现在的位置: 首页 > 综合 > 正文

利用正则表达式除去html得到纯文本

2012年09月18日 ⁄ 综合 ⁄ 共 3010字 ⁄ 字号 评论关闭

public static string DelHTML(string Htmlstring)//将HTML去除

         {     
             #region
             //删除脚本

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"<script[^>]*?>.*?</script>","",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             //删除HTML

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"<(.[^>]*)>","",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"([\r\n])[\s]+","",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"-->","",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"<!--.*","",System.Text.RegularExpressions.RegexOptions.IgnoreCase);
           
             //Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"<A>.*</A>","");
       
             //Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"<[a-zA-Z]*=\.[a-zA-Z]*\?[a-zA-Z]+=\d&\w=%[a-zA-Z]*|[A-Z0-9]","");

                       

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"&(quot|#34);","\"",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"&(amp|#38);","&",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"&(lt|#60);","<",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"&(gt|#62);",">",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"&(nbsp|#160);"," ",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"&(iexcl|#161);","\xa1",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring,@"&(cent|#162);","\xa2",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"&(pound|#163);","\xa3",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring,@"&(copy|#169);","\xa9",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

             Htmlstring =System.Text.RegularExpressions. Regex.Replace(Htmlstring, @"&#(\d+);","",System.Text.RegularExpressions.RegexOptions.IgnoreCase);

           
             Htmlstring.Replace("<","");

             Htmlstring.Replace(">","");

             Htmlstring.Replace("\r\n","");

             //Htmlstring=HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();
             #endregion

             return Htmlstring;

         }


抱歉!评论已关闭.