public object FeedJob(string pages)
{
try
{
var node = ServerManager.Get(Request.RequestUri.Authority);
if (node.NodeType == Node.NodeTypeEnum.FEEDPROXY)
{
var ps = pages.Split(',').Select(m => Convert.ToInt32(m)).ToArray();
var feeds = FeedLiteDb.GetFeedModels(ps, 50);
feeds.RemoveAll(m => m.Status == FeedStatus.OFF);
var compile = new CompileFeedAddress();
foreach (var feed in feeds)
{
feed.Address = compile.Compile(feed.Address);
}
return(feeds);
}
}
catch { }
return(new { });
}
public object Run(object t, ParallelTask task)
{
var model = t as CrawlTaskModel;
var results = new List <ExtractResult>();
var reporter = task.Progress as IProgress <string>;
reporter.Report("正在读取Feed记录");
var feed = FeedLiteDb.GetFeed(model.FeedId);
reporter.Report("正在下载 Feed");
var compile = new CompileFeedAddress();
feed.Address = compile.Compile(feed.Address);
var job = new FeedJob();
var snap = job.DoTask(feed, false);
reporter.Report("Feed 下载完成");
var block = RuiJiExpression.ParserBlock(feed.RuiJiExpression);
var feedResult = RuiJiExtracter.Extract(snap.Content, block);
results.Add(feedResult);
reporter.Report("正在提取Feed地址");
var j = new FeedExtractJob();
var urls = j.ExtractAddress(snap);
reporter.Report("Feed地址提取完成");
foreach (var url in urls)
{
reporter.Report("正在提取地址 " + url);
var r = ContentQueue.Instance.Extract(url);
results.AddRange(r);
}
reporter.Report("计算完成");
if (!model.IncludeContent)
{
results.ForEach((m) =>
{
ClearContent(m);
});
}
return(results);
}
