C# 判断字符编码是否为中文_转

通过对字符的unicode编码进行判断来确定字符是否为中文：在unicode 字符串中，中文的范围是在4E00~9FFF:CJK Unified Ideographs

 protected bool   IsChineseLetter(string input,int index)
    {

        int code = 0;
        int chfrom = Convert.ToInt32("4e00", 16);    //范围（0x4e00～0x9fff）转换成int（chfrom～chend）
        int chend = Convert.ToInt32("9fff", 16);
        if (input != "")
        {
             code = Char.ConvertToUtf32(input, index);    //获得字符串input中指定索引index处字符unicode编码
            
           if (code >= chfrom && code <= chend)     
            {
                 return true;     //当code在中文范围内返回true

             }
            else
            {
                  return false ;    //当code不在中文范围内返回false
             }
         }

          return false;
 }

public bool IsChina(string CString)
          {
              bool BoolValue = false;
              for (int i = 0; i < CString.Length; i++)
              {
                  if (Convert.ToInt32(Convert.ToChar(CString.Substring(i, 1))) < Convert.ToInt32(Convert.ToChar(128)))
                  {
                      BoolValue = false;
                  }
                  else
                  {
                      return BoolValue = true;
                  }
              }
              return BoolValue;
          }

通过正则表达式：/u4e00-/u9fa5 汉字的范围

for (int i=0; i<s.length; i++)
{
    Regex rx = new Regex("^[/u4e00-/u9fa5]$");
    if (rx.IsMatch(s[i]))
    {
        // 是
    }
    else
    {
        // 否
    }
}

unicodeencoding unicodeencoding = new unicodeencoding(); 
byte [] unicodebytearray = unicodeencoding.getbytes( inputstring ); 
for( int i = 0; i < unicodebytearray.length; i++ ) 
{ 
i++; 
//如果是中文字符那么高位不为0 
if ( unicodebytearray[i] != 0 ) 
{ 
} 
……  

        /// <summary>
        /// 给定一个字符串，判断其是否只包含有汉字
        /// </summary>
        /// <param name="testStr"></param>
        /// <returns></returns>
        public bool IsOnlyContainsChinese(string testStr)
        {
            char[] words = testStr.ToCharArray();
            foreach (char word in words)
            {
                if ( IsGBCode(word.ToString()) || IsGBKCode(word.ToString()) ) // it is a GB2312 or GBK chinese word
                {
                    continue;
                }
                else
                {
                    return false;
                }
            }
            return true;
        }

        /// <summary>
        /// 判断一个word是否为GB2312编码的汉字
        /// </summary>
        /// <param name="word"></param>
        /// <returns></returns>
        private bool IsGBCode(string word)
        {
            byte[] bytes = Encoding.GetEncoding("GB2312").GetBytes(word);
            if (bytes.Length <= 1) // if there is only one byte, it is ASCII code or other code
            {
                return false;
            }
            else
            {
                byte byte1 = bytes[0];
                byte byte2 = bytes[1];
                if (byte1 >= 176 && byte1 <= 247 && byte2 >= 160 && byte2 <= 254)    //判断是否是GB2312
                {
                    return true;
                }
                else
                {
                    return false;
                }
            }
        }

        /// <summary>
        /// 判断一个word是否为GBK编码的汉字
        /// </summary>
        /// <param name="word"></param>
        /// <returns></returns>
        private bool IsGBKCode(string word)
        {
            byte[] bytes = Encoding.GetEncoding("GBK").GetBytes(word.ToString());
            if (bytes.Length <= 1) // if there is only one byte, it is ASCII code
            {
                return false;
            }
            else
            {
                byte byte1 = bytes[0];
                byte byte2 = bytes[1];
                if ( byte1 >= 129 && byte1 <= 254 && byte2 >= 64 && byte2 <= 254)     //判断是否是GBK编码
                {
                    return true;
                }
                else
                {
                    return false;
                }
            }
        }

        /// <summary>
        /// 判断一个word是否为Big5编码的汉字
        /// </summary>
        /// <param name="word"></param>
        /// <returns></returns>
        private bool IsBig5Code(string word)
        {
            byte[] bytes = Encoding.GetEncoding("Big5").GetBytes(word.ToString());
            if (bytes.Length <= 1) // if there is only one byte, it is ASCII code
            {
                return false;
            }
            else
            {
                byte byte1 = bytes[0];
                byte byte2 = bytes[1];
                if ( (byte1 >= 129 && byte1 <= 254) && ((byte2 >= 64 && byte2 <= 126) || (byte2 >= 161 && byte2 <= 254)) )     //判断是否是Big5编码
                {
                    return true;
                }
                else
                {
                    return false;
                }
            }
        }