.net正则过滤html

最新推荐文章于 2023-01-04 15:08:04 发布

yandong19861103

最新推荐文章于 2023-01-04 15:08:04 发布

阅读量515

点赞数

分类专栏： ASP.NET 文章标签： html .net string 脚本

本文链接：https://blog.csdn.net/yandong19861103/article/details/3131895

版权

ASP.NET 专栏收录该内容

160 篇文章 0 订阅

订阅专栏

/**/
    ///   <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, @"([/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;