internal static void IgnoreIrrelevantResults
(Match match, IMDbRegEx imDbRegex, IMDbSearchResult result)
{
//TODO: Also ignore "Making Of" titles.
//TODO: Also ignore documentaries (option).
string extra = imDbRegex.GetMatchValue
(match, "Extra", true);
IgnoreVideoGameResult(result, extra);
}
internal static List<IIMDbSearchResult> SingleMatchAddToSearchResults
(string html, IMDbRegEx imDbRegex, List<IIMDbSearchResult> results)
{
var titleIndexFound = LocateTitleIndex(html);
if (!titleIndexFound)
return results;
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"A single result was found.");
//TODO: Are these RegExes the same as the 'movie result' ones?
const string titleAndYearPattern = @"<title>(?<Title>.*?)\((?<Year>.*?)\)</title>";
const string ImdbIdPattern = @"<a\shref=""/title/(?<IMDbID>[^/]*)/fullcredits""";
Match match = imDbRegex.GetRegExMatch(html, titleAndYearPattern);
#region Return if no matches were found
if (match == null || match.Length == 0)
return results;
#endregion
var result = new IMDbSearchResult
{
Title = imDbRegex.GetMatchValue(match, "Title", true),
Year = imDbRegex.GetMatchValue(match, "Year", true)
};
MineImdbIdFromSingleFilmMatch
(html, imDbRegex, ImdbIdPattern, result);
AddSingleFilmResultToSearchResults
(results, result);
return results;
}
internal static IMDbSearchResult MultipleMatchesMineDetailsOfSingleFilmResult
(Match match, IMDbRegEx imDbRegex)
{
#region Result's details
Debugger.LogMessageToFile(
"[IMDb Conventional Film Search Engine] " +
"New search result was found. " +
"Proceeding to add result to the list of total search results...");
var result = new IMDbSearchResult();
result.IMDb_ID = imDbRegex.GetMatchValue
(match, "IMDbID", true);
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Result IMDb ID: " + result.IMDb_ID);
result.Title = imDbRegex.GetMatchValue
(match, "Title", true);
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Result Title: " + result.Title);
result.Year = imDbRegex.GetMatchValue
(match, "Year", true);
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Result Year: " + result.Year);
string URL = imDbRegex.GetMatchValue
(match, "URL", true);
result.URL = "http://www.imdb.com" + URL;
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Result URL: " + result.URL);
#endregion
return result;
}
internal static void MineImdbIdFromSingleFilmMatch
(string html, IMDbRegEx imDbRegex,
string imdbIdPattern, IIMDbSearchResult result)
{
Match match = imDbRegex.GetRegExMatch(html, imdbIdPattern);
if (match != null && match.Length > 0)
result.IMDb_ID =
imDbRegex.GetMatchValue
(match, "IMDbID", true);
}
internal static void GetRuntime
(IIMDbMovie movie,
string trimmedHTML,
IMDbRegEx imDbRegEx)
{
Debugger.LogMessageToFile
("[IMDb film details downloader]" +
" Extracting Runtime...");
Match match = imDbRegEx.GetRegExMatch
(trimmedHTML, imDbRegEx.RuntimePattern);
string runtime = imDbRegEx.GetMatchValue
(match, "Runtime", true);
runtime = IMDbMovieDetailsDownloaderHelpers.FixRuntime(runtime);
movie.Runtime = runtime;
Debugger.LogMessageToFile("IMDb returned Runtime: " + movie.Runtime);
//MessageBox.Show("IMDb returned Runtime: " + movie.Runtime);
}
internal static void GetRating
(IIMDbMovie movie,
string trimmedHtml, IMDbRegEx imDbRegEx)
{
if (!ImdbFilmDetailsIndividualChoices.GetIMDbMovieRatings)
return;
Debugger.LogMessageToFile
("[IMDb film details downloader]" +
" Extracting Rating...");
Match match = imDbRegEx.GetRegExMatch
(trimmedHtml, imDbRegEx.RatingPattern);
movie.Rating = imDbRegEx.GetMatchValue
(match, "Rating", true);
Debugger.LogMessageToFile("IMDb returned Rating: "
+ movie.Rating);
}
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;
}
private static List<IIMDbSearchResult> MultipleMatchesAddToSearchResults
(Match match, IMDbRegEx imDbRegex, List<IIMDbSearchResult> results)
{
while (match != null && match.Length > 0)
{
var result = IMDbConventionalFilmSearchEngineHelpers
.MultipleMatchesMineDetailsOfSingleFilmResult(match, imDbRegex);
IMDbConventionalFilmSearchEngineHelpers
.IgnoreIrrelevantResults
(match, imDbRegex, result);
results.Add(result);
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Result was added to list.");
match = match.NextMatch();
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine]" +
" Proceeding to next result...");
}
return results;
}
internal static void GetReleaseDate
(IIMDbMovie movie,
string trimmedHTML, IMDbRegEx imDbRegEx)
{
if (!ImdbFilmDetailsIndividualChoices.GetIMDbMovieReleaseDate)
return;
Debugger.LogMessageToFile("[IMDb film details downloader]" +
" Extracting Release Date...");
Match match = imDbRegEx.GetRegExMatch
(trimmedHTML, imDbRegEx.ReleaseDatePattern);
movie.Release_Date = imDbRegEx.GetMatchValue
(match, "ReleaseDate", true);
Debugger.LogMessageToFile("[IMDb film details downloader]" +
" IMDb returned Release Date: " + movie.Release_Date);
//MessageBox.Show("IMDb returned Release Date: " + movie.Release_Date);
}
internal static void GetStudio
(IMDbMovie movie,
string trimmedHtml,
IMDbRegEx imDbRegEx)
{
if (!ImdbFilmDetailsIndividualChoices
.GetIMDbMovieProductionStudio)
return;
Debugger.LogMessageToFile
("[IMDb film details downloader] " +
"Extracting Studio...");
Match match
= imDbRegEx
.GetRegExMatch
(trimmedHtml,
imDbRegEx.StudioPattern);
movie.Studio = imDbRegEx.GetMatchValue
(match, "Studio", true);
Debugger.LogMessageToFile
("[IMDb film details downloader] " +
"IMDb returned Studio: "
+ movie.Studio);
}
internal static void GetReview
(IMDbMovie movie,
string trimmedHTML,
IMDbRegEx imDbRegEx)
{
if (!ImdbFilmDetailsIndividualChoices.GetIMDbMovieReviews)
return;
Debugger.LogMessageToFile("[IMDb film details downloader] Extracting Review score...");
Match match = imDbRegEx.GetRegExMatch(trimmedHTML, imDbRegEx.ReviewPattern);
movie.Review = imDbRegEx.GetMatchValue(match, "Review", true);
if (movie.Review.Contains("/"))
movie.Review = movie.Review.Substring(0, movie.Review.IndexOf("/", System.StringComparison.Ordinal));
Debugger.LogMessageToFile("IMDb returned Review: " + movie.Review);
}
internal static void ExtractRatingDescription
(IMDbMovie movie,
string trimmedHTML, IMDbRegEx imDbRegEx)
{
if (!ImdbFilmDetailsIndividualChoices.GetIMDbMovieRatingDescription)
return;
Debugger.LogMessageToFile("[IMDb film details downloader] Extracting Rating Description...");
Match match =
imDbRegEx.GetRegExMatch
(trimmedHTML, imDbRegEx.RatingDescriptionPattern);
movie.RatingDescription =
imDbRegEx.GetMatchValue
(match, "RatingDescription", true);
Debugger.LogMessageToFile("[IMDb film details downloader] IMDb returned Rating Description: " +
movie.RatingDescription);
}
internal static void MineOverviewUsingRegex
(IMDbMovie movie,
string trimmedHTML,
IMDbRegEx imDbRegEx )
{
if (!ImdbFilmDetailsIndividualChoices.GetIMDbMovieShortOverview)
return;
Debugger.LogMessageToFile
("[IMDb film details downloader]" +
" Extracting Short Overview...");
Match match = imDbRegEx.GetRegExMatch
(trimmedHTML, imDbRegEx.FilmDescriptionPattern);
movie.OverviewShort =
imDbRegEx.GetMatchValue
(match, "ShortOverview", true);
if (movie.OverviewShort.ToLower().EndsWith("more"))
movie.OverviewShort =
movie.OverviewShort.Substring
(0, movie.OverviewShort.Length - 4).Trim();
movie.OverviewShort = movie.OverviewShort.Trim() + "...";
Debugger.LogMessageToFile("IMDb returned Overview: " + movie.OverviewShort);
}
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;
}
}
}
internal static void MineProductionYearUsingRegex
(IIMDbMovie movie, string trimmedHtml, IMDbRegEx imDbRegEx)
{
Debugger.LogMessageToFile("[IMDb film details downloader]" +
" Extracting Year...");
Match match = imDbRegEx.GetRegExMatch
(trimmedHtml, imDbRegEx.YearPattern);
movie.Year = imDbRegEx.GetMatchValue
(match, "Year", true);
movie.Year = movie.Year.TrimEnd('/');
Debugger.LogMessageToFile
("[IMDb film details downloader] " +
"IMDb returned Year: " + movie.Year);
//MessageBox.Show
// ("IMDb returned Year: " + movie.Year);
}
AddSingleOrMultipleMatchesToSearchResults
(IMDbRegEx imDbRegex, string html,
bool multipleResultsFound,
List<IIMDbSearchResult> results)
{
if (!multipleResultsFound)
{
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Multiple results were not found.");
results = IMDbConventionalFilmSearchEngineHelpers.SingleMatchAddToSearchResults
(html, imDbRegex, results);
}
else
{
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Multiple results were found.");
#region Result Pattern 1
Match match = imDbRegex.GetRegExMatch(html, imDbRegex.MovieResultPattern1);
results = MultipleMatchesAddToSearchResults(match, imDbRegex, results);
#endregion
#region Result Pattern 2
match = imDbRegex.GetRegExMatch(html, imDbRegex.MovieResultPattern2);
results = MultipleMatchesAddToSearchResults(match, imDbRegex, results);
#endregion
}
return results;
}
internal static void GetTagline
(IIMDbMovie movie,
string trimmedHTML, IMDbRegEx imDbRegEx)
{
if (!ImdbFilmDetailsIndividualChoices.GetIMDbMovieTaglines)
return;
Debugger.LogMessageToFile("Extracting Tagline...");
Match match = imDbRegEx.GetRegExMatch
(trimmedHTML, imDbRegEx.TaglinePattern);
movie.Tagline = imDbRegEx.GetMatchValue
(match, "Tagline", true);
Debugger.LogMessageToFile("IMDb returned Tagline: " + movie.Tagline);
//MessageBox.Show("IMDb returned Tagline: " + movie.Tagline);
}
public static IList<IIMDbSearchResult> SearchForTitleReturnResults(string title)
{
#region Vars
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Initializing variables...");
var searchResults = new List<IIMDbSearchResult>();
var imDbRegex = new IMDbRegEx();
//TODO: Test if SetPatterns() is correctly called and regexes are succesfully loaded!
string titleEncoded = ToolBox.Utils.WebUtils.EncodeURL(title);
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Title to search for: " + title);
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"Encoded title: " + titleEncoded);
string SearchURL = "http://www.imdb.com/find?s=title&q=" + titleEncoded;
//string SearchURL = "http://m.imdb.com/find?q=" + titleEncoded + "&button=Search";
Debugger.LogMessageToFile
("[IMDb Conventional Film Search Engine] " +
"IMDb search URL: " + SearchURL);
#endregion
var watch = new Stopwatch();
watch.Start();
long millisecondsnow;
string searchResultsHtml
= DownloadSearchResultsHtml
(watch, SearchURL);
if ( String.IsNullOrEmpty
( searchResultsHtml ))
return searchResults;
var multipleResultsFound
= IMDbConventionalFilmSearchEngineHelpers.DretectMultipleResults
(searchResultsHtml);
searchResults = AddSingleOrMultipleMatchesToSearchResults
(imDbRegex, searchResultsHtml,
multipleResultsFound, searchResults);
//MessageBox.Show("Results: " + results.Count.ToString() );
millisecondsnow = watch.ElapsedMilliseconds;
//MessageBox.Show
// ("Extraction of all" +
// " results needed "
// + millisecondsnow
// + " ms");
return searchResults;
}
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;
}