private string GetPlot(string page)
{
string _res = BaseCollector.GetItem(page, m_CountryFactory.RE_Plot + "</h2>.*?<p>(?<Plot>[^<]+)", "Plot", RegexOptions.Singleline | RegexOptions.IgnoreCase);
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex(string.Format("<h5>{0}</h5>.*?info-content\">(.*?)</div>", m_CountryFactory.RE_Plot), RegexOptions.Multiline | RegexOptions.IgnoreCase).Match(page.Replace("\n", ""));
if (match.Success)
{
string _result = HttpUtility.HtmlDecode(Regex.Replace(match.Groups[1].Value, "<a class.*?</a>", "")).Trim().TrimEnd(new char[] { '|' }).TrimEnd(new char[0]);
_result = Regex.Replace(_result, "<a href=\"/title.*?</a>", "");
int _start = _result.LastIndexOf(".");
if (_start >= 0)
{
try
{
_result = _result.Substring(0, _start + 1);
}
catch { }
}
return(_result);
}
return(string.Empty);
}
}
public static BaseCollector GetNewCollectorObject(string collectorName)
{
// be careful here coz due to obfuscation u can't rely on type name
Type t = BaseCollector.GetCollectorType(collectorName);
//Create the object, cast it and return it to the caller
return((BaseCollector)Activator.CreateInstance(t));
}
public CollectorThreadParams(BaseCollector collector, string keywords, string imdbid, bool skipImages, AutoResetEvent eventDone)
{
Collector = collector;
Keywords = keywords;
ImdbId = imdbid;
Event = eventDone;
SkipImages = skipImages;
}
public BaseCollector GetCollectorByParam(CollectorType itemtype, BaseParamter baseParamter)
{
BaseCollector collector = null;
string collectType = EnumExtension.GetCollectorType(itemtype);
string collectNameSpace = collectType.Substring(0, collectType.LastIndexOf("."));
collector = CollectorFactory.CreateInstance(collectNameSpace + "," + collectType, baseParamter);
return(collector);
}
private string GetMetascore(string page)
{
string _res = BaseCollector.GetItem(page, @"Metascore: <a href=""criticreviews\?ref_=tt_ov_rt"" title=""(?<Metascore>\d+)\s", "Metascore", RegexOptions.Singleline).Trim();
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
return(string.Empty);
}
private string GetTrailerLink(string page)
{
string _res = BaseCollector.GetItem(page, @"<a href=""(?<Trailer>/video/imdb/vi\d+/\?ref_=tt_ov_vi)"" class=""btn2 btn2_text_on large title-trailer", "Trailer");
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
return(string.Empty);
}
}
public CollectorNode(List <ResultItemBase> results, bool isSelected, BaseCollector collector)
{
Results = new ObservableCollection <ResultItemBase>();
if (results != null && results.Count > 0)
{
Results.AddRange(results);
}
IsExpanded = true;
IsSelected = isSelected;
IsEnabled = true;
SearchTime = string.Empty;
Collector = collector;
Name = collector != null ? collector.CollectorName : Name;
}
//public bool Insert(BaseCollector collector)
//{
// try
// {
// BaseParamter paramter = collector.Paramter;
// //string typeStr =paramter.GetType().ToString();
// using (MySql.Data.MySqlClient.MySqlConnection conn = new MySql.Data.MySqlClient.MySqlConnection())
// {
// conn.ConnectionString = ConfigurationManager.AppSettings["MySqlConnectionStr"];
// MySql.Data.MySqlClient.MySqlCommand cmd = new MySql.Data.MySqlClient.MySqlCommand();
// cmd.CommandText = "insert into tb_monitoritem(name,ip,note,itemtype,paramter) values(@name,@ip,@note,@itemtype,@paramter)";
// cmd.CommandType = CommandType.Text;
// cmd.Parameters.Add("@name", MySql.Data.MySqlClient.MySqlDbType.VarChar);
// cmd.Parameters.Add("@ip", MySql.Data.MySqlClient.MySqlDbType.VarChar);
// cmd.Parameters.Add("@note", MySql.Data.MySqlClient.MySqlDbType.VarChar);
// cmd.Parameters.Add("@itemtype", MySql.Data.MySqlClient.MySqlDbType.Int32);
// cmd.Parameters.Add("@paramter", MySql.Data.MySqlClient.MySqlDbType.VarBinary);
// cmd.Parameters[0].Value = Guid.NewGuid();
// cmd.Parameters[1].Value = paramter.Host;
// cmd.Parameters[2].Value = cmd.Parameters[0].Value;
// cmd.Parameters[3].Value = 1;
// cmd.Parameters[4].Value = _ser.Serialize(paramter);
// cmd.Connection = conn;
// conn.Open();
// int affectedrows = cmd.ExecuteNonQuery();
// cmd.Dispose();//此处可以不用调用,
// conn.Close();// 离开 using 块, connection 会自行关闭
// }
// return true;
// }
// catch (Exception ex)
// {
// throw new Exception(ex.Message);
// }
//}
//public List<CollectorInfo> GetList(string condition)
//{
// List<CollectorInfo> list = new List<CollectorInfo>();
// using (MySql.Data.MySqlClient.MySqlConnection conn = new MySql.Data.MySqlClient.MySqlConnection())
// {
// conn.ConnectionString = ConfigurationManager.AppSettings["MySqlConnectionStr"];
// conn.Open();
// MySql.Data.MySqlClient.MySqlCommand cmd = new MySql.Data.MySqlClient.MySqlCommand();
// cmd.CommandType = CommandType.Text;
// cmd.CommandText = "select name,ip,note,itemtype,paramter from tb_monitoritem";
// cmd.Connection = conn;
// System.Data.Common.DbDataReader reader = cmd.ExecuteReader();
// byte[] buffer = null;
// if (reader.HasRows)
// {
// while (reader.Read())
// {
// CollectorInfo info = new CollectorInfo();
// int itemtype = reader.GetInt32(3);
// long len = reader.GetBytes(4, 0, null, 0, 0);
// buffer = new byte[len];
// len = reader.GetBytes(4, 0, buffer, 0, (int)len);
// info.Collector = GetCollectorByBytes(itemtype, buffer);
// info.LastTime = DateTime.MinValue;
// list.Add(info);
// }
// }
// }
// return list;
//}
#endregion
public BaseCollector GetCollectorByBytes(CollectorType itemtype, byte[] bytes)
{
BaseCollector collector = null;
string collectType = EnumExtension.GetCollectorType(itemtype);
string collectNameSpace = collectType.Substring(0, collectType.LastIndexOf("."));
BaseParamter baseParamter = (BaseParamter)Activator.CreateInstance(EnumExtension.GetParamterType((CollectorType)itemtype));
baseParamter = (DBParamter)_ser.Deserialize(bytes);
collector = CollectorFactory.CreateInstance(collectNameSpace + "," + collectType, baseParamter);
return(collector);
}
private List <string> GetCountries(string page)
{
List <string> list = new List <string>();
list = BaseCollector.GetItems(page, @"<a href=""/country/[^\?]+\?ref_=tt_dt_dt"" itemprop='url'>(?<Country>[^<]+)</a>", "Country");
if (list.Count != 0)
{
return(list);
}
else
{
Regex regex = new Regex(string.Format("<h5>{0}</h5>.*?info-content\">(.*?)</div>", m_CountryFactory.RE_Countries), RegexOptions.Multiline | RegexOptions.IgnoreCase);
Match match = regex.Match(page.Replace("\n", "").Replace("\r", ""));
if (match.Success)
{
//regex = new Regex("<a\\shref=\"/Sections/Countries.*?>(.*?)</a>", RegexOptions.Multiline | RegexOptions.IgnoreCase);
regex = new Regex("<a href=\"/country.*?>(.*?)</a>", RegexOptions.Multiline | RegexOptions.IgnoreCase);
if (regex.Matches(match.Groups[1].Value).Count > 0)
{
foreach (Match match2 in regex.Matches(match.Groups[1].Value))
{
list.Add(HttpUtility.HtmlDecode(match2.Groups[1].Value));
}
}
else
{
regex = new Regex("<a\\shref=\"/Sections/Countries.*?>(.*?)</a>", RegexOptions.Multiline | RegexOptions.IgnoreCase);
if (regex.Matches(match.Groups[1].Value).Count > 0)
{
foreach (Match match2 in regex.Matches(match.Groups[1].Value))
{
list.Add(HttpUtility.HtmlDecode(match2.Groups[1].Value));
}
}
else
{
string[] _s = match.Groups[1].Value.Split('|');
foreach (string _item in _s)
{
list.Add(HttpUtility.HtmlDecode(_item.Trim().Trim('\r').Trim()));
}
}
}
}
return(list);
}
}
private string GetRunTime(string page)
{
string _res = BaseCollector.GetItem(page, "(?<Runtime>\\d{1,3})\\smin", "Runtime");
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex(@"(\d{1,3})\smin", RegexOptions.Multiline | RegexOptions.IgnoreCase).Match(page);
if (match.Success)
{
return(match.Groups[1].Value);
}
return(string.Empty);
}
}
private string GetReleaseDateFast(string page)
{
string _res = BaseCollector.GetItem(page, m_CountryFactory.RE_ReleaseDateFast + @"</h4>.*?(?<ReleaseDate>[^\(]*)", "ReleaseDate");
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex(m_CountryFactory.RE_ReleaseDateFast + @"</h5>.*\n.*\n([^\(]*)", RegexOptions.Multiline | RegexOptions.IgnoreCase).Match(page);
if (match.Success)
{
return(match.Groups[1].Value);
}
return(string.Empty);
}
}
private string GetMPAA(string page)
{
string _res = BaseCollector.GetItem(page, "contentRating\">(?<MPAA>[^<]+)<", "MPAA").Trim().Replace("\r", "").Replace("\n", "").Trim();
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex("MPAA</a>:</h5>[\\r\\n]*<div class=\"info-content\">([^<]+)</div", RegexOptions.Singleline | RegexOptions.IgnoreCase).Match(page);
if (match.Success)
{
return(HttpUtility.HtmlDecode(match.Groups[1].Value.Replace("\r", "").Replace("\n", "").Trim()));
}
return(string.Empty);
}
}
private string GetTagline(string page)
{
string _res = BaseCollector.GetItem(page, "Taglines:</h4>.*?(?<Tagline>[^<]+)<", "Tagline").Trim().Replace("\r", "").Replace("\n", "").Trim();
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex("Tagline:</h5>.*?info-content\">(.*?)<", RegexOptions.Singleline | RegexOptions.IgnoreCase).Match(page);
if (match.Success)
{
return(HttpUtility.HtmlDecode(match.Groups[1].Value.Replace("\r", "").Replace("\n", "").Trim()));
}
return(string.Empty);
}
}
private string GetRating(string page)
{
string _res = BaseCollector.GetItem(page, @"itemprop=""ratingValue"">(?<Rating>[\d\.]+)</span></strong><span class=""mellow"">/<span itemprop=""bestRating"">10", "Rating");
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex(@"<b>(\d.*?)/10</b>", RegexOptions.Multiline | RegexOptions.IgnoreCase).Match(page);
if (match.Success)
{
return(match.Groups[1].Value);
}
return(string.Empty);
}
}
private string GetYear(string page)
{
string _res = BaseCollector.GetItem(page, @"<title>[^\(]+\((?<Year>\d{4})|<title>[^\(]+\(.*?(?<Year>\d{4})", "Year");
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex(@"<title>.*?\s\((\d{4})", RegexOptions.Multiline | RegexOptions.IgnoreCase).Match(page);
if (match.Success)
{
return(match.Groups[1].Value);
}
return(string.Empty);
}
}
private string GetOriginalTitle(string page)
{
string _res = BaseCollector.GetItem(page, "class=\"title-extra\">(?<OriginalTitle>[^<^\\(]+)", "OriginalTitle").Trim().Trim(new char[] { '"' });
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex("class=\"title-extra\">(?<OriginalTitle>[^<^\\(]+)", RegexOptions.Singleline | RegexOptions.IgnoreCase).Match(page);
if (match.Success)
{
return(HttpUtility.HtmlDecode(match.Groups["OriginalTitle"].Value.Trim()).Trim(new char[] { '"' }));
}
return(string.Empty);
}
}
private string GetTitle(string page)
{
string _res = BaseCollector.GetItem(page, @"<title>(?<Title>[^\(]+)", "Title").Trim().Trim(new char[] { '"' }).Replace("IMDb - ", "");
if (!string.IsNullOrEmpty(_res))
{
return(_res);
}
else
{
Match match = new Regex(@"<title>(.*?)\s\(\d{4}", RegexOptions.Multiline | RegexOptions.IgnoreCase).Match(page);
if (match.Success)
{
return(HttpUtility.HtmlDecode(match.Groups[1].Value.Trim()).Trim(new char[] { '"' }));
}
return(string.Empty);
}
}
private List <string> GetActors(string page)
{
List <string> _res = BaseCollector.GetItems(page, @"href=""/name/nm[0-9]+/\?ref_=tt_cl_t\d+"" itemprop='[^']+'>\s*(<span class=""itemprop"" itemprop=""name"">)?(?<Actor>[^<]+)</", "Actor", RegexOptions.Singleline | RegexOptions.IgnoreCase);
if (_res.Count != 0)
{
return(_res);
}
else
{
MatchCollection matchs = new Regex("<td class=\"nm\"><a href=\"(.*?)\">(.*?)</a>", RegexOptions.Multiline | RegexOptions.IgnoreCase).Matches(page);
List <string> list = new List <string>();
foreach (Match match in matchs)
{
list.Add(HttpUtility.HtmlDecode(match.Groups[2].Value.Trim()));
}
return(list);
}
}
public MovieInfo QueryPreferredCollector(string imdbid, string keywords)
{
MovieInfo _result = null;
// find the type of the preferred collector
BaseCollector _tmp = BaseCollector.GetMovieCollector(FileManager.Configuration.Options.PreferedInfoCollector);
if (_tmp != null)
{
// create a new collector (to not affect the old one)
BaseCollector _prefCollector = GetNewCollectorObject(FileManager.Configuration.Options.PreferedInfoCollector); //BaseCollector.GetMovieCollector(FileManager.Configuration.Options.PreferedInfoCollector);
Loggy.Logger.Debug("Detected prefinfocollector: " + _prefCollector.CollectorName);
//BaseCollector _prefCollector = BaseCollector.GetMovieCollectorObject(FileManager.Configuration.Options.PreferedInfoCollector);
if (_prefCollector != null)
{
_prefCollector.IMDBID = imdbid;
_prefCollector.CurrentMovie = new MovieItem(MoviePath);
_result = _prefCollector.QueryMovieInfo(_prefCollector.IMDBID);
if ((_result == null) || (string.IsNullOrEmpty(_result.Name)))
{
// ask user
_prefCollector.GetResults(keywords, _prefCollector.IMDBID, true);
BindingList <MovieInfo> _candidates = new BindingList <MovieInfo>();
foreach (ResultItemBase _rib in _prefCollector.ResultsList)
{
if (!string.IsNullOrEmpty(imdbid) && !string.IsNullOrEmpty(_rib.MovieInfo.IMDBID) && imdbid != _rib.MovieInfo.IMDBID)
{
continue; // IMDBID does not match with ours
}
if (!_candidates.Contains(_rib.MovieInfo, new MovieInfoComparer()))
{
_candidates.Add(_rib.MovieInfo);
}
}
ChooseMovieDialogResult _dresult = ChooseMovieFromIMDb.GetCorrectMovie(null, _candidates, "", false);
_result = _dresult != null ? _dresult.MovieInfo : null;
}
}
}
return(_result);
}
private List <string> GetCompanies(string page)
{
List <string> list = BaseCollector.GetItems(page, @"href=""/company/co\d+\?ref_=tt_dt_co"" itemprop='url'>(<span class=""itemprop"" itemprop=""name"">)?(?<Company>[^<]+)</", "Company", RegexOptions.Singleline);
if (list.Count == 0)
{
// old layout
list = new List <string>();
Regex regex = new Regex(string.Format("<h5>{0}</h5>.*?info-content\">(.*?)</div>", m_CountryFactory.RE_Companies), RegexOptions.Multiline | RegexOptions.IgnoreCase);
Match match = regex.Match(page.Replace("\n", ""));
if (match.Success)
{
regex = new Regex("<a href=\"/company.*?>(.*?)</a>", RegexOptions.Multiline | RegexOptions.IgnoreCase);
foreach (Match match2 in regex.Matches(match.Groups[1].Value))
{
list.Add(HttpUtility.HtmlDecode(match2.Groups[1].Value));
}
}
}
return(list);
}
private List <string> GetDirectors(string page)
{
// new layout
List <string> list = new List <string>();
//Regex regex = new Regex(string.Format("{0}(.*?)</div>", m_CountryFactory.RE_Directors), RegexOptions.Singleline | RegexOptions.IgnoreCase);
//Match match = regex.Match(page.Replace("\n", ""));
//if (match.Success)
//{
list = BaseCollector.GetItems(page, @"href=""/name/nm\d+/\?ref_=tt_ov_\wr"" itemprop='url'>(<span class=""itemprop"" itemprop=""name"">)?(?<Director>[^<]+)</", "Director");
//}
if (list.Count != 0)
{
return(list.Take(1).ToList());
}
else
{
list = BaseCollector.GetItems(page, "itemprop=\"director\".*?>(?<Director>[^<]+)</a", "Director", RegexOptions.Singleline | RegexOptions.IgnoreCase);
if (list.Count != 0)
{
return(list);
}
else
{
List <string> list2 = new List <string>();
Regex regex2 = new Regex(string.Format("<h5>{0}</h5>.*?info-content\">(.*?)</div>", m_CountryFactory.RE_Directors), RegexOptions.Multiline | RegexOptions.IgnoreCase);
Match match2 = regex2.Match(page.Replace("\n", ""));
if (match2.Success)
{
Regex regex3 = new Regex("<a\\shref=\"/name/nm\\d{7}/\".*?\">(.*?)</a>", RegexOptions.Multiline | RegexOptions.IgnoreCase);
foreach (Match match22 in regex3.Matches(match2.Groups[2].Value))
{
list2.Add(HttpUtility.HtmlDecode(match22.Groups[1].Value));
}
}
return(list2);
}
}
}
private List <string> GetGenres(string page)
{
// new layout
List <string> _result = BaseCollector.GetItems(page, @"<a href=""/genre/[^\?]+\?ref_=tt_ov_inf"" >(<span class=""itemprop"" itemprop=""genre"">)?(?<Genre>[^<]+)</", "Genre", RegexOptions.Singleline);
if (_result.Count != 0)
{
return(_result);
}
else
{
// old layout
List <string> list = new List <string>();
MatchCollection matchs = new Regex("<a href=\"/Sections/Genres/[^/]*/\">([^\"]*)</a>", RegexOptions.Singleline | RegexOptions.IgnoreCase).Matches(page);
if (matchs.Count != 0)
{
foreach (Match match in matchs)
{
list.Add(HttpUtility.HtmlDecode(match.Groups[1].Value.Trim()));
}
}
else
{
matchs = new Regex(m_CountryFactory.RE_Genres, RegexOptions.Singleline | RegexOptions.IgnoreCase).Matches(page);
foreach (Match match in matchs)
{
string[] _s = match.Groups[1].Value.Split('|');
foreach (string _item in _s)
{
list.Add(HttpUtility.HtmlDecode(_item.Trim().Trim('\r').Trim()));
}
}
}
return(list);
}
}
public void SetCollector(BaseCollector collector)
{
Collector = collector;
Name = collector != null ? collector.CollectorName : Name;
}
