HTMLHelper

public class HTMLHelper
{
/**/
/// <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, @"([\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);
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("\r\n", "");
}
htmlstring = System.Web.HttpUtility.HtmlEncode(htmlstring).Trim();
return htmlstring;
}


/**/
/// <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 NoHTMLByTextarea(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, @"-->", "", 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, "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, "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 = System.Web.HttpUtility.HtmlEncode(htmlstring).Trim();
if (htmlstring.Contains("\r\n"))
{
htmlstring = htmlstring.Replace("\r\n", "<br/>");
}
if (htmlstring.Contains("\n"))
{
htmlstring = htmlstring.Replace("\n", "<br/>");
}
return htmlstring;
}
}