[c/c++]判断一个字符串是不是UTF-8字符串

#define CHECK_LENGTH 20       //检查是否为utf8编码时所检查的字符长度   



int is_utf8_string(char *utf)  
    {  
        int length = strlen(utf);  
        int check_sub = 0;  
        int i = 0;  
  
        if ( length > CHECK_LENGTH )  //只取前面特定长度的字符来验证即可  
        {  
            length = CHECK_LENGTH; 
        }  
  
        for ( ; i < length; i ++ )  
        {  
            if ( check_sub == 0 )  
            {  
                if ( (utf[i] >> 7) == 0 )         //0xxx xxxx  
                {  
                    continue;  
                }  
                else if ( (utf[i] & 0xE0) == 0xC0 ) //110x xxxx  
                {  
                    check_sub = 1;  
                }  
                else if ( (utf[i] & 0xF0) == 0xE0 ) //1110 xxxx  
                {  
                    check_sub = 2;  
                }  
                else if ( (utf[i] & 0xF8) == 0xF0 ) //1111 0xxx  
                {  
                    check_sub = 3;  
                }  
                else if ( (utf[i] & 0xFC) == 0xF8 ) //1111 10xx  
                {  
                    check_sub = 4;  
                }  
                else if ( (utf[i] & 0xFE) == 0xFC ) //1111 110x  
                {  
                    check_sub = 5;  
                }  
                else  
                {  
                    return 0;  
                }  
            }  
            else  
            {  
                if ( (utf[i] & 0xC0) != 0x80 )  
                {  
                    return 0;  
                }  
                check_sub --;  
            }  
        }  
        return 1;  
    }  

 

转载于:https://www.cnblogs.com/AkazaAkari/p/8686327.html