.net去除html标签代码 

public  string NoHTML(string Htmlstring)
        {
            //删除脚本
            Htmlstring = Regex.Replace(Htmlstring, @"<script[^>]*?>.*?</script>", "", RegexOptions.IgnoreCase);
            //删除HTML
            Htmlstring = Regex.Replace(Htmlstring, @"<br>", "$br$", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"<(.[^>]*)>", "", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"$br$", "<br>", RegexOptions.IgnoreCase);
            Htmlstring = Regex.Replace(Htmlstring, @"([\r\n])[\s]+", "<br>", 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", "");
            return Htmlstring;
        }

posted on 2009-04-17 12:03 sanmao 阅读(734) 评论(1)  编辑  收藏

FeedBack:
# re: .net去除html标签代码[未登录]
2014-05-17 09:20 | bobby
正则表达式在这个程度上己经不够了,比如清除未知的属性,和标记,建议使用wuxiu.safehelper  回复  更多评论
  

只有注册用户登录后才能发表评论。


网站导航:
 

常用链接

留言簿(5)

随笔分类

随笔档案

搜索

  •  

最新评论

阅读排行榜

评论排行榜