internal void RecieveCrawlResult(CrawlResult pResult)
{
PlotWaterLine sPlotWaterLine;
lock (mLocker)
{
if (mLineDic.TryGetValue(pResult.LineID, out sPlotWaterLine) == false) //丢弃任务
{
return;
}
}
sPlotWaterLine.RecieveCrawlResult(pResult);
}
/// <summary>
/// 线程池抓取任务
/// </summary>
/// <param name="obj"></param>
/// <returns></returns>
public static void ExeTask(object obj)
{
CrawlTask sCrawlTask = obj as CrawlTask;
CrawlResult sCrawlResult = new CrawlResult(sCrawlTask.ID, sCrawlTask.PlotKey, sCrawlTask.LineID);
sCrawlTask.List.ForEach(t =>
{
try
{
item.URL = t.Url;
item.Method = "get";
result = httpHelper.GetHtml(item);
sCrawlResult.List.Add(new CrawlResultDetail
{
Result = true,
ID = t.ID,
Ext = "html",
Content = result.Html,
Info = null
});
if (DelayMin >= DelayMax)
{
DelayMax = DelayMin + 5000;
}
Thread.Sleep(new Random().Next(DelayMin, DelayMax));
}
catch (Exception ee)
{
sCrawlResult.List.Add(new CrawlResultDetail
{
Result = false,
ID = t.ID,
Ext = "Error",
Content = null,
Info = ee.Message
});
}
});
lock (mLocker)
{
mTaskPool.Remove(sCrawlTask.ID);
//界面设计
HostStatus sHostStatus;
if (mHostDic.TryGetValue(sCrawlTask.Host, out sHostStatus))
{
//sHostStatus.Total += sCrawlResult.List.Count;
sHostStatus.TaskCount--;
}
}
//发送任务回数据中心
WCFServer.SendingCrawlResult(sCrawlResult, sCrawlTask.Authority);
}
internal static void ReceiveCrawlResult(CrawlResult pCrawlResult)
{
OpenPlot sOpenPlot;
lock (mLocker)
{
if (mPlotPool.TryGetValue(pCrawlResult.PlotKey, out sOpenPlot) == false)
{
return;
}
}
sOpenPlot.RecieveCrawlResult(pCrawlResult);
}
internal void RecieveCrawlResult(CrawlResult pResult)
{
CrawlTask sCrawlTask;
lock (mLocker)
{
if (mRunningTaskDic.TryGetValue(pResult.TaskID, out sCrawlTask) == false || sCrawlTask.List.Count != pResult.List.Count) //未注册的任务
{
return;
}
else
{
//该任务算是完成了
mRunningTaskDic.Remove(pResult.TaskID);
for (int i = 0; i < sCrawlTask.List.Count; i++)
{
mRunningTaskDetailDic.Remove(sCrawlTask.List[i].Key);
}
}
}
//最后时间
mLastProduceDt = DateTime.Now;
//将任务和结果合并成一个对象
List<CrawlOriData> sCrawlOriDataList = new List<CrawlOriData>();
//任务包完成的细节任务集合ID
List<int> mFinishedIdList = new List<int>();
for (int i = 0; i < sCrawlTask.List.Count; i++)
{
CrawlResultDetail sCrawlResultDetail = pResult.List[i];
CrawlTaskDetail sCrawlTaskDetail = sCrawlTask.List[i];
//将完成ID
mFinishedIdList.Add(sCrawlTaskDetail.ID);
CrawlOriData sCrawlOriData = new CrawlOriData(sCrawlTaskDetail, sCrawlResultDetail, (byte)("utf-8".Equals(Chaset, StringComparison.OrdinalIgnoreCase) ? 0 : 1));
//存储原始数据
crawlDbAdapter.InsertCrawlResult(sCrawlResultDetail);
//
sCrawlOriDataList.Add(sCrawlOriData);
}
//更新数据库标示，已经完成任务
mFinishedIdList.ForEach(t => {
crawlDbAdapter.UpdateCTask(t);
});
AddCrawlDatasWaitDrill(sCrawlOriDataList);
}
/// <summary>
/// 爬虫获取到html发送回服务中心
/// </summary>
/// <param name="pCrawlResult"></param>
public static void SendingCrawlResult(CrawlResult pCrawlResult, string pAuthority)
{
LwbProcess(pCrawlResult, (int)CrawlCmd.发送爬行任务, pAuthority);
}