zoukankan      html  css  js  c++  java
  • 去除HTML标记

            ///   <summary>   
            ///    去除HTML标记   
            ///   </summary>   
            ///   <param    name="NoHTML">包括HTML的源码   </param>   
            ///   <returns>已经去除后的文字</returns>   
            public static string NoHTML(string Htmlstring)
            {
                //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, @"([
    ])[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("
    ", "");
                Htmlstring = HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();
    
                return Htmlstring;
            }
    

      

  • 相关阅读:
    爱的感悟
    连点成图:享受创建图形的乐趣
    python实现动态更新远程机器列表的SSH登录脚本
    python生成数据库中所有表的DESC描述
    生活之美
    克服“测试怠惰”的习惯
    使用git和github托管个人项目
    连点成线
    一次合并数据库的经历
    python使用装饰器捕获异常
  • 原文地址:https://www.cnblogs.com/lilb/p/5666305.html
Copyright © 2011-2022 走看看