/// <summary>
/// Sets up PageManager threads and begin search
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
private void Work(object sender, DoWorkEventArgs e)
{
_working = true;
if (Thread.CurrentThread.Name == null)
Thread.CurrentThread.Name = "Search Manager: " +
_searchCriteria.SearchText;
System.Diagnostics.Stopwatch stopwatch = new System.Diagnostics.Stopwatch();
stopwatch.Start();
// Set up the page managers, each running on an async thread
_pageManagers = new PageManager[_threadCount];
for (int i = 0; i < _pageManagers.Count(); i++)
{
// Keep track of the highest page we've
// attempted to scrape (page numbers are not zero-based)
_pageNumber = i+1;
// PageManagers internally fire off an async worker which each
// call the ResultReturned / WorkerFinished event handlers.
_pageManagers[i] = new PageManager(
_pageNumber,
_searchCriteria,
Scraper.LoadSearchPage, // inject method for testing here
ResultReturned,
WorkerFinished);
_pageManagers[i].RunWorkerAsync();
}
while (_working)
{
// User can cancel a search through the UI.
if (CancellationPending)
{
HaltAllOngoingWork();
}
}
stopwatch.Stop();
string msg = "Search time : {0} ms" + Environment.NewLine;
msg = string.Format(msg, stopwatch.ElapsedMilliseconds);
Debug.WriteLine(msg);
}
/// <summary>
/// Called by PageManagers when their task is complete.
/// If there is still work to be done, replaces the finished
/// pageManager with a new one (creates a new thread)
/// Exits if all work is complete.
/// </summary>
/// <param name="obj">sender object</param>
/// <param name="args">Result<AmazonItem></param>
public void WorkerFinished(object obj, RunWorkerCompletedEventArgs args)
{
// See if any of the workers are reporting that they're out
// of results.
bool outOfResults = _pageManagers.Any(i => i.WorkStatus ==
PageManager.Status.NoResults);
// If so, don't deploy another thread.
if (outOfResults)
{
string msg = "PageManager reporting no more results;";
msg += " not deploying another thread.";
Debug.WriteLine(msg);
return;
}
// Or if there are no threads that are marked "Working",
// we are done
if (IsWorkFinished())
{
HaltAllOngoingWork();
return;
}
if (args == null)
{
return;
}
if (args.Error != null)
{
Debug.WriteLine(args.Error.Message);
return;
}
if (args.Result == null ||
args.Result.GetType() != typeof(PageManager))
{
return;
}
// If this PageManager is done but we haven't hit our
// target number of results, we should spawn a new thread
PageManager finished = (PageManager)args.Result;
// Get the index of the PageManager whose search page number
// matches the one that just finished (we're going to replace
// it with a new PageManager)
int index = _pageManagers.ToList().FindIndex(i =>
i.PageNumber == finished.PageNumber);
// Increment the variable that tracks the
// highest page number we've searched so far
// TODO: since page number is shared state, there is a
// slight chance that two PageManagers might hit this
// code
_pageNumber += 1;
// Start searching a new page
PageManager newPageManager = new PageManager(
_pageNumber,
_searchCriteria,
Scraper.LoadSearchPage, // inject method for testing here
ResultReturned,
WorkerFinished);
_pageManagers[index].Dispose(); // get rid of old one
_pageManagers[index] = newPageManager;
_pageManagers[index].RunWorkerAsync();
}
/// <summary>
/// Called by PageManagers when their task is complete.
/// If there is still work to be done, replaces the finished
/// pageManager with a new one (creates a new thread)
/// Exits if all work is complete.
/// </summary>
/// <param name="obj">sender object</param>
/// <param name="args">Result<AmazonItem></param>
public void WorkerFinished(object obj, RunWorkerCompletedEventArgs args)
{
// See if any of the workers are reporting that they're out
// of results.
bool outOfResults = _pageManagers.Any(i => i.WorkStatus ==
PageManager.Status.NoResults);
// If so, don't deploy another thread.
if (outOfResults)
{
string msg = "PageManager reporting no more results;";
msg += " not deploying another thread.";
Debug.WriteLine(msg);
return;
}
// Or if there are no threads that are marked "Working",
// we are done
if (IsWorkFinished())
{
HaltAllOngoingWork();
return;
}
if (args == null) return;
if (args.Error != null)
{
Debug.WriteLine(args.Error.Message);
return;
}
if (args.Result == null ||
args.Result.GetType() != typeof(PageManager)) return;
// If this PageManager is done but we haven't hit our
// target number of results, we should spawn a new thread
PageManager finished = (PageManager)args.Result;
// Get the index of the PageManager whose search page number
// matches the one that just finished (we're going to replace
// it with a new PageManager)
int index =_pageManagers.ToList().FindIndex(i =>
i.PageNumber == finished.PageNumber);
// Increment the variable that tracks the
// highest page number we've searched so far
// TODO: since page number is shared state, there is a
// slight chance that two PageManagers might hit this
// code
_pageNumber += 1;
// Start searching a new page
PageManager newPageManager = new PageManager(
_pageNumber,
_searchCriteria,
Scraper.LoadSearchPage, // inject method for testing here
ResultReturned,
WorkerFinished);
_pageManagers[index].Dispose(); // get rid of old one
_pageManagers[index] = newPageManager;
_pageManagers[index].RunWorkerAsync();
}
