public async Task Execute()
{
if (this.SiteDb == null || this.TransferTask == null)
{
return;
}
List <TransferPage> pagelist = SiteDb.TransferPages.TableScan.Where(o => o.taskid == TransferTask.Id && o.done == false).SelectAll();
DownloadManager manager = new DownloadManager()
{
SiteDb = SiteDb, UserId = this.TransferTask.UserId
};
List <TransferPage> transpages = new List <TransferPage>();
string baseurl = null;
bool defaultstart = true;
foreach (var item in pagelist)
{
if (baseurl == null)
{
baseurl = item.absoluteUrl;
}
var down = await DownloadHelper.DownloadUrlAsync(item.absoluteUrl, manager.CookieContainer, "GET", null, null);
if (down == null || string.IsNullOrEmpty(down.GetString()))
{
item.done = true;
SiteDb.TransferPages.AddOrUpdate(item);
continue;
}
Page page = null;
string downloadbody = down.GetString();
Guid sourcehash = Lib.Security.Hash.ComputeHashGuid(downloadbody);
item.HtmlSourceHash = sourcehash;
if (!string.IsNullOrEmpty(downloadbody))
{
var result = SiteDb.TransferPages.Query.Where(o => o.HtmlSourceHash == sourcehash).SelectAll();
if (result != null && result.Count > 0)
{
var transferpage = result[0];
TransferHelper.AddPageRoute(SiteDb, transferpage.PageId, item.absoluteUrl, baseurl);
item.done = true;
item.PageId = transferpage.PageId;
SiteDb.TransferPages.AddOrUpdate(item);
continue;
}
}
transpages.Add(item);
SiteObject downloadobject = TransferHelper.AddDownload(manager, down, item.absoluteUrl, defaultstart, true, baseurl);
if (downloadobject != null && downloadobject is Page)
{
page = downloadobject as Page;
}
if (page != null)
{
item.PageId = page.Id;
}
UpdateTransferPage(transpages, manager);
if (defaultstart)
{
defaultstart = false;
}
// DownloadOnePage(manager, item);
}
while (!manager.IsComplete)
{
System.Threading.Thread.Sleep(20);
}
this.SiteDb.TransferTasks.SetDone(this.TransferTask.Id);
}
private async Task Downloads(SiteDb siteDb, TransferProgress progress, DownloadManager manager)
{
List <TransferPage> transferingPages = new List <TransferPage>();
List <TransferPage> lowerPriorityPages = new List <TransferPage>();
var query = siteDb.TransferPages.Query.Where(o => o.taskid == progress.TaskId && o.done == false);
while (true)
{
List <TransferPage> pagelist = query.SelectAll();
pagelist.RemoveAll(o => DoneUrlHash.Contains(o.Id));
if (pagelist == null || pagelist.Count == 0)
{
if (progress.counter < progress.TotalPages && lowerPriorityPages.Count() > 0)
{
var needed = progress.TotalPages - progress.counter;
var neededpages = lowerPriorityPages.Take(needed);
foreach (var item in neededpages)
{
progress.counter += 1;
siteDb.TransferPages.AddOrUpdate(item);
lowerPriorityPages.Remove(item);
}
continue;
}
else
{
break;
}
}
foreach (var item in pagelist)
{
DoneUrlHash.Add(item.Id);
var down = await DownloadHelper.DownloadUrlAsync(item.absoluteUrl, manager.CookieContainer);
siteDb.TransferTasks.UpdateCookie(progress.TaskId, manager.CookieContainer);
if (down == null || string.IsNullOrEmpty(down.GetString()))
{
item.done = true;
siteDb.TransferPages.AddOrUpdate(item);
continue;
}
Page page = null;
string downloadbody = down.GetString();
Guid sourcehash = Lib.Security.Hash.ComputeHashGuid(downloadbody);
item.HtmlSourceHash = sourcehash;
if (!string.IsNullOrEmpty(downloadbody))
{
var result = SiteDb.TransferPages.Query.Where(o => o.HtmlSourceHash == sourcehash).SelectAll();
if (result != null && result.Count > 0)
{
var transferpage = result[0];
TransferHelper.AddPageRoute(SiteDb, transferpage.PageId, item.absoluteUrl, progress.BaseUrl);
item.done = true;
item.PageId = transferpage.PageId;
SiteDb.TransferPages.AddOrUpdate(item);
continue;
}
}
transferingPages.Add(item);
SiteObject downloadobject = TransferHelper.AddDownload(manager, down, item.absoluteUrl, item.DefaultStartPage, true, progress.BaseUrl);
if (downloadobject != null && downloadobject is Page)
{
page = downloadobject as Page;
}
if (page != null)
{
item.PageId = page.Id;
}
if (page == null || page.Dom == null)
{
item.done = true;
manager.SiteDb.TransferPages.AddOrUpdate(item);
continue;
}
if (progress.counter < progress.TotalPages && item.depth < progress.Levels)
{
page.Dom.URL = item.absoluteUrl;
var links = TransferHelper.GetAbsoluteLinks(page.Dom, page.Dom.getBaseUrl());
foreach (var linkitem in links)
{
if (progress.counter >= progress.TotalPages)
{
break;
}
if (!UrlHelper.isSameHost(linkitem, progress.BaseUrl))
{
continue;
}
if (!TransferHelper.IsPageUrl(linkitem))
{
continue;
}
TransferPage newpage = new TransferPage();
newpage.absoluteUrl = linkitem;
newpage.depth = item.depth + 1;
newpage.taskid = progress.TaskId;
if (!IsDuplicate(siteDb, newpage))
{
if (TransferHelper.IsLowerPrioUrl(linkitem))
{
if (lowerPriorityPages.Find(o => o.Id == newpage.Id) == null)
{
lowerPriorityPages.Add(newpage);
}
}
else
{
progress.counter += 1;
siteDb.TransferPages.AddOrUpdate(newpage);
}
}
}
}
UpdateTransferPage(transferingPages, manager);
}
}
while (transferingPages.Count() > 0)
{
System.Threading.Thread.Sleep(1000);
UpdateTransferPage(transferingPages, manager);
}
}