public void Crawl(Uri url)
{
_linkFilters = new List<ILinkFilter>()
{
new NoAnchorsFilter(),
new NoMailToFilter(),
new CacheHitFilter(_cache),
};
//Could be done with subclassing
//(class NoExternalLinkCrawler : Crawler)
//but adds way more complexity for what it is worth.
if (!_followExternalLinks) _linkFilters.Add(new NoExternalLinks(url));
var fetcher = new Fetcher(url);
//adds the starting url to the queue (even if already downloaded)
_queue.Enqueue(fetcher);
foreach (Page cachedPage in _cache)
{
//Each page already in cache is supplied to the OnPageLoaded
//method, so their links can be processed. This basically
//resumes (with a bit of overhead) the process where it
//was stopped.
OnPageLoaded(cachedPage);
}
_queue.Process();
}
public void Enqueue(Fetcher fetcher)
{
lock (_lockObject)
{
//If a fetcher with the same url is already in queue it is not
//enqueued - it is assumed that two pages with the same
//url are the same page, so no need to download them twice.
if (_queue.All(x => x.Uri.AbsoluteUri != fetcher.Uri.AbsoluteUri))
{
_queue.Add(fetcher);
}
}
}
//Callback invoked on error - it just removes the
//fetcher silently.
private void RemoveFetcher(Fetcher fetcher)
{
lock (_lockObject)
{
_executing.Remove(fetcher);
}
}
//Callback invoked on successful completion.
private void OnCompleted(Fetcher fetcher)
{
RemoveFetcher(fetcher);
_callback(fetcher.DownloadedPage);
//If, after invoking the parent's callback, no jobs are in queue
//and no jobs are executing it means that the crawling is over,
//so the ProcessingOver event is fired.
if (!_queue.Any() && !_executing.Any() && ProcessingOver != null)
{
ProcessingOver();
}
}
