zoukankan      html  css  js  c++  java
  • 去除Html标签

     
    1. /// <summary>
    2.         /// 去除HTML标记
    3.         /// </summary>
    4.         /// <param name="NoHTML">包括HTML的源码 </param>
    5.         /// <returns>已经去除后的文字</returns>
    6.         public string NoHTML(string Htmlstring)
    7.         {   
    8.             //删除脚本
    9.             Htmlstring = Regex.Replace(Htmlstring,@"<script[^>]*?>.*?</script>","",RegexOptions.IgnoreCase);
    10.             //删除HTML
    11.             Htmlstring = Regex.Replace(Htmlstring,@"<(.[^>]*)>","",RegexOptions.IgnoreCase);
    12.             Htmlstring = Regex.Replace(Htmlstring,@"([/r/n])[/s]+","",RegexOptions.IgnoreCase);
    13.             Htmlstring = Regex.Replace(Htmlstring,@"-->","",RegexOptions.IgnoreCase);
    14.             Htmlstring = Regex.Replace(Htmlstring,@"<!--.*","",RegexOptions.IgnoreCase);
    15.             Htmlstring = Regex.Replace(Htmlstring,@"&(quot|#34);","/"",RegexOptions.IgnoreCase);
    16.             Htmlstring = Regex.Replace(Htmlstring,@"&(amp|#38);","&",RegexOptions.IgnoreCase);
    17.             Htmlstring = Regex.Replace(Htmlstring,@"&(lt|#60);","<",RegexOptions.IgnoreCase);
    18.             Htmlstring = Regex.Replace(Htmlstring,@"&(gt|#62);",">",RegexOptions.IgnoreCase);
    19.             Htmlstring = Regex.Replace(Htmlstring,@"&(nbsp|#160);"," ",RegexOptions.IgnoreCase);
    20.             Htmlstring = Regex.Replace(Htmlstring,@"&(iexcl|#161);","/xa1",RegexOptions.IgnoreCase);
    21.             Htmlstring = Regex.Replace(Htmlstring,@"&(cent|#162);","/xa2",RegexOptions.IgnoreCase);
    22.             Htmlstring = Regex.Replace(Htmlstring,@"&(pound|#163);","/xa3",RegexOptions.IgnoreCase);
    23.             Htmlstring = Regex.Replace(Htmlstring,@"&(copy|#169);","/xa9",RegexOptions.IgnoreCase);
    24.             Htmlstring = Regex.Replace(Htmlstring, @"&#(/d+);","",RegexOptions.IgnoreCase);
    25.             Htmlstring.Replace("<","");
    26.             Htmlstring.Replace(">","");
    27.             Htmlstring.Replace("/r/n","");
    28.             Htmlstring=HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();
    29.             return Htmlstring;
    30.         } 
  • 相关阅读:
    BZOJ3674:可持久化并查集加强版
    BZOJ3772:精神污染
    BZOJ3932:[CQOI2015]任务查询系统
    BZOJ3123:[SDOI2013]森林
    BZOJ1926:[SDOI2010]粟粟的书架
    029 列表类型内置方法
    02 Python爬虫之盗亦有道
    01 Python爬虫之Requests库入门
    028 字符串类型内置方法
    027 数字类型内置方法
  • 原文地址:https://www.cnblogs.com/dingdingmao/p/3146591.html
Copyright © 2011-2022 走看看