C# 移除富文本多余标签

进阶中的小檀

已于 2022-04-28 17:05:32 修改

阅读量783

点赞数

文章标签： .net

于 2022-04-27 16:48:24 首次发布

本文链接：https://blog.csdn.net/qq_35260798/article/details/124454917

版权

/// <summary>
        /// 移除富文本所有标签
        /// </summary>
        /// <param name="txt"></param>
        /// <returns></returns>
        public static string ReplaceAllTag(this string txt)
        {
            txt = Regex.Replace(txt, "<style[^>]*?>[\\s\\S]*?<\\/style>", "");//删除css
            txt = Regex.Replace(txt, "<script[^>]*?>[\\s\\S]*?<\\/script>", "");//删除js
            txt = Regex.Replace(txt, "<[^>]+>", "");//删除html标记
            txt = Regex.Replace(txt, "\\s*|\t|\r|\n", "");//去除tab、空格、空行
            txt = Regex.Replace(txt, "&nbsp;", "");
            txt = txt.Replace(" ", "");
            txt = txt.Replace("\"", "");//去除异常的引号" " "
            txt = txt.Replace("\"", "");
            return txt;
        }
        /// <summary>
        /// 移除富文本多余的属性
        /// </summary>
        /// <param name="txt"></param>
        /// <returns></returns>
        public static string ReplaceContent(this string txt)
        {
            txt = Regex.Replace(txt, "<source.*?>", "");
            txt = Regex.Replace(txt, "<video.*?>", "");
            txt = Regex.Replace(txt, "</video>", "");
            txt = Regex.Replace(txt, "class[^=]*=[\"']*[^\"'>]+[\"']*", "");
            txt = Regex.Replace(txt, "style[^=]*=[\"']*[^\"'>]+[\"']*", "");
            txt = Regex.Replace(txt, "width[^=]*=[\"']*[^\"'>]+[\"']*", "");
            txt = Regex.Replace(txt, "height[^=]*=[\"']*[^\"'>]+[\"']*", "");
            txt = Regex.Replace(txt, "href[^=]*=[\"']*[^\"'>]+[\"']*", "");//去除a标签 href
            txt = Regex.Replace(txt, "<style[^>]*?>[\\s\\S]*?<\\/style>", "");//去除style
            txt = Regex.Replace(txt, "<script[^>]*?>[\\s\\S]*?<\\/script>", "");//去除script
            txt = Regex.Replace(txt, "&nbsp;", "");
            txt = Regex.Replace(txt, "<p></p>", "");
            txt = Regex.Replace(txt, "figure", "p");
            return txt;
        }