static void Main(string[] args)
{
const int POISON_QUEUE_MAX_TRY = 5;
Console.WriteLine("====== WORKER ======");
using var fromFrontQueue = new PullSocket($">{ThreadSafeFactory.FrontQueue}");
using var toBackQueue = new PushSocket($">{ThreadSafeFactory.BackUrlQueue}");
using var toPoisonQueue = new PushSocket($">{ThreadSafeFactory.BackPosionQueue}");
Console.WriteLine("Initialized Socket");
//process tasks forever
var frontQueueProcessor = Task.Run(() =>
{
while (true)
{
try
{
Console.WriteLine($"Waiting for front queue");
var workload = fromFrontQueue.ReceiveFrameString();
UrlQueueItem queueItem = workload.FromJson <UrlQueueItem>();
Console.WriteLine($"Request from frontqueue {queueItem.Url}");
_urlQueue.Enqueue(queueItem);
}
catch (Exception ex)
{
Console.WriteLine($"Error in Front Queue Processor {ex.Message}");
}
}
});
var processUrlTask = Task.Run(() =>
{
while (true)
{
Parallel.For(0, 2 * maxThread / 3, (threadIndex) =>
{
//while (_urlQueue.IsEmpty)
//{
// Task.Delay(5000).Wait();
//}
if (_urlQueue.TryDequeue(out UrlQueueItem url))
{
Console.WriteLine($"Processing {url.Url}");
url.CrawlStart = DateTime.UtcNow;
try
{
foreach (var childUrl in _htmlRenderer.FindChildLinks(url))
{
if (!_visitedUrls.ContainsKey(childUrl))
{
_visitedUrls.TryAdd(childUrl, false);
Console.WriteLine($"Sending to backqueue");
toBackQueue.SendFrame(childUrl);
Console.WriteLine($"Sending to backqueue");
}
}
}
catch (Exception ex)
{
var u = url.Clone() as UrlPosionQueueItem;
u.TryCount = 0;
u.CrawlFinished = DateTime.UtcNow;
u.Errors = new List <string>
{
ex.Message + "|" + ex?.InnerException?.Message
};
_urlPoisonQueue.Enqueue(u);
Console.WriteLine($"Error in Url Processor {ex.Message}");
}
url.CrawlFinished = DateTime.UtcNow;
}
});
}
static void Main(string[] seed)
{
Console.WriteLine("====== FRONT QUEUE ======");
using var toWorkers = new PushSocket($"@{ThreadSafeFactory.FrontQueue}");
using var fromBackQueue = new PullSocket($">{ThreadSafeFactory.BackQueue}");
Console.WriteLine("Socket initialized");
foreach (var s in seed)
{
_urlQueue.Enqueue(new UrlQueueItem {
Url = s
});
}
var backQueueProcessor = Task.Run(() =>
{
// Receive from backqueue
while (true)
{
Console.WriteLine($"Waiting for backqueue");
var workload = fromBackQueue.ReceiveFrameString();
Console.WriteLine($"Got message from backqueue");
UrlQueueItem queueItem = new UrlQueueItem
{
Url = workload
};
if (!_urlVisiting.ContainsKey(queueItem.Url))
{
_urlQueue.Enqueue(queueItem);
}
else
{
Console.WriteLine($"Url {queueItem.Url} already sent to processing");
}
}
});
var frontQueueProcessor = Task.Run(() =>
{
// send to worker
while (true)
{
do
{
Parallel.For(0, maxThread, (threadIndex) =>
{
//while (_urlQueue.IsEmpty)
//{
// Task.Delay(5000).Wait();
//}
if (_urlQueue.TryDequeue(out UrlQueueItem url))
{
if (_politenessPolicy.CanIGoThere(url, out long crawlDelay))
{
Task.Delay((int)crawlDelay).Wait();
url.QueuedOn = DateTime.UtcNow;
Console.WriteLine($"Sending to worker [{url.Url}]");
toWorkers.SendFrame(url.ToJson());
Console.WriteLine($"Sent to worker");
_urlVisiting.AddOrUpdate(url.Url.ToLower(), false, (key, val) => false);
}
}
});
} while (!_urlQueue.IsEmpty);
}
});
