去除字符串中的HTML標簽

 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>"; //定義style的正則表達式
14
15     string regEx_script = "<script[^>]*?>[\\s\\S]*?<\\/script>"; //定義script的正則表達式  
16
17     string regEx_html = "<[^>]+>"; //定義HTML標簽的正則表達式  
18
19     htmlStr = Regex.Replace(htmlStr, regEx_style, "");//刪除css
20
21 htmlStr = Regex.Replace(htmlStr, regEx_script, "");//刪除js
22
23     htmlStr = Regex.Replace(htmlStr, regEx_html, "");//刪除html標記
24
25     htmlStr = Regex.Replace(htmlStr, "\\s*|\t|\r|\n", "");//去除tab、空格、空行
26
27     htmlStr = htmlStr.Replace(" ", "");
28
29     htmlStr = htmlStr.Replace(""", "");//去除異常的引號" " "
30
31     htmlStr = htmlStr.Replace(""", "");
32
33     return htmlStr.Trim();
34
35 }