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:
    [YTU]_2436( C++ 习题 输出日期时间--友元类)
    [YTU]_2435 ( C++ 习题 输出日期时间--友元函数)
    病毒侵袭
    石子合并(区间DP经典例题)
    AC自动机模板2
    【模板】最近公共祖先(LCA)
    华华给月月出题
    线性筛素数
    华华开始学信息学
    华华和月月种树
  • 原文地址:https://www.cnblogs.com/lilb/p/5666305.html
Copyright © 2011-2022 走看看