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, "");//删除js
22
23 htmlStr = Regex.Replace(htmlStr, regEx_html, "");//删除html标记
24
25 htmlStr = Regex.Replace(htmlStr, "\s*| |
|
", "");//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 }