Special Episode parsing support in ParsingService

Added ParsingService.ParseSpecialEpisodeTitle
Added SeriesService.FindByNameInexact
Added EpisodeService.FindSpecialEpisodeByName
Added IsPossibleSpecialEpisode method to parse info
DownloadDecisionMaker will try to find special episodes if a parse fails or is a possible special episode

src/NzbDrone.Core/DecisionEngine/DownloadDecisionMaker.cs

@@ -52,6 +52,13 @@ private IEnumerable<DownloadDecision> GetDecisions(List<ReleaseInfo> reports, Se
                 _logger.ProgressInfo("No reports found");
             }
 
+            // get series from search criteria
+            Tv.Series series = null;
+            if (searchCriteria != null)
+            {
+                series = searchCriteria.Series;
+            }
+
             var reportNumber = 1;
 
             foreach (var report in reports)
@@ -61,8 +68,21 @@ private IEnumerable<DownloadDecision> GetDecisions(List<ReleaseInfo> reports, Se
 
                 try
                 {
+                    // use parsing service to parse episode info (this allows us to do episode title searches against the episode repository)
                     var parsedEpisodeInfo = Parser.Parser.ParseTitle(report.Title);
 
+                    // do we have a possible special episode?
+                    if (parsedEpisodeInfo == null || parsedEpisodeInfo.IsPossibleSpecialEpisode())
+                    {
+                        // try to parse as a special episode
+                        var specialEpisodeInfo = _parsingService.ParseSpecialEpisodeTitle(report.Title, series);
+                        if (specialEpisodeInfo != null)
+                        {
+                            // use special episode
+                            parsedEpisodeInfo = specialEpisodeInfo;
+                        }
+                    }
+
                     if (parsedEpisodeInfo != null && !string.IsNullOrWhiteSpace(parsedEpisodeInfo.SeriesTitle))
                     {
                         var remoteEpisode = _parsingService.Map(parsedEpisodeInfo, report.TvRageId, searchCriteria);

src/NzbDrone.Core/Parser/Model/ParsedEpisodeInfo.cs

@@ -33,6 +33,12 @@ public bool IsAbsoluteNumbering()
             return AbsoluteEpisodeNumbers.Any();
         }
 
+        public bool IsPossibleSpecialEpisode()
+        {
+            // if we dont have eny episode numbers we are likely a special episode and need to do a search by episode title
+            return string.IsNullOrEmpty(AirDate) && (EpisodeNumbers.Length == 0 || SeasonNumber == 0 || String.IsNullOrWhiteSpace(SeriesTitle));
+        }
+
         public override string ToString()
         {
             string episodeString = "[Unknown Episode]";

src/NzbDrone.Core/Parser/Parser.cs

@@ -114,6 +114,11 @@ public static class Parser
         private static readonly Regex YearInTitleRegex = new Regex(@"^(?<title>.+?)(?:\W|_)?(?<year>\d{4})",
                                                                    RegexOptions.IgnoreCase | RegexOptions.Compiled);
 
+        private static readonly Regex NonWordRegex = new Regex(@"\W+", RegexOptions.Compiled);
+        private static readonly Regex CommonWordRegex = new Regex(@"\b(a|an|the|and|or|of|part)\b\s?",
+                                                         RegexOptions.IgnoreCase | RegexOptions.Compiled);
+
+
         public static ParsedEpisodeInfo ParsePath(string path)
         {
             var fileInfo = new FileInfo(path);
@@ -220,6 +225,15 @@ public static string CleanupEpisodeTitle(string title)
             return MultiPartCleanupRegex.Replace(title, string.Empty).Trim();
         }
 
+        public static string NormalizeEpisodeTitle(string title)
+        {
+            // convert any non-word characters to a single space
+            string normalizedSpaces = NonWordRegex.Replace(title, " ").ToLower();
+            // remove common words
+            string normalized = CommonWordRegex.Replace(normalizedSpaces, String.Empty);
+            return normalized;
+        }
+
         public static string ParseReleaseGroup(string title)
         {
             const string defaultReleaseGroup = "DRONE";

src/NzbDrone.Core/Parser/ParsingService.cs

@@ -12,6 +12,7 @@ namespace NzbDrone.Core.Parser
 {
     public interface IParsingService
     {
+        ParsedEpisodeInfo ParseSpecialEpisodeTitle(string title, Series series);
         LocalEpisode GetEpisodes(string filename, Series series, bool sceneSource);
         Series GetSeries(string title);
         RemoteEpisode Map(ParsedEpisodeInfo parsedEpisodeInfo, int tvRageId, SearchCriteriaBase searchCriteria = null);
@@ -39,10 +40,68 @@ public ParsingService(IEpisodeService episodeService,
             _logger = logger;
         }
 
+        public ParsedEpisodeInfo ParseSpecialEpisodeTitle(string title, Series series)
+        {
+            try
+            {
+                if (series == null)
+                {
+                    // find series if we dont have it already
+                    // we use an inexact match here since the series name is often mangled with the episode title
+                    series = _seriesService.FindByTitleInexact(title);
+                    if (series == null)
+                    {
+                        // no series matched
+                        return null;
+                    }
+                }
+
+                // find special episode in series season 0
+                Episode episode = _episodeService.FindEpisodeByName(series.Id, 0, title);
+                if (episode != null)
+                {
+                    // created parsed info from tv episode that we found
+                    var info = new ParsedEpisodeInfo();
+                    info.SeriesTitle = series.Title;
+                    info.SeriesTitleInfo = new SeriesTitleInfo();
+                    info.SeriesTitleInfo.Title = info.SeriesTitle;
+                    info.SeasonNumber = episode.SeasonNumber;
+                    info.EpisodeNumbers = new int[1] { episode.EpisodeNumber };
+                    info.FullSeason = false;
+                    info.Quality = QualityParser.ParseQuality(title);
+                    info.ReleaseGroup = Parser.ParseReleaseGroup(title);
+
+                    _logger.Info("Found special episode {0} for title '{1}'", info, title);
+                    return info;
+                }
+            }
+            catch (Exception e)
+            {
+                _logger.ErrorException("An error has occurred while trying to parse special episode " + title, e);
+            }
+
+            return null;
+        }
+
+
+
         public LocalEpisode GetEpisodes(string filename, Series series, bool sceneSource)
         {
             var parsedEpisodeInfo = Parser.ParsePath(filename);
 
+            // do we have a possible special episode?
+            if (parsedEpisodeInfo == null || parsedEpisodeInfo.IsPossibleSpecialEpisode())
+            {
+                // try to parse as a special episode
+                var title = System.IO.Path.GetFileNameWithoutExtension(filename);
+                var specialEpisodeInfo = ParseSpecialEpisodeTitle(title, series);
+                if (specialEpisodeInfo != null)
+                {
+                    // use special episode
+                    parsedEpisodeInfo = specialEpisodeInfo;
+                }
+            }
+
             if (parsedEpisodeInfo == null)
             {
                 return null;

src/NzbDrone.Core/Tv/EpisodeService.cs

@@ -15,6 +15,7 @@ public interface IEpisodeService
         Episode GetEpisode(int id);
         Episode FindEpisode(int seriesId, int seasonNumber, int episodeNumber, bool useScene = false);
         Episode FindEpisode(int seriesId, int absoluteEpisodeNumber);
+        Episode FindEpisodeByName(int seriesId, int seasonNumber, string episodeTitle);
         Episode GetEpisode(int seriesId, String date);
         Episode FindEpisode(int seriesId, String date);
         List<Episode> GetEpisodeBySeries(int seriesId);
@@ -88,6 +89,21 @@ public List<Episode> GetEpisodesBySeason(int seriesId, int seasonNumber)
             return _episodeRepository.GetEpisodes(seriesId, seasonNumber);
         }
 
+        public Episode FindEpisodeByName(int seriesId, int seasonNumber, string episodeTitle) 
+        {
+            // TODO: can replace this search mechanism with something smarter/faster/better
+            var search = Parser.Parser.NormalizeEpisodeTitle(episodeTitle);
+            return _episodeRepository.GetEpisodes(seriesId, seasonNumber)
+                .FirstOrDefault(e => 
+                {
+                    // normalize episode title
+                    string title = Parser.Parser.NormalizeEpisodeTitle(e.Title);
+                    // find episode title within search string
+                    return (title.Length > 0) && search.Contains(title); 
+                });
+        }
+
+
         public PagingSpec<Episode> EpisodesWithoutFiles(PagingSpec<Episode> pagingSpec)
         {
             var episodeResult = _episodeRepository.EpisodesWithoutFiles(pagingSpec, false);

src/NzbDrone.Core/Tv/SeriesService.cs

@@ -20,6 +20,7 @@ public interface ISeriesService
         Series FindByTvRageId(int tvRageId);
         Series FindByTitle(string title);
         Series FindByTitle(string title, int year);
+        Series FindByTitleInexact(string title);
         void SetSeriesType(int seriesId, SeriesTypes seriesTypes);
         void DeleteSeries(int seriesId, bool deleteFiles);
         List<Series> GetAllSeries();
@@ -100,6 +101,55 @@ public Series FindByTitle(string title)
             return _seriesRepository.FindByTitle(Parser.Parser.CleanSeriesTitle(title));
         }
 
+        public Series FindByTitleInexact(string title)
+        {
+            // perform fuzzy matching of series name
+            // TODO: can replace this search mechanism with something smarter/faster/better
+
+            // find any series clean title within the provided release title
+            string cleanTitle = Parser.Parser.CleanSeriesTitle(title);
+            var list = _seriesRepository.All().Where(s => cleanTitle.Contains(s.CleanTitle)).ToList();
+            if (!list.Any())
+            {
+                // no series matched
+                return null;
+            }
+            else if (list.Count == 1)
+            {
+                // return the first series if there is only one 
+                return list.Single();
+            }
+            else 
+            {
+                // build ordered list of series by position in the search string
+                var query = 
+                        list.Select(series => new
+                            {
+                                position = cleanTitle.IndexOf(series.CleanTitle),
+                                length = series.CleanTitle.Length,
+                                series = series
+                            })
+                            .Where(s => (s.position>=0))
+                            .ToList()
+                            .OrderBy(s => s.position)
+                            .ThenByDescending(s => s.length)
+                            .ToList();
+
+                // get the leftmost series that is the longest
+                // series are usually the first thing in release title, so we select the leftmost and longest match
+                // we could have multiple matches for series which have a common prefix like "Love it", "Love it Too" so we pick the longest one
+                var match = query.First().series;
+
+                _logger.Trace("Multiple series matched {0} from title {1}", match.Title, title);
+                foreach (var entry in list)
+                {
+                    _logger.Trace("Multiple series match candidate: {0} cleantitle: {1}", entry.Title, entry.CleanTitle);
+                }
+
+                return match;
+            }
+        }
+
         public Series FindByTitle(string title, int year)
         {
             return _seriesRepository.FindByTitle(title, year);