protected void DownloadFile(string urllink)
{
Url url = new Url(new Uri(urllink), -1, UrlStatus.Queue);
UrlsSeen.Add(url);
Download(url);
}
protected void DownloadFile(string urllink, Page page)
{
Url url = new Url(new Uri(urllink), page.Url.depth, UrlStatus.Queue);
UrlsSeen.Add(url);
Download(url);
}
/// <summary>
/// Enqueue a new Url to the Urls Queue
/// Add to the UrlsSeen list
/// </summary>
/// <returns></returns>
protected bool AddProcess(Url url)
{
string link = url.link;
if (UrlsSeen.Any(u => u.link == link))
{
_logger.LogInformation("Skipping...URL already queued", link);
}
else if (_config.UseWhiteList && !IsWhiteListedDomain(url.uri.Authority))
{
_logger.LogInformation("URL domain not on whitelist", link);
}
else if (IsExcludedDomain(link))
{
_logger.LogInformation("Skipping...URL domain is excluded", link);
}
else if (IsExcludedFileType(link))
{
_logger.LogInformation("Skipping...file type is excluded", link);
}
else if (ShouldDownload(link))
{
UrlsSeen.Add(url);
Download(url);
}
else if (ShouldContinue(url.depth))
{
UrlsSeen.Add(url);
ParallelManager.Add(FetchNewPage, url);
return(true);
}
return(false);
}