c# 去除文本的html标签

 1 public static string ContentReplace(string input)
 2         {
 3             input = Regex.Replace(input, @"<(.[^>]*)>", "", RegexOptions.IgnoreCase);
 4             input = Regex.Replace(input, @"([\r\n])[\s]+", "", RegexOptions.IgnoreCase);
 5             input = Regex.Replace(input, @"-->", "", RegexOptions.IgnoreCase);
 6             input = Regex.Replace(input, @"<!--.*", "", RegexOptions.IgnoreCase);
 7 
 8             input = Regex.Replace(input, @"&(quot|#34);", "\"", RegexOptions.IgnoreCase);
 9             input = Regex.Replace(input, @"&(amp|#38);", "&", RegexOptions.IgnoreCase);
10             input = Regex.Replace(input, @"&(lt|#60);", "<", RegexOptions.IgnoreCase);
11             input = Regex.Replace(input, @"&(gt|#62);", ">", RegexOptions.IgnoreCase);
12             input = Regex.Replace(input, @"&(nbsp|#160);", " ", RegexOptions.IgnoreCase);
13             input = Regex.Replace(input, @"&(iexcl|#161);", "\xa1", RegexOptions.IgnoreCase);
14             input = Regex.Replace(input, @"&(cent|#162);", "\xa2", RegexOptions.IgnoreCase);
15             input = Regex.Replace(input, @"&(pound|#163);", "\xa3", RegexOptions.IgnoreCase);
16             input = Regex.Replace(input, @"&(copy|#169);", "\xa9", RegexOptions.IgnoreCase);
17             input = Regex.Replace(input, @"&#(\d+);", "", RegexOptions.IgnoreCase);
18 
19             input.Replace("<", "");
20             input.Replace(">", "");
21             input.Replace("\r\n", "");
22             //去两端空格，中间多余空格
23             input = Regex.Replace(input.Trim(), "\\s+", " ");
24             return input;
25         }

 

转载于:https://www.cnblogs.com/ghelement/p/4511499.html