public void Run(CrawlerWorkerArgs args)
{
_parent = args.Parent;
_callback = args.Function;
if (_callback == null)
{
throw new ArgumentException("No callback was provided so no work would be done.");
}
_robots = args.Robots;
_bannedExts = args.BannedExtensions;
_visited = args.Visited;
_visitedLock = args.VisitedLock;
_sites = args.Sites;
_respectRobots = args.RespectRobots;
_userAgent = args.UserAgent;
_maxDepth = args.MaxDepth;
_id = args.ID;
_web = new HtmlWeb();
_web.UserAgent = _userAgent;
while (true)
{
Site next = null;
if (!_sites.GetNextAvailableWorker(out next))
{
Interlocked.Increment(ref _parent.PausedWorkers);
if (_sites.Empty() && _parent.PausedWorkers == _parent.TotalWorkers)
{
return;
}
Thread.Sleep(500);
Interlocked.Decrement(ref _parent.PausedWorkers);
continue;
}
String url = next.Url;
_visitedLock.EnterReadLock();
bool visited = _visited.Contains(url);
_visitedLock.ExitReadLock();
if (next.Depth < _maxDepth && !visited)
{
VisitOneSite(next);
}
}
}
private void RunWorker(object obj)
{
if (obj.GetType() != typeof(Tuple <CrawlerWorkerArgs, CrawlerWorker>))
{
String error = "Incorrect arguments provided to RunWorker, can't run crawler.";
_log.Fatal(error);
throw new InvalidOperationException(error);
}
var tuple = (Tuple <CrawlerWorkerArgs, CrawlerWorker>)obj;
CrawlerWorkerArgs args = tuple.Item1;
CrawlerWorker worker = tuple.Item2;
worker.Run(args);
}
public void Crawl()
{
List <Thread> workerThreads = new List <Thread>();
for (int i = 0; i < _settings.WorkerCount; ++i)
{
ParameterizedThreadStart ts = new ParameterizedThreadStart(RunWorker);
Thread temp = new Thread(ts);
CrawlerWorkerArgs args = new CrawlerWorkerArgs()
{
Parent = this,
Function = _settings.Function,
Robots = _robots,
Visited = _visited,
VisitedLock = _visitedLock,
Sites = _sites,
RespectRobots = _settings.RespectRobots,
UserAgent = _settings.UserAgent,
MaxDepth = _settings.MaxDepth,
BannedExtensions = _settings.BannedExtensions,
ID = i,
};
var functionArgs = Tuple.Create(args, new CrawlerWorker());
_log.Debug($"Starting worker {i}");
temp.Start(functionArgs);
workerThreads.Add(temp);
}
foreach (Thread t in workerThreads)
{
t.Join();
}
}