internal static IMDbMovie MineFilmDetailsFromAdditionalPages
(bool showProgress, IMDbFilmDetails filmDetails,
IMDbMovie movie, string movieUrl)
{
if (!showProgress)
return movie;
string creditsUrl = movieUrl + "fullcredits";
string longOverviewUrl = movieUrl + "plotsummary";
string goofUrl = movieUrl + "goofs";
string triviaUrl = movieUrl + "trivia";
string quotesUrl = movieUrl + "quotes";
MainImportingEngine.ThisProgress.Progress
(MainImportingEngine.CurrentProgress, "Getting actors...");
//filmDetails.GetActorsUsingRegex(ref movie, creditsUrl);
MainImportingEngine.ThisProgress.Progress
(MainImportingEngine.CurrentProgress, "Getting plot summary...");
//filmDetails.GetLongOverview(ref movie, longOverviewUrl);
//MainImportingEngine.ThisProgress.Progress
// (MainImportingEngine.CurrentProgress, "Getting trivia...");
//filmDetails.GetTrivia(ref movie, triviaUrl);
//MainImportingEngine.ThisProgress.Progress
// (MainImportingEngine.CurrentProgress, "Getting goofs...");
//filmDetails.GetGoofs(ref movie, goofUrl);
//MainImportingEngine.ThisProgress.Progress
// (MainImportingEngine.CurrentProgress, "Getting quotes...");
//filmDetails.GetQuotes(ref movie, quotesUrl);
//TODO: Get these additional IMDb film details
//Get Stars
//Get Awards
//Get Credits
//Get Plot Keywords
//Get Official Websites
//Get Production Country
//Get Language
//Get Box Office Budget
//Get Box Office Gross
//Get Color
//Get Aspect Ratio
//Get Connections
//Get Soundtracks
//Get User Review
return movie;
}
private static IMDbMovie MineFilmDetailsFromMainPage
(IMDbFilmDetails filmDetails,
string trimmedHtml, IMDbMovie movie )
{
IMDbRegEx imDbRegEx = new IMDbRegEx();
imDbRegEx.SetRegExPatterns();
IMDbFilmDetails.MineProductionYearUsingRegex(movie, trimmedHtml, imDbRegEx);
//IMDbFilmDetails.MineProductionYearUsingXpath(movie, trimmedHtml);
IMDbFilmDetails.GetActorsUsingXpath(movie, trimmedHtml);
IMDbFilmDetails.GetReleaseDate(movie, trimmedHtml, imDbRegEx);
IMDbFilmDetails.GetTagline(movie, trimmedHtml, imDbRegEx);
IMDbFilmDetails.GetRuntime(movie, trimmedHtml, imDbRegEx);
IMDbFilmDetails.GetRating(movie, trimmedHtml, imDbRegEx);
IMDbFilmDetails.ExtractRatingDescription(movie, trimmedHtml, imDbRegEx);
IMDbFilmDetails.GetReview(movie, trimmedHtml, imDbRegEx);
IMDbFilmDetails.GetStudio(movie, trimmedHtml, imDbRegEx);
//IMDbFilmDetails.MineOverviewUsingRegex(movie, trimmedHtml, imDbRegEx);
IMDbFilmDetails.MineOverviewUsingXpath(movie, trimmedHtml);
//filmDetails.MineDirectorUsingRegex(ref movie, trimmedHtml);
filmDetails.MineDirectorUsingXpath(ref movie, trimmedHtml);
//filmDetails.MineWriterUsingRegex(ref movie, trimmedHtml);
filmDetails.MineWriterUsingXpath(ref movie, trimmedHtml);
filmDetails.GetGenres(ref movie, trimmedHtml);
return movie;
}
//TODO: Remove Debugger logs for all movie details.
//TODO: Add indivindual settings for each extra film detail type.
internal static IIMDbMovie DownloadMovieDetailsFromIMDb
(string imdbID, bool showProgress )
{
var movie = new IMDbMovie();
var filmDetails = new IMDbFilmDetails();
var webClient = new WebClient();
try
{
var movieUrl = IMDbMovieDetailsDownloaderHelpers
.ConstructMovieUrl(imdbID);
var html = IMDbMovieDetailsDownloaderHelpers
.DownloadBaseTitleHtml
(movieUrl, webClient);
if (IMDbMovieDetailsDownloaderHelpers
.CheckForValidHtmlData(html))
return null;
//TODO: IMDb Movie Details Extraction does not work if the main html page trimming is activated. The html trimming regex should be fixed.
//string trimmedHTML = CleanHtmlPage(html);
//if (String.IsNullOrEmpty(trimmedHTML))
// trimmedHTML = html;
string trimmedHtml = html;
//Debugger.LogMessageToFile("Base Title html: " + trimmedHtml);
movie.Title
= IMDbMovieDetailsDownloaderHelpers
.ExtractFilmTitle
(imdbID, trimmedHtml);
movie = MineFilmDetails
(showProgress, trimmedHtml,
movieUrl, movie, filmDetails);
}
catch (Exception e)
{
Debugger.LogMessageToFile
("The GetMovieInfo() method returned an exception: "
+ Environment.NewLine + e );
}
return movie;
}
private static IMDbMovie MineFilmDetails
(bool showProgress,
string trimmedHtml,
string movieUrl,
IMDbMovie movie,
IMDbFilmDetails filmDetails)
{
if (showProgress)
{
MainImportingEngine.ThisProgress.Progress
(MainImportingEngine.CurrentProgress,
"Downloading details from IMDb for '"
+ movie.Title + "'...");
}
movie = MineFilmDetailsFromMainPage
(filmDetails, trimmedHtml, movie);
movie = IMDbMovieDetailsDownloaderHelpers
.MineFilmDetailsFromAdditionalPages
(showProgress, filmDetails, movie, movieUrl);
return movie;
}
