/// <summary>
/// 抓取页面时使用代理
/// </summary>
/// <param name="url"></param>
/// <returns></returns>
private string DownloadWithProxy(string url)
{
try
{
for (var i = 0; i < 3; i++)
{
WebProxy webProxy = null;
var service = new ProxyService();
Proxy proxy = service.GetProxy();
if (proxy != null)
{
webProxy = new WebProxy(proxy.Adress + ":" + proxy.Port);
}
var html = HttpHelper.DownloadHtml(url, webProxy, TimeOut);
return(html);
}
}
catch (Exception ex)
{
}
return("");
}
public override void Start()
{
LogHelper.LogMsg("》》》开始抓取西刺代理");
try
{
var service = new ProxyService();
List <string> list = new List <string>()
{
"http://www.xicidaili.com/nt/",
"http://www.xicidaili.com/nn/",
"http://www.xicidaili.com/wn/",
"http://www.xicidaili.com/wt/"
};
foreach (var utlitem in list)
{
string url = utlitem;
string html = DownloadProxyPage(url);
if (!CheckHtml(html, url, "西刺代理"))
{
continue;
}
HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(html);
HtmlNode node = doc.DocumentNode;
//获取总页数
var totalPage = GetTotalPage(node);
for (var i = 1; i <= totalPage; i++)
{
var pageUrl = url + i;
//如果是第一页，已经取过不用再取
if (i != 1)
{
html = DownloadProxyPage(pageUrl);
if (!CheckHtml(html, pageUrl, "西刺代理"))
{
return;
}
doc.LoadHtml(html);
node = doc.DocumentNode;
}
string xpathstring = "//table[@id='ip_list']/tr[@class]";
HtmlNodeCollection collection = node.SelectNodes(xpathstring);
if (collection == null)
{
LogHelper.LogError("无获取西刺代理ip，请检查网站结构是否有改动");
continue;
}
var proxyList = new List <Proxy>();
//提取代理
Parallel.ForEach(collection, item =>
{
Proxy proxy = new Proxy();
string xpath = "td[2]";
proxy.Adress = item.SelectSingleNode(xpath).InnerHtml.Trim();
xpath = "td[3]";
int port = 0;
if (!int.TryParse(item.SelectSingleNode(xpath).InnerHtml.Trim(), out port))
{
LogHelper.LogError("西刺代理，取端口号时出错：" + item.InnerHtml);
return;
}
proxy.Port = port;
proxy.Source = pageUrl;
proxyList.Add(proxy);
});
VerifyAndSave(proxyList);
}
}
}
catch (Exception ex)
{
LogHelper.LogError("抓取西刺代理时出错：" + ex);
}
LogHelper.LogMsg("《《《西刺代理抓取完成");
}
