private async void Worker()
{
for (int i = parserSettings.StartPoint; i <= parserSettings.EndPoint; i++)
{
// raise completed event and exit if parser is deactivate
if (!isActive)
{
OnCompleted?.Invoke(this);
return;
}
// get page source by page id
var source = await loader.GetSourceByPageId(i);
// create html document from source
var document = await new HtmlParser().ParseDocumentAsync(source);
// get urls for this page
var result = parser.CollectUrls(document);
// if urls is empty then get next page
if (result.Count() == 0)
{
continue;
}
// create urls queue
var urlQueue = new ConcurrentQueue <string>(result);
// parse each url in multithreaded mode
Task[] tasks = new Task[parserSettings.ThreadCount];
for (int j = 0; j < parserSettings.ThreadCount; j++)
{
tasks[j] = ParseEachUrl(urlQueue);
}
// wait for all tasks to complete
Task.WaitAll(tasks);
// raise page completed event
OnPageCompleted?.Invoke(this, i);
}
// raise completed event
OnCompleted?.Invoke(this);
// parser is deactivate
isActive = false;
}
private async void Worker()
{
for (int i = parserSettings.StartPoint; i <= parserSettings.EndPoint; i++)
{
if (!isActive)
{
OnCompleted?.Invoke(this);
return;
}
var source = await loader.GetSourceByPageId(i);
var domParser = new HtmlParser();
var document = await domParser.ParseAsync(source);
var result = parser.Parse(document);
OnNewData.Invoke(this, result);
}
OnCompleted?.Invoke(this);
isActive = false;
}