• 过滤标记


    当看到贴吧里面这个东西的时候我瞬间石化  好吧 有想法  呵呵 果断挖过来留着 

     
    /// <summary>
    /// 过滤标记
    /// </summary>
    /// <param name="NoHTML">包括HTML,脚本,数据库关键字,特殊字符的源码 </param>
    /// <returns>已经去除标记后的文字</returns>
    public static string NoHTML(string Htmlstring)
    {
    if (Htmlstring == null)
    {
    return "";
    }
    else
    {
    //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, "xp_cmdshell", "", RegexOptions.IgnoreCase);

    //删除与数据库相关的词
    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);


    return Htmlstring ;

    }

    }
  • 相关阅读:
    [ APIO 2015 ] 雅加达的摩天楼
    「POI2011 R1」Conspiracy
    「COCI2016/2017 Contest #2」Bruza
    「THUWC 2017」随机二分图
    「HAOI2015」按位或
    Topcoder Srm 726 Div1 Hard
    「LOJ6482」LJJ爱数数
    「2017 山东一轮集训 Day4」基因
    「Codechef April Lunchtime 2015」Palindromeness
    「UOJ207」共价大爷游长沙
  • 原文地址:https://www.cnblogs.com/mxxblog/p/2389335.html
Copyright © 2020-2023  润新知