用C#去除字符串中HTML的格式

 

想去掉除了段落标记之外的所有html标记，只要页面的文字，好比是我把代码贴到记事本里面的效果，去掉了链接等代码。 可以试试。  public static string DelHTML(string Htmlstring)//将HTML去除         {             #region             //删除样式             Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring, @"<STYLE>[/s/S]*<\/STYLE>", "", System.Text.RegularExpressions.RegexOptions.IgnoreCase);             //删除WORD样式 begin             Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring, @"Normal[^>]*?Mso", "", System.Text.RegularExpressions.RegexOptions.IgnoreCase); 　　　　　　Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring, @"Normal[^>]*?Mso", "", System.Text.RegularExpressions.RegexOptions.IgnoreCase);             Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring, @"v[^>]*?}", "", System.Text.RegularExpressions.RegexOptions.IgnoreCase);             Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring, @"o[^>]*?beha", "", System.Text.RegularExpressions.RegexOptions.IgnoreCase);             Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring, @"w[^>]*?beha", "", System.Text.RegularExpressions.RegexOptions.IgnoreCase); Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring, @".shape[^>]*?beha", "", System.Text.RegularExpressions.RegexOptions.IgnoreCase);             //删除WORD样式 end             //删除脚本             Htmlstring = System.Text.RegularExpressions.Regex.Replace(Htmlstring, @"<script[^>]*?>.*?</script>", "", System.Text.RegularExpressions.RegexOptions.IgnoreCase);             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, @"&(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 = Htmlstring.Replace("<", "");             Htmlstring = Htmlstring.Replace(">", "");             Htmlstring = Htmlstring.Replace("\r\n", "");             #endregion             //Htmlstring = Regex.Replace(Htmlstring, @"[^一-龥]", ""); //只留汉字             return Htmlstring;         } 调用时传入原数据．返回的就是除去后的．

转载于:https://www.cnblogs.com/jxcia_Lai/archive/2010/05/27/1744984.html