c#正则过滤html标记

 public static string NoHTML(stringHtmlstring)

        {
            //删除脚本
            Htmlstring = Regex.Replace(Htmlstring, @"<script[^>]*?>.*?</script>", "",
               RegexOptions.IgnoreCase);
             //删除HTML
            Htmlstring = Regex.Replace(Htmlstring, @"<(.[^>]*)>", "",
             RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"([\r\n])[\s]+", "",
             RegexOptions.IgnoreCase);
          Htmlstring = Regex.Replace(Htmlstring, @"-->", "", RegexOptions.IgnoreCase);
         Htmlstring = Regex.Replace(Htmlstring, @"<!--.*", "", RegexOptions.IgnoreCase);
           Htmlstring = Regex.Replace(Htmlstring, @"&(quot|#34);", "\"",
            RegexOptions.IgnoreCase);
          Htmlstring = Regex.Replace(Htmlstring, @"&(amp|#38);", "&",
           RegexOptions.IgnoreCase);
         Htmlstring = Regex.Replace(Htmlstring, @"&(lt|#60);", "<",
          RegexOptions.IgnoreCase);
          Htmlstring = Regex.Replace(Htmlstring, @"&(gt|#62);", ">",
             RegexOptions.IgnoreCase);
         Htmlstring = Regex.Replace(Htmlstring, @"&(nbsp|#160);", "   ",
            RegexOptions.IgnoreCase);
         Htmlstring = Regex.Replace(Htmlstring, @"&(iexcl|#161);", "\xa1",
           RegexOptions.IgnoreCase);
          Htmlstring = Regex.Replace(Htmlstring, @"&(cent|#162);", "\xa2",
            RegexOptions.IgnoreCase);
           Htmlstring = Regex.Replace(Htmlstring, @"&(pound|#163);", "\xa3",
             RegexOptions.IgnoreCase);
          Htmlstring = Regex.Replace(Htmlstring, @"&(copy|#169);", "\xa9",
           RegexOptions.IgnoreCase);
          Htmlstring = Regex.Replace(Htmlstring, @"&#(\d+);", "",
           RegexOptions.IgnoreCase);
 
         Htmlstring.Replace("<", "");
         Htmlstring.Replace(">", "");
          Htmlstring.Replace("\r\n", "");
          Htmlstring = HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();
 
          return Htmlstring;
     }

 

 1:     public static string NoHTML(stringHtmlstring)

   2:      {

   3:          //删除脚本

   4:          Htmlstring = Regex.Replace(Htmlstring, @"<script[^>]*?>.*?</script>", "",

   5:            RegexOptions.IgnoreCase);

   6:          //删除HTML

   7:          Htmlstring = Regex.Replace(Htmlstring, @"<(.[^>]*)>", "",

   8:            RegexOptions.IgnoreCase);

   9:          Htmlstring = Regex.Replace(Htmlstring, @"([\r\n])[\s]+", "",

  10:            RegexOptions.IgnoreCase);

  11:          Htmlstring = Regex.Replace(Htmlstring, @"-->", "", RegexOptions.IgnoreCase);

  12:          Htmlstring = Regex.Replace(Htmlstring, @"<!--.*", "", RegexOptions.IgnoreCase);

  13:          Htmlstring = Regex.Replace(Htmlstring, @"&(quot|#34);", "\"",

  14:            RegexOptions.IgnoreCase);

  15:          Htmlstring = Regex.Replace(Htmlstring, @"&(amp|#38);", "&",

  16:            RegexOptions.IgnoreCase);

  17:          Htmlstring = Regex.Replace(Htmlstring, @"&(lt|#60);", "<",

  18:            RegexOptions.IgnoreCase);

  19:          Htmlstring = Regex.Replace(Htmlstring, @"&(gt|#62);", ">",

  20:            RegexOptions.IgnoreCase);

  21:          Htmlstring = Regex.Replace(Htmlstring, @"&(nbsp|#160);", "   ",

  22:            RegexOptions.IgnoreCase);

  23:          Htmlstring = Regex.Replace(Htmlstring, @"&(iexcl|#161);", "\xa1",

  24:            RegexOptions.IgnoreCase);

  25:          Htmlstring = Regex.Replace(Htmlstring, @"&(cent|#162);", "\xa2",

  26:            RegexOptions.IgnoreCase);

  27:          Htmlstring = Regex.Replace(Htmlstring, @"&(pound|#163);", "\xa3",

  28:            RegexOptions.IgnoreCase);

  29:          Htmlstring = Regex.Replace(Htmlstring, @"&(copy|#169);", "\xa9",

  30:            RegexOptions.IgnoreCase);

  31:          Htmlstring = Regex.Replace(Htmlstring, @"&#(\d+);", "",

  32:            RegexOptions.IgnoreCase);

  33:   

  34:          Htmlstring.Replace("<", "");

  35:          Htmlstring.Replace(">", "");

  36:          Htmlstring.Replace("\r\n", "");

  37:          Htmlstring = HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();

  38:   

  39:          return Htmlstring;

  40:      }

 

转载于:https://www.cnblogs.com/wangchuang/archive/2012/05/23/2515280.html