asp.net去掉HTML標記

using System;
using System.Web;
using System.Text.RegularExpressions;

 

public static string NoHTML(string Htmlstring)  
  {  
   //刪除指令碼  
   Htmlstring   =   Regex.Replace(Htmlstring,@"<script[^>]*?>.*?</script>","",RegexOptions.IgnoreCase);  
   //刪除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.Replace("<","");  
   Htmlstring.Replace(">","");  
   Htmlstring.Replace("\r\n","");  
   Htmlstring=HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();  
   
   return   Htmlstring;  
  }

 

取出文本中的圖片地址:

public   static   string   GetImgUrl(string   HTMLStr)  
  {  
    string   str   =   string.Empty;  
    string   sPattern   =   @"^<img\s+[^>]*>";  
    Regex   r   =   new   Regex(@"<img\s+[^>]*\s*src\s*=\s*([']?)(?<url>\S+)'?[^>]*>",  
            RegexOptions.Compiled);  
    Match   m   =   r.Match(HTMLStr.ToLower());  
    if   (m.Success)  
        str   =   m.Result("${url}");  
    return   str;  
 }

 

本文來自CSDN部落格，轉載請標明出處：http://blog.csdn.net/TQY2008/archive/2008/09/26/2983594.aspx