private static string CleanHtmlPage(string html)
{
IMDbRegEx imDbRegEx = new IMDbRegEx();
imDbRegEx.SetRegExPatterns();
Debugger.LogMessageToFile
("Cleaning IMDb result html ");
string trimmedHtml = imDbRegEx.GetRegExString
(html, imDbRegEx.TrimmedHTMLpattern);
return trimmedHtml;
}
internal static Match GetMovieTitleRegexMatch
(string imdbID, string trimmedHtml)
{
Debugger.LogMessageToFile
("[IMDb Movie Details Downloader]" +
" Getting Title regex match...");
var imDbRegEx = new IMDbRegEx();
imDbRegEx.SetRegExPatterns();
Match match = imDbRegEx.GetRegExMatch
(trimmedHtml,
IMDbRegEx.TitlePatternOriginal);
try
{
string filmTitle
= match.Groups[1].Captures[0].Value;
return match;
}
catch (Exception)
{
match = imDbRegEx.GetRegExMatch
(trimmedHtml,
IMDbRegEx.TitlePatternPrimary);
try
{
string filmTitle
= match.Groups[1].Captures[0].Value;
return match;
}
catch (Exception e)
{
Debugger.LogMessageToFile
("[IMDb Movie Details Downloader] " +
"The IMDb Movie Details Downloader was unable" +
" to extract the movie title " +
"for the movie with IMDb ID: "
+ imdbID + ".");
return match;
}
}
}
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;
}
