.net正则匹配a标签

2009-08-26 16:00:51|  分类： Asp.Net |字号 订阅
//c#通过正则匹配html中的a标签 的方法
public string matchA(string teststr)
    {
        StringBuilder sb = new StringBuilder();
        //关键就是这个里面的这则表达式
        Regex reg = new Regex(@"<\s*a\shref=*[^>]*>([^<]|<(?!/a))*<\s*/a\s*>");
         MatchCollection match = reg.Matches(teststr);
         foreach (Match var in match)
         {
             if (var != null)
             {
                 sb.Append(Server.HtmlEncode(var.Value));
                 sb.Append("<br />");
             }
         }
         return sb.ToString();
    }

//这个是测试方法获取网易首页的a标签所有信息
public void test()
{
        string temp=string.Empty;
        WebRequest webReq = WebRequest.Create(" http://www.163.com");
        WebResponse webRes = webReq.GetResponse();
        Stream resStream = webRes.GetResponseStream();
        StreamReader sr = new StreamReader(resStream, Encoding.Default);
        StringBuilder sb = new StringBuilder();
        while ((temp = sr.ReadLine()) != null)
        {
            sb.Append(temp);
        }
        webRes.Close();
        sr.Close();
       //打印出所有匹配信息
       Response.Write(matchA(sb.ToString()));
   
}