C#过滤html标签

 1 public static String Html2Text(String htmlStr)
 2 
 3 {
 4 
 5     if (String.IsNullOrEmpty(htmlStr))
 6 
 7     {
 8 
 9         return "";
10 
11     }
12 
13     String regEx_style = "<style[^>]*?>[\\s\\S]*?<\\/style>"; //定义style的正则表达式 
14 
15     String regEx_script = <script[^>]*?>[\\s\\S]*?<\\/script>定义script的正则表达式   
16 
17     String regEx_html = <[^>]+>定义HTML标签的正则表达式   
18 
19     htmlStr = Regex.replace(htmlStr,regEx_style,"");删除css
20 
21     htmlStr = Regex.replace(htmlStr,regEx_script,1)">删除js
22 
23     htmlStr = Regex.replace(htmlStr,regEx_html,1)">删除html标记
24 
25     htmlStr = Regex.replace(htmlStr,1)">\\s*|\t|\r|\n",1)">去除tab、空格、空行
26 
27     htmlStr = htmlStr.replace(" );
28 
29     htmlStr = htmlStr.replace(""","");//去除异常的引号"
30 
31     htmlStr = htmlStr.replace();
32 
33     return htmlStr.Trim();
34 
35 }