public CrawledGiftEventArgs(string crawlerName, Gift gift)
{
CrawlerName = crawlerName;
Gift = gift;
}
void CrawlOnePage()
{
_watch.Start();
// 新建web client
var webClient = CreateWebClient();
Debug("创建了Webclient");
// 获取要爬取的url
var roomNumber = DouyuService.NextRoom();
var url = DouyuService.GetRoomApiUrl(roomNumber);
Debug("获取了房间页面: {0}", url);
// 获取页面内容
var page = "";
while (true)
{
// 申请代理
while (Proxy == null)
{
Proxy = ProxyPool.GetProxy();
if (Proxy == null)
{
MyThread.Wait(1000);
continue;
}
Debug("找到了代理 {0}", Proxy.Address);
// 使用代理
webClient.Proxy = Proxy;
Debug("设置了代理");
break;
}
// 获取网页
try {
// Action<int> crawlingRoom = new Action<int>(GiftCrawler.OnCrawlingRoom);
//IAsyncResult result= crawlingRoom.BeginInvoke(roomNumber, null, null);
// crawlingRoom.EndInvoke(result);
GiftCrawler.OnCrawlingRoom(roomNumber);
Debug("on crawing room");
page = Encoding.UTF8.GetString(webClient.DownloadData(url));
Debug("dowanload data");
if (page.Contains("error") == false)
{
webClient = CreateWebClient();
ProxyPool.RemoveProxy(Proxy);
Proxy = null;
Debug("页面都是乱码, 重新创建Webclient, 移除代理");
continue;
}
//Action<int> crawledRoom = GiftCrawler.OnCrawledRoom;
//crawledRoom.BeginInvoke(roomNumber, null, null);
GiftCrawler.OnCrawledRoom(roomNumber);
Debug("on crawled room");
} catch (WebException webEx) {
// 代理无效了?
Debug("爬取礼物页面发生WebException, 异常信息 = {0}, url = {1}, proxy = {2}",
webEx.Message, url, Proxy.Address);
ProxyPool.RemoveProxy(Proxy);
Debug("移除了无效的代理 - {0}", Proxy.Address);
Proxy = null;
continue;
} catch (Exception ex) {
Debug("爬取礼物页面发生Exception, 异常信息 = {0}, url = {1}, proxy = {2}",
ex.Message, url, Proxy.Address);
}
// 解析礼物
try {
// 解析礼物页面
dynamic roomInfo = JsonConvert.DeserializeObject <dynamic>(page);
Debug("deserialize object");
// 没有礼物
if (roomInfo["error"].Value != 0)
{
Debug("没有爬到礼物, 错误信息 = {0}, url = {1}, proxy = {2}",
roomInfo["data"].Value, url, Proxy.Address);
return;
}
// 爬到礼物
foreach (dynamic item in roomInfo["data"]["gift"])
{
Gift gift = new Gift(
int.Parse(item["id"].Value),
item["name"].Value,
item["type"].Value,
item["pc"].Value,
item["gx"].Value,
item["desc"].Value,
item["intro"].Value,
item["mimg"].Value,
item["himg"].Value
);
//Action<Gift> crawlingGift = GiftCrawler.OnCrawlingGift;
//crawlingGift.BeginInvoke(gift, null, null);
GiftCrawler.OnCrawledGift(gift);
Debug("on crawled gift: {0}", gift.Name);
GiftService.SaveGift(gift);
Debug("save gift: {0}", gift.Name);
}
MyThread.Wait(3000);
Debug("delayed 3000ms...");
return;
} catch (Exception ex) {
Debug("解析礼物页面出现Exception, 异常信息 = {0}, url = {1}, proxy = {2}",
ex.Message, url, Proxy.Address);
continue;
}
}
}