NoHtml

标签：ace   rip   context   htm   脚本   reg   trim   public   regex   

public static string NoHTML(string Htmlstring)
{
//删除脚本
Htmlstring = Regex.Replace(Htmlstring, @"<script[^>]*?>.*?</script>", string.Empty, RegexOptions.IgnoreCase);
//删除HTML
Htmlstring = Regex.Replace(Htmlstring, @"<(.[^>]*)>", string.Empty, RegexOptions.IgnoreCase);
Htmlstring = Regex.Replace(Htmlstring, @"([\r\n])[\s]+", string.Empty, RegexOptions.IgnoreCase);
Htmlstring = Regex.Replace(Htmlstring, @"-->", string.Empty, RegexOptions.IgnoreCase);
Htmlstring = Regex.Replace(Htmlstring, @"<!--.*", string.Empty, 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 = Htmlstring.Replace("<", string.Empty);
Htmlstring = Htmlstring.Replace(">", string.Empty);
Htmlstring = Htmlstring.Replace("\r\n", string.Empty);
Htmlstring = HttpContext.Current.Server.HtmlEncode(Htmlstring).Trim();

return Htmlstring;
}

NoHtml

标签：ace   rip   context   htm   脚本   reg   trim   public   regex   

原文地址：http://www.cnblogs.com/iss0101/p/6898063.html