refactor: add to_json function

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);

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
 }

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
   }
 };

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]);

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); }
+
+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;
+}