c#使用正则表达式替换html标签

  /// <summary>
        /// 去除HTML标记(用正则彻底去除HTMLCSSscript代码 )
        /// </summary>
        /// <param name="Htmlstring">包括HTML的源码 </param>
        /// <returns>已经去除后的文字</returns>
        public static string NoHTML(this 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;
        }
View Code
原文地址:https://www.cnblogs.com/lihongchen/p/4178748.html