLeftCol->pData, i);
-      double rx = getVectorDoubleValueFnRight(pRightCol->pData, i);
+      double lx = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i);
+      double rx = getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
       if (isnan(lx) || isinf(lx) || isnan(rx) || isinf(rx)) {
         colDataAppendNULL(pOutputCol, i);
         continue;
@@ -835,7 +1019,7 @@ void vectorMathRemainder(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
       *output = lx - ((int64_t)(lx / rx)) * rx;
     }
   } else if (pLeft->numOfRows == 1) {
-    double lx = getVectorDoubleValueFnLeft(pLeftCol->pData, 0);
+    double lx = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), 0);
     if (colDataIsNull_f(pLeftCol->nullbitmap, 0) || isnan(lx) || isinf(lx)) {  // Set pLeft->numOfRows NULL value
       colDataAppendNNULL(pOutputCol, 0, pRight->numOfRows);
     } else {
@@ -845,7 +1029,7 @@ void vectorMathRemainder(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
           continue;
         }
 
-        double rx = getVectorDoubleValueFnRight(pRightCol->pData, i);
+        double rx = getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
         if (isnan(rx) || isinf(rx) || FLT_EQUAL(rx, 0)) {
           colDataAppendNULL(pOutputCol, i);
           continue;
@@ -855,7 +1039,7 @@ void vectorMathRemainder(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
       }
     }
   } else if (pRight->numOfRows == 1) {
-    double rx = getVectorDoubleValueFnRight(pRightCol->pData, 0);
+    double rx = getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0);
     if (colDataIsNull_f(pRightCol->nullbitmap, 0) || FLT_EQUAL(rx, 0)) {  // Set pLeft->numOfRows NULL value
       colDataAppendNNULL(pOutputCol, 0, pLeft->numOfRows);
     } else {
@@ -865,7 +1049,7 @@ void vectorMathRemainder(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
           continue;
         }
 
-        double lx = getVectorDoubleValueFnLeft(pLeftCol->pData, i);
+        double lx = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i);
         if (isnan(lx) || isinf(lx)) {
           colDataAppendNULL(pOutputCol, i);
           continue;
@@ -895,7 +1079,11 @@ void vectorMathMinus(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pO
 
   double *output = (double *)pOutputCol->pData;
   for (; i < pLeft->numOfRows && i >= 0; i += step, output += 1) {
-    *output = - getVectorDoubleValueFnLeft(pLeftCol->pData, i);
+    if (colDataIsNull_s(pLeft->columnData, i)) {
+      colDataAppendNULL(pOutputCol, i);
+      continue;  // TODO set null or ignore
+    }
+    *output = - getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i);
   }
 
   pOutputCol->hasNull = pLeftCol->hasNull;
@@ -1099,30 +1287,39 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
 
       char *pLeftData = colDataGetData(pLeft->columnData, i);
       char *pRightData = colDataGetData(pRight->columnData, i);
+
+      int64_t leftOut = 0;
+      int64_t rightOut = 0;
+      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut);
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
       colDataAppendInt8(pOut->columnData, i, (int8_t*)&res);
     }
   } else if (pRight->numOfRows == 1) {
-    char *pRightData = colDataGetData(pRight->columnData, 0);
     ASSERT(pLeft->pHashFilter == NULL);
-
     for (; i >= 0 && i < pLeft->numOfRows; i += step) {
       if (colDataIsNull_s(pLeft->columnData, i)) {
         continue;
       }
 
       char *pLeftData = colDataGetData(pLeft->columnData, i);
+      char *pRightData = colDataGetData(pRight->columnData, 0);
+      int64_t leftOut = 0;
+      int64_t rightOut = 0;
+      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut);
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
       colDataAppendInt8(pOut->columnData, i, (int8_t*)&res);
     }
   } else if (pLeft->numOfRows == 1) {
-    char *pLeftData = colDataGetData(pLeft->columnData, 0);
     for (; i >= 0 && i < pRight->numOfRows; i += step) {
       if (colDataIsNull_s(pRight->columnData, i)) {
         continue;
       }
 
+      char *pLeftData = colDataGetData(pLeft->columnData, 0);
       char *pRightData = colDataGetData(pLeft->columnData, i);
+      int64_t leftOut = 0;
+      int64_t rightOut = 0;
+      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut);
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
       colDataAppendInt8(pOut->columnData, i, (int8_t*)&res);
     }
@@ -1203,6 +1400,10 @@ void vectorNotMatch(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOu
   vectorCompare(pLeft, pRight, pOut, _ord, OP_TYPE_NMATCH);
 }
 
+void vectorJsonContains(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
+  vectorCompare(pLeft, pRight, pOut, _ord, OP_TYPE_JSON_CONTAINS);
+}
+
 void vectorIsNull(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
   for(int32_t i = 0; i < pLeft->numOfRows; ++i) {
     int8_t v = colDataIsNull_s(pLeft->columnData, i)? 1:0;
@@ -1271,6 +1472,10 @@ _bin_scalar_fn_t getBinScalarOperatorFn(int32_t binFunctionId) {
       return vectorBitOr;
     case OP_TYPE_IS_TRUE:
       return vectorIsTrue;
+    case OP_TYPE_JSON_GET_VALUE:
+      return vectorJsonArrow;
+    case OP_TYPE_JSON_CONTAINS:
+      return vectorJsonContains;
     default:
       assert(0);
       return NULL;
diff --git a/source/util/src/tcompare.c b/source/util/src/tcompare.c
index 93022de021..0ddc76e415 100644
--- a/source/util/src/tcompare.c
+++ b/source/util/src/tcompare.c
@@ -222,6 +222,11 @@ int32_t compareLenPrefixedWStrDesc(const void *pLeft, const void *pRight) {
   return compareLenPrefixedWStr(pRight, pLeft);
 }
 
+int32_t compareJsonContainsKey(const void* pLeft, const void* pRight) {
+  if(pLeft) return 0;
+  return 1;
+}
+
 /*
  * Compare two strings
  *    TSDB_MATCH:            Match

From e2c6bf41df1d33ba3698fe0aef0f32b3e52bec0a Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Mon, 18 Apr 2022 13:36:01 +0800
Subject: [PATCH 04/15] refactor: add compare logic for json value

---
 include/common/tdatablock.h        |  20 +-
 source/libs/scalar/src/sclvector.c | 289 +++++++++++++++++------------
 2 files changed, 186 insertions(+), 123 deletions(-)

diff --git a/include/common/tdatablock.h b/include/common/tdatablock.h
index 15f3246013..8b9cf74d7a 100644
--- a/include/common/tdatablock.h
+++ b/include/common/tdatablock.h
@@ -54,12 +54,15 @@ SEpSet getEpSet_s(SCorEpSet* pEpSet);
     BMCharPos(bm_, r_) |= (1u << (7u - BitPos(r_))); \
   } while (0)
 
+#define colDataIsNull_var(pColumnInfoData, row) (pColumnInfoData->varmeta.offset[row] == -1)
+#define colDataSetNull_var(pColumnInfoData, row)  (pColumnInfoData->varmeta.offset[row] = -1)
+
 static FORCE_INLINE bool colDataIsNull_s(const SColumnInfoData* pColumnInfoData, uint32_t row) {
   if (!pColumnInfoData->hasNull) {
     return false;
   }
   if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
-    return pColumnInfoData->varmeta.offset[row] == -1;
+    return colDataIsNull_var(pColumnInfoData, row);
   } else {
     if (pColumnInfoData->nullbitmap == NULL) {
       return false;
@@ -86,7 +89,7 @@ static FORCE_INLINE bool colDataIsNull(const SColumnInfoData* pColumnInfoData, u
   }
 
   if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
-    return pColumnInfoData->varmeta.offset[row] == -1;
+    return colDataIsNull_var(pColumnInfoData, row);
   } else {
     if (pColumnInfoData->nullbitmap == NULL) {
       return false;
@@ -98,15 +101,18 @@ static FORCE_INLINE bool colDataIsNull(const SColumnInfoData* pColumnInfoData, u
 
 #define BitmapLen(_n) (((_n) + ((1 << NBIT) - 1)) >> NBIT)
 
+#define colDataGetVarData(p1_, r_) ((p1_)->pData + (p1_)->varmeta.offset[(r_)])
+
+#define colDataGetNumData(p1_, r_) ((p1_)->pData + ((r_) * (p1_)->info.bytes))
 // SColumnInfoData, rowNumber
-#define colDataGetData(p1_, r_)                                                        \
-  ((IS_VAR_DATA_TYPE((p1_)->info.type)) ? ((p1_)->pData + (p1_)->varmeta.offset[(r_)]) \
-                                        : ((p1_)->pData + ((r_) * (p1_)->info.bytes)))
+#define colDataGetData(p1_, r_)                                        \
+  ((IS_VAR_DATA_TYPE((p1_)->info.type)) ?  colDataGetVarData(p1_, r_)  \
+                                        :  colDataGetNumData(p1_, r_)
 
 static FORCE_INLINE void colDataAppendNULL(SColumnInfoData* pColumnInfoData, uint32_t currentRow) {
   // There is a placehold for each NULL value of binary or nchar type.
   if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
-    pColumnInfoData->varmeta.offset[currentRow] = -1;  // it is a null value of VAR type.
+    colDataSetNull_var(pColumnInfoData, currentRow);  // it is a null value of VAR type.
   } else {
     colDataSetNull_f(pColumnInfoData->nullbitmap, currentRow);
   }
@@ -117,7 +123,7 @@ static FORCE_INLINE void colDataAppendNULL(SColumnInfoData* pColumnInfoData, uin
 static FORCE_INLINE void colDataAppendNNULL(SColumnInfoData* pColumnInfoData, uint32_t start, size_t nRows) {
   if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
     for (int32_t i = start; i < start + nRows; ++i) {
-      pColumnInfoData->varmeta.offset[i] = -1;  // it is a null value of VAR type.
+      colDataSetNull_var(pColumnInfoData,i);  // it is a null value of VAR type.
     }
   } else {
     for (int32_t i = start; i < start + nRows; ++i) {
diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c
index 11dc60dab3..aa51c9b631 100644
--- a/source/libs/scalar/src/sclvector.c
+++ b/source/libs/scalar/src/sclvector.c
@@ -25,6 +25,9 @@
 #include "tdatablock.h"
 #include "ttypes.h"
 
+#define LEFT_COL ((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData))
+#define RIGHT_COL ((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData))
+
 typedef int64_t (*_getBigintValue_fn_t)(void *src, int32_t index);
 
 int64_t getVectorBigintValue_TINYINT(void *src, int32_t index) {
@@ -189,65 +192,6 @@ static FORCE_INLINE void varToBool(char *buf, SScalarParam* pOut, int32_t rowInd
 //  taosMemoryFree(t);
 //}
 
-//TODO opt performance, tmp is not needed.
-int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, int32_t inType, int32_t outType) {
-  int32_t bufSize = pIn->columnData->info.bytes;
-  char *tmp = taosMemoryMalloc(bufSize + VARSTR_HEADER_SIZE);
-
-//  bool vton = false;
-
-  _bufConverteFunc func = NULL;
-  if (TSDB_DATA_TYPE_BOOL == outType) {
-    func = varToBool;
-  } else if (IS_SIGNED_NUMERIC_TYPE(outType) || TSDB_DATA_TYPE_TIMESTAMP == outType) {
-    func = varToSigned;
-  } else if (IS_UNSIGNED_NUMERIC_TYPE(outType)) {
-    func = varToUnsigned;
-  } else if (IS_FLOAT_TYPE(outType)) {
-    func = varToFloat;
-//  } else if (outType == TSDB_DATA_TYPE_NCHAR) {   // can not be nchar or binary
-//    func = varToNchar;
-//    vton = true;
-  } else {
-    sclError("invalid convert outType:%d", outType);
-    return TSDB_CODE_QRY_APP_ERROR;
-  }
-
-  pOut->numOfRows = pIn->numOfRows;
-  for (int32_t i = 0; i < pIn->numOfRows; ++i) {
-    if (colDataIsNull_s(pIn->columnData, i)) {
-      colDataAppendNULL(pOut->columnData, i);
-      continue;
-    }
-
-    char* data = colDataGetData(pIn->columnData, i);
-//    if (vton) {
-//      memcpy(tmp, data, varDataTLen(data));
-//    } else {
-      if (TSDB_DATA_TYPE_VARCHAR == inType) {
-        memcpy(tmp, varDataVal(data), varDataLen(data));
-        tmp[varDataLen(data)] = 0;
-      } else {
-        ASSERT(varDataLen(data) <= bufSize);
-
-        int len = taosUcs4ToMbs((TdUcs4 *)varDataVal(data), varDataLen(data), tmp);
-        if (len < 0) {
-          sclError("castConvert taosUcs4ToMbs error 1");
-          taosMemoryFreeClear(tmp);
-          return TSDB_CODE_QRY_APP_ERROR;
-        }
-
-        tmp[len] = 0;
-      }
-//    }
-    
-    (*func)(tmp, pOut, i);
-  }
-  
-  taosMemoryFreeClear(tmp);
-  return TSDB_CODE_SUCCESS;
-}
-
 void convertNumberToNumber(const void *inData, void *outData, int8_t inType, int8_t outType){
   switch (outType) {
     case TSDB_DATA_TYPE_BOOL: {
@@ -303,18 +247,13 @@ void convertNumberToNumber(const void *inData, void *outData, int8_t inType, int
 
 void convertStringToDouble(const void *inData, void *outData, int8_t inType, int8_t outType){
   char *tmp = taosMemoryMalloc(varDataTLen(inData));
-  if (inType == TSDB_DATA_TYPE_NCHAR) {
-    int len = taosUcs4ToMbs((TdUcs4 *)varDataVal(inData), varDataLen(inData), tmp);
-    if (len < 0) {
-      sclError("castConvert taosUcs4ToMbs error 1");
-    }
-
-    tmp[len] = 0;
-  } else {
-    memcpy(tmp, varDataVal(inData), varDataLen(inData));
-    tmp[varDataLen(inData)] = 0;
+  int len = taosUcs4ToMbs((TdUcs4 *)varDataVal(inData), varDataLen(inData), tmp);
+  if (len < 0) {
+    sclError("castConvert taosUcs4ToMbs error 1");
   }
 
+  tmp[len] = 0;
+
   ASSERT(outType == TSDB_DATA_TYPE_DOUBLE);
   double value = strtod(tmp, NULL);
 
@@ -322,11 +261,85 @@ void convertStringToDouble(const void *inData, void *outData, int8_t inType, int
   taosMemoryFreeClear(tmp);
 }
 
+//TODO opt performance, tmp is not needed.
+int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, int32_t inType, int32_t outType) {
+  ASSERT(!IS_VAR_DATA_TYPE(outType));
+  int32_t bufSize = pIn->columnData->info.bytes;
+  char *tmp = taosMemoryMalloc(bufSize + VARSTR_HEADER_SIZE);
+
+//  bool vton = false;
+
+  _bufConverteFunc func = NULL;
+  if (TSDB_DATA_TYPE_BOOL == outType) {
+    func = varToBool;
+  } else if (IS_SIGNED_NUMERIC_TYPE(outType) || TSDB_DATA_TYPE_TIMESTAMP == outType) {
+    func = varToSigned;
+  } else if (IS_UNSIGNED_NUMERIC_TYPE(outType)) {
+    func = varToUnsigned;
+  } else if (IS_FLOAT_TYPE(outType)) {
+    func = varToFloat;
+//  } else if (outType == TSDB_DATA_TYPE_NCHAR) {   // can not be nchar or binary
+//    func = varToNchar;
+//    vton = true;
+  } else {
+    sclError("invalid convert outType:%d", outType);
+    return TSDB_CODE_QRY_APP_ERROR;
+  }
+
+  pOut->numOfRows = pIn->numOfRows;
+  for (int32_t i = 0; i < pIn->numOfRows; ++i) {
+    if (colDataIsNull_s(pIn->columnData, i)) {
+      colDataAppendNULL(pOut->columnData, i);
+      continue;
+    }
+
+    char* data = colDataGetVarData(pIn->columnData, i);
+    int32_t convertType = inType;
+    if(inType == TSDB_DATA_TYPE_JSON){
+      if(*data == TSDB_DATA_TYPE_NULL) {
+        colDataAppendNULL(pOut->columnData, i);
+        continue;
+      }
+      else if(*data == TSDB_DATA_TYPE_NCHAR) {
+        data += CHAR_BYTES;
+        convertType = TSDB_DATA_TYPE_NCHAR;
+      } else {
+        convertNumberToNumber(data+CHAR_BYTES, colDataGetNumData(pOut->columnData, i), *data, outType);
+        continue;
+      }
+    }
+//    if (vton) {
+//      memcpy(tmp, data, varDataTLen(data));
+//    } else {
+      if (TSDB_DATA_TYPE_VARCHAR == convertType) {
+        memcpy(tmp, varDataVal(data), varDataLen(data));
+        tmp[varDataLen(data)] = 0;
+      } else if (TSDB_DATA_TYPE_NCHAR == convertType){
+        ASSERT(varDataLen(data) <= bufSize);
+
+        int len = taosUcs4ToMbs((TdUcs4 *)varDataVal(data), varDataLen(data), tmp);
+        if (len < 0) {
+          sclError("castConvert taosUcs4ToMbs error 1");
+          taosMemoryFreeClear(tmp);
+          return TSDB_CODE_QRY_APP_ERROR;
+        }
+
+        tmp[len] = 0;
+      }
+//    }
+    
+    (*func)(tmp, pOut, i);
+  }
+  
+  taosMemoryFreeClear(tmp);
+  return TSDB_CODE_SUCCESS;
+}
+
 double getVectorDoubleValue_JSON(void *src, int32_t index){
-  ASSERT(!colDataIsNull_s(((SColumnInfoData*)src), index));
-  char *data = colDataGetData((SColumnInfoData*)src, index);
+  ASSERT(!colDataIsNull_var(((SColumnInfoData*)src), index));
+  char *data = colDataGetVarData((SColumnInfoData*)src, index);
   double out = 0;
-  if(*data == TSDB_DATA_TYPE_NCHAR) {
+  if(*data == TSDB_DATA_TYPE_NCHAR) {   // json inner type can not be BINARY
     convertStringToDouble(data+CHAR_BYTES, &out, *data, TSDB_DATA_TYPE_DOUBLE);
   } {
     convertNumberToNumber(data+CHAR_BYTES, &out, *data, TSDB_DATA_TYPE_DOUBLE);
@@ -685,8 +698,12 @@ static void vectorMathAddHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRig
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
-      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
-                + getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0);
+      if (colDataIsNull_s(pLeftCol, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
+                + getVectorDoubleValueFnRight(RIGHT_COL, 0);
     }
     pOutputCol->hasNull = pLeftCol->hasNull;
     if (pOutputCol->hasNull) {
@@ -729,17 +746,17 @@ void vectorJsonArrow(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pO
 
   pOut->numOfRows = TMAX(pLeft->numOfRows, pRight->numOfRows);
 
-  char *pRightData = colDataGetData(pRight->columnData, 0);
+  char *pRightData = colDataGetVarData(pRight->columnData, 0);
   for (; i >= 0 && i < pLeft->numOfRows; i += step) {
-    if (pLeft->columnData->varmeta.offset[i] == -1) {
-      pOutputCol->varmeta.offset[i] = -1;
+    if (colDataIsNull_var(pLeft->columnData, i)) {
+      colDataSetNull_var(pOutputCol, i);
       pOutputCol->hasNull = true;
       continue;
     }
-    char *pLeftData = colDataGetData(pLeft->columnData, i);
+    char *pLeftData = colDataGetVarData(pLeft->columnData, i);
     char *value = getJsonValue(pLeftData, pRightData);
     if (!value || *value == TSDB_DATA_TYPE_NULL) {
-      pOutputCol->varmeta.offset[i] = -1;
+      colDataSetNull_var(pOutputCol, i);
       pOutputCol->hasNull = true;
       continue;
     }
@@ -769,8 +786,8 @@ void vectorMathAdd(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
         colDataAppendNULL(pOutputCol, i);
         continue;  // TODO set null or ignore
       }
-      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
-                + getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
+      *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
+                + getVectorDoubleValueFnRight(RIGHT_COL, i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -802,8 +819,12 @@ static void vectorMathSubHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRig
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
-      *output = (getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
-                 - getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0)) * factor;
+      if (colDataIsNull_s(pLeftCol, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = (getVectorDoubleValueFnLeft(LEFT_COL, i)
+                 - getVectorDoubleValueFnRight(RIGHT_COL, 0)) * factor;
     }
     pOutputCol->hasNull = pLeftCol->hasNull;
     if (pOutputCol->hasNull) {
@@ -834,8 +855,8 @@ void vectorMathSub(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
         colDataAppendNULL(pOutputCol, i);
         continue;  // TODO set null or ignore
       }
-      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
-                - getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
+      *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
+                - getVectorDoubleValueFnRight(RIGHT_COL, i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -867,8 +888,12 @@ static void vectorMathMultiplyHelper(SColumnInfoData* pLeftCol, SColumnInfoData*
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
-      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
-                * getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0);
+      if (colDataIsNull_s(pLeftCol, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
+                * getVectorDoubleValueFnRight(RIGHT_COL, 0);
     }
     pOutputCol->hasNull = pLeftCol->hasNull;
     if (pOutputCol->hasNull) {
@@ -898,8 +923,8 @@ void vectorMathMultiply(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
         colDataAppendNULL(pOutputCol, i);
         continue;  // TODO set null or ignore
       }
-      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
-                * getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
+      *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
+                * getVectorDoubleValueFnRight(RIGHT_COL, i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -941,8 +966,8 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
         colDataAppendNULL(pOutputCol, i);
         continue;  // TODO set null or ignore
       }
-      *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
-                 /getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
+      *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
+                 /getVectorDoubleValueFnRight(RIGHT_COL, i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -954,12 +979,16 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
     }
 
   } else if (pLeft->numOfRows == 1) {
-    if (colDataIsNull_f(pLeftCol->nullbitmap, 0)) {  // Set pLeft->numOfRows NULL value
+    if (colDataIsNull_s(pLeftCol, 0)) {  // Set pLeft->numOfRows NULL value
       colDataAppendNNULL(pOutputCol, 0, pRight->numOfRows);
     } else {
       for (; i >= 0 && i < pRight->numOfRows; i += step, output += 1) {
-        *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), 0)
-                  / getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
+        if (colDataIsNull_s(pRightCol, i)) {
+          colDataAppendNULL(pOutputCol, i);
+          continue;  // TODO set null or ignore
+        }
+        *output = getVectorDoubleValueFnLeft(LEFT_COL, 0)
+                  / getVectorDoubleValueFnRight(RIGHT_COL, i);
       }
       pOutputCol->hasNull = pRightCol->hasNull;
       if (pOutputCol->hasNull) {
@@ -967,12 +996,16 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
       }
     }
   } else if (pRight->numOfRows == 1) {
-    if (colDataIsNull_f(pRightCol->nullbitmap, 0)) {  // Set pLeft->numOfRows NULL value
+    if (colDataIsNull_s(pRightCol, 0)) {  // Set pLeft->numOfRows NULL value
       colDataAppendNNULL(pOutputCol, 0, pLeft->numOfRows);
     } else {
       for (; i >= 0 && i < pLeft->numOfRows; i += step, output += 1) {
-        *output = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i)
-                  / getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0);
+        if (colDataIsNull_s(pLeftCol, i)) {
+          colDataAppendNULL(pOutputCol, i);
+          continue;  // TODO set null or ignore
+        }
+        *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
+                  / getVectorDoubleValueFnRight(RIGHT_COL, 0);
       }
       pOutputCol->hasNull = pLeftCol->hasNull;
       if (pOutputCol->hasNull) {
@@ -1004,13 +1037,13 @@ void vectorMathRemainder(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
 
   if (pLeft->numOfRows == pRight->numOfRows) {
     for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-      if (colDataIsNull_f(pLeftCol->nullbitmap, i) || colDataIsNull_f(pRightCol->nullbitmap, i)) {
+      if (colDataIsNull_s(pLeftCol, i) || colDataIsNull_s(pRightCol, i)) {
         colDataAppendNULL(pOutputCol, i);
         continue;
       }
 
-      double lx = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i);
-      double rx = getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
+      double lx = getVectorDoubleValueFnLeft(LEFT_COL, i);
+      double rx = getVectorDoubleValueFnRight(RIGHT_COL, i);
       if (isnan(lx) || isinf(lx) || isnan(rx) || isinf(rx)) {
         colDataAppendNULL(pOutputCol, i);
         continue;
@@ -1019,17 +1052,17 @@ void vectorMathRemainder(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
       *output = lx - ((int64_t)(lx / rx)) * rx;
     }
   } else if (pLeft->numOfRows == 1) {
-    double lx = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), 0);
-    if (colDataIsNull_f(pLeftCol->nullbitmap, 0) || isnan(lx) || isinf(lx)) {  // Set pLeft->numOfRows NULL value
+    double lx = getVectorDoubleValueFnLeft(LEFT_COL, 0);
+    if (colDataIsNull_s(pLeftCol, 0) || isnan(lx) || isinf(lx)) {  // Set pLeft->numOfRows NULL value
       colDataAppendNNULL(pOutputCol, 0, pRight->numOfRows);
     } else {
       for (; i >= 0 && i < pRight->numOfRows; i += step, output += 1) {
-        if (colDataIsNull_f(pRightCol->nullbitmap, i)) {
+        if (colDataIsNull_s(pRightCol, i)) {
           colDataAppendNULL(pOutputCol, i);
           continue;
         }
 
-        double rx = getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), i);
+        double rx = getVectorDoubleValueFnRight(RIGHT_COL, i);
         if (isnan(rx) || isinf(rx) || FLT_EQUAL(rx, 0)) {
           colDataAppendNULL(pOutputCol, i);
           continue;
@@ -1039,17 +1072,17 @@ void vectorMathRemainder(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
       }
     }
   } else if (pRight->numOfRows == 1) {
-    double rx = getVectorDoubleValueFnRight((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData), 0);
-    if (colDataIsNull_f(pRightCol->nullbitmap, 0) || FLT_EQUAL(rx, 0)) {  // Set pLeft->numOfRows NULL value
+    double rx = getVectorDoubleValueFnRight(RIGHT_COL, 0);
+    if (colDataIsNull_s(pRightCol, 0) || FLT_EQUAL(rx, 0)) {  // Set pLeft->numOfRows NULL value
       colDataAppendNNULL(pOutputCol, 0, pLeft->numOfRows);
     } else {
       for (; i >= 0 && i < pLeft->numOfRows; i += step, output += 1) {
-        if (colDataIsNull_f(pLeftCol->nullbitmap, i)) {
+        if (colDataIsNull_s(pLeftCol, i)) {
           colDataAppendNULL(pOutputCol, i);
           continue;
         }
 
-        double lx = getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i);
+        double lx = getVectorDoubleValueFnLeft(LEFT_COL, i);
         if (isnan(lx) || isinf(lx)) {
           colDataAppendNULL(pOutputCol, i);
           continue;
@@ -1083,7 +1116,7 @@ void vectorMathMinus(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pO
       colDataAppendNULL(pOutputCol, i);
       continue;  // TODO set null or ignore
     }
-    *output = - getVectorDoubleValueFnLeft((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData), i);
+    *output = - getVectorDoubleValueFnLeft(LEFT_COL, i);
   }
 
   pOutputCol->hasNull = pLeftCol->hasNull;
@@ -1151,11 +1184,15 @@ static void vectorBitAndHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRigh
 
   double *output = (double *)pOutputCol->pData;
 
-  if (colDataIsNull_f(pRightCol->nullbitmap, 0)) {  // Set pLeft->numOfRows NULL value
+  if (colDataIsNull_s(pRightCol, 0)) {  // Set pLeft->numOfRows NULL value
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
-      *output = getVectorBigintValueFnLeft(pLeftCol->pData, i) & getVectorBigintValueFnRight(pRightCol->pData, 0);
+      if (colDataIsNull_s(pLeftCol, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorBigintValueFnLeft(LEFT_COL, i) & getVectorBigintValueFnRight(RIGHT_COL, 0);
     }
     pOutputCol->hasNull = pLeftCol->hasNull;
     if (pOutputCol->hasNull) {
@@ -1181,7 +1218,11 @@ void vectorBitAnd(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut,
   int64_t *output = (int64_t *)pOutputCol->pData;
   if (pLeft->numOfRows == pRight->numOfRows) {
     for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-      *output = getVectorBigintValueFnLeft(pLeftCol->pData, i) & getVectorBigintValueFnRight(pRightCol->pData, i);
+      if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorBigintValueFnLeft(LEFT_COL, i) & getVectorBigintValueFnRight(RIGHT_COL, i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -1208,12 +1249,16 @@ static void vectorBitOrHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRight
 
   int64_t *output = (int64_t *)pOutputCol->pData;
 
-  if (colDataIsNull_f(pRightCol->nullbitmap, 0)) {  // Set pLeft->numOfRows NULL value
+  if (colDataIsNull_s(pRightCol, 0)) {  // Set pLeft->numOfRows NULL value
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
-    int64_t rx = getVectorBigintValueFnRight(pRightCol->pData, 0);
+    int64_t rx = getVectorBigintValueFnRight(RIGHT_COL, 0);
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
-      *output = getVectorBigintValueFnLeft(pLeftCol->pData, i) | rx;
+      if (colDataIsNull_s(pLeftCol, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorBigintValueFnLeft(LEFT_COL, i) | rx;
     }
     pOutputCol->hasNull = pLeftCol->hasNull;
     if (pOutputCol->hasNull) {
@@ -1239,7 +1284,11 @@ void vectorBitOr(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut,
   int64_t *output = (int64_t *)pOutputCol->pData;
   if (pLeft->numOfRows == pRight->numOfRows) {
     for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-      *output = getVectorBigintValueFnLeft(pLeftCol->pData, i) | getVectorBigintValueFnRight(pRightCol->pData, i);
+      if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
+      *output = getVectorBigintValueFnLeft(LEFT_COL, i) | getVectorBigintValueFnRight(RIGHT_COL, i);
     }
 
     pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -1407,6 +1456,10 @@ void vectorJsonContains(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
 void vectorIsNull(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
   for(int32_t i = 0; i < pLeft->numOfRows; ++i) {
     int8_t v = colDataIsNull_s(pLeft->columnData, i)? 1:0;
+    if (v && pLeft->columnData->info.type == TSDB_DATA_TYPE_JSON){
+      char *data = colDataGetVarData(pLeft->columnData, i);
+      v = (*data == TSDB_DATA_TYPE_NULL)? 1:0;
+    }
     colDataAppendInt8(pOut->columnData, i, &v);
   }
   pOut->numOfRows = pLeft->numOfRows;
@@ -1415,6 +1468,10 @@ void vectorIsNull(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut,
 void vectorNotNull(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
   for(int32_t i = 0; i < pLeft->numOfRows; ++i) {
     int8_t v = colDataIsNull_s(pLeft->columnData, i)? 0:1;
+    if (v && pLeft->columnData->info.type == TSDB_DATA_TYPE_JSON){
+      char *data = colDataGetVarData(pLeft->columnData, i);
+      v = (*data == TSDB_DATA_TYPE_NULL)? 0:1;
+    }
     colDataAppendInt8(pOut->columnData, i, &v);
   }
   pOut->numOfRows = pLeft->numOfRows;

From d2d29d992ff8c155e09e389191c51ba2713a9209 Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Mon, 18 Apr 2022 14:25:19 +0800
Subject: [PATCH 05/15] refactor: add compare logic for json value

---
 include/common/tdatablock.h        |  2 +-
 source/libs/scalar/src/sclvector.c | 26 +++++++++++++++++++++-----
 2 files changed, 22 insertions(+), 6 deletions(-)

diff --git a/include/common/tdatablock.h b/include/common/tdatablock.h
index 8b9cf74d7a..e40efd815d 100644
--- a/include/common/tdatablock.h
+++ b/include/common/tdatablock.h
@@ -107,7 +107,7 @@ static FORCE_INLINE bool colDataIsNull(const SColumnInfoData* pColumnInfoData, u
 // SColumnInfoData, rowNumber
 #define colDataGetData(p1_, r_)                                        \
   ((IS_VAR_DATA_TYPE((p1_)->info.type)) ?  colDataGetVarData(p1_, r_)  \
-                                        :  colDataGetNumData(p1_, r_)
+                                        :  colDataGetNumData(p1_, r_))
 
 static FORCE_INLINE void colDataAppendNULL(SColumnInfoData* pColumnInfoData, uint32_t currentRow) {
   // There is a placehold for each NULL value of binary or nchar type.
diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c
index aa51c9b631..56787b229b 100644
--- a/source/libs/scalar/src/sclvector.c
+++ b/source/libs/scalar/src/sclvector.c
@@ -347,7 +347,7 @@ double getVectorDoubleValue_JSON(void *src, int32_t index){
   return out;
 }
 
-void convertJsonValue(__compar_fn_t *fp, int32_t optr, int8_t typeLeft, int8_t typeRight, char **pLeftData, char **pRightData, void *pLeftOut, void *pRightOut){
+void convertJsonValue(__compar_fn_t *fp, int32_t optr, int8_t typeLeft, int8_t typeRight, char **pLeftData, char **pRightData, void *pLeftOut, void *pRightOut, bool *isNull){
   if(optr == OP_TYPE_JSON_CONTAINS) {
     return;
   }
@@ -364,6 +364,10 @@ void convertJsonValue(__compar_fn_t *fp, int32_t optr, int8_t typeLeft, int8_t t
     typeRight = **pRightData;
     (*pRightData) ++;
   }
+  if(typeLeft == TSDB_DATA_TYPE_NULL || typeRight == TSDB_DATA_TYPE_NULL){
+    *isNull = true;
+    return;
+  }
   int8_t type = vectorGetConvertType(typeLeft, typeRight);
 
   if(type == 0) {
@@ -381,7 +385,7 @@ void convertJsonValue(__compar_fn_t *fp, int32_t optr, int8_t typeLeft, int8_t t
     *pLeftData = pLeftOut;
   }
 
-  if(IS_VAR_DATA_TYPE(typeRight)) {
+  if(typeRight == TSDB_DATA_TYPE_NCHAR) {
     convertStringToDouble(*pRightData, pRightOut, typeRight, type);
     *pRightData = pRightOut;
   } else if(typeRight != type) {
@@ -1339,7 +1343,11 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
 
       int64_t leftOut = 0;
       int64_t rightOut = 0;
-      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut);
+      bool isJsonnull = false;
+      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut, &isJsonnull);
+      if(isJsonnull){
+        continue;  // TODO set null or ignore
+      }
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
       colDataAppendInt8(pOut->columnData, i, (int8_t*)&res);
     }
@@ -1354,7 +1362,11 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
       char *pRightData = colDataGetData(pRight->columnData, 0);
       int64_t leftOut = 0;
       int64_t rightOut = 0;
-      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut);
+      bool isJsonnull = false;
+      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut, &isJsonnull);
+      if(isJsonnull){
+        continue;  // TODO set null or ignore
+      }
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
       colDataAppendInt8(pOut->columnData, i, (int8_t*)&res);
     }
@@ -1368,7 +1380,11 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
       char *pRightData = colDataGetData(pLeft->columnData, i);
       int64_t leftOut = 0;
       int64_t rightOut = 0;
-      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut);
+      bool isJsonnull = false;
+      convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut, &isJsonnull);
+      if(isJsonnull){
+        continue;  // TODO set null or ignore
+      }
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
       colDataAppendInt8(pOut->columnData, i, (int8_t*)&res);
     }

From 24f805c5c2ebfbb1ed8fe5dc8c967b210eca84d2 Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Mon, 18 Apr 2022 15:54:40 +0800
Subject: [PATCH 06/15] refactor: add to_json function

---
 include/libs/scalar/scalar.h        |  1 +
 include/util/tjson.h                |  2 +-
 source/libs/function/src/builtins.c |  2 +-
 source/libs/scalar/src/sclfunc.c    | 54 +++++++++++++++++++++++++++++
 source/util/src/tjson.c             | 35 ++++++++++++++++++-
 5 files changed, 91 insertions(+), 3 deletions(-)

diff --git a/include/libs/scalar/scalar.h b/include/libs/scalar/scalar.h
index 49af10cd2c..6f4a87b02b 100644
--- a/include/libs/scalar/scalar.h
+++ b/include/libs/scalar/scalar.h
@@ -76,6 +76,7 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
 /* Time related functions */
 int32_t toISO8601Function(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 int32_t toUnixtimestampFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t toJsonFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 int32_t timeTruncateFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 int32_t timeDiffFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 int32_t nowFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
diff --git a/include/util/tjson.h b/include/util/tjson.h
index 9f347e5896..28bcb3cfc6 100644
--- a/include/util/tjson.h
+++ b/include/util/tjson.h
@@ -76,7 +76,7 @@ char* tjsonToString(const SJson* pJson);
 char* tjsonToUnformattedString(const SJson* pJson);
 
 SJson* tjsonParse(const char* pStr);
-bool   tjsonIsObject(const SJson* pJson);
+bool   tjsonValidateJson(const char* pJson);
 
 #ifdef __cplusplus
 }
diff --git a/source/libs/function/src/builtins.c b/source/libs/function/src/builtins.c
index bba5145812..342eb5990b 100644
--- a/source/libs/function/src/builtins.c
+++ b/source/libs/function/src/builtins.c
@@ -893,7 +893,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
     .translateFunc = translateToJson,
     .getEnvFunc   = NULL,
     .initFunc     = NULL,
-    .sprocessFunc = NULL,
+    .sprocessFunc = toJsonFunction,
     .finalizeFunc = NULL
   }
 };
diff --git a/source/libs/scalar/src/sclfunc.c b/source/libs/scalar/src/sclfunc.c
index 97cc80b946..05269989cf 100644
--- a/source/libs/scalar/src/sclfunc.c
+++ b/source/libs/scalar/src/sclfunc.c
@@ -4,6 +4,7 @@
 #include "ttime.h"
 #include "sclInt.h"
 #include "sclvector.h"
+#include "tjson.h"
 
 typedef float (*_float_fn)(float);
 typedef double (*_double_fn)(double);
@@ -920,6 +921,59 @@ int32_t toUnixtimestampFunction(SScalarParam *pInput, int32_t inputNum, SScalarP
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t toJsonFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput) {
+  int32_t type = GET_PARAM_TYPE(pInput);
+  if (type != TSDB_DATA_TYPE_BINARY && type != TSDB_DATA_TYPE_NCHAR) {
+    return TSDB_CODE_FAILED;
+  }
+
+  if (inputNum != 1) {
+    return TSDB_CODE_FAILED;
+  }
+
+  char *input = pInput[0].columnData->pData + pInput[0].columnData->varmeta.offset[0];
+  char *tmp = taosMemoryCalloc(pInput[0].columnData->info.bytes + 1, 1);
+  for (int32_t i = 0; i < pInput[0].numOfRows; ++i) {
+    if (colDataIsNull_s(pInput[0].columnData, i)) {
+      colDataAppendNULL(pOutput->columnData, i);
+      continue;
+    }
+
+    if(type == TSDB_DATA_TYPE_NCHAR){
+      if (varDataTLen(input) > TSDB_MAX_JSON_TAG_LEN){
+        colDataAppendNULL(pOutput->columnData, i);
+        continue;
+      }
+      int32_t len  = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), tmp);
+      if (len < 0) {
+        colDataAppendNULL(pOutput->columnData, i);
+        continue;
+      }
+      tmp[len] = 0;
+    }else{
+      if (varDataLen(input) > (TSDB_MAX_JSON_TAG_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE){
+        colDataAppendNULL(pOutput->columnData, i);
+        continue;
+      }
+      memcpy(tmp, varDataVal(input), varDataLen(input));
+      tmp[varDataTLen(input)] = 0;
+    }
+
+    if(!tjsonValidateJson(tmp)){
+      colDataAppendNULL(pOutput->columnData, i);
+      continue;
+    }
+
+    colDataAppend(pOutput->columnData, i, input, false);
+    input += varDataTLen(input);
+  }
+  taosMemoryFree(tmp);
+
+  pOutput->numOfRows = pInput->numOfRows;
+
+  return TSDB_CODE_SUCCESS;
+}
+
 int32_t timeTruncateFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput) {
   int32_t type = GET_PARAM_TYPE(&pInput[0]);
   int32_t timePrec = GET_PARAM_PRECISON(&pInput[0]);
diff --git a/source/util/src/tjson.c b/source/util/src/tjson.c
index 8df404f85f..59dab76989 100644
--- a/source/util/src/tjson.c
+++ b/source/util/src/tjson.c
@@ -276,4 +276,37 @@ int32_t tjsonToArray(const SJson* pJson, const char* pName, FToObject func, void
 }
 
 SJson* tjsonParse(const char* pStr) { return cJSON_Parse(pStr); }
-bool tjsonIsObject(const SJson* pJson) { return cJSON_IsObject(pJson); }
\ No newline at end of file
+
+bool tjsonValidateJson(const char *jIn) {
+  if (!jIn){
+    return false;
+  }
+
+  // set json real data
+  cJSON *root = cJSON_Parse(jIn);
+  if (root == NULL){
+    return false;
+  }
+
+  if(!cJSON_IsObject(root)){
+    return false;
+  }
+  int size = cJSON_GetArraySize(root);
+  for(int i = 0; i < size; i++) {
+    cJSON* item = cJSON_GetArrayItem(root, i);
+    if (!item) {
+      return false;
+    }
+
+    char* jsonKey = item->string;
+    if (!jsonKey) return false;
+    for (size_t j = 0; j < strlen(jsonKey); ++i) {
+      if (isprint(jsonKey[i]) == 0) return false;
+    }
+
+    if (item->type == cJSON_Object || item->type == cJSON_Array) {
+      return false;
+    }
+  }
+  return true;
+}
\ No newline at end of file

From ad299634232e4c07dec38eab35ed6461f250b2bc Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Mon, 18 Apr 2022 21:00:35 +0800
Subject: [PATCH 07/15] refactor: modify json compare in where condition

---
 source/client/src/clientImpl.c     | 72 ++++++++++++++++++++++++++-
 source/libs/scalar/inc/filterInt.h |  2 -
 source/libs/scalar/src/filter.c    | 80 +++++++++---------------------
 3 files changed, 94 insertions(+), 60 deletions(-)

diff --git a/source/client/src/clientImpl.c b/source/client/src/clientImpl.c
index 74b8e711dc..c1351d0fc8 100644
--- a/source/client/src/clientImpl.c
+++ b/source/client/src/clientImpl.c
@@ -619,7 +619,7 @@ void doSetOneRowPtr(SReqResultInfo* pResultInfo) {
     int32_t bytes = pResultInfo->fields[i].bytes;
 
     if (IS_VAR_DATA_TYPE(type)) {
-      if (pCol->offset[pResultInfo->current] != -1) {
+      if (!colDataIsNull_var(pCol,pResultInfo->current)) {
         char* pStart = pResultInfo->pCol[i].offset[pResultInfo->current] + pResultInfo->pCol[i].pData;
 
         pResultInfo->length[i] = varDataLen(pStart);
@@ -725,6 +725,76 @@ static int32_t doConvertUCS4(SReqResultInfo* pResultInfo, int32_t numOfRows, int
       pResultInfo->pCol[i].pData = pResultInfo->convertBuf[i];
       pResultInfo->row[i] = pResultInfo->pCol[i].pData;
     }
+
+    if (type == TSDB_DATA_TYPE_JSON) {
+      char* p = taosMemoryRealloc(pResultInfo->convertBuf[i], colLength[i]);
+      if (p == NULL) {
+        return TSDB_CODE_OUT_OF_MEMORY;
+      }
+
+      pResultInfo->convertBuf[i] = p;
+      int32_t len = 0;
+      SResultColumn* pCol = &pResultInfo->pCol[i];
+      for (int32_t j = 0; j < numOfRows; ++j) {
+        if (pCol->offset[j] != -1) {
+          char* pStart = pCol->offset[j] + pCol->pData;
+
+          int32_t jsonInnerType = *pStart;
+          char *jsonInnerData = pStart + CHAR_BYTES;
+          char dst[TSDB_MAX_JSON_TAG_LEN] = {0};
+          if(jsonInnerType == TSDB_DATA_TYPE_NULL){
+            sprintf(varDataVal(dst), "%s", TSDB_DATA_NULL_STR_L);
+            varDataSetLen(dst, strlen(varDataVal(dst)));
+          }else if(jsonInnerType == TSDB_DATA_TYPE_JSON){
+            int32_t length = taosUcs4ToMbs((TdUcs4 *)varDataVal(jsonInnerData), varDataLen(jsonInnerData), varDataVal(dst));
+
+            if (length <= 0) {
+              tscError("charset:%s to %s. val:%s convert failed.", DEFAULT_UNICODE_ENCODEC, tsCharset, varDataVal(jsonInnerData));
+              length = 0;
+            }
+            varDataSetLen(dst, length);
+          }else if (jsonInnerType == TSDB_DATA_TYPE_NCHAR) {   // value -> "value"
+            *(char*)varDataVal(dst) = '\"';
+            int32_t length = taosUcs4ToMbs((TdUcs4 *)varDataVal(jsonInnerData), varDataLen(jsonInnerData), varDataVal(dst) + CHAR_BYTES);
+            if (length <= 0) {
+              tscError("charset:%s to %s. val:%s convert failed.", DEFAULT_UNICODE_ENCODEC, tsCharset, varDataVal(jsonInnerData));
+              length = 0;
+            }
+            varDataSetLen(dst, length + CHAR_BYTES*2);
+            *(char*)(varDataVal(dst), length + CHAR_BYTES) = '\"';
+          }else if(jsonInnerType == TSDB_DATA_TYPE_DOUBLE){
+            double jsonVd = *(double*)(jsonInnerData);
+            sprintf(varDataVal(dst), "%.9lf", jsonVd);
+            varDataSetLen(dst, strlen(varDataVal(dst)));
+          }else if(jsonInnerType == TSDB_DATA_TYPE_BIGINT){
+            int64_t jsonVd = *(int64_t*)(jsonInnerData);
+            sprintf(varDataVal(dst), "%" PRId64, jsonVd);
+            varDataSetLen(dst, strlen(varDataVal(dst)));
+          }else if(jsonInnerType == TSDB_DATA_TYPE_BOOL){
+            sprintf(varDataVal(dst), "%s", (*((char *)jsonInnerData) == 1) ? "true" : "false");
+            varDataSetLen(dst, strlen(varDataVal(dst)));
+          }else {
+            ASSERT(0);
+          }
+
+          if(len + varDataTLen(dst) > colLength[i]){
+            p = taosMemoryRealloc(pResultInfo->convertBuf[i], len + varDataTLen(dst));
+            if (p == NULL) {
+              return TSDB_CODE_OUT_OF_MEMORY;
+            }
+
+            pResultInfo->convertBuf[i] = p;
+          }
+          p = pResultInfo->convertBuf[i] + len;
+          memcpy(p, dst, varDataTLen(dst));
+          pCol->offset[j] = len;
+          len += varDataTLen(dst);
+        }
+      }
+
+      pResultInfo->pCol[i].pData = pResultInfo->convertBuf[i];
+      pResultInfo->row[i] = pResultInfo->pCol[i].pData;
+    }
   }
 
   return TSDB_CODE_SUCCESS;
diff --git a/source/libs/scalar/inc/filterInt.h b/source/libs/scalar/inc/filterInt.h
index 977fab2e85..2b9e3c7f39 100644
--- a/source/libs/scalar/inc/filterInt.h
+++ b/source/libs/scalar/inc/filterInt.h
@@ -308,7 +308,6 @@ struct SFilterInfo {
 #define FILTER_GET_COL_FIELD_DATA(fi, ri) (colDataGetData(((SColumnInfoData *)(fi)->data), (ri)))
 #define FILTER_GET_VAL_FIELD_TYPE(fi) (((SValueNode *)((fi)->desc))->node.resType.type)
 #define FILTER_GET_VAL_FIELD_DATA(fi) ((char *)(fi)->data)
-#define FILTER_GET_JSON_VAL_FIELD_DATA(fi) ((char *)(fi)->desc)
 #define FILTER_GET_TYPE(fl) ((fl) & FLD_TYPE_MAX)
 
 #define FILTER_GROUP_UNIT(i, g, uid) ((i)->units + (g)->unitIdxs[uid])
@@ -321,7 +320,6 @@ struct SFilterInfo {
 #define FILTER_UNIT_COL_SIZE(i, u) FILTER_GET_COL_FIELD_SIZE(FILTER_UNIT_LEFT_FIELD(i, u))
 #define FILTER_UNIT_COL_ID(i, u) FILTER_GET_COL_FIELD_ID(FILTER_UNIT_LEFT_FIELD(i, u))
 #define FILTER_UNIT_VAL_DATA(i, u) FILTER_GET_VAL_FIELD_DATA(FILTER_UNIT_RIGHT_FIELD(i, u))
-#define FILTER_UNIT_JSON_VAL_DATA(i, u) FILTER_GET_JSON_VAL_FIELD_DATA(FILTER_UNIT_RIGHT_FIELD(i, u))
 #define FILTER_UNIT_COL_IDX(u) ((u)->left.idx)
 #define FILTER_UNIT_OPTR(u) ((u)->compare.optr)
 #define FILTER_UNIT_COMP_FUNC(u) ((u)->compare.func)
diff --git a/source/libs/scalar/src/filter.c b/source/libs/scalar/src/filter.c
index ea249005b4..ee85b99ad5 100644
--- a/source/libs/scalar/src/filter.c
+++ b/source/libs/scalar/src/filter.c
@@ -63,22 +63,18 @@ OptrStr gOptrStr[] = {
 
 bool filterRangeCompGi (const void *minv, const void *maxv, const void *minr, const void *maxr, __compar_fn_t cfunc) {
   int32_t result = cfunc(maxv, minr);
-  //if (result == TSDB_DATA_JSON_CAN_NOT_COMPARE) return false;
   return result >= 0;
 }
 bool filterRangeCompGe (const void *minv, const void *maxv, const void *minr, const void *maxr, __compar_fn_t cfunc) {
   int32_t result = cfunc(maxv, minr);
-  //if (result == TSDB_DATA_JSON_CAN_NOT_COMPARE) return false;
   return result > 0;
 }
 bool filterRangeCompLi (const void *minv, const void *maxv, const void *minr, const void *maxr, __compar_fn_t cfunc) {
   int32_t result = cfunc(minv, maxr);
-  //if (result == TSDB_DATA_JSON_CAN_NOT_COMPARE) return false;
   return result <= 0;
 }
 bool filterRangeCompLe (const void *minv, const void *maxv, const void *minr, const void *maxr, __compar_fn_t cfunc) {
   int32_t result = cfunc(minv, maxr);
-  //if (result == TSDB_DATA_JSON_CAN_NOT_COMPARE) return false;
   return result < 0;
 }
 bool filterRangeCompii (const void *minv, const void *maxv, const void *minr, const void *maxr, __compar_fn_t cfunc) {
@@ -1783,9 +1779,9 @@ int32_t fltInitValFieldData(SFilterInfo *info) {
       bytes = (len + 1) * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE;
 
       fi->data = taosMemoryCalloc(1, bytes);
-    } else if (type != TSDB_DATA_TYPE_JSON){
+    } else{
       if (dType->type == TSDB_DATA_TYPE_VALUE_ARRAY) {  //TIME RANGE
-/*      
+/*
         fi->data = taosMemoryCalloc(dType->bytes, tDataTypes[type].bytes);
         for (int32_t a = 0; a < dType->bytes; ++a) {
           int64_t *v = taosArrayGet(var->arr, a);
@@ -1796,31 +1792,27 @@ int32_t fltInitValFieldData(SFilterInfo *info) {
       } else {
         fi->data = taosMemoryCalloc(1, sizeof(int64_t));
       }
-    } else{     // type == TSDB_DATA_TYPE_JSON
-      // fi->data = null;  use fi->desc as data, because json value is variable, so use tVariant (fi->desc)
     }
 
-    if(type != TSDB_DATA_TYPE_JSON) {
-      if (dType->type == type) {
-        assignVal(fi->data, nodesGetValueFromNode(var), dType->bytes, type);
+    if (dType->type == type) {
+      assignVal(fi->data, nodesGetValueFromNode(var), dType->bytes, type);
+    } else {
+      SScalarParam out = {.columnData = taosMemoryCalloc(1, sizeof(SColumnInfoData))};
+      out.columnData->info.type = type;
+      if (IS_VAR_DATA_TYPE(type)) {
+        out.columnData->info.bytes = bytes;
       } else {
-        SScalarParam out = {.columnData = taosMemoryCalloc(1, sizeof(SColumnInfoData))};
-        out.columnData->info.type = type;
-        if (IS_VAR_DATA_TYPE(type)) {
-          out.columnData->info.bytes = bytes;
-        } else {
-          out.columnData->info.bytes = tDataTypes[type].bytes;
-        }
-
-        // todo refactor the convert
-        int32_t code = doConvertDataType(var, &out);
-        if (code != TSDB_CODE_SUCCESS) {
-          qError("convert value to type[%d] failed", type);
-          return TSDB_CODE_TSC_INVALID_OPERATION;
-        }
-
-        memcpy(fi->data, out.columnData->pData, out.columnData->info.bytes);
+        out.columnData->info.bytes = tDataTypes[type].bytes;
       }
+
+      // todo refactor the convert
+      int32_t code = doConvertDataType(var, &out);
+      if (code != TSDB_CODE_SUCCESS) {
+        qError("convert value to type[%d] failed", type);
+        return TSDB_CODE_TSC_INVALID_OPERATION;
+      }
+
+      memcpy(fi->data, out.columnData->pData, out.columnData->info.bytes);
     }
 
     // match/nmatch for nchar type need convert from ucs4 to mbs
@@ -2561,11 +2553,7 @@ int32_t filterGenerateComInfo(SFilterInfo *info) {
     info->cunits[i].colId = FILTER_UNIT_COL_ID(info, unit);
     
     if (unit->right.type == FLD_TYPE_VALUE) {
-      if(FILTER_UNIT_DATA_TYPE(unit) == TSDB_DATA_TYPE_JSON){   // json value is tVariant
-        info->cunits[i].valData = FILTER_UNIT_JSON_VAL_DATA(info, unit);
-      }else{
-        info->cunits[i].valData = FILTER_UNIT_VAL_DATA(info, unit);
-      }
+      info->cunits[i].valData = FILTER_UNIT_VAL_DATA(info, unit);
     } else {
       info->cunits[i].valData = NULL;
     }
@@ -2891,18 +2879,8 @@ static FORCE_INLINE bool filterExecuteImplIsNull(void *pinfo, int32_t numOfRows,
   for (int32_t i = 0; i < numOfRows; ++i) {
     uint32_t uidx = info->groups[0].unitIdxs[0];
     void *colData = colDataGetData((SColumnInfoData *)info->cunits[uidx].colData, i);
-    if(info->cunits[uidx].dataType == TSDB_DATA_TYPE_JSON){
-      if (!colData){  // for json->'key' is null
-        (*p)[i] = 1;
-      }else if( *(char*)colData == TSDB_DATA_TYPE_JSON){  // for json is null
-        colData = POINTER_SHIFT(colData, CHAR_BYTES);
-        (*p)[i] = colDataIsNull((SColumnInfoData *)info->cunits[uidx].colData, 0, i, NULL);
-      }else{
-        (*p)[i] = 0;
-      }
-    }else{
-      (*p)[i] = ((colData == NULL) || colDataIsNull((SColumnInfoData *)info->cunits[uidx].colData, 0, i, NULL));
-    }
+    (*p)[i] = ((colData == NULL) || colDataIsNull((SColumnInfoData *)info->cunits[uidx].colData, 0, i, NULL));
+
     if ((*p)[i] == 0) {
       all = false;
     }    
@@ -2926,19 +2904,7 @@ static FORCE_INLINE bool filterExecuteImplNotNull(void *pinfo, int32_t numOfRows
     uint32_t uidx = info->groups[0].unitIdxs[0];
     void *colData = colDataGetData((SColumnInfoData *)info->cunits[uidx].colData, i);
 
-    if(info->cunits[uidx].dataType == TSDB_DATA_TYPE_JSON){
-      if (!colData) {   // for json->'key' is not null
-        (*p)[i] = 0;
-      }else if( *(char*)colData == TSDB_DATA_TYPE_JSON){   // for json is not null
-        colData = POINTER_SHIFT(colData, CHAR_BYTES);
-        (*p)[i] = !colDataIsNull((SColumnInfoData *)info->cunits[uidx].colData, 0, i, NULL);
-      }else{    // for json->'key' is not null
-        (*p)[i] = 1;
-      }
-    }else {
-      (*p)[i] = ((colData != NULL) && !colDataIsNull((SColumnInfoData *)info->cunits[uidx].colData, 0, i, NULL));
-    }
-
+    (*p)[i] = ((colData != NULL) && !colDataIsNull((SColumnInfoData *)info->cunits[uidx].colData, 0, i, NULL));
     if ((*p)[i] == 0) {
       all = false;
     }

From 134a161c7205157157395b3adc23a48be8525d9e Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Mon, 18 Apr 2022 21:05:34 +0800
Subject: [PATCH 08/15] refactor: modify json compare in where condition

---
 source/client/src/clientImpl.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/source/client/src/clientImpl.c b/source/client/src/clientImpl.c
index c1351d0fc8..49f663772d 100644
--- a/source/client/src/clientImpl.c
+++ b/source/client/src/clientImpl.c
@@ -619,7 +619,7 @@ void doSetOneRowPtr(SReqResultInfo* pResultInfo) {
     int32_t bytes = pResultInfo->fields[i].bytes;
 
     if (IS_VAR_DATA_TYPE(type)) {
-      if (!colDataIsNull_var(pCol,pResultInfo->current)) {
+      if (pCol->offset[pResultInfo->current] != -1) {
         char* pStart = pResultInfo->pCol[i].offset[pResultInfo->current] + pResultInfo->pCol[i].pData;
 
         pResultInfo->length[i] = varDataLen(pStart);

From b96082fe1b36468466e78e76c6ff9017da8f280a Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Tue, 19 Apr 2022 09:46:12 +0800
Subject: [PATCH 09/15] refactor: fix compare logic

---
 source/libs/scalar/src/filter.c    |  9 +++++++
 source/libs/scalar/src/scalar.c    |  1 +
 source/libs/scalar/src/sclvector.c | 39 +++++++++++++++---------------
 3 files changed, 30 insertions(+), 19 deletions(-)

diff --git a/source/libs/scalar/src/filter.c b/source/libs/scalar/src/filter.c
index ee85b99ad5..23047c2930 100644
--- a/source/libs/scalar/src/filter.c
+++ b/source/libs/scalar/src/filter.c
@@ -1050,6 +1050,8 @@ int32_t fltAddGroupUnitFromNode(SFilterInfo *info, SNode* tree, SArray *group) {
 
       cell = cell->pNext;
     }
+    colDataDestroy(out.columnData);
+    taosMemoryFree(out.columnData);
   } else {
     filterAddFieldFromNode(info, node->pRight, &right);
     
@@ -1813,6 +1815,8 @@ int32_t fltInitValFieldData(SFilterInfo *info) {
       }
 
       memcpy(fi->data, out.columnData->pData, out.columnData->info.bytes);
+      colDataDestroy(out.columnData);
+      taosMemoryFree(out.columnData);
     }
 
     // match/nmatch for nchar type need convert from ucs4 to mbs
@@ -3537,6 +3541,11 @@ EDealRes fltReviseRewriter(SNode** pNode, void* pContext) {
         return DEAL_RES_CONTINUE;
       }
 
+      if (OP_TYPE_JSON_CONTAINS == node->opType) {
+        stat->scalarMode = true;
+        return DEAL_RES_CONTINUE;
+      }
+
       if (QUERY_NODE_COLUMN != nodeType(node->pLeft)) {
         SNode *t = node->pLeft;
         node->pLeft = node->pRight;
diff --git a/source/libs/scalar/src/scalar.c b/source/libs/scalar/src/scalar.c
index 5f1e08dbcf..9c4d2d94dd 100644
--- a/source/libs/scalar/src/scalar.c
+++ b/source/libs/scalar/src/scalar.c
@@ -132,6 +132,7 @@ void sclFreeRes(SHashObj *res) {
 void sclFreeParam(SScalarParam *param) {
   if (param->columnData != NULL) {
     colDataDestroy(param->columnData);
+    taosMemoryFree(param->columnData);
   }
 
   if (param->pHashFilter != NULL) {
diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c
index e6deef5824..e97831b1f6 100644
--- a/source/libs/scalar/src/sclvector.c
+++ b/source/libs/scalar/src/sclvector.c
@@ -188,17 +188,17 @@ static FORCE_INLINE void varToBool(char *buf, SScalarParam* pOut, int32_t rowInd
   colDataAppendInt8(pOut->columnData, rowIndex, (int8_t*) &v);
 }
 
-//static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIndex) {
-//  int32_t len = 0;
-//  int32_t inputLen = varDataLen(buf);
-//
-//  char* t = taosMemoryCalloc(1,(inputLen + 1) * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE);
-//  /*int32_t resLen = */taosMbsToUcs4(varDataVal(buf), inputLen, (TdUcs4*) varDataVal(t), pOut->columnData->info.bytes, &len);
-//  varDataSetLen(t, len);
-//
-//  colDataAppend(pOut->columnData, rowIndex, t, false);
-//  taosMemoryFree(t);
-//}
+static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIndex) {
+  int32_t len = 0;
+  int32_t inputLen = varDataLen(buf);
+
+  char* t = taosMemoryCalloc(1,(inputLen + 1) * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE);
+  /*int32_t resLen = */taosMbsToUcs4(varDataVal(buf), inputLen, (TdUcs4*) varDataVal(t), pOut->columnData->info.bytes, &len);
+  varDataSetLen(t, len);
+
+  colDataAppend(pOut->columnData, rowIndex, t, false);
+  taosMemoryFree(t);
+}
 
 void convertNumberToNumber(const void *inData, void *outData, int8_t inType, int8_t outType){
   switch (outType) {
@@ -275,7 +275,7 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
   int32_t bufSize = pIn->columnData->info.bytes;
   char *tmp = taosMemoryMalloc(bufSize + VARSTR_HEADER_SIZE);
 
-//  bool vton = false;
+  bool vton = false;
 
   _bufConverteFunc func = NULL;
   if (TSDB_DATA_TYPE_BOOL == outType) {
@@ -286,9 +286,10 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
     func = varToUnsigned;
   } else if (IS_FLOAT_TYPE(outType)) {
     func = varToFloat;
-//  } else if (outType == TSDB_DATA_TYPE_NCHAR) {   // can not be nchar or binary
-//    func = varToNchar;
-//    vton = true;
+  } else if (outType == TSDB_DATA_TYPE_NCHAR) {   // binary -> nchar
+    ASSERT(inType == TSDB_DATA_TYPE_VARCHAR);
+    func = varToNchar;
+    vton = true;
   } else {
     sclError("invalid convert outType:%d", outType);
     return TSDB_CODE_QRY_APP_ERROR;
@@ -316,9 +317,9 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
         continue;
       }
     }
-//    if (vton) {
-//      memcpy(tmp, data, varDataTLen(data));
-//    } else {
+    if (vton) {
+      memcpy(tmp, data, varDataTLen(data));
+    } else {
       if (TSDB_DATA_TYPE_VARCHAR == convertType) {
         memcpy(tmp, varDataVal(data), varDataLen(data));
         tmp[varDataLen(data)] = 0;
@@ -334,7 +335,7 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
 
         tmp[len] = 0;
       }
-//    }
+    }
     
     (*func)(tmp, pOut, i);
   }

From 412e2cbb0d3631017c81783fdfb8d18ca67f304f Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Tue, 19 Apr 2022 17:52:04 +0800
Subject: [PATCH 10/15] refactor: add test casese for json compare logic

---
 include/util/tdef.h                           |   5 -
 source/libs/parser/src/parUtil.c              |   4 +-
 source/libs/scalar/src/scalar.c               |  88 +++++++++------
 source/libs/scalar/src/sclvector.c            |  23 +++-
 .../libs/scalar/test/scalar/scalarTests.cpp   | 103 +++++++++++++++++-
 5 files changed, 177 insertions(+), 46 deletions(-)

diff --git a/include/util/tdef.h b/include/util/tdef.h
index eb7a51cd7e..4c3126caa9 100644
--- a/include/util/tdef.h
+++ b/include/util/tdef.h
@@ -272,11 +272,6 @@ typedef enum ELogicConditionType {
 #define TSDB_MAX_TAG_CONDITIONS 1024
 
 #define TSDB_MAX_JSON_TAG_LEN   16384
-#define TSDB_JSON_PLACEHOLDER      0x7F
-#define TSDB_JSON_null             0x00
-#define TSDB_JSON_NOT_NULL         0x01
-#define TSDB_JSON_NULL             0x00
-
 
 #define TSDB_AUTH_LEN          16
 #define TSDB_PASSWORD_LEN      32
diff --git a/source/libs/parser/src/parUtil.c b/source/libs/parser/src/parUtil.c
index f3eb0c5ebc..0ff5843736 100644
--- a/source/libs/parser/src/parUtil.c
+++ b/source/libs/parser/src/parUtil.c
@@ -276,7 +276,7 @@ static bool isValidateTag(char *input) {
 
 int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* pMsgBuf, int16_t startColId){
   // set json NULL data
-  uint8_t jsonNULL = TSDB_JSON_NULL;
+  uint8_t jsonNULL = TSDB_DATA_TYPE_NULL;
   int jsonIndex = startColId + 1;
   if (!json || strcasecmp(json, TSDB_DATA_NULL_STR_L) == 0){
     tdAddColToKVRow(kvRowBuilder, jsonIndex, &jsonNULL, CHAR_BYTES);
@@ -328,7 +328,7 @@ int parseJsontoTagData(const char* json, SKVRowBuilder* kvRowBuilder, SMsgBuf* p
     strncpy(varDataVal(tagKV), jsonKey, keyLen);
     varDataSetLen(tagKV, keyLen);
     if(taosHashGetSize(keyHash) == 0){
-      uint8_t jsonNotNULL = TSDB_JSON_NOT_NULL;
+      uint8_t jsonNotNULL = TSDB_DATA_TYPE_JSON;
       tdAddColToKVRow(kvRowBuilder, jsonIndex++, &jsonNotNULL, CHAR_BYTES);   // add json type
     }
     taosHashPut(keyHash, jsonKey, keyLen, &keyLen, CHAR_BYTES);  // add key to hash to remove dumplicate, value is useless
diff --git a/source/libs/scalar/src/scalar.c b/source/libs/scalar/src/scalar.c
index 9c4d2d94dd..2fd688a233 100644
--- a/source/libs/scalar/src/scalar.c
+++ b/source/libs/scalar/src/scalar.c
@@ -604,39 +604,55 @@ EDealRes sclWalkOperator(SNode* pNode, SScalarCtx *ctx) {
   return DEAL_RES_CONTINUE;
 }
 
-//EDealRes sclWalkTarget(SNode* pNode, SScalarCtx *ctx) {
-//  STargetNode *target = (STargetNode *)pNode;
-//
-//  if (target->dataBlockId >= taosArrayGetSize(ctx->pBlockList)) {
-//    sclError("target tupleId is too big, tupleId:%d, dataBlockNum:%d", target->dataBlockId, (int32_t)taosArrayGetSize(ctx->pBlockList));
-//    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
-//    return DEAL_RES_ERROR;
-//  }
-//
-//  SSDataBlock *block = *(SSDataBlock **)taosArrayGet(ctx->pBlockList, target->dataBlockId);
-//  if (target->slotId >= taosArrayGetSize(block->pDataBlock)) {
-//    sclError("target slot not exist, dataBlockId:%d, slotId:%d, dataBlockNum:%d", target->dataBlockId, target->slotId, (int32_t)taosArrayGetSize(block->pDataBlock));
-//    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
-//    return DEAL_RES_ERROR;
-//  }
-//
-//  // if block->pDataBlock is not enough, there are problems if target->slotId bigger than the size of block->pDataBlock,
-//  SColumnInfoData *col = taosArrayGet(block->pDataBlock, target->slotId);
-//
-//  SScalarParam *res = (SScalarParam *)taosHashGet(ctx->pRes, (void *)&target->pExpr, POINTER_BYTES);
-//  if (NULL == res) {
-//    sclError("no valid res in hash, node:%p, type:%d", target->pExpr, nodeType(target->pExpr));
-//    ctx->code = TSDB_CODE_QRY_APP_ERROR;
-//    return DEAL_RES_ERROR;
-//  }
-//
-//  colDataAssign(col, res->columnData, res->numOfRows);
-//  block->info.rows = res->numOfRows;
-//
-//  sclFreeParam(res);
-//  taosHashRemove(ctx->pRes, (void *)&target->pExpr, POINTER_BYTES);
-//  return DEAL_RES_CONTINUE;
-//}
+EDealRes sclWalkTarget(SNode* pNode, SScalarCtx *ctx) {
+  STargetNode *target = (STargetNode *)pNode;
+
+  if (target->dataBlockId >= taosArrayGetSize(ctx->pBlockList)) {
+    sclError("target tupleId is too big, tupleId:%d, dataBlockNum:%d", target->dataBlockId, (int32_t)taosArrayGetSize(ctx->pBlockList));
+    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
+    return DEAL_RES_ERROR;
+  }
+
+  int32_t index = -1;
+  for(int32_t i = 0; i < taosArrayGetSize(ctx->pBlockList); ++i) {
+    SSDataBlock* pb = taosArrayGetP(ctx->pBlockList, i);
+    if (pb->info.blockId == target->dataBlockId) {
+      index = i;
+      break;
+    }
+  }
+
+  if (index == -1) {
+    sclError("column tupleId is too big, tupleId:%d, dataBlockNum:%d", target->dataBlockId, (int32_t)taosArrayGetSize(ctx->pBlockList));
+    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
+    return DEAL_RES_ERROR;
+  }
+
+  SSDataBlock *block = *(SSDataBlock **)taosArrayGet(ctx->pBlockList, index);
+
+  if (target->slotId >= taosArrayGetSize(block->pDataBlock)) {
+    sclError("target slot not exist, dataBlockId:%d, slotId:%d, dataBlockNum:%d", target->dataBlockId, target->slotId, (int32_t)taosArrayGetSize(block->pDataBlock));
+    ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
+    return DEAL_RES_ERROR;
+  }
+
+  // if block->pDataBlock is not enough, there are problems if target->slotId bigger than the size of block->pDataBlock,
+  SColumnInfoData *col = taosArrayGet(block->pDataBlock, target->slotId);
+
+  SScalarParam *res = (SScalarParam *)taosHashGet(ctx->pRes, (void *)&target->pExpr, POINTER_BYTES);
+  if (NULL == res) {
+    sclError("no valid res in hash, node:%p, type:%d", target->pExpr, nodeType(target->pExpr));
+    ctx->code = TSDB_CODE_QRY_APP_ERROR;
+    return DEAL_RES_ERROR;
+  }
+
+  colDataAssign(col, res->columnData, res->numOfRows);
+  block->info.rows = res->numOfRows;
+
+  sclFreeParam(res);
+  taosHashRemove(ctx->pRes, (void *)&target->pExpr, POINTER_BYTES);
+  return DEAL_RES_CONTINUE;
+}
 
 EDealRes sclCalcWalker(SNode* pNode, void* pContext) {
   if (QUERY_NODE_VALUE == nodeType(pNode) || QUERY_NODE_NODE_LIST == nodeType(pNode) || QUERY_NODE_COLUMN == nodeType(pNode)) {
@@ -656,9 +672,9 @@ EDealRes sclCalcWalker(SNode* pNode, void* pContext) {
     return sclWalkOperator(pNode, ctx);
   }
 
-//  if (QUERY_NODE_TARGET == nodeType(pNode)) {
-//    return sclWalkTarget(pNode, ctx);
-//  }
+  if (QUERY_NODE_TARGET == nodeType(pNode)) {
+    return sclWalkTarget(pNode, ctx);
+  }
 
   sclError("invalid node type for scalar calculating, type:%d", nodeType(pNode));
   ctx->code = TSDB_CODE_QRY_INVALID_INPUT;
diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c
index e97831b1f6..7ad200befc 100644
--- a/source/libs/scalar/src/sclvector.c
+++ b/source/libs/scalar/src/sclvector.c
@@ -767,6 +767,17 @@ static void doReleaseVec(SColumnInfoData* pCol, int32_t type) {
 }
 
 char *getJsonValue(char *json, char *key){    //todo
+  int16_t cols = kvRowNCols(json);
+  for (int i = 0; i < cols; ++i) {
+    SColIdx *pColIdx = kvRowColIdxAt(json, i);
+    char *data = kvRowColVal(json, pColIdx);
+    if(i == 0 && *data == TSDB_DATA_TYPE_NULL){
+      return NULL;
+    }
+    if(strncmp(key, varDataVal(data), varDataLen(data)) == 0){
+      return data + varDataTLen(data);
+    }
+  }
   return NULL;
 }
 
@@ -844,7 +855,11 @@ void vectorMathAdd(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
 
     if (pLeft->numOfRows == pRight->numOfRows) {
       for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-        *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) + getVectorDoubleValueFnRight(pRightCol->pData, i);
+        if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+          colDataAppendNULL(pOutputCol, i);
+          continue;  // TODO set null or ignore
+        }
+        *output = getVectorDoubleValueFnLeft(LEFT_COL, i) + getVectorDoubleValueFnRight(RIGHT_COL, i);
       }
 
       pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
@@ -957,7 +972,11 @@ void vectorMathSub(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
 
     if (pLeft->numOfRows == pRight->numOfRows) {
       for (; i < pRight->numOfRows && i >= 0; i += step, output += 1) {
-        *output = getVectorDoubleValueFnLeft(pLeftCol->pData, i) - getVectorDoubleValueFnRight(pRightCol->pData, i);
+        if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+          colDataAppendNULL(pOutputCol, i);
+          continue;  // TODO set null or ignore
+        }
+        *output = getVectorDoubleValueFnLeft(LEFT_COL, i) - getVectorDoubleValueFnRight(RIGHT_COL, i);
       }
 
       pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
diff --git a/source/libs/scalar/test/scalar/scalarTests.cpp b/source/libs/scalar/test/scalar/scalarTests.cpp
index 61ef2fdce2..975effc63a 100644
--- a/source/libs/scalar/test/scalar/scalarTests.cpp
+++ b/source/libs/scalar/test/scalar/scalarTests.cpp
@@ -92,6 +92,7 @@ void scltAppendReservedSlot(SArray *pBlockList, int16_t *dataBlockId, int16_t *s
     blockDataEnsureCapacity(res, rows);
 
     *dataBlockId = taosArrayGetSize(pBlockList) - 1;
+    res->info.blockId = *dataBlockId;
     *slotId = 0;
   } else {
     SSDataBlock *res = *(SSDataBlock **)taosArrayGetLast(pBlockList);
@@ -124,6 +125,56 @@ void scltMakeValueNode(SNode **pNode, int32_t dataType, void *value) {
   
   *pNode = (SNode *)vnode;
 }
+void scltMakeJsonColumnNode(SNode **pNode, SSDataBlock **block, int32_t dataType, int32_t dataBytes, int32_t rowNum, void *value) {
+  SNode *node = (SNode*)nodesMakeNode(QUERY_NODE_COLUMN);
+  SColumnNode *rnode = (SColumnNode *)node;
+  rnode->node.resType.type = dataType;
+  rnode->node.resType.bytes = dataBytes;
+  rnode->dataBlockId = 0;
+
+  if (NULL == *block) {
+    SSDataBlock *res = (SSDataBlock *)taosMemoryCalloc(1, sizeof(SSDataBlock));
+    res->info.numOfCols = 3;
+    res->info.rows = rowNum;
+    res->pDataBlock = taosArrayInit(3, sizeof(SColumnInfoData));
+    for (int32_t i = 0; i < 2; ++i) {
+      SColumnInfoData idata = {{0}};
+      idata.info.type  = TSDB_DATA_TYPE_NULL;
+      idata.info.bytes = 10;
+      idata.info.colId = i + 1;
+
+      int32_t size = idata.info.bytes * rowNum;
+      idata.pData = (char *)taosMemoryCalloc(1, size);
+      taosArrayPush(res->pDataBlock, &idata);
+    }
+
+    SColumnInfoData idata = {{0}};
+    idata.info.type  = dataType;
+    idata.info.bytes = dataBytes;
+    idata.info.colId = 3;
+    int32_t size = idata.info.bytes * rowNum;
+    idata.pData = (char *)taosMemoryCalloc(1, size);
+    taosArrayPush(res->pDataBlock, &idata);
+
+    blockDataEnsureCapacity(res, rowNum);
+
+    SColumnInfoData *pColumn = (SColumnInfoData *)taosArrayGetLast(res->pDataBlock);
+    for (int32_t i = 0; i < rowNum; ++i) {
+      uint32_t len = pColumnInfoData->varmeta.length;
+      pColumnInfoData->varmeta.offset[i] = len;
+
+      memcpy(pColumnInfoData->pData + len, value, varDataLen(value));
+      pColumnInfoData->varmeta.length += varDataLen(value);
+    }
+
+    rnode->slotId = 2;
+    rnode->colId = 3;
+
+    *block = res;
+  }
+
+  *pNode = (SNode *)rnode;
+}
 
 void scltMakeColumnNode(SNode **pNode, SSDataBlock **block, int32_t dataType, int32_t dataBytes, int32_t rowNum, void *value) {
   SNode *node = (SNode*)nodesMakeNode(QUERY_NODE_COLUMN);
@@ -909,6 +960,56 @@ TEST(constantTest, greater_and_lower) {
   nodesDestroyNode(res);
 }
 
+TEST(columnTest, smallint_value_add_json_column) {
+  scltInitLogFile();
+
+  SNode *pLeft = NULL, *pRight = NULL, *opNode = NULL;
+  int32_t leftv = 1;
+  char *rightv= "{"k1":4,"k2":"hello","k3":null,"k4":true,"k5":5.44}";
+  char key[10] = {0};
+  memcpy(varDataVal(key), "k1", 2);
+  varDataLen(key) = 2;
+
+  SKVRowBuilder kvRowBuilder;
+  tdInitKVRowBuilder(&kvRowBuilder);
+  parseJsontoTagData(rightv, &kvRowBuilder, NULL, 0);
+  SKVRow row = tdGetKVRowFromBuilder(&kvRowBuilder);
+
+  double eRes[5] = {1.0, -4, -3, 24, 101};
+
+  SSDataBlock *src = NULL;
+  int32_t rowNum = 1;
+
+  scltMakeValueNode(&pRight, TSDB_DATA_TYPE_BINARY, key);
+  scltMakeJsonColumnNode(&pLeft, &src, TSDB_DATA_TYPE_JSON, varDataLen(row), rowNum, row);
+  scltMakeOpNode(&opNode, OP_TYPE_JSON_GET_VALUE, TSDB_DATA_TYPE_JSON, pLeft, pRight);
+
+  SNode *right = NULL;
+  scltMakeValueNode(&right, TSDB_DATA_TYPE_INT, &leftv);
+  scltMakeOpNode(&opNode, OP_TYPE_ADD, TSDB_DATA_TYPE_DOUBLE, opNode, right);
+
+  SArray *blockList = taosArrayInit(2, POINTER_BYTES);
+  taosArrayPush(blockList, &src);
+  SColumnInfo colInfo = createColumnInfo(1, TSDB_DATA_TYPE_DOUBLE, sizeof(double));
+  int16_t dataBlockId = 0, slotId = 0;
+  scltAppendReservedSlot(blockList, &dataBlockId, &slotId, true, rowNum, &colInfo);
+  scltMakeTargetNode(&opNode, dataBlockId, slotId, opNode);
+
+  int32_t code = scalarCalculate(opNode, blockList, NULL);
+  ASSERT_EQ(code, 0);
+
+  SSDataBlock *res = *(SSDataBlock **)taosArrayGetLast(blockList);
+  ASSERT_EQ(res->info.rows, rowNum);
+  SColumnInfoData *column = (SColumnInfoData *)taosArrayGetLast(res->pDataBlock);
+  ASSERT_EQ(column->info.type, TSDB_DATA_TYPE_DOUBLE);
+  for (int32_t i = 0; i < rowNum; ++i) {
+    ASSERT_EQ(*((double *)colDataGetData(column, i)), eRes[i]);
+  }
+
+  taosArrayDestroyEx(blockList, scltFreeDataBlock);
+  nodesDestroyNode(opNode);
+}
+
 TEST(columnTest, smallint_value_add_int_column) {
   scltInitLogFile();
   
@@ -928,7 +1029,7 @@ TEST(columnTest, smallint_value_add_int_column) {
   int16_t dataBlockId = 0, slotId = 0;
   scltAppendReservedSlot(blockList, &dataBlockId, &slotId, true, rowNum, &colInfo);
   scltMakeTargetNode(&opNode, dataBlockId, slotId, opNode);
-  
+
   int32_t code = scalarCalculate(opNode, blockList, NULL);
   ASSERT_EQ(code, 0);
 

From 2133a2de1d5ab0c761bf9fb9b85c0f5c139beba8 Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Tue, 19 Apr 2022 20:30:06 +0800
Subject: [PATCH 11/15] refactor: add test casese for json compare logic

---
 include/common/tdataformat.h                  |   2 +-
 source/common/src/tdatablock.c                |  21 ++-
 source/libs/scalar/src/sclvector.c            |   4 +-
 source/libs/scalar/test/scalar/CMakeLists.txt |   3 +-
 .../libs/scalar/test/scalar/scalarTests.cpp   | 143 +++++++-----------
 5 files changed, 80 insertions(+), 93 deletions(-)

diff --git a/include/common/tdataformat.h b/include/common/tdataformat.h
index 2dd887ae74..a0f4351bbf 100644
--- a/include/common/tdataformat.h
+++ b/include/common/tdataformat.h
@@ -502,7 +502,7 @@ typedef struct {
 
 #define TD_KV_ROW_HEAD_SIZE (sizeof(uint16_t) + sizeof(int16_t))
 
-#define kvRowLen(r)            (*(TDRowLenT *)(r))
+#define kvRowLen(r)            (*(uint16_t *)(r))
 #define kvRowNCols(r)          (*(int16_t *)POINTER_SHIFT(r, sizeof(uint16_t)))
 #define kvRowSetLen(r, len)    kvRowLen(r) = (len)
 #define kvRowSetNCols(r, n)    kvRowNCols(r) = (n)
diff --git a/source/common/src/tdatablock.c b/source/common/src/tdatablock.c
index 14b7fe5d0e..5bdf7ddb85 100644
--- a/source/common/src/tdatablock.c
+++ b/source/common/src/tdatablock.c
@@ -113,14 +113,27 @@ int32_t colDataAppend(SColumnInfoData* pColumnInfoData, uint32_t currentRow, con
 
   int32_t type = pColumnInfoData->info.type;
   if (IS_VAR_DATA_TYPE(type)) {
+    int32_t dataLen = varDataTLen(pData);
+    if(type == TSDB_DATA_TYPE_JSON) {
+      if(*pData == TSDB_DATA_TYPE_NULL) {
+        dataLen = 0;
+      }else if(*pData == TSDB_DATA_TYPE_NCHAR) {
+        dataLen = varDataTLen(pData+CHAR_BYTES);
+      }else if(*pData == TSDB_DATA_TYPE_BIGINT || *pData == TSDB_DATA_TYPE_DOUBLE) {
+        dataLen = LONG_BYTES;
+      }else if(*pData == TSDB_DATA_TYPE_BOOL) {
+        dataLen = CHAR_BYTES;
+      }
+      dataLen += CHAR_BYTES;
+    }
     SVarColAttr* pAttr = &pColumnInfoData->varmeta;
-    if (pAttr->allocLen < pAttr->length + varDataTLen(pData)) {
+    if (pAttr->allocLen < pAttr->length + dataLen) {
       uint32_t newSize = pAttr->allocLen;
       if (newSize == 0) {
         newSize = 8;
       }
 
-      while (newSize < pAttr->length + varDataTLen(pData)) {
+      while (newSize < pAttr->length + dataLen) {
         newSize = newSize * 1.5;
       }
 
@@ -136,8 +149,8 @@ int32_t colDataAppend(SColumnInfoData* pColumnInfoData, uint32_t currentRow, con
     uint32_t len = pColumnInfoData->varmeta.length;
     pColumnInfoData->varmeta.offset[currentRow] = len;
 
-    memcpy(pColumnInfoData->pData + len, pData, varDataTLen(pData));
-    pColumnInfoData->varmeta.length += varDataTLen(pData);
+    memcpy(pColumnInfoData->pData + len, pData, dataLen);
+    pColumnInfoData->varmeta.length += dataLen;
   } else {
     memcpy(pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow, pData, pColumnInfoData->info.bytes);
   }
diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c
index 7ad200befc..c2eafedf9b 100644
--- a/source/libs/scalar/src/sclvector.c
+++ b/source/libs/scalar/src/sclvector.c
@@ -774,7 +774,7 @@ char *getJsonValue(char *json, char *key){    //todo
     if(i == 0 && *data == TSDB_DATA_TYPE_NULL){
       return NULL;
     }
-    if(strncmp(key, varDataVal(data), varDataLen(data)) == 0){
+    if(memcmp(key, data, varDataTLen(data)) == 0){
       return data + varDataTLen(data);
     }
   }
@@ -803,7 +803,7 @@ void vectorJsonArrow(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pO
       pOutputCol->hasNull = true;
       continue;
     }
-    colDataAppend(pOutputCol, i, pLeftData, false);
+    colDataAppend(pOutputCol, i, value, false);
   }
 }
 
diff --git a/source/libs/scalar/test/scalar/CMakeLists.txt b/source/libs/scalar/test/scalar/CMakeLists.txt
index 6cbac1e25d..03418d17ac 100644
--- a/source/libs/scalar/test/scalar/CMakeLists.txt
+++ b/source/libs/scalar/test/scalar/CMakeLists.txt
@@ -8,11 +8,12 @@ AUX_SOURCE_DIRECTORY(${CMAKE_CURRENT_SOURCE_DIR} SOURCE_LIST)
 ADD_EXECUTABLE(scalarTest ${SOURCE_LIST})
 TARGET_LINK_LIBRARIES(
         scalarTest
-        PUBLIC os util common gtest qcom function nodes scalar
+        PUBLIC os util common gtest qcom function nodes scalar parser
 )
 
 TARGET_INCLUDE_DIRECTORIES(
         scalarTest
         PUBLIC "${TD_SOURCE_DIR}/include/libs/scalar/"
+        PUBLIC "${TD_SOURCE_DIR}/source/libs/parser/inc"
         PRIVATE "${TD_SOURCE_DIR}/source/libs/scalar/inc"
 )
diff --git a/source/libs/scalar/test/scalar/scalarTests.cpp b/source/libs/scalar/test/scalar/scalarTests.cpp
index 975effc63a..8049e80590 100644
--- a/source/libs/scalar/test/scalar/scalarTests.cpp
+++ b/source/libs/scalar/test/scalar/scalarTests.cpp
@@ -125,56 +125,6 @@ void scltMakeValueNode(SNode **pNode, int32_t dataType, void *value) {
   
   *pNode = (SNode *)vnode;
 }
-void scltMakeJsonColumnNode(SNode **pNode, SSDataBlock **block, int32_t dataType, int32_t dataBytes, int32_t rowNum, void *value) {
-  SNode *node = (SNode*)nodesMakeNode(QUERY_NODE_COLUMN);
-  SColumnNode *rnode = (SColumnNode *)node;
-  rnode->node.resType.type = dataType;
-  rnode->node.resType.bytes = dataBytes;
-  rnode->dataBlockId = 0;
-
-  if (NULL == *block) {
-    SSDataBlock *res = (SSDataBlock *)taosMemoryCalloc(1, sizeof(SSDataBlock));
-    res->info.numOfCols = 3;
-    res->info.rows = rowNum;
-    res->pDataBlock = taosArrayInit(3, sizeof(SColumnInfoData));
-    for (int32_t i = 0; i < 2; ++i) {
-      SColumnInfoData idata = {{0}};
-      idata.info.type  = TSDB_DATA_TYPE_NULL;
-      idata.info.bytes = 10;
-      idata.info.colId = i + 1;
-
-      int32_t size = idata.info.bytes * rowNum;
-      idata.pData = (char *)taosMemoryCalloc(1, size);
-      taosArrayPush(res->pDataBlock, &idata);
-    }
-
-    SColumnInfoData idata = {{0}};
-    idata.info.type  = dataType;
-    idata.info.bytes = dataBytes;
-    idata.info.colId = 3;
-    int32_t size = idata.info.bytes * rowNum;
-    idata.pData = (char *)taosMemoryCalloc(1, size);
-    taosArrayPush(res->pDataBlock, &idata);
-
-    blockDataEnsureCapacity(res, rowNum);
-
-    SColumnInfoData *pColumn = (SColumnInfoData *)taosArrayGetLast(res->pDataBlock);
-    for (int32_t i = 0; i < rowNum; ++i) {
-      uint32_t len = pColumnInfoData->varmeta.length;
-      pColumnInfoData->varmeta.offset[i] = len;
-
-      memcpy(pColumnInfoData->pData + len, value, varDataLen(value));
-      pColumnInfoData->varmeta.length += varDataLen(value);
-    }
-
-    rnode->slotId = 2;
-    rnode->colId = 3;
-
-    *block = res;
-  }
-
-  *pNode = (SNode *)rnode;
-}
 
 void scltMakeColumnNode(SNode **pNode, SSDataBlock **block, int32_t dataType, int32_t dataBytes, int32_t rowNum, void *value) {
   SNode *node = (SNode*)nodesMakeNode(QUERY_NODE_COLUMN);
@@ -960,54 +910,77 @@ TEST(constantTest, greater_and_lower) {
   nodesDestroyNode(res);
 }
 
-TEST(columnTest, smallint_value_add_json_column) {
-  scltInitLogFile();
+void makeJsonArrow(SSDataBlock **src, SNode **opNode, void *json, char *key){
+  char key[32] = {0};
+  memcpy(varDataVal(key), key, strlen(key));
+  varDataLen(key) = strlen(key);
 
   SNode *pLeft = NULL, *pRight = NULL, *opNode = NULL;
-  int32_t leftv = 1;
-  char *rightv= "{"k1":4,"k2":"hello","k3":null,"k4":true,"k5":5.44}";
-  char key[10] = {0};
-  memcpy(varDataVal(key), "k1", 2);
-  varDataLen(key) = 2;
+  scltMakeValueNode(&pRight, TSDB_DATA_TYPE_BINARY, key);
+  scltMakeColumnNode(&pLeft, src, TSDB_DATA_TYPE_JSON, varDataLen(json), 1, json);
+  scltMakeOpNode(opNode, OP_TYPE_JSON_GET_VALUE, TSDB_DATA_TYPE_JSON, pLeft, pRight);
+}
+
+void makeOperator(SNode **opNode, SArray *blockList, int32_t resType, EOperatorType opType, int32_t rightType, void *rightData){
+  SNode *right = NULL;
+  scltMakeValueNode(&right, rightType, rightData);
+  scltMakeOpNode(opNode, opType, resType, *opNode, right);
+
+  SColumnInfo colInfo = createColumnInfo(1, resType, tDataTypes[resType].bytes);
+  int16_t dataBlockId = 0, slotId = 0;
+  scltAppendReservedSlot(blockList, &dataBlockId, &slotId, true, 1, &colInfo);
+  scltMakeTargetNode(opNode, dataBlockId, slotId, *opNode);
+}
+
+void makeCalculate(void *json, void *key, void *value, double exceptValue, EOperatorType opType){
+  SArray *blockList = taosArrayInit(2, POINTER_BYTES);
+  SSDataBlock *src = NULL;
+  SNode *opNode = NULL;
+
+  makeJsonArrow(&src, &opNode, row, key);
+  taosArrayPush(blockList, &src);
+
+  makeOperator(&opNode, blockList, TSDB_DATA_TYPE_DOUBLE, opType, TSDB_DATA_TYPE_INT, value);
+
+  int32_t code = scalarCalculate(opNode, blockList, NULL);
+  ASSERT_EQ(code, 0);
+
+  SSDataBlock *res = *(SSDataBlock **)taosArrayGetLast(blockList);
+  ASSERT_EQ(res->info.rows, 1);
+  SColumnInfoData *column = (SColumnInfoData *)taosArrayGetLast(res->pDataBlock);
+  ASSERT_EQ(column->info.type, TSDB_DATA_TYPE_DOUBLE);
+  ASSERT_EQ(*((double *)colDataGetData(column, 0)), exceptValue);
+
+  taosArrayDestroyEx(blockList, scltFreeDataBlock);
+  nodesDestroyNode(opNode);
+}
+
+TEST(columnTest, int_value_add_json_column) {
+  scltInitLogFile();
+  char *rightv= "{\"k1\":4,\"k2\":\"hello\",\"k3\":null,\"k4\":true,\"k5\":5.44}";
+  char *key = "k1";
 
   SKVRowBuilder kvRowBuilder;
   tdInitKVRowBuilder(&kvRowBuilder);
   parseJsontoTagData(rightv, &kvRowBuilder, NULL, 0);
   SKVRow row = tdGetKVRowFromBuilder(&kvRowBuilder);
 
-  double eRes[5] = {1.0, -4, -3, 24, 101};
+  int32_t leftv = 1;
+  makeCalculate(row, key, &leftv, 5.0, OP_TYPE_ADD);
 
-  SSDataBlock *src = NULL;
-  int32_t rowNum = 1;
+  leftv = 8;
+  makeCalculate(row, key, &leftv, -4, OP_TYPE_SUB);
 
-  scltMakeValueNode(&pRight, TSDB_DATA_TYPE_BINARY, key);
-  scltMakeJsonColumnNode(&pLeft, &src, TSDB_DATA_TYPE_JSON, varDataLen(row), rowNum, row);
-  scltMakeOpNode(&opNode, OP_TYPE_JSON_GET_VALUE, TSDB_DATA_TYPE_JSON, pLeft, pRight);
+  leftv = 2;
+  makeCalculate(row, key, &leftv, 8.0, OP_TYPE_MULTI);
 
-  SNode *right = NULL;
-  scltMakeValueNode(&right, TSDB_DATA_TYPE_INT, &leftv);
-  scltMakeOpNode(&opNode, OP_TYPE_ADD, TSDB_DATA_TYPE_DOUBLE, opNode, right);
+  leftv = 2;
+  makeCalculate(row, key, &leftv, 2.0, OP_TYPE_DIV);
 
-  SArray *blockList = taosArrayInit(2, POINTER_BYTES);
-  taosArrayPush(blockList, &src);
-  SColumnInfo colInfo = createColumnInfo(1, TSDB_DATA_TYPE_DOUBLE, sizeof(double));
-  int16_t dataBlockId = 0, slotId = 0;
-  scltAppendReservedSlot(blockList, &dataBlockId, &slotId, true, rowNum, &colInfo);
-  scltMakeTargetNode(&opNode, dataBlockId, slotId, opNode);
+  leftv = 3;
+  makeCalculate(row, key, &leftv, 1.0, OP_TYPE_MODE);
 
-  int32_t code = scalarCalculate(opNode, blockList, NULL);
-  ASSERT_EQ(code, 0);
-
-  SSDataBlock *res = *(SSDataBlock **)taosArrayGetLast(blockList);
-  ASSERT_EQ(res->info.rows, rowNum);
-  SColumnInfoData *column = (SColumnInfoData *)taosArrayGetLast(res->pDataBlock);
-  ASSERT_EQ(column->info.type, TSDB_DATA_TYPE_DOUBLE);
-  for (int32_t i = 0; i < rowNum; ++i) {
-    ASSERT_EQ(*((double *)colDataGetData(column, i)), eRes[i]);
-  }
-
-  taosArrayDestroyEx(blockList, scltFreeDataBlock);
-  nodesDestroyNode(opNode);
+  makeCalculate(row, key, &leftv, -4, OP_TYPE_MINUS);
 }
 
 TEST(columnTest, smallint_value_add_int_column) {

From a016f1157a0abaa2001a7706bdb5ae3d8db9572f Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Wed, 20 Apr 2022 16:52:32 +0800
Subject: [PATCH 12/15] refactor: add unit test for json compare

---
 include/common/tdatablock.h                   |  32 +-
 source/libs/scalar/src/sclvector.c            | 335 +++++++-----------
 .../libs/scalar/test/scalar/scalarTests.cpp   | 235 ++++++++++--
 3 files changed, 358 insertions(+), 244 deletions(-)

diff --git a/include/common/tdatablock.h b/include/common/tdatablock.h
index e40efd815d..1d6f496eea 100644
--- a/include/common/tdatablock.h
+++ b/include/common/tdatablock.h
@@ -57,11 +57,30 @@ SEpSet getEpSet_s(SCorEpSet* pEpSet);
 #define colDataIsNull_var(pColumnInfoData, row) (pColumnInfoData->varmeta.offset[row] == -1)
 #define colDataSetNull_var(pColumnInfoData, row)  (pColumnInfoData->varmeta.offset[row] = -1)
 
+#define BitmapLen(_n) (((_n) + ((1 << NBIT) - 1)) >> NBIT)
+
+#define colDataGetVarData(p1_, r_) ((p1_)->pData + (p1_)->varmeta.offset[(r_)])
+
+#define colDataGetNumData(p1_, r_) ((p1_)->pData + ((r_) * (p1_)->info.bytes))
+// SColumnInfoData, rowNumber
+#define colDataGetData(p1_, r_)                                        \
+  ((IS_VAR_DATA_TYPE((p1_)->info.type)) ?  colDataGetVarData(p1_, r_)  \
+                                        :  colDataGetNumData(p1_, r_))
+
 static FORCE_INLINE bool colDataIsNull_s(const SColumnInfoData* pColumnInfoData, uint32_t row) {
+  if (pColumnInfoData->info.type == TSDB_DATA_TYPE_JSON){
+    if(colDataIsNull_var(pColumnInfoData, row)){
+      return true;
+    }
+    char *data = colDataGetVarData(pColumnInfoData, row);
+    return (*data == TSDB_DATA_TYPE_NULL);
+  } 
+  
   if (!pColumnInfoData->hasNull) {
     return false;
   }
-  if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
+
+  if (pColumnInfoData->info.type== TSDB_DATA_TYPE_VARCHAR || pColumnInfoData->info.type == TSDB_DATA_TYPE_NCHAR) {
     return colDataIsNull_var(pColumnInfoData, row);
   } else {
     if (pColumnInfoData->nullbitmap == NULL) {
@@ -99,16 +118,6 @@ static FORCE_INLINE bool colDataIsNull(const SColumnInfoData* pColumnInfoData, u
   }
 }
 
-#define BitmapLen(_n) (((_n) + ((1 << NBIT) - 1)) >> NBIT)
-
-#define colDataGetVarData(p1_, r_) ((p1_)->pData + (p1_)->varmeta.offset[(r_)])
-
-#define colDataGetNumData(p1_, r_) ((p1_)->pData + ((r_) * (p1_)->info.bytes))
-// SColumnInfoData, rowNumber
-#define colDataGetData(p1_, r_)                                        \
-  ((IS_VAR_DATA_TYPE((p1_)->info.type)) ?  colDataGetVarData(p1_, r_)  \
-                                        :  colDataGetNumData(p1_, r_))
-
 static FORCE_INLINE void colDataAppendNULL(SColumnInfoData* pColumnInfoData, uint32_t currentRow) {
   // There is a placehold for each NULL value of binary or nchar type.
   if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
@@ -271,3 +280,4 @@ static FORCE_INLINE void blockCompressEncode(const SSDataBlock* pBlock, char* da
 #endif
 
 #endif /*_TD_COMMON_EP_H_*/
+
diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c
index c2eafedf9b..ae8c395997 100644
--- a/source/libs/scalar/src/sclvector.c
+++ b/source/libs/scalar/src/sclvector.c
@@ -28,6 +28,75 @@
 #define LEFT_COL ((pLeftCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pLeftCol : pLeftCol->pData))
 #define RIGHT_COL ((pRightCol->info.type == TSDB_DATA_TYPE_JSON ? (void*)pRightCol : pRightCol->pData))
 
+void convertNumberToNumber(const void *inData, void *outData, int8_t inType, int8_t outType){
+  switch (outType) {
+    case TSDB_DATA_TYPE_BOOL: {
+      GET_TYPED_DATA(*((bool *)outData), bool, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_TINYINT: {
+      GET_TYPED_DATA(*((int8_t *)outData), int8_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_SMALLINT: {
+      GET_TYPED_DATA(*((int16_t *)outData), int16_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_INT: {
+      GET_TYPED_DATA(*((int32_t *)outData), int32_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_BIGINT:
+    case TSDB_DATA_TYPE_TIMESTAMP: {
+      GET_TYPED_DATA(*((int64_t *)outData), int64_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_UTINYINT: {
+      GET_TYPED_DATA(*((uint8_t *)outData), uint8_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_USMALLINT: {
+      GET_TYPED_DATA(*((uint16_t *)outData), uint16_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_UINT: {
+      GET_TYPED_DATA(*((uint32_t *)outData), uint32_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_UBIGINT: {
+      GET_TYPED_DATA(*((uint64_t *)outData), uint64_t, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_FLOAT: {
+      GET_TYPED_DATA(*((float *)outData), float, inType, inData);
+      break;
+    }
+    case TSDB_DATA_TYPE_DOUBLE: {
+      GET_TYPED_DATA(*((double *)outData), double, inType, inData);
+      break;
+    }
+    default:{
+      ASSERT(0);
+    }
+  }
+}
+
+void convertStringToDouble(const void *inData, void *outData, int8_t inType, int8_t outType){
+  char *tmp = taosMemoryMalloc(varDataTLen(inData));
+  int len = taosUcs4ToMbs((TdUcs4 *)varDataVal(inData), varDataLen(inData), tmp);
+  if (len < 0) {
+    sclError("castConvert taosUcs4ToMbs error 1");
+  }
+
+  tmp[len] = 0;
+
+  ASSERT(outType == TSDB_DATA_TYPE_DOUBLE);
+  double value = strtod(tmp, NULL);
+
+  *((double *)outData) = value;
+  taosMemoryFreeClear(tmp);
+}
+
 typedef int64_t (*_getBigintValue_fn_t)(void *src, int32_t index);
 
 int64_t getVectorBigintValue_TINYINT(void *src, int32_t index) {
@@ -64,6 +133,20 @@ int64_t getVectorBigintValue_BOOL(void *src, int32_t index) {
   return (int64_t)*((bool *)src + index);
 }
 
+int64_t getVectorBigintValue_JSON(void *src, int32_t index){
+  ASSERT(!colDataIsNull_var(((SColumnInfoData*)src), index));
+  char *data = colDataGetVarData((SColumnInfoData*)src, index);
+  double out = 0;
+  if (*data == TSDB_DATA_TYPE_NULL){
+    return 0;
+  } else if(*data == TSDB_DATA_TYPE_NCHAR) {   // json inner type can not be BINARY
+    convertStringToDouble(data+CHAR_BYTES, &out, *data, TSDB_DATA_TYPE_DOUBLE);
+  } else {
+    convertNumberToNumber(data+CHAR_BYTES, &out, *data, TSDB_DATA_TYPE_DOUBLE);
+  }
+  return (int64_t)out;
+}
+
 _getBigintValue_fn_t getVectorBigintValueFn(int32_t srcType) {
     _getBigintValue_fn_t p = NULL;
     if(srcType==TSDB_DATA_TYPE_TINYINT) {
@@ -90,6 +173,8 @@ _getBigintValue_fn_t getVectorBigintValueFn(int32_t srcType) {
         p = getVectorBigintValue_BIGINT;
     }else if(srcType==TSDB_DATA_TYPE_BOOL) {
         p = getVectorBigintValue_BOOL;
+    }else if(srcType==TSDB_DATA_TYPE_JSON) {
+        p = getVectorBigintValue_JSON;
     }else {
         assert(0);
     }
@@ -200,75 +285,6 @@ static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIn
   taosMemoryFree(t);
 }
 
-void convertNumberToNumber(const void *inData, void *outData, int8_t inType, int8_t outType){
-  switch (outType) {
-    case TSDB_DATA_TYPE_BOOL: {
-      GET_TYPED_DATA(*((bool *)outData), bool, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_TINYINT: {
-      GET_TYPED_DATA(*((int8_t *)outData), int8_t, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_SMALLINT: {
-      GET_TYPED_DATA(*((int16_t *)outData), int16_t, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_INT: {
-      GET_TYPED_DATA(*((int32_t *)outData), int32_t, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_BIGINT:
-    case TSDB_DATA_TYPE_TIMESTAMP: {
-      GET_TYPED_DATA(*((int64_t *)outData), int64_t, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_UTINYINT: {
-      GET_TYPED_DATA(*((uint8_t *)outData), uint8_t, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_USMALLINT: {
-      GET_TYPED_DATA(*((uint16_t *)outData), uint16_t, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_UINT: {
-      GET_TYPED_DATA(*((uint32_t *)outData), uint32_t, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_UBIGINT: {
-      GET_TYPED_DATA(*((uint64_t *)outData), uint64_t, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_FLOAT: {
-      GET_TYPED_DATA(*((float *)outData), float, inType, inData);
-      break;
-    }
-    case TSDB_DATA_TYPE_DOUBLE: {
-      GET_TYPED_DATA(*((double *)outData), double, inType, inData);
-      break;
-    }
-    default:{
-      ASSERT(0);
-    }
-  }
-}
-
-void convertStringToDouble(const void *inData, void *outData, int8_t inType, int8_t outType){
-  char *tmp = taosMemoryMalloc(varDataTLen(inData));
-  int len = taosUcs4ToMbs((TdUcs4 *)varDataVal(inData), varDataLen(inData), tmp);
-  if (len < 0) {
-    sclError("castConvert taosUcs4ToMbs error 1");
-  }
-
-  tmp[len] = 0;
-
-  ASSERT(outType == TSDB_DATA_TYPE_DOUBLE);
-  double value = strtod(tmp, NULL);
-
-  *((double *)outData) = value;
-  taosMemoryFreeClear(tmp);
-}
-
 //TODO opt performance, tmp is not needed.
 int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, int32_t inType, int32_t outType) {
   ASSERT(!IS_VAR_DATA_TYPE(outType));
@@ -345,12 +361,13 @@ int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, in
 }
 
 double getVectorDoubleValue_JSON(void *src, int32_t index){
-  ASSERT(!colDataIsNull_var(((SColumnInfoData*)src), index));
   char *data = colDataGetVarData((SColumnInfoData*)src, index);
   double out = 0;
-  if(*data == TSDB_DATA_TYPE_NCHAR) {   // json inner type can not be BINARY
+  if (*data == TSDB_DATA_TYPE_NULL){
+    return out;
+  } else if(*data == TSDB_DATA_TYPE_NCHAR) {   // json inner type can not be BINARY
     convertStringToDouble(data+CHAR_BYTES, &out, *data, TSDB_DATA_TYPE_DOUBLE);
-  } {
+  } else {
     convertNumberToNumber(data+CHAR_BYTES, &out, *data, TSDB_DATA_TYPE_DOUBLE);
   }
   return out;
@@ -569,24 +586,24 @@ int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut) {
 }
 
 int8_t gConvertTypes[TSDB_DATA_TYPE_BLOB+1][TSDB_DATA_TYPE_BLOB+1] = {
-/*         NULL BOOL TINY SMAL INT  BIG  FLOA DOUB VARC TIME NCHA UTIN USMA UINT UBIG VARB JSON DECI BLOB */
+/*         NULL BOOL TINY SMAL INT  BIG  FLOA DOUB VARC TIME NCHA UTIN USMA UINT UBIG JSON VARB DECI BLOB */
 /*NULL*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
-/*BOOL*/   0,   0,   0,   3,   4,   5,   6,   7,   7,   9,   7,   0,   12,  13,  14,  7,   0,   0,   0,
-/*TINY*/   0,   0,   0,   3,   4,   5,   6,   7,   7,   9,   7,   3,   4,   5,   7,   7,   0,   0,   0,
-/*SMAL*/   0,   0,   0,   0,   4,   5,   6,   7,   7,   9,   7,   3,   4,   5,   7,   7,   0,   0,   0,
-/*INT */   0,   0,   0,   0,   0,   5,   6,   7,   7,   9,   7,   4,   4,   5,   7,   7,   0,   0,   0,
-/*BIGI*/   0,   0,   0,   0,   0,   0,   6,   7,   7,   0,   7,   5,   5,   5,   7,   7,   0,   0,   0,
-/*FLOA*/   0,   0,   0,   0,   0,   0,   0,   7,   7,   6,   7,   6,   6,   6,   6,   7,   0,   0,   0,
-/*DOUB*/   0,   0,   0,   0,   0,   0,   0,   0,   7,   7,   7,   7,   7,   7,   7,   7,   0,   0,   0,
+/*BOOL*/   0,   0,   0,   3,   4,   5,   6,   7,   7,   9,   7,   0,   12,  13,  14,  0,   7,   0,   0,
+/*TINY*/   0,   0,   0,   3,   4,   5,   6,   7,   7,   9,   7,   3,   4,   5,   7,   0,   7,   0,   0,
+/*SMAL*/   0,   0,   0,   0,   4,   5,   6,   7,   7,   9,   7,   3,   4,   5,   7,   0,   7,   0,   0,
+/*INT */   0,   0,   0,   0,   0,   5,   6,   7,   7,   9,   7,   4,   4,   5,   7,   0,   7,   0,   0,
+/*BIGI*/   0,   0,   0,   0,   0,   0,   6,   7,   7,   0,   7,   5,   5,   5,   7,   0,   7,   0,   0,
+/*FLOA*/   0,   0,   0,   0,   0,   0,   0,   7,   7,   6,   7,   6,   6,   6,   6,   0,   7,   0,   0,
+/*DOUB*/   0,   0,   0,   0,   0,   0,   0,   0,   7,   7,   7,   7,   7,   7,   7,   0,   7,   0,   0,
 /*VARC*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   7,   0,   7,   7,   7,   7,   0,   0,   0,   0,
-/*TIME*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   7,   9,   9,   9,   7,   7,   0,   0,   0,
+/*TIME*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   7,   9,   9,   9,   7,   0,   7,   0,   0,
 /*NCHA*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   7,   7,   7,   7,   0,   0,   0,   0,
-/*UTIN*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   12,  13,  14,  7,   0,   0,   0,
-/*USMA*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   13,  14,  7,   0,   0,   0,
-/*UINT*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   14,  7,   0,   0,   0,
-/*UBIG*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   7,   0,   0,   0,
-/*VARB*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
+/*UTIN*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   12,  13,  14,  0,   7,   0,   0,
+/*USMA*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   13,  14,  0,   7,   0,   0,
+/*UINT*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   14,  0,   7,   0,   0,
+/*UBIG*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   7,   0,   0,
 /*JSON*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
+/*VARB*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
 /*DECI*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
 /*BLOB*/   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0
 };
@@ -677,7 +694,7 @@ enum {
 static int32_t doConvertHelper(SScalarParam* pDest, int32_t* convert, const SScalarParam* pParam, int32_t type) {
   SColumnInfoData* pCol = pParam->columnData;
 
-  if (IS_VAR_DATA_TYPE(pCol->info.type) || pCol->info.type != TSDB_DATA_TYPE_JSON) {
+  if (IS_VAR_DATA_TYPE(pCol->info.type) && pCol->info.type != TSDB_DATA_TYPE_JSON) {
     pDest->numOfRows = pParam->numOfRows;
 
     SDataType t = {.type = type, .bytes = tDataTypes[type].bytes};
@@ -718,10 +735,6 @@ static void vectorMathAddHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRig
       *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
                 + getVectorDoubleValueFnRight(RIGHT_COL, 0);
     }
-    pOutputCol->hasNull = pLeftCol->hasNull;
-    if (pOutputCol->hasNull) {
-      memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(numOfRows));
-    }
   }
 }
 
@@ -735,12 +748,12 @@ static void vectorMathBigintAddHelper(SColumnInfoData* pLeftCol, SColumnInfoData
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
+      if (colDataIsNull_s(pLeftCol, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
       *output = getVectorBigintValueFnLeft(pLeftCol->pData, i) + getVectorBigintValueFnRight(pRightCol->pData, 0);
     }
-    pOutputCol->hasNull = pLeftCol->hasNull;
-    if (pOutputCol->hasNull) {
-      memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(numOfRows));
-    }
   }
 }
 
@@ -771,8 +784,11 @@ char *getJsonValue(char *json, char *key){    //todo
   for (int i = 0; i < cols; ++i) {
     SColIdx *pColIdx = kvRowColIdxAt(json, i);
     char *data = kvRowColVal(json, pColIdx);
-    if(i == 0 && *data == TSDB_DATA_TYPE_NULL){
-      return NULL;
+    if(i == 0){
+      if(*data == TSDB_DATA_TYPE_NULL) {
+        return NULL;
+      }
+      continue;
     }
     if(memcmp(key, data, varDataTLen(data)) == 0){
       return data + varDataTLen(data);
@@ -798,7 +814,7 @@ void vectorJsonArrow(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pO
     }
     char *pLeftData = colDataGetVarData(pLeft->columnData, i);
     char *value = getJsonValue(pLeftData, pRightData);
-    if (!value || *value == TSDB_DATA_TYPE_NULL) {
+    if (!value) {
       colDataSetNull_var(pOutputCol, i);
       pOutputCol->hasNull = true;
       continue;
@@ -835,14 +851,6 @@ void vectorMathAdd(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
         }
         *output = getVectorBigintValueFnLeft(pLeftCol->pData, i) + getVectorBigintValueFnRight(pRightCol->pData, i);
       }
-      pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
-      if (pOutputCol->hasNull) {
-        int32_t numOfBitLen = BitmapLen(pLeft->numOfRows);
-        for (int32_t j = 0; j < numOfBitLen; ++j) {
-          pOutputCol->nullbitmap[j] = pLeftCol->nullbitmap[j] | pRightCol->nullbitmap[j];
-        }
-      }
-
     } else if (pLeft->numOfRows == 1) {
       vectorMathBigintAddHelper(pRightCol, pLeftCol, pOutputCol, pRight->numOfRows, step, i);
     } else if (pRight->numOfRows == 1) {
@@ -861,15 +869,6 @@ void vectorMathAdd(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
         }
         *output = getVectorDoubleValueFnLeft(LEFT_COL, i) + getVectorDoubleValueFnRight(RIGHT_COL, i);
       }
-
-      pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
-      if (pOutputCol->hasNull) {
-        int32_t numOfBitLen = BitmapLen(pLeft->numOfRows);
-        for (int32_t j = 0; j < numOfBitLen; ++j) {
-          pOutputCol->nullbitmap[j] = pLeftCol->nullbitmap[j] | pRightCol->nullbitmap[j];
-        }
-      }
-
     } else if (pLeft->numOfRows == 1) {
       vectorMathAddHelper(pRightCol, pLeftCol, pOutputCol, pRight->numOfRows, step, i);
     } else if (pRight->numOfRows == 1) {
@@ -899,10 +898,6 @@ static void vectorMathSubHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRig
       *output = (getVectorDoubleValueFnLeft(LEFT_COL, i)
                  - getVectorDoubleValueFnRight(RIGHT_COL, 0)) * factor;
     }
-    pOutputCol->hasNull = pLeftCol->hasNull;
-    if (pOutputCol->hasNull) {
-      memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(numOfRows));
-    }
   }
 }
 
@@ -916,12 +911,12 @@ static void vectorMathBigintSubHelper(SColumnInfoData* pLeftCol, SColumnInfoData
     colDataAppendNNULL(pOutputCol, 0, numOfRows);
   } else {
     for (; i >= 0 && i < numOfRows; i += step, output += 1) {
+      if (colDataIsNull_s(pLeftCol, i)) {
+        colDataAppendNULL(pOutputCol, i);
+        continue;  // TODO set null or ignore
+      }
       *output = (getVectorBigintValueFnLeft(pLeftCol->pData, i) - getVectorBigintValueFnRight(pRightCol->pData, 0)) * factor;
     }
-    pOutputCol->hasNull = pLeftCol->hasNull;
-    if (pOutputCol->hasNull) {
-      memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(numOfRows));
-    }
   }
 }
 
@@ -951,15 +946,6 @@ void vectorMathSub(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
         }
         *output = getVectorBigintValueFnLeft(pLeftCol->pData, i) - getVectorBigintValueFnRight(pRightCol->pData, i);
       }
-
-      pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
-      if (pOutputCol->hasNull) {
-        int32_t numOfBitLen = BitmapLen(pLeft->numOfRows);
-        for (int32_t j = 0; j < numOfBitLen; ++j) {
-          pOutputCol->nullbitmap[j] = pLeftCol->nullbitmap[j] | pRightCol->nullbitmap[j];
-        }
-      }
-
     } else if (pLeft->numOfRows == 1) {
       vectorMathBigintSubHelper(pRightCol, pLeftCol, pOutputCol, pRight->numOfRows, step, -1, i);
     } else if (pRight->numOfRows == 1) {
@@ -978,15 +964,6 @@ void vectorMathSub(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut
         }
         *output = getVectorDoubleValueFnLeft(LEFT_COL, i) - getVectorDoubleValueFnRight(RIGHT_COL, i);
       }
-
-      pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
-      if (pOutputCol->hasNull) {
-        int32_t numOfBitLen = BitmapLen(pLeft->numOfRows);
-        for (int32_t j = 0; j < numOfBitLen; ++j) {
-          pOutputCol->nullbitmap[j] = pLeftCol->nullbitmap[j] | pRightCol->nullbitmap[j];
-        }
-      }
-
     } else if (pLeft->numOfRows == 1) {
       vectorMathSubHelper(pRightCol, pLeftCol, pOutputCol, pRight->numOfRows, step, -1, i);
     } else if (pRight->numOfRows == 1) {
@@ -1016,10 +993,6 @@ static void vectorMathMultiplyHelper(SColumnInfoData* pLeftCol, SColumnInfoData*
       *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
                 * getVectorDoubleValueFnRight(RIGHT_COL, 0);
     }
-    pOutputCol->hasNull = pLeftCol->hasNull;
-    if (pOutputCol->hasNull) {
-      memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(numOfRows));
-    }
   }
 }
 
@@ -1047,15 +1020,6 @@ void vectorMathMultiply(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
       *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
                 * getVectorDoubleValueFnRight(RIGHT_COL, i);
     }
-
-    pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
-    if (pOutputCol->hasNull) {
-      int32_t numOfBitLen = BitmapLen(pLeft->numOfRows);
-      for (int32_t j = 0; j < numOfBitLen; ++j) {
-        pOutputCol->nullbitmap[j] = pLeftCol->nullbitmap[j] | pRightCol->nullbitmap[j];
-      }
-    }
-
   } else if (pLeft->numOfRows == 1) {
     vectorMathMultiplyHelper(pRightCol, pLeftCol, pOutputCol, pRight->numOfRows, step, i);
   } else if (pRight->numOfRows == 1) {
@@ -1090,15 +1054,6 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
       *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
                  /getVectorDoubleValueFnRight(RIGHT_COL, i);
     }
-
-    pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
-    if (pOutputCol->hasNull) {
-      int32_t numOfBitLen = BitmapLen(pLeft->numOfRows);
-      for (int32_t j = 0; j < numOfBitLen; ++j) {
-        pOutputCol->nullbitmap[j] = pLeftCol->nullbitmap[j] | pRightCol->nullbitmap[j];
-      }
-    }
-
   } else if (pLeft->numOfRows == 1) {
     if (colDataIsNull_s(pLeftCol, 0)) {  // Set pLeft->numOfRows NULL value
       colDataAppendNNULL(pOutputCol, 0, pRight->numOfRows);
@@ -1111,10 +1066,6 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
         *output = getVectorDoubleValueFnLeft(LEFT_COL, 0)
                   / getVectorDoubleValueFnRight(RIGHT_COL, i);
       }
-      pOutputCol->hasNull = pRightCol->hasNull;
-      if (pOutputCol->hasNull) {
-        memcpy(pOutputCol->nullbitmap, pRightCol->nullbitmap, BitmapLen(pRight->numOfRows));
-      }
     }
   } else if (pRight->numOfRows == 1) {
     if (colDataIsNull_s(pRightCol, 0)) {  // Set pLeft->numOfRows NULL value
@@ -1128,10 +1079,6 @@ void vectorMathDivide(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *p
         *output = getVectorDoubleValueFnLeft(LEFT_COL, i)
                   / getVectorDoubleValueFnRight(RIGHT_COL, 0);
       }
-      pOutputCol->hasNull = pLeftCol->hasNull;
-      if (pOutputCol->hasNull) {
-        memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(pLeft->numOfRows));
-      }
     }
   }
 
@@ -1240,11 +1187,6 @@ void vectorMathMinus(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pO
     *output = - getVectorDoubleValueFnLeft(LEFT_COL, i);
   }
 
-  pOutputCol->hasNull = pLeftCol->hasNull;
-  if (pOutputCol->hasNull) {
-    memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(pLeft->numOfRows));
-  }
-
   doReleaseVec(pLeftCol,  leftConvert);
 }
 
@@ -1315,10 +1257,6 @@ static void vectorBitAndHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRigh
       }
       *output = getVectorBigintValueFnLeft(LEFT_COL, i) & getVectorBigintValueFnRight(RIGHT_COL, 0);
     }
-    pOutputCol->hasNull = pLeftCol->hasNull;
-    if (pOutputCol->hasNull) {
-      memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(numOfRows));
-    }
   }
 }
 
@@ -1345,15 +1283,6 @@ void vectorBitAnd(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut,
       }
       *output = getVectorBigintValueFnLeft(LEFT_COL, i) & getVectorBigintValueFnRight(RIGHT_COL, i);
     }
-
-    pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
-    if (pOutputCol->hasNull) {
-      int32_t numOfBitLen = BitmapLen(pLeft->numOfRows);
-      for (int32_t j = 0; j < numOfBitLen; ++j) {
-        pOutputCol->nullbitmap[j] = pLeftCol->nullbitmap[j] & pRightCol->nullbitmap[j];
-      }
-    }
-
   } else if (pLeft->numOfRows == 1) {
     vectorBitAndHelper(pRightCol, pLeftCol, pOutputCol, pRight->numOfRows, step, i);
   } else if (pRight->numOfRows == 1) {
@@ -1381,10 +1310,6 @@ static void vectorBitOrHelper(SColumnInfoData* pLeftCol, SColumnInfoData* pRight
       }
       *output = getVectorBigintValueFnLeft(LEFT_COL, i) | rx;
     }
-    pOutputCol->hasNull = pLeftCol->hasNull;
-    if (pOutputCol->hasNull) {
-      memcpy(pOutputCol->nullbitmap, pLeftCol->nullbitmap, BitmapLen(numOfRows));
-    }
   }
 }
 
@@ -1411,14 +1336,6 @@ void vectorBitOr(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut,
       }
       *output = getVectorBigintValueFnLeft(LEFT_COL, i) | getVectorBigintValueFnRight(RIGHT_COL, i);
     }
-
-    pOutputCol->hasNull = (pLeftCol->hasNull || pRightCol->hasNull);
-    if (pOutputCol->hasNull) {
-      int32_t numOfBitLen = BitmapLen(pLeft->numOfRows);
-      for (int32_t j = 0; j < numOfBitLen; ++j) {
-        pOutputCol->nullbitmap[j] = pLeftCol->nullbitmap[j] | pRightCol->nullbitmap[j];
-      }
-    }
   } else if (pLeft->numOfRows == 1) {
     vectorBitOrHelper(pRightCol, pLeftCol, pOutputCol, pRight->numOfRows, step, i);
   } else if (pRight->numOfRows == 1) {
@@ -1439,6 +1356,7 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
   if (pRight->pHashFilter != NULL) {
     for (; i >= 0 && i < pLeft->numOfRows; i += step) {
       if (colDataIsNull_s(pLeft->columnData, i)) {
+        colDataAppendNULL(pOut->columnData, i);
         continue;
       }
 
@@ -1452,6 +1370,7 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
   if (pLeft->numOfRows == pRight->numOfRows) {
     for (; i < pRight->numOfRows && i >= 0; i += step) {
       if (colDataIsNull_s(pLeft->columnData, i) || colDataIsNull_s(pRight->columnData, i)) {
+        colDataAppendNULL(pOut->columnData, i);
         continue;  // TODO set null or ignore
       }
 
@@ -1463,6 +1382,7 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
       bool isJsonnull = false;
       convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut, &isJsonnull);
       if(isJsonnull){
+        colDataAppendNULL(pOut->columnData, i);
         continue;  // TODO set null or ignore
       }
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
@@ -1472,6 +1392,7 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
     ASSERT(pLeft->pHashFilter == NULL);
     for (; i >= 0 && i < pLeft->numOfRows; i += step) {
       if (colDataIsNull_s(pLeft->columnData, i)) {
+        colDataAppendNULL(pOut->columnData, i);
         continue;
       }
 
@@ -1482,6 +1403,7 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
       bool isJsonnull = false;
       convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut, &isJsonnull);
       if(isJsonnull){
+        colDataAppendNULL(pOut->columnData, i);
         continue;  // TODO set null or ignore
       }
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
@@ -1490,6 +1412,7 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
   } else if (pLeft->numOfRows == 1) {
     for (; i >= 0 && i < pRight->numOfRows; i += step) {
       if (colDataIsNull_s(pRight->columnData, i)) {
+        colDataAppendNULL(pOut->columnData, i);
         continue;
       }
 
@@ -1500,6 +1423,7 @@ void vectorCompareImpl(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *
       bool isJsonnull = false;
       convertJsonValue(&fp, optr, GET_PARAM_TYPE(pLeft), GET_PARAM_TYPE(pRight), &pLeftData, &pRightData, &leftOut, &rightOut, &isJsonnull);
       if(isJsonnull){
+        colDataAppendNULL(pOut->columnData, i);
         continue;  // TODO set null or ignore
       }
       bool  res = filterDoCompare(fp, optr, pLeftData, pRightData);
@@ -1589,10 +1513,6 @@ void vectorJsonContains(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam
 void vectorIsNull(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
   for(int32_t i = 0; i < pLeft->numOfRows; ++i) {
     int8_t v = colDataIsNull_s(pLeft->columnData, i)? 1:0;
-    if (v && pLeft->columnData->info.type == TSDB_DATA_TYPE_JSON){
-      char *data = colDataGetVarData(pLeft->columnData, i);
-      v = (*data == TSDB_DATA_TYPE_NULL)? 1:0;
-    }
     colDataAppendInt8(pOut->columnData, i, &v);
   }
   pOut->numOfRows = pLeft->numOfRows;
@@ -1601,10 +1521,6 @@ void vectorIsNull(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut,
 void vectorNotNull(SScalarParam* pLeft, SScalarParam* pRight, SScalarParam *pOut, int32_t _ord) {
   for(int32_t i = 0; i < pLeft->numOfRows; ++i) {
     int8_t v = colDataIsNull_s(pLeft->columnData, i)? 0:1;
-    if (v && pLeft->columnData->info.type == TSDB_DATA_TYPE_JSON){
-      char *data = colDataGetVarData(pLeft->columnData, i);
-      v = (*data == TSDB_DATA_TYPE_NULL)? 0:1;
-    }
     colDataAppendInt8(pOut->columnData, i, &v);
   }
   pOut->numOfRows = pLeft->numOfRows;
@@ -1671,3 +1587,4 @@ _bin_scalar_fn_t getBinScalarOperatorFn(int32_t binFunctionId) {
       return NULL;
   }
 }
+
diff --git a/source/libs/scalar/test/scalar/scalarTests.cpp b/source/libs/scalar/test/scalar/scalarTests.cpp
index 8049e80590..502ab7edfe 100644
--- a/source/libs/scalar/test/scalar/scalarTests.cpp
+++ b/source/libs/scalar/test/scalar/scalarTests.cpp
@@ -38,6 +38,7 @@
 #include "scalar.h"
 #include "nodes.h"
 #include "tlog.h"
+#include "parUtil.h"
 
 #define _DEBUG_PRINT_ 0
 
@@ -911,17 +912,32 @@ TEST(constantTest, greater_and_lower) {
 }
 
 void makeJsonArrow(SSDataBlock **src, SNode **opNode, void *json, char *key){
-  char key[32] = {0};
-  memcpy(varDataVal(key), key, strlen(key));
-  varDataLen(key) = strlen(key);
+  char keyVar[32] = {0};
+  memcpy(varDataVal(keyVar), key, strlen(key));
+  varDataLen(keyVar) = strlen(key);
 
-  SNode *pLeft = NULL, *pRight = NULL, *opNode = NULL;
-  scltMakeValueNode(&pRight, TSDB_DATA_TYPE_BINARY, key);
+  SNode *pLeft = NULL, *pRight = NULL;
+  scltMakeValueNode(&pRight, TSDB_DATA_TYPE_BINARY, keyVar);
   scltMakeColumnNode(&pLeft, src, TSDB_DATA_TYPE_JSON, varDataLen(json), 1, json);
   scltMakeOpNode(opNode, OP_TYPE_JSON_GET_VALUE, TSDB_DATA_TYPE_JSON, pLeft, pRight);
 }
 
-void makeOperator(SNode **opNode, SArray *blockList, int32_t resType, EOperatorType opType, int32_t rightType, void *rightData){
+void makeOperator(SNode **opNode, SArray *blockList, EOperatorType opType, int32_t rightType, void *rightData){
+  int32_t resType = TSDB_DATA_TYPE_NULL;
+  if(opType == OP_TYPE_ADD || opType == OP_TYPE_SUB || opType == OP_TYPE_MULTI ||
+      opType == OP_TYPE_DIV || opType == OP_TYPE_MOD || opType == OP_TYPE_MINUS){
+    resType = TSDB_DATA_TYPE_DOUBLE;
+  }else if(opType == OP_TYPE_BIT_AND || opType == OP_TYPE_BIT_OR){
+    resType = TSDB_DATA_TYPE_BIGINT;
+  }else if(opType == OP_TYPE_GREATER_THAN || opType == OP_TYPE_GREATER_EQUAL ||
+             opType == OP_TYPE_LOWER_THAN || opType == OP_TYPE_LOWER_EQUAL ||
+             opType == OP_TYPE_EQUAL || opType == OP_TYPE_NOT_EQUAL ||
+             opType == OP_TYPE_IS_NULL || opType == OP_TYPE_IS_NOT_NULL || opType == OP_TYPE_IS_TRUE ||
+             opType == OP_TYPE_LIKE || opType == OP_TYPE_NOT_LIKE || opType == OP_TYPE_MATCH ||
+             opType == OP_TYPE_NMATCH){
+    resType = TSDB_DATA_TYPE_BOOL;
+  }
+
   SNode *right = NULL;
   scltMakeValueNode(&right, rightType, rightData);
   scltMakeOpNode(opNode, opType, resType, *opNode, right);
@@ -932,15 +948,15 @@ void makeOperator(SNode **opNode, SArray *blockList, int32_t resType, EOperatorT
   scltMakeTargetNode(opNode, dataBlockId, slotId, *opNode);
 }
 
-void makeCalculate(void *json, void *key, void *value, double exceptValue, EOperatorType opType){
+void makeCalculate(void *json, void *key, int32_t rightType, void *rightData, double exceptValue, EOperatorType opType){
   SArray *blockList = taosArrayInit(2, POINTER_BYTES);
   SSDataBlock *src = NULL;
   SNode *opNode = NULL;
 
-  makeJsonArrow(&src, &opNode, row, key);
+  makeJsonArrow(&src, &opNode, json, (char*)key);
   taosArrayPush(blockList, &src);
 
-  makeOperator(&opNode, blockList, TSDB_DATA_TYPE_DOUBLE, opType, TSDB_DATA_TYPE_INT, value);
+  makeOperator(&opNode, blockList, opType, rightType, rightData);
 
   int32_t code = scalarCalculate(opNode, blockList, NULL);
   ASSERT_EQ(code, 0);
@@ -948,39 +964,209 @@ void makeCalculate(void *json, void *key, void *value, double exceptValue, EOper
   SSDataBlock *res = *(SSDataBlock **)taosArrayGetLast(blockList);
   ASSERT_EQ(res->info.rows, 1);
   SColumnInfoData *column = (SColumnInfoData *)taosArrayGetLast(res->pDataBlock);
-  ASSERT_EQ(column->info.type, TSDB_DATA_TYPE_DOUBLE);
-  ASSERT_EQ(*((double *)colDataGetData(column, 0)), exceptValue);
+
+  if(colDataIsNull_f(column->nullbitmap, 0)){
+    ASSERT_EQ(DBL_MAX, exceptValue);
+    printf("result:NULL\n");
+
+  }else if(opType == OP_TYPE_ADD || opType == OP_TYPE_SUB || opType == OP_TYPE_MULTI || opType == OP_TYPE_DIV ||
+             opType == OP_TYPE_MOD || opType == OP_TYPE_MINUS){
+    double tmp = *((double *)colDataGetData(column, 0));
+    ASSERT_TRUE(tmp == exceptValue);
+    printf("result:%lf\n", tmp);
+  }else if(opType == OP_TYPE_BIT_AND || opType == OP_TYPE_BIT_OR){
+    ASSERT_EQ(*((int64_t *)colDataGetData(column, 0)), exceptValue);
+    printf("result:%ld\n", *((int64_t *)colDataGetData(column, 0)));
+  }else if(opType == OP_TYPE_GREATER_THAN || opType == OP_TYPE_GREATER_EQUAL || opType == OP_TYPE_LOWER_THAN ||
+             opType == OP_TYPE_LOWER_EQUAL || opType == OP_TYPE_EQUAL || opType == OP_TYPE_NOT_EQUAL ||
+             opType == OP_TYPE_IS_NULL || opType == OP_TYPE_IS_NOT_NULL || opType == OP_TYPE_IS_TRUE ||
+             opType == OP_TYPE_LIKE || opType == OP_TYPE_NOT_LIKE || opType == OP_TYPE_MATCH || opType == OP_TYPE_NMATCH){
+    ASSERT_EQ(*((bool *)colDataGetData(column, 0)), exceptValue);
+    printf("result:%d\n", *((bool *)colDataGetData(column, 0)));
+  }
 
   taosArrayDestroyEx(blockList, scltFreeDataBlock);
   nodesDestroyNode(opNode);
 }
 
-TEST(columnTest, int_value_add_json_column) {
+TEST(columnTest, json_column_arith_op) {
   scltInitLogFile();
   char *rightv= "{\"k1\":4,\"k2\":\"hello\",\"k3\":null,\"k4\":true,\"k5\":5.44}";
-  char *key = "k1";
 
   SKVRowBuilder kvRowBuilder;
   tdInitKVRowBuilder(&kvRowBuilder);
   parseJsontoTagData(rightv, &kvRowBuilder, NULL, 0);
   SKVRow row = tdGetKVRowFromBuilder(&kvRowBuilder);
 
-  int32_t leftv = 1;
-  makeCalculate(row, key, &leftv, 5.0, OP_TYPE_ADD);
+  const int32_t len = 8;
+  EOperatorType op[len] = {OP_TYPE_ADD, OP_TYPE_SUB, OP_TYPE_MULTI, OP_TYPE_DIV,
+                         OP_TYPE_MOD, OP_TYPE_MINUS, OP_TYPE_BIT_AND, OP_TYPE_BIT_OR};
+  int32_t input[len] = {1, 8, 2, 2, 3, 0, -4, 9};
 
-  leftv = 8;
-  makeCalculate(row, key, &leftv, -4, OP_TYPE_SUB);
+  printf("--------------------json int---------------------\n");
+  char *key = "k1";
+  double eRes[len] = {5.0, -4, 8.0, 2.0, 1.0, -4, 4&-4, 4|9};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes[i], op[i]);
+  }
 
-  leftv = 2;
-  makeCalculate(row, key, &leftv, 8.0, OP_TYPE_MULTI);
+  printf("--------------------json string---------------------\n");
 
-  leftv = 2;
-  makeCalculate(row, key, &leftv, 2.0, OP_TYPE_DIV);
+  key = "k2";
+  double eRes1[len] = {1.0, -8, 0, 0, 0, 0, 0, 9};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes1[i], op[i]);
+  }
 
-  leftv = 3;
-  makeCalculate(row, key, &leftv, 1.0, OP_TYPE_MODE);
+  printf("---------------------json null--------------------\n");
 
-  makeCalculate(row, key, &leftv, -4, OP_TYPE_MINUS);
+  key = "k3";
+  double eRes2[len] = {DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes2[i], op[i]);
+  }
+
+  printf("---------------------json bool--------------------\n");
+
+  key = "k4";
+  double eRes3[len] = {2.0, -7, 2, 0.5, 1, -1, 1&-4, 1|9};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes3[i], op[i]);
+  }
+
+  printf("----------------------json double-------------------\n");
+
+  key = "k5";
+  double eRes4[len] = {6.44, -2.56, 10.88, 2.72, 2.44, -5.44, 5&-4, 5|9};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes4[i], op[i]);
+  }
+
+  printf("---------------------json not exist--------------------\n");
+
+  key = "k10";
+  double eRes5[len] = {DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes5[i], op[i]);
+  }
+}
+
+void *prepareNchar(char* rightData){
+  int32_t len = 0;
+  int32_t inputLen = strlen(rightData);
+
+  char* t = (char*)taosMemoryCalloc(1,(inputLen + 1) * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE);
+  taosMbsToUcs4(rightData, inputLen, (TdUcs4*) varDataVal(t), inputLen * TSDB_NCHAR_SIZE, &len);
+  varDataSetLen(t, len);
+  return t;
+}
+
+TEST(columnTest, json_column_logic_op) {
+  scltInitLogFile();
+  char *rightv= "{\"k1\":4,\"k2\":\"hello\",\"k3\":null,\"k4\":true,\"k5\":5.44,\"k6\":\"6.6hello\"}";
+
+  SKVRowBuilder kvRowBuilder;
+  tdInitKVRowBuilder(&kvRowBuilder);
+  parseJsontoTagData(rightv, &kvRowBuilder, NULL, 0);
+  SKVRow row = tdGetKVRowFromBuilder(&kvRowBuilder);
+
+  const int32_t len = 9;
+  const int32_t len1 = 4;
+  EOperatorType op[len+len1] = {OP_TYPE_GREATER_THAN, OP_TYPE_GREATER_EQUAL, OP_TYPE_LOWER_THAN, OP_TYPE_LOWER_EQUAL, OP_TYPE_EQUAL, OP_TYPE_NOT_EQUAL,
+                          OP_TYPE_IS_NULL, OP_TYPE_IS_NOT_NULL, OP_TYPE_IS_TRUE, OP_TYPE_LIKE, OP_TYPE_NOT_LIKE, OP_TYPE_MATCH, OP_TYPE_NMATCH};
+
+  int32_t input[len] = {1, 8, 2, 2, 3, 0, 0, 0, 0};
+  char *inputNchar[len1] = {"hell_", "hel%", "hell", "llll"};
+
+  printf("--------------------json int---------------------\n");
+  char *key = "k1";
+  bool eRes[len+len1] = {true, false, false, false, false, true, false, true, true, false, false, false, false};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes[i], op[i]);
+  }
+  for(int i = len; i < len + len1; i++){
+    void* rightData = prepareNchar(inputNchar[i-len]);
+    makeCalculate(row, key, TSDB_DATA_TYPE_NCHAR, rightData, eRes[i], op[i]);
+    taosMemoryFree(rightData);
+  }
+
+  printf("--------------------json string---------------------\n");
+
+  key = "k2";
+  bool eRes1[len+len1] = {false, false, true, true, false, false, false, true, false, true, false, true, true};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes1[i], op[i]);
+  }
+  for(int i = len; i < len + len1; i++){
+    void* rightData = prepareNchar(inputNchar[i-len]);
+    makeCalculate(row, key, TSDB_DATA_TYPE_NCHAR, rightData, eRes1[i], op[i]);
+    taosMemoryFree(rightData);
+  }
+
+  printf("--------------------json null---------------------\n");
+
+  key = "k3";
+  double eRes2[len+len1] = {DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, true, false, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes2[i], op[i]);
+  }
+  for(int i = len; i < len + len1; i++){
+    void* rightData = prepareNchar(inputNchar[i-len]);
+    makeCalculate(row, key, TSDB_DATA_TYPE_NCHAR, rightData, eRes2[i], op[i]);
+    taosMemoryFree(rightData);
+  }
+
+  printf("--------------------json bool---------------------\n");
+
+  key = "k4";
+  bool eRes3[len+len1] = {false, false, true, true, false, true, false, true, true, false, false, false, false};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes3[i], op[i]);
+  }
+  for(int i = len; i < len + len1; i++){
+    void* rightData = prepareNchar(inputNchar[i-len]);
+    makeCalculate(row, key, TSDB_DATA_TYPE_NCHAR, rightData, eRes3[i], op[i]);
+    taosMemoryFree(rightData);
+  }
+
+  printf("--------------------json double---------------------\n");
+
+  key = "k5";
+  bool eRes4[len+len1] = {true, false, false, false, false, true, false, true, true, false, false, false, false};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes4[i], op[i]);
+  }
+  for(int i = len; i < len + len1; i++){
+    void* rightData = prepareNchar(inputNchar[i-len]);
+    makeCalculate(row, key, TSDB_DATA_TYPE_NCHAR, rightData, eRes4[i], op[i]);
+    taosMemoryFree(rightData);
+  }
+
+  printf("--------------------json double---------------------\n");
+
+  key = "k6";
+  bool eRes5[len+len1] = {true, false, false, false, false, true, false, true, true, false, false, false, true};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes5[i], op[i]);
+  }
+  for(int i = len; i < len + len1; i++){
+    void* rightData = prepareNchar(inputNchar[i-len]);
+    makeCalculate(row, key, TSDB_DATA_TYPE_NCHAR, rightData, eRes5[i], op[i]);
+    taosMemoryFree(rightData);
+  }
+
+  printf("---------------------json not exist--------------------\n");
+
+  key = "k10";
+  double eRes10[len+len1] = {DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, true, false, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX, DBL_MAX};
+  for(int i = 0; i < len; i++){
+    makeCalculate(row, key, TSDB_DATA_TYPE_INT, &input[i], eRes10[i], op[i]);
+  }
+  for(int i = len; i < len + len1; i++){
+    void* rightData = prepareNchar(inputNchar[i-len]);
+    makeCalculate(row, key, TSDB_DATA_TYPE_NCHAR, rightData, eRes10[i], op[i]);
+    taosMemoryFree(rightData);
+  }
 }
 
 TEST(columnTest, smallint_value_add_int_column) {
@@ -3171,3 +3357,4 @@ int main(int argc, char** argv) {
 }
 
 #pragma GCC diagnostic pop
+

From 1ab6342018927dfcbce4c7f30f7f3ec0c08a44ca Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Wed, 20 Apr 2022 18:54:15 +0800
Subject: [PATCH 13/15] refactor: fix error in merge

---
 source/libs/parser/src/parUtil.c   | 1 -
 source/libs/scalar/src/sclvector.c | 1 -
 tools/shell/src/shellEngine.c      | 3 ++-
 3 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/source/libs/parser/src/parUtil.c b/source/libs/parser/src/parUtil.c
index 0ff5843736..7f945eb70c 100644
--- a/source/libs/parser/src/parUtil.c
+++ b/source/libs/parser/src/parUtil.c
@@ -262,7 +262,6 @@ int32_t trimString(const char* src, int32_t len, char* dst, int32_t dlen) {
     j++;
   }
   dst[j] = '\0';
-  strtrim(dst);
   return j;
 }
 
diff --git a/source/libs/scalar/src/sclvector.c b/source/libs/scalar/src/sclvector.c
index ae8c395997..d484738328 100644
--- a/source/libs/scalar/src/sclvector.c
+++ b/source/libs/scalar/src/sclvector.c
@@ -287,7 +287,6 @@ static FORCE_INLINE void varToNchar(char* buf, SScalarParam* pOut, int32_t rowIn
 
 //TODO opt performance, tmp is not needed.
 int32_t vectorConvertFromVarData(const SScalarParam* pIn, SScalarParam* pOut, int32_t inType, int32_t outType) {
-  ASSERT(!IS_VAR_DATA_TYPE(outType));
   int32_t bufSize = pIn->columnData->info.bytes;
   char *tmp = taosMemoryMalloc(bufSize + VARSTR_HEADER_SIZE);
 
diff --git a/tools/shell/src/shellEngine.c b/tools/shell/src/shellEngine.c
index e8f53ef981..ca5bf69795 100644
--- a/tools/shell/src/shellEngine.c
+++ b/tools/shell/src/shellEngine.c
@@ -761,7 +761,8 @@ static int calcColWidth(TAOS_FIELD *field, int precision) {
         return TMAX(field->bytes, width);
       }
 
-    case TSDB_DATA_TYPE_NCHAR: {
+    case TSDB_DATA_TYPE_NCHAR:
+    case TSDB_DATA_TYPE_JSON:{
       int16_t bytes = field->bytes * TSDB_NCHAR_SIZE;
       if (bytes > tsMaxBinaryDisplayWidth) {
         return TMAX(tsMaxBinaryDisplayWidth, width);

From da232cbeaf2a906fe9c8cf74537cb869b6543baa Mon Sep 17 00:00:00 2001
From: plum-lihui <huili@taosdata.com>
Date: Wed, 20 Apr 2022 20:02:10 +0800
Subject: [PATCH 14/15] [test: check error from valgrind log file]

---
 tests/script/sh/checkValgrind.sh          | 38 ++++++++++++
 tests/script/tsim/valgrind/checkError.sim | 76 +++++++++++++++++++++++
 tests/tsim/src/simExe.c                   |  6 ++
 3 files changed, 120 insertions(+)
 create mode 100755 tests/script/sh/checkValgrind.sh
 create mode 100644 tests/script/tsim/valgrind/checkError.sim

diff --git a/tests/script/sh/checkValgrind.sh b/tests/script/sh/checkValgrind.sh
new file mode 100755
index 0000000000..17d7ce1aff
--- /dev/null
+++ b/tests/script/sh/checkValgrind.sh
@@ -0,0 +1,38 @@
+#!/bin/bash
+
+set +e
+#set -x
+
+NODE_NAME=
+
+while getopts "n:" arg
+do
+  case $arg in
+    n)
+      NODE_NAME=$OPTARG
+      ;;
+    ?)
+      echo "unkown argument"
+      ;;
+  esac
+done
+
+SCRIPT_DIR=`dirname $0`
+cd $SCRIPT_DIR/../
+SCRIPT_DIR=`pwd`
+
+IN_TDINTERNAL="community"
+if [[ "$SCRIPT_DIR" == *"$IN_TDINTERNAL"* ]]; then
+  cd ../../..
+else
+  cd ../../
+fi
+
+TAOS_DIR=`pwd`
+LOG_DIR=$TAOS_DIR/sim/$NODE_NAME/log
+#CFG_DIR=$TAOS_DIR/sim/$NODE_NAME/cfg
+
+#echo ---- $LOG_DIR
+
+errors=`grep "ERROR SUMMARY:" ${LOG_DIR}/valgrind-taosd-*.log | cut -d ' ' -f 2,3,4,5 | tr -d "\n"`
+echo $errors
diff --git a/tests/script/tsim/valgrind/checkError.sim b/tests/script/tsim/valgrind/checkError.sim
new file mode 100644
index 0000000000..3b66cb4a8c
--- /dev/null
+++ b/tests/script/tsim/valgrind/checkError.sim
@@ -0,0 +1,76 @@
+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+#system sh/deploy.sh -n dnode2 -i 2
+#system sh/deploy.sh -n dnode3 -i 3
+#system sh/deploy.sh -n dnode4 -i 4
+#system sh/cfg.sh -n dnode1 -c supportVnodes -v 0
+system sh/exec.sh -n dnode1 -s start
+#system sh/exec.sh -n dnode2 -s start
+#system sh/exec.sh -n dnode3 -s start
+#system sh/exec.sh -n dnode4 -s start
+
+sleep 2000
+
+#$loop_cnt = 0
+#check_dnode_ready:
+#	$loop_cnt = $loop_cnt + 1
+#	sleep 200
+#	if $loop_cnt == 10 then
+#	  print ====> dnode not ready!
+#		return -1
+#	endi
+#sql show dnodes
+#print ===> $rows $data[0][0] $data[0][1] $data[0][2] $data[0][3] $data[0][4] $data[0][5] $data[0][6]
+#print ===> $rows $data[1][0] $data[1][1] $data[1][2] $data[1][3] $data[1][4] $data[1][5] $data[1][6]
+#print ===> $rows $data[2][0] $data[2][1] $data[2][2] $data[2][3] $data[2][4] $data[2][5] $data[2][6]
+#print ===> $rows $data[3][0] $data[3][1] $data[3][2] $data[3][3] $data[3][4] $data[3][5] $data[3][6]
+#if $data[0][0] != 1 then
+#  return -1
+#endi
+#if $data[0][4] != ready then
+#  goto check_dnode_ready
+#endi
+#
+##sql connect
+#sql create dnode $hostname port 7200
+#sql create dnode $hostname port 7300
+#sql create dnode $hostname port 7400
+#
+#$loop_cnt = 0
+#check_dnode_ready_1:
+#$loop_cnt = $loop_cnt + 1
+#sleep 200
+#if $loop_cnt == 10 then
+#  print ====> dnodes not ready!
+#	return -1
+#endi
+#sql show dnodes
+#print ===> $rows $data[0][0] $data[0][1] $data[0][2] $data[0][3] $data[0][4] $data[0][5] $data[0][6]
+#print ===> $rows $data[1][0] $data[1][1] $data[1][2] $data[1][3] $data[1][4] $data[1][5] $data[1][6]
+#print ===> $rows $data[2][0] $data[2][1] $data[2][2] $data[2][3] $data[2][4] $data[2][5] $data[2][6]
+#print ===> $rows $data[3][0] $data[3][1] $data[3][2] $data[3][3] $data[3][4] $data[3][5] $data[3][6]
+#if $data[0][4] != ready then
+#  goto check_dnode_ready_1
+#endi
+#if $data[1][4] != ready then
+#  goto check_dnode_ready_1
+#endi
+#if $data[2][4] != ready then
+#  goto check_dnode_ready_1
+#endi
+#if $data[3][4] != ready then
+#  goto check_dnode_ready_1
+#endi
+
+#=========== please add any actions above =================
+
+print ====> stop all dondes to output valgrind log file
+system sh/exec.sh -n dnode1 -s stop -x SIGINT
+
+print ====> start to check if there are ERRORS in vagrind log file for each dnode
+# -n : dnode[x] be check
+system_content sh/checkValgrind.sh -n dnode1 
+print cmd return result----> [ $system_content ]
+if $system_content != @ERROR SUMMARY: 0 errors@ then
+  return -1
+endi 
diff --git a/tests/tsim/src/simExe.c b/tests/tsim/src/simExe.c
index 8b3edb6035..dbc3c2c460 100644
--- a/tests/tsim/src/simExe.c
+++ b/tests/tsim/src/simExe.c
@@ -441,6 +441,12 @@ void simStoreSystemContentResult(SScript *script, char *filename) {
   // if ((fd = fopen(filename, "r")) != NULL) {
   if ((pFile = taosOpenFile(filename, TD_FILE_READ)) != NULL) {
     taosReadFile(pFile, script->system_ret_content, MAX_SYSTEM_RESULT_LEN - 1);
+    int32_t len = strlen(script->system_ret_content);
+    for (int32_t i = 0; i < len; ++i) {
+      if (script->system_ret_content[i] == '\n' || script->system_ret_content[i] == '\r') {
+        script->system_ret_content[i] = 0;
+      }
+    }
     taosCloseFile(&pFile);
     char rmCmd[MAX_FILE_NAME_LEN] = {0};
     sprintf(rmCmd, "rm -f %s", filename);

From 3ecf6df9e362ba5b18a54985048b6462775a7fb0 Mon Sep 17 00:00:00 2001
From: plum-lihui <huili@taosdata.com>
Date: Wed, 20 Apr 2022 20:29:50 +0800
Subject: [PATCH 15/15] [test: add valgrind case]

---
 tests/script/jenkins/basic.txt            |  3 +++
 tests/script/tsim/valgrind/checkError.sim | 11 +++++++++--
 2 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/tests/script/jenkins/basic.txt b/tests/script/jenkins/basic.txt
index 64ea437648..506abf3fc5 100644
--- a/tests/script/jenkins/basic.txt
+++ b/tests/script/jenkins/basic.txt
@@ -93,4 +93,7 @@
 # --- sma
 ./test.sh -f tsim/sma/tsmaCreateInsertData.sim
 
+# --- valgrind
+./test.sh -f tsim/valgrind/checkError.sim -v
+
 #======================b1-end===============
diff --git a/tests/script/tsim/valgrind/checkError.sim b/tests/script/tsim/valgrind/checkError.sim
index 3b66cb4a8c..80700dcbce 100644
--- a/tests/script/tsim/valgrind/checkError.sim
+++ b/tests/script/tsim/valgrind/checkError.sim
@@ -71,6 +71,13 @@ print ====> start to check if there are ERRORS in vagrind log file for each dnod
 # -n : dnode[x] be check
 system_content sh/checkValgrind.sh -n dnode1 
 print cmd return result----> [ $system_content ]
-if $system_content != @ERROR SUMMARY: 0 errors@ then
-  return -1
+if $system_content == @ERROR SUMMARY: 0 errors@ then
+  return 0
 endi 
+
+$null=
+if $system_content == $null then
+  return 0
+endi 
+
+return -1
\ No newline at end of file