private void RetrievePages(string url, int deep)
{
try
{
HtmlWeb hw = new HtmlWeb();
HtmlAgilityPack.HtmlDocument doc = hw.Load(url);
//foreach (HtmlNode link in doc.DocumentNode.SelectNodes("//a[@href]"))
HtmlNodeCollection htmlNodes = doc.DocumentNode.SelectNodes("//a[@href]");
htmlNodes.ProperQuickShuffle(random);
for (int i = 0; i < htmlNodes.Count - 1; i++)
{
if (bw.CancellationPending)
{
break;
}
Mre.WaitOne();
HtmlNode link = htmlNodes[i];
if (link != null)
{
HtmlAttribute att = link.Attributes["href"];
if ((att.Value.Length > 12) &&
(att.Value.Substring(0, 4).ToUpper() == "HTTP") &&
!EndsWithForbiddenExtension(att.Value) &&
!ContainsForbiden(att.Value) &&
!UrlList.EndWithPart(att.Value.Substring(att.Value.Length - Math.Min(att.Value.Length, 20)))
)
{
//if ((!UrlList.Contains(att.Value)) &&
// (!UrlList.EndWithPart(att.Value.Substring(att.Value.Length - Math.Min(att.Value.Length, 20)))))
UrlList.Add(att.Value);
UrlListToProcess.Add(att.Value);
NoOfLinks++;
bw.ReportProgress(NoOfLinks);
Log.Information(NoOfLinks + ": " + att.Value);
//Console.WriteLine(NoOfLinks.ToString() + ": " + att.Value);
if (random.Next(i, htmlNodes.Count) % 2 == 0)
{
//Log.Information("Soda");
LoadPage(att.Value);
RetrieveNextPages(att.Value, deep);
}
else
{
//Log.Information("Liha");
RetrieveNextPages(att.Value, deep);
LoadPage(att.Value);
}
if (MaxDepth != deep)
{
RetrievePages(att.Value, deep + 1);
}
}
}
}
}
catch (Exception ex)
{
Log.Error($"Error: {ex}", ex);
}
}
