Fixed: Improved special episode parsing for multiple matching titles

src/NzbDrone.Core.Test/TvTests/EpisodeServiceTests/FindEpisodeByTitleFixture.cs

@@ -14,32 +14,50 @@ namespace NzbDrone.Core.Test.TvTests.EpisodeServiceTests
     [TestFixture]
     public class FindEpisodeByTitleFixture : CoreTest<EpisodeService>
     {
-        private Episode _episode;
+        private List<Episode> _episodes;
 
         [SetUp]
         public void Setup()
         {
-            _episode = Builder<Episode>.CreateNew().Build();
+            _episodes = Builder<Episode>.CreateListOfSize(5)
+                                        .Build()
+                                        .ToList();
         }
 
-        private void GivenEpisodeTitle(string title)
+        private void GivenEpisodesWithTitles(params string[] titles)
         {
-            _episode.Title = title;
+            for (int i = 0; i < titles.Count(); i++)
+            {
+                _episodes[i].Title = titles[i];
+            }
 
             Mocker.GetMock<IEpisodeRepository>()
                   .Setup(s => s.GetEpisodes(It.IsAny<int>(), It.IsAny<int>()))
-                  .Returns(new List<Episode> { _episode });
+                  .Returns(_episodes);
         }
 
         [Test]
         public void should_find_episode_by_title()
         {
-            GivenEpisodeTitle("A Journey to the Highlands");
+            const string expectedTitle = "A Journey to the Highlands";
+            GivenEpisodesWithTitles(expectedTitle);
 
             Subject.FindEpisodeByTitle(1, 1, "Downton.Abbey.A.Journey.To.The.Highlands.720p.BluRay.x264-aAF")
                    .Title
                    .Should()
-                   .Be(_episode.Title);
+                   .Be(expectedTitle);
+        }
+
+        [Test]
+        public void should_prefer_longer_match()
+        {
+            const string expectedTitle = "Inside The Walking Dead: Walker University";
+            GivenEpisodesWithTitles("Inside The Walking Dead", expectedTitle);
+
+            Subject.FindEpisodeByTitle(1, 1, "The.Walking.Dead.S04.Special.Inside.The.Walking.Dead.Walker.University.720p.HDTV.x264-W4F")
+                   .Title
+                   .Should()
+                   .Be(expectedTitle);
         }
     }
 }

src/NzbDrone.Core/Parser/Parser.cs

@@ -357,9 +357,12 @@ namespace NzbDrone.Core.Parser
 
         public static string NormalizeEpisodeTitle(string title)
         {
-            return SpecialEpisodeWordRegex.Replace(title, String.Empty)
+            title = SpecialEpisodeWordRegex.Replace(title, String.Empty);
-                                          .Trim()
+            title = PunctuationRegex.Replace(title, " ");
-                                          .ToLower();
+            title = DuplicateSpacesRegex.Replace(title, " ");
+
+            return title.Trim()
+                        .ToLower();
         }
 
         public static string NormalizeTitle(string title)

src/NzbDrone.Core/Tv/EpisodeService.cs

@@ -17,7 +17,7 @@ namespace NzbDrone.Core.Tv
         List<Episode> GetEpisodes(IEnumerable<Int32> ids);
         Episode FindEpisode(int seriesId, int seasonNumber, int episodeNumber);
         Episode FindEpisode(int seriesId, int absoluteEpisodeNumber);
-        Episode FindEpisodeByTitle(int seriesId, int seasonNumber, string episodeTitle);
+        Episode FindEpisodeByTitle(int seriesId, int seasonNumber, string releaseTitle);
         List<Episode> FindEpisodesBySceneNumbering(int seriesId, int seasonNumber, int episodeNumber);
         Episode FindEpisodeBySceneNumbering(int seriesId, int sceneAbsoluteEpisodeNumber);
         Episode GetEpisode(int seriesId, String date);
@@ -103,19 +103,25 @@ namespace NzbDrone.Core.Tv
             return _episodeRepository.GetEpisodes(seriesId, seasonNumber);
         }
         
-        public Episode FindEpisodeByTitle(int seriesId, int seasonNumber, string episodeTitle) 
+        public Episode FindEpisodeByTitle(int seriesId, int seasonNumber, string releaseTitle) 
         {
             // TODO: can replace this search mechanism with something smarter/faster/better
-            var search = Parser.Parser.NormalizeEpisodeTitle(episodeTitle).Replace(".", " ");
+            var normalizedReleaseTitle = Parser.Parser.NormalizeEpisodeTitle(releaseTitle).Replace(".", " ");
-
+            var episodes = _episodeRepository.GetEpisodes(seriesId, seasonNumber);
-            return _episodeRepository.GetEpisodes(seriesId, seasonNumber)
+
-                .FirstOrDefault(e => 
+            var query = episodes.Select(
-                {
+                episode => new
-                    // normalize episode title
+                           {
-                    var title = Parser.Parser.NormalizeEpisodeTitle(e.Title);
+                               Position = normalizedReleaseTitle.IndexOf(Parser.Parser.NormalizeEpisodeTitle(episode.Title), StringComparison.CurrentCultureIgnoreCase),
-                    // find episode title within search string
+                               Length = Parser.Parser.NormalizeEpisodeTitle(episode.Title).Length,
-                    return (title.Length > 0) && search.Contains(title); 
+                               Episode = episode
-                });
+                           })
+                                .Where(e => e.Episode.Title.Length > 0 && e.Position >= 0)
+                                .OrderBy(e => e.Position)
+                                .ThenByDescending(e => e.Length)
+                                .ToList();
+
+            return query.First().Episode;
         }
 
         public List<Episode> EpisodesWithFiles(int seriesId)