public void ScanFacebookExecuted(object param)
{
// Okay I apologize in advance for this shit.... Read on if you dare...
//
// Purpose of this block of code:
// To scrape facebook as fast as possible, asynchronously.
//
// How it's achieved:
// Parallel foreach loops iterating over urls and async web requests
//
// Why this is garbage:
// Parallel foreach is used to iterate over the urls as fast as possible. However, I'm using
// async web requests inside of the parallel loops. Unfortunately this breaks C# apparently.
// When you await an async request inside of a parallel foreach, the parallel loop is no longer
// blocking and exits immediately. This means that that part at the bottom, the part that updates the
// UI, will run prior to the results being returned.... Yeah... fml
//
// How I worked around this:
// I create a List<Object>. When a request goes out we add one to the list. When a response comes back
// we remove one from the list. At the base of the function we have an infinite non-blocking loop that
// waits for the requests count to equal zero. When it's zero it sorts our list based on confidence score
// and updates the UI.
//
// Problems with this method:
// Currently I'm not accounting for failed requests.... I have no idea what would happen. Probably nothing good.
// Like seriously... I don't expect the application to suddenly get better if a request fails... Fix this...
List <Object> requests = new List <Object>();
List <UsernameLinkModel> tempLinks = new List <UsernameLinkModel>();
try
{
IEnumerable <DBLinkModel> tmpCache = DatabaseSearchResults.Where(pm => pm.IsSelected);
foreach (var item in tmpCache)
{
var temp = TargetLinks.Where(tl => tl.id == item.Id && tl.usernames.Count > 0);
if (temp.Count() > 0)
{
tempLinks.Add(temp.FirstOrDefault());
tmpCache = tmpCache.Where(tl => tl.Id != item.Id);
}
}
TargetLinks.Clear();
TopTargets.Clear();
object outerLock = new object();
object innerLock = new object();
// First we run over our list of voter db targets
Parallel.ForEach(tmpCache, async pm =>
{
DBLinkModel target = pm;
Console.WriteLine("Running search on target: " + target.firstname + " " + target.lastname);
lock (outerLock)
{
requests.Add(new Object());
}
// foreach voter db target, try and see if there are any facebook users in ohio with that name
string source = await ExternalBrowser.CallExternalBrowser(FacebookStuff.FormatQueryURL(target), true);
// lock adding and removing requests, because you know, threads, shared resources, fml
lock (outerLock)
{
requests.RemoveAt(requests.Count - 1);
}
// Link made
TargetLinks.Add(FacebookStuff.ExtractUsernamesFromSource(target, source));
if (requests.Count == 0)
{
// For each link made
Parallel.ForEach(TargetLinks, targetLink =>
{
// Run over every possible user and score it
Parallel.ForEach(targetLink.usernames, async username =>
{
lock (innerLock)
{
requests.Add(new Object());
}
// Get the source for this possible facebook match
string profileSource = await ExternalBrowser.CallExternalBrowser(username.ProfileLink);
lock (innerLock)
{
requests.RemoveAt(requests.Count - 1);
}
// Parse and score the source
username.ConfidenceScore = RelationshipStuff.GetConfidenceLevel(SelectedPerson, profileSource);
username.UserModelLinkId = targetLink.id;
username.FullName = targetLink.name;
// wait until alllllllll the requests are done. If you're curious why in 2017 I need to do this, read the
// above giant comment block
while (requests.Count > 0)
{
System.Windows.Forms.Application.DoEvents();
}
Console.WriteLine("SCAN COMPLETED AT: " + DateTime.Now.ToString());
Sort();
});
});
}
});
}
catch (Exception err) { Console.WriteLine(err.Message); /*Nom nom nom*/ }
}
