zoukankan      html  css  js  c++  java
  • 去除HTML标记

    直接操作下面代码即可获取你想要的:

            ///   <summary>
            ///   去除HTML标记
            ///   </summary>
            ///   <param   name="NoHTML">包括HTML的源码   </param>
            ///   <param   name="isClearSpechars">去除特殊字符   </param>
            ///   <returns>已经去除后的文字</returns>
            public static string NoHTML(string htmlstring, bool isClearSpechars = true)
            {
                if (String.IsNullOrEmpty(htmlstring))
                {
                    return string.Empty;
                }
    
                //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 = Regex.Replace(htmlstring, "xp_cmdshell", "", RegexOptions.IgnoreCase);
                //删除与数据库相关的词--可以根据个人要求进行删除添加
                htmlstring = Regex.Replace(htmlstring, "select", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "insert", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "delete from", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "count''", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "drop table", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "truncate", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "asc", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "mid", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "char", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "xp_cmdshell", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "exec master", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "net localgroup administrators", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "and", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "net user", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "or", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "net", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "*", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "-", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "delete", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "drop", "", RegexOptions.IgnoreCase);
                htmlstring = Regex.Replace(htmlstring, "script", "", RegexOptions.IgnoreCase);
                if (isClearSpechars)
                {
                    //特殊的字符
                    htmlstring = htmlstring.Replace("<", "");
                    htmlstring = htmlstring.Replace(">", "");
                    htmlstring = htmlstring.Replace("*", "");
                    htmlstring = htmlstring.Replace("-", "");
                    htmlstring = htmlstring.Replace("?", "");
                    htmlstring = htmlstring.Replace("'", "''");
                    htmlstring = htmlstring.Replace(",", "");
                    htmlstring = htmlstring.Replace("/", "");
                    htmlstring = htmlstring.Replace(";", "");
                    htmlstring = htmlstring.Replace("*/", "");
                    htmlstring = htmlstring.Replace("
    ", "");
                }
                htmlstring = System.Web.HttpUtility.HtmlEncode(htmlstring).Trim();
                return htmlstring;
            }
  • 相关阅读:
    print(end=" ") 滚动输出到屏幕
    写入到csv文件的两种方式(pd.DaaFrame 和 csv.writerow)
    pytorch 损失函数(nn.BCELoss 和 nn.CrossEntropyLoss)(思考多标签分类问题)
    pytorch实战(二)hw2——预测收入是否高于50000,分类问题
    信用卡新颖的攻击方式,黑客可非接触式卡进行交易
    微软结束对SolarWinds黑客的内部调查
    Apple M1芯片首款恶意软件曝光
    新的Office恶意软件,黑客可远程窃取信息
    如何审核Active Directory中的密码更改
    特斯拉恶意软件使用新的传送和规避技术
  • 原文地址:https://www.cnblogs.com/yueyongsheng/p/12622417.html
Copyright © 2011-2022 走看看