Merge pull request #14232 from taosdata/enh/cast_function

enh(query): enhance cast function to support more types

source/libs/function/src/builtins.c

@@ -1220,18 +1220,19 @@ static int32_t translateSubstr(SFunctionNode* pFunc, char* pErrBuf, int32_t len)
 
 static int32_t translateCast(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
   // The number of parameters has been limited by the syntax definition
-  uint8_t para1Type = ((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType.type;
+  //uint8_t para1Type = ((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType.type;
+
   // The function return type has been set during syntax parsing
   uint8_t para2Type = pFunc->node.resType.type;
-  if (para2Type != TSDB_DATA_TYPE_BIGINT && para2Type != TSDB_DATA_TYPE_UBIGINT &&
-      para2Type != TSDB_DATA_TYPE_VARCHAR && para2Type != TSDB_DATA_TYPE_NCHAR &&
-      para2Type != TSDB_DATA_TYPE_TIMESTAMP) {
-    return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
-  }
-  if ((para2Type == TSDB_DATA_TYPE_TIMESTAMP && IS_VAR_DATA_TYPE(para1Type)) ||
-      (para2Type == TSDB_DATA_TYPE_BINARY && para1Type == TSDB_DATA_TYPE_NCHAR)) {
-    return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
-  }
+  //if (para2Type != TSDB_DATA_TYPE_BIGINT && para2Type != TSDB_DATA_TYPE_UBIGINT &&
+  //    para2Type != TSDB_DATA_TYPE_VARCHAR && para2Type != TSDB_DATA_TYPE_NCHAR &&
+  //    para2Type != TSDB_DATA_TYPE_TIMESTAMP) {
+  //  return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
+  //}
+  //if ((para2Type == TSDB_DATA_TYPE_TIMESTAMP && IS_VAR_DATA_TYPE(para1Type)) ||
+  //    (para2Type == TSDB_DATA_TYPE_BINARY && para1Type == TSDB_DATA_TYPE_NCHAR)) {
+  //  return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
+  //}
 
   int32_t para2Bytes = pFunc->node.resType.bytes;
   if (IS_VAR_DATA_TYPE(para2Type)) {

source/libs/scalar/src/sclfunc.c

@@ -735,6 +735,60 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
     char *input = colDataGetData(pInput[0].columnData, i);
 
     switch(outputType) {
+      case TSDB_DATA_TYPE_TINYINT: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(int8_t *)output = taosStr2Int8(varDataVal(input), NULL, 10);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len  = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(int8_t *)output = taosStr2Int8(newBuf, NULL, 10);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(int8_t *)output, int8_t, inputType, input);
+        }
+        break;
+      }
+      case TSDB_DATA_TYPE_SMALLINT: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(int16_t *)output = taosStr2Int16(varDataVal(input), NULL, 10);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len  = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(int16_t *)output = taosStr2Int16(newBuf, NULL, 10);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(int16_t *)output, int16_t, inputType, input);
+        }
+        break;
+      }
+      case TSDB_DATA_TYPE_INT: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(int32_t *)output = taosStr2Int32(varDataVal(input), NULL, 10);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len  = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(int32_t *)output = taosStr2Int32(newBuf, NULL, 10);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(int32_t *)output, int32_t, inputType, input);
+        }
+        break;
+      }
       case TSDB_DATA_TYPE_BIGINT: {
         if (inputType == TSDB_DATA_TYPE_BINARY) {
           *(int64_t *)output = taosStr2Int64(varDataVal(input), NULL, 10);
@@ -753,6 +807,60 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
         }
         break;
       }
+      case TSDB_DATA_TYPE_UTINYINT: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(uint8_t *)output = taosStr2UInt8(varDataVal(input), NULL, 10);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(uint8_t *)output = taosStr2UInt8(newBuf, NULL, 10);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(uint8_t *)output, uint8_t, inputType, input);
+        }
+        break;
+      }
+      case TSDB_DATA_TYPE_USMALLINT: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(uint16_t *)output = taosStr2UInt16(varDataVal(input), NULL, 10);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(uint16_t *)output = taosStr2UInt16(newBuf, NULL, 10);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(uint16_t *)output, uint16_t, inputType, input);
+        }
+        break;
+      }
+      case TSDB_DATA_TYPE_UINT: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(uint32_t *)output = taosStr2UInt32(varDataVal(input), NULL, 10);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(uint32_t *)output = taosStr2UInt32(newBuf, NULL, 10);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(uint32_t *)output, uint32_t, inputType, input);
+        }
+        break;
+      }
       case TSDB_DATA_TYPE_UBIGINT: {
         if (inputType == TSDB_DATA_TYPE_BINARY) {
           *(uint64_t *)output = taosStr2UInt64(varDataVal(input), NULL, 10);
@@ -771,10 +879,64 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
         }
         break;
       }
+      case TSDB_DATA_TYPE_FLOAT: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(float *)output = taosStr2Float(varDataVal(input), NULL);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len  = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(float *)output = taosStr2Float(newBuf, NULL);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(float *)output, float, inputType, input);
+        }
+        break;
+      }
+      case TSDB_DATA_TYPE_DOUBLE: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(double *)output = taosStr2Double(varDataVal(input), NULL);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len  = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(double *)output = taosStr2Double(newBuf, NULL);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(double *)output, double, inputType, input);
+        }
+        break;
+      }
+      case TSDB_DATA_TYPE_BOOL: {
+        if (inputType == TSDB_DATA_TYPE_BINARY) {
+          *(bool *)output = taosStr2Int8(varDataVal(input), NULL, 10);
+        } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len  = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          newBuf[len] = 0;
+          *(bool *)output = taosStr2Int8(newBuf, NULL, 10);
+          taosMemoryFree(newBuf);
+        } else {
+          GET_TYPED_DATA(*(bool *)output, bool, inputType, input);
+        }
+        break;
+      }
       case TSDB_DATA_TYPE_TIMESTAMP: {
         if (inputType == TSDB_DATA_TYPE_BINARY || inputType == TSDB_DATA_TYPE_NCHAR) {
-          //not support
-          return TSDB_CODE_FAILED;
+          //convert to 0
+          *(int64_t *)output = 0;
         } else {
           GET_TYPED_DATA(*(int64_t *)output, int64_t, inputType, input);
         }
@@ -789,8 +951,16 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
           len = sprintf(varDataVal(output), "%.*s", len, varDataVal(input));
           varDataSetLen(output, len);
         } else if (inputType == TSDB_DATA_TYPE_NCHAR) {
-          //not support
-          return TSDB_CODE_FAILED;
+          char *newBuf = taosMemoryCalloc(1, inputLen);
+          int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
+          if (len < 0) {
+            taosMemoryFree(newBuf);
+            return TSDB_CODE_FAILED;
+          }
+          len = TMIN(len, outputLen - VARSTR_HEADER_SIZE);
+          memcpy(varDataVal(output), newBuf, len);
+          varDataSetLen(output, len);
+          taosMemoryFree(newBuf);
         } else {
           char tmp[400] = {0};
           NUM_TO_STRING(inputType, input, sizeof(tmp), tmp);

tests/system-test/2-query/cast.py

@@ -630,27 +630,27 @@ class TDTestCase:
         ( tdSql.checkData(i, 0, '12') for i in range(tdSql.queryRows) )
 
         tdLog.printNoPrefix("==========step40: error cast condition, should return error ")
-        tdSql.error("select cast(c1 as int) as b from ct4")
-        tdSql.error("select cast(c1 as bool) as b from ct4")
-        tdSql.error("select cast(c1 as tinyint) as b from ct4")
-        tdSql.error("select cast(c1 as smallint) as b from ct4")
-        tdSql.error("select cast(c1 as float) as b from ct4")
-        tdSql.error("select cast(c1 as double) as b from ct4")
-        tdSql.error("select cast(c1 as tinyint unsigned) as b from ct4")
-        tdSql.error("select cast(c1 as smallint unsigned) as b from ct4")
-        tdSql.error("select cast(c1 as int unsigned) as b from ct4")
-
-        tdSql.error("select cast(c2 as int) as b from ct4")
-        tdSql.error("select cast(c3 as bool) as b from ct4")
-        tdSql.error("select cast(c4 as tinyint) as b from ct4")
-        tdSql.error("select cast(c5 as smallint) as b from ct4")
-        tdSql.error("select cast(c6 as float) as b from ct4")
-        tdSql.error("select cast(c7 as double) as b from ct4")
-        tdSql.error("select cast(c8 as tinyint unsigned) as b from ct4")
-
-        tdSql.error("select cast(c8 as timestamp ) as b from ct4")
-        tdSql.error("select cast(c9 as timestamp ) as b from ct4")
-        tdSql.error("select cast(c9 as binary(64) ) as b from ct4")
+        #tdSql.error("select cast(c1 as int) as b from ct4")
+        #tdSql.error("select cast(c1 as bool) as b from ct4")
+        #tdSql.error("select cast(c1 as tinyint) as b from ct4")
+        #tdSql.error("select cast(c1 as smallint) as b from ct4")
+        #tdSql.error("select cast(c1 as float) as b from ct4")
+        #tdSql.error("select cast(c1 as double) as b from ct4")
+        #tdSql.error("select cast(c1 as tinyint unsigned) as b from ct4")
+        #tdSql.error("select cast(c1 as smallint unsigned) as b from ct4")
+        #tdSql.error("select cast(c1 as int unsigned) as b from ct4")
+
+        #tdSql.error("select cast(c2 as int) as b from ct4")
+        #tdSql.error("select cast(c3 as bool) as b from ct4")
+        #tdSql.error("select cast(c4 as tinyint) as b from ct4")
+        #tdSql.error("select cast(c5 as smallint) as b from ct4")
+        #tdSql.error("select cast(c6 as float) as b from ct4")
+        #tdSql.error("select cast(c7 as double) as b from ct4")
+        #tdSql.error("select cast(c8 as tinyint unsigned) as b from ct4")
+
+        #tdSql.error("select cast(c8 as timestamp ) as b from ct4")
+        #tdSql.error("select cast(c9 as timestamp ) as b from ct4")
+        #tdSql.error("select cast(c9 as binary(64) ) as b from ct4")
         pass
 
     def run(self):