private static void TestTaskmanager()
{
var tm = new TaskManager();
var brContextFactory = new BrowsingContextFactory();
var link = "https://www.olx.ua/nedvizhimost/arenda-kvartir/dolgosrochnaya-arenda-kvartir/dnepr/";
var getPagesCountTask = new GetPagesCountActivity(link, brContextFactory.GetNew()).AsSimpleTask(tm);
var printPagesCountTask = new SimpleTask(() => System.Console.WriteLine("PagesCount: {0}", getPagesCountTask.Result));
printPagesCountTask.ContinueWith(() =>
{
var linkPage = "?page={0}";
var tasks = new ITask <IEnumerable <PreviewAdModel> > [getPagesCountTask.Result];
for (int i = 1; i <= getPagesCountTask.Result; i++)
{
var procLink = i == 1 ? link : link + string.Format(linkPage, i);
tasks[i - 1] = new GetPreviewModelsActivity(procLink, brContextFactory).AsSimpleTask(tm);
}
tasks.ForEach(s => s.ContinueWith(() => { return(new SimpleTask(() => { s.Result.ForEach(k => System.Console.WriteLine(k.Price)); })); }));
/* tasks.ContinueWith(new SimpleTask(() =>
* {
* System.Console.WriteLine("Completed");
* }));*/
return(tasks);
}).ContinueWith(() => { return(new SimpleTask(() => System.Console.WriteLine("Completed"))); });
getPagesCountTask.ContinueWith(printPagesCountTask);
tm.AddTask(getPagesCountTask);
tm.Start();
}
private List <PreviewAdModel> GetPreviewModels()
{
List <PreviewAdModel> res = new List <PreviewAdModel>();
for (int j = 0; j < _maxEfforts; j++)
{
//TODO: Create logging
Console.WriteLine("Processing page: {0}", _link);
var regexPattern = "ID(.*).html";
var brContext = _brContextFactory.GetNew();
var task = brContext.OpenAsync(_link);
var doc = task.Result;
var offers = doc.QuerySelectorAll(".offer");
if (!offers.Any())
{
Thread.Sleep(3000);
}
else
{
var k = 1;
foreach (var offer in offers)
{
try
{
// TODO: Create logging
Console.WriteLine("\tProcessing offer: {0}", k);
var resLink = new PreviewAdModel();
resLink.AdId = uint.Parse(offer.Children.First().Attributes["data-id"].Value);
resLink.TempId =
Regex.Match(offer.QuerySelector(".detailsLink").Attributes["href"].Value, regexPattern).Groups[1
].Value;
var priceString = offer.QuerySelector(".price").Children.First().InnerHtml;
resLink.Price = double.Parse(Regex.Replace(priceString, "[А-Яа-яA-Za-z$ .]", ""));
resLink.PhoneNumbers.AddRange(GetPhoneNumbers(resLink.TempId, _brContextFactory.GetNew()));
res.Add(resLink);
k++;
}
catch (Exception ex)
{
//TODO: Create logging
Console.WriteLine(ex.Message);
System.Diagnostics.Debug.WriteLine(ex.ToString());
}
}
break;
}
}
return(res);
}
private static List <PreviewAdModel> GetPreviewModels(string link)
{
var res = new List <PreviewAdModel>();
var brContextFactory = new BrowsingContextFactory();
var linkPage = "?page={0}";
var pagesCount = GetPagesCount(link, brContextFactory.GetNew());
for (int i = 1; i <= pagesCount; i++)
{
var procLink = i == 1 ? link : link + string.Format(linkPage, i);
res.AddRange(GetPreviewModels(brContextFactory, procLink));
System.Console.WriteLine();
}
return(res);
}
private static List <AdModel> GetLinks(string link)
{
var res = new List <AdModel>();
var brContextFactory = new BrowsingContextFactory();
Logger.Info(String.Format("Start processing for: {0}", link));
var linkPage = "?page={0}";
var regexPattern = "ID(.*).html";
var brContext = brContextFactory.GetNew();
var processedIds = new List <string>();
var pagesCount = GetPagesCount(link, brContextFactory.GetNew());
Logger.Info(String.Format("Total pages: {0}", pagesCount));
for (int i = 1; i <= pagesCount; i++)
{
for (int j = 0; j < 3; j++)
{
Logger.Info("Processing page: {0}", i);
var procLink = i == 1 ? link : link + String.Format(linkPage, i);
var task = brContext.OpenAsync(procLink);
var doc = task.Result;
var rawLinks = doc.QuerySelectorAll(".marginright5.link.linkWithHash.detailsLink");
if (rawLinks.Count() == 0)
{
Thread.Sleep(3000);
}
else
{
var k = 1;
foreach (var rawLink in rawLinks)
{
try
{
Logger.Debug(String.Format("\tProcessing link: {0}", k));
var resLink = new AdModel {
Link = rawLink.Attributes["href"].Value
};
resLink.TempId = Regex.Match(resLink.Link, regexPattern).Groups[1].Value;
if (!processedIds.Contains(resLink.TempId))
{
var priceTask = brContextFactory.GetNew().OpenAsync(rawLink.Attributes["href"].Value);
resLink.PhoneNumbers.AddRange(GetPhoneNumbers(resLink.TempId, brContextFactory.GetNew()));
var docAd = priceTask.Result;
var priceString = docAd.QuerySelector(".pricelabel.tcenter").Children[0].InnerHtml;
var isPrivateString = docAd.QuerySelector("#offerdescription > div.clr.descriptioncontent.marginbott20 > table > tbody > tr:nth-child(1) > td:nth-child(1) > table > tbody > tr > td > strong > a").InnerHtml;
isPrivateString = Regex.Replace(isPrivateString, "[\t\n]", "");
resLink.IsPrivate = !isPrivateString.Equals("Бизнес");
string value = Regex.Replace(priceString, "[А-Яа-яA-Za-z$ .]", "");
resLink.Price = Double.Parse(value);
string rooms = docAd.QuerySelector("#offerdescription > div.clr.descriptioncontent.marginbott20 > table > tbody > tr:nth-child(2) > td.col > table > tbody > tr > td > strong").InnerHtml;
resLink.Rooms = Int32.Parse(Regex.Replace(rooms, "[А-Яа-яA-Za-z .]", ""));
res.Add(resLink);
processedIds.Add(resLink.TempId);
k++;
}
else
{
Logger.Warn("Dublicate");
}
}
catch (Exception ex)
{
Logger.Error(ex.ToString());
}
}
break;
}
}
Console.WriteLine();
}
return(res);
}