c#去除html标签

public static string DislodgeHtml(string strHtml)
       {
           string[] aryReg ={
                       @"<script[^>]*?>.*?</script>",
                       @"<(\/\s*)?!?((\w+:)?\w+)(\w+(\s*=?\s*(([""'])(\\[""'tbnr]|[^\7])*?\7|\w+)|.{0})|\s)*?(\/\s*)?>",
                       @"([\r\n])[\s]+",
                       @"&(quot|#34);",
                       @"&(amp|#38);",
                       @"&(lt|#60);",
                       @"&(gt|#62);",
                       @"&(nbsp|#160);",
                       @"&(iexcl|#161);",
                       @"&(cent|#162);",
                       @"&(pound|#163);",
                       @"&(copy|#169);",
                       @"&#(\d+);",
                       @"-->",
                       @"<!--.*\n"
                       };
           string newReg = aryReg[0];
           string strOutput = strHtml;
           for (int i = 0; i < aryReg.Length; i++)
           {
               Regex regex = new Regex(aryReg[i], RegexOptions.IgnoreCase);
               strOutput = regex.Replace(strOutput, string.Empty);
           }
           strOutput.Replace("<", "");
           strOutput.Replace(">", "");
           strOutput.Replace("\r\n", "");
           return strOutput;
       }
原文地址:https://www.cnblogs.com/objectboy/p/2987768.html