/* //альтернативное решение проблемы с кодировкой
* WebClient wc = new WebClient();
* string str = wc.DownloadString(url);
* HtmlDocument htmlDocument = new HtmlDocument();
* htmlDocument.LoadHtml(str);*/
public IEnumerable <TorrentMovie> GetMovies()
{
Log.Information($"MoviesDownloader.MoviesGetterService.GetMovies() start at {DateTime.Now}{Environment.NewLine}");
var urls = GetPaths();
foreach (var url in urls)
{
var nodes = GetHtmlNodesTagTr(url);
if (nodes != null)
{
foreach (var node in nodes)
{
var NodeId = node.Id.ToString();
if (NodeId == "tr-5836841")//ignore post"Открылся новый сайт новинок"
{
continue;
}
var SubPathDownLoad = "";
try
{
SubPathDownLoad = "https://rutracker.org/forum/"
+ node.SelectSingleNode(".//a[@class='torTopic bold tt-text']").Attributes["href"].Value;
}
catch (Exception ex)
{
Log.Error($"MoviesDownloader.MoviesGetterService.GetMovies(){Environment.NewLine}" +
$"Get path of movies form url was fail with exception:{Environment.NewLine}{ex.Message}" +
$"{Environment.NewLine}url={url}{Environment.NewLine}NodeId={NodeId}");
continue;
}
Log.Information($"MoviesDownloader.MoviesGetterService.GetMovies(){Environment.NewLine}" +
$"Parent's page address{Environment.NewLine}{url}{Environment.NewLine}" +
$"Address is being processed{Environment.NewLine}{SubPathDownLoad}{Environment.NewLine}" +
$"{DateTime.Now}{Environment.NewLine}");
var processedMovie = _unitOfWork.torrentMove.AsQueryable().FirstOrDefault(p => p.pathDownLoad.Equals(SubPathDownLoad)); //Check to be movie in DB
if (processedMovie != null) //if movie exist in the DB
{
var lastPost = node.SelectSingleNode(".//td[@class='vf-col-last-post tCenter small nowrap']").SelectSingleNode(".//p").InnerText;
if (processedMovie.lastPost != lastPost || processedMovie.amountOfComments == 0 || processedMovie.commentIndex > 2000)
{
processedMovie.lastPost = lastPost;
_saveCommentsToDb.SaveCommens(processedMovie);
_commentHanlder.GetCommentIndex(processedMovie);
}
continue;
}
else
{
try
{
var textFromTegA = node.SelectSingleNode(".//a[@class='torTopic bold tt-text']").InnerText.ToString();
var additInformFromTitle = Regex.Match(textFromTegA, @"(\[.+\])").ToString();
var _title = Regex.Replace(textFromTegA, @"(\[.+\].+)", "");//delete of addition information
var _genre = Regex.Replace(Regex.Match(additInformFromTitle, @"\s[а-я].+,").ToString(), @"(,$)", "");
var _country = Regex.Match(additInformFromTitle, @"[А-Я]{1,3}[a-я]+").ToString();
var _earOfIssue = Regex.Match(additInformFromTitle, @"\d{4}").ToString();
var _videoQuality = Regex.Replace(Regex.Match(additInformFromTitle, @"[A-Z]+.+]").ToString(), "]", "");
var _size = Regex.Replace(node.SelectSingleNode(".//a[@class='small f-dl dl-stub']").InnerText.ToString(), @" ", "");
var _commentIndex = 0;
string _strDownloads = Regex.Replace(node.SelectSingleNode(".//p[@title='Торрент скачан']").InnerText, @"\D+", "").ToString();
int _downloads = Convert.ToInt32(_strDownloads);
var _pathDownLoad = "https://rutracker.org/forum/" + node.SelectSingleNode(".//a[@class='torTopic bold tt-text']").Attributes["href"].Value;
var _lastPost = node.SelectSingleNode(".//td[@class='vf-col-last-post tCenter small nowrap']").Element("p").InnerText;
if (_genre != null)
{
torrentMoves.Add(new TorrentMovie
{
title = _title,
genre = _genre,
country = _country,
earOfIssue = _earOfIssue,
videoQuality = _videoQuality,
size = _size,
commentIndex = _commentIndex,
downloads = _downloads,
pathDownLoad = _pathDownLoad,
movieId = NodeId,
lastPost = _lastPost
});
}
}
catch (Exception ex)
{
Log.Information($"MoviesDownloader.MoviesGetterService.GetMovies(){Environment.NewLine}Error reading html tags or/and creating new TorrentMovie object :{Environment.NewLine}{ex.Message}" +
$"{Environment.NewLine}url={url}{Environment.NewLine}SubPath={SubPathDownLoad}{Environment.NewLine}{DateTime.Now}{Environment.NewLine}");
}
}
}
}
}
Log.Information($"MoviesDownloader.MoviesGetterService.GetMovies() finished at {DateTime.Now}{Environment.NewLine}");
return(torrentMoves);
}
