01
2012
06

Asp.Net C# 删除文本中所有的Html标记元素

        public string NoHTML(string Htmlstring)  //替换HTML标记,将所有HTML标记删除
        {
            try
            {
                //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 = Regex.Replace(Htmlstring, @"<img[^>]*>;", "", RegexOptions.IgnoreCase);
                Htmlstring.Replace("<", "");
                Htmlstring.Replace(">", "");
                Htmlstring.Replace("\r\n", "");
                //Htmlstring = HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();
                return Htmlstring;
            }
            catch (Exception ex)
            {
                return "";
            }
        }

« 上一篇下一篇 »

发表评论:

◎欢迎参与讨论,请在这里发表您的看法、交流您的观点。