c#远程取得指定url的网页内容

最新推荐文章于 2021-05-30 19:56:21 发布

weixin_30412013

最新推荐文章于 2021-05-30 19:56:21 发布

阅读量152

点赞数

文章标签： c#

原文链接：http://www.cnblogs.com/jishu/archive/2009/03/13/1940127.html

版权

public static string QueryWebPageContent(string url)
{
try
{
string content = "";
WebClient wc = new WebClient();
byte[] buffer = wc.DownloadData(new Uri(url));

//用默认编码转换网页内容
Encoding encoding = Encoding.Default;
content = Encoding.Default.GetString(buffer);

//得到网页的实际编码
Regex regex = new Regex(@"<meta\s+http-equiv\s*=\s*[""']*Content-Type[""']*\s+content\s*=\s*[""']*text/html;\s*charset=([^""']*)[""']*\s*[/]*>|<meta\s+content\s*=\s*[""']*text/html;\s*charset=([^""']*)[""']*\s+http-equiv\s*=\s*[""']*Content-Type[""']*\s*[/]*>", RegexOptions.IgnoreCase);
Match match = regex.Match(info.Content);

if (match.Groups.Count > 2)
{
try
{
encoding = Encoding.GetEncoding(match.Groups[1].Value + match.Groups[2].Value);
content = encoding.GetString(buffer);
}
catch
{ }
}

return content;
}
catch
{
return "";
}
}