Fixed: Escape regex in ParseAlbumWithSearchCriteria (#244)

* Add regex escape to fix #231

* Add escape to artist name

* Fix test case

* Use single album parameter and add test cases

* Add artist test cases

* Add qualities to release titles

* Create albums in ParserFixture

* Added missing case in QualityParser. Handle escaping regex better for artists/albums that are just symbols.

* Removed custom code to escape slashes. Enhanced regex to support more test cases.

* Fixed Regex for other test cases.

* Small enhancements to code. Removed log statement.

* Tweaked one of my regex to account for not stripping ? from SimpleTitleRegex.

src/NzbDrone.Core.Test/ParserTests/ParserFixture.cs

@@ -16,7 +16,7 @@ namespace NzbDrone.Core.Test.ParserTests
     public class ParserFixture : CoreTest
     {
         Artist _artist = new Artist();
-        private List<Album> _albums = new List<Album>{new Album()};
+        private List<Album> _albums = new List<Album> { new Album() };
 
         [SetUp]
         public void Setup()
@@ -24,12 +24,17 @@ namespace NzbDrone.Core.Test.ParserTests
             _artist = Builder<Artist>
                 .CreateNew()
                 .Build();
+            _albums = Builder<List<Album>>
+                .CreateNew()
+                .Build();
         }
 
         private void GivenSearchCriteria(string artistName, string albumTitle)
         {
             _artist.Name = artistName;
-            _albums.First().Title = albumTitle;
+            var a = new Album();
+            a.Title = albumTitle;
+            _albums.Add(a);
         }
 
         [TestCase("Bad Format", "badformat")]
@@ -43,7 +48,7 @@ namespace NzbDrone.Core.Test.ParserTests
         public void should_remove_accents_from_title()
         {
             const string title = "Carniv\u00E0le";
-            
+
             title.CleanArtistName().Should().Be("carnivale");
         }
 
@@ -82,16 +87,16 @@ namespace NzbDrone.Core.Test.ParserTests
             Parser.Parser.ParseAlbumTitle(postTitle).ArtistName.Should().Be(title);
         }
 
-        [TestCase("02 Unchained.flac")]
-        [TestCase("Fall Out Boy - 02 - Title.wav")]
+        [TestCase("02 Unchained.flac")] // This isn't valid on any regex we have. We must always have an artist
+        [TestCase("Fall Out Boy - 02 - Title.wav")] // This isn't valid on any regex we have. We don't support Artist - Track - TrackName
         public void should_parse_quality_from_extension(string title)
         {
             Parser.Parser.ParseAlbumTitle(title).Quality.Quality.Should().NotBe(Quality.Unknown);
             Parser.Parser.ParseAlbumTitle(title).Quality.QualitySource.Should().Be(QualitySource.Extension);
         }
 
-        [TestCase("of Montreal-Hissing Fauna, Are You The Destroyer? 2007", "Hissing Fauna, Are You The Destroyer", "of Montreal", "2007")]
-        [TestCase("of Montreal - 2007 - Hissing Fauna, Are You The Destroyer?", "Hissing Fauna, Are You The Destroyer", "of Montreal", "2007")]
+        [TestCase("of Montreal-Hissing Fauna, Are You The Destroyer? 2007", "Hissing Fauna, Are You The Destroyer?", "of Montreal", "2007")]
+        [TestCase("of Montreal - 2007 - Hissing Fauna, Are You The Destroyer?", "Hissing Fauna, Are You The Destroyer?", "of Montreal", "2007")]
         public void should_parse_album(string title, string correctAlbum, string correctArtist, string correctYear)
         {
             ParsedAlbumInfo result = Parser.Parser.ParseAlbumTitle(title);
@@ -203,5 +208,29 @@ namespace NzbDrone.Core.Test.ParserTests
             var parseResult = Parser.Parser.ParseAlbumTitleWithSearchCriteria("Black Sabbath  Black Sabbath FLAC", _artist, _albums);
             parseResult.Should().BeNull();
         }
+
+        [TestCase("Ed Sheeran", "I See Fire", "Ed Sheeran I See Fire[Mimp3.eu].mp3 FLAC")]
+        [TestCase("Ed Sheeran", "Divide", "Ed Sheeran   ? Divide FLAC")]
+        [TestCase("Ed Sheeran", "+", "Ed Sheeran + FLAC")]
+        //[TestCase("Glasvegas", @"EUPHORIC /// HEARTBREAK \\\", @"EUPHORIC /// HEARTBREAK \\\ FLAC")] // slashes not being escaped properly
+        [TestCase("XXXTENTACION", "?", "XXXTENTACION ? FLAC")]
+        [TestCase("Hey", "BŁYSK", "Hey - BŁYSK FLAC")]
+        public void should_escape_albums(string artist, string album, string releaseTitle)
+        {
+            GivenSearchCriteria(artist, album);
+            var parseResult = Parser.Parser.ParseAlbumTitleWithSearchCriteria(releaseTitle, _artist, _albums);
+            parseResult.AlbumTitle.Should().Be(album);
+        }
+
+        [TestCase("???", "Album", "??? Album FLAC")]
+        [TestCase("+", "Album", "+ Album FLAC")]
+        [TestCase(@"/\", "Album", @"/\ Album FLAC")]
+        [TestCase("+44", "When Your Heart Stops Beating", "+44 When Your Heart Stops Beating FLAC")]
+        public void should_escape_artists(string artist, string album, string releaseTitle)
+        {
+            GivenSearchCriteria(artist, album);
+            var parseResult = Parser.Parser.ParseAlbumTitleWithSearchCriteria(releaseTitle, _artist, _albums);
+            parseResult.ArtistName.Should().Be(artist);
+        }
     }
 }

src/NzbDrone.Core/Parser/Parser.cs

@@ -122,7 +122,7 @@ namespace NzbDrone.Core.Parser
 
             //Artist-Album Year
             //Hyphen no space between artist and album
-            new Regex(@"^(?:(?<artist>.+?)(?:-)+)(?<album>.+?)\W*(?<releaseyear>\d{4})",
+            new Regex(@"^(?:(?<artist>.+?)(?:-)+)(?<album>.+?)\b(?<releaseyear>\d{4})",
                 RegexOptions.IgnoreCase | RegexOptions.Compiled),
 
             
@@ -163,7 +163,7 @@ namespace NzbDrone.Core.Parser
         private static readonly Regex FileExtensionRegex = new Regex(@"\.[a-z0-9]{2,4}$",
                                                                 RegexOptions.IgnoreCase | RegexOptions.Compiled);
 
-        private static readonly Regex SimpleTitleRegex = new Regex(@"(?:(480|720|1080|2160|320)[ip]|[xh][\W_]?26[45]|DD\W?5\W1|[<>?*:|]|848x480|1280x720|1920x1080|3840x2160|4096x2160|(8|10)b(it)?)\s*",
+        private static readonly Regex SimpleTitleRegex = new Regex(@"(?:(480|720|1080|2160|320)[ip]|[xh][\W_]?26[45]|DD\W?5\W1|[<>*:|]|848x480|1280x720|1920x1080|3840x2160|4096x2160|(8|10)b(it)?)\s*",
                                                                 RegexOptions.IgnoreCase | RegexOptions.Compiled);
 
         private static readonly Regex WebsitePrefixRegex = new Regex(@"^\[\s*[a-z]+(\.[a-z]+)+\s*\][- ]*",
@@ -351,8 +351,11 @@ namespace NzbDrone.Core.Parser
 
                 simpleTitle = CleanTorrentSuffixRegex.Replace(simpleTitle, string.Empty);
 
-                var releaseRegex = new Regex(@"\b(?<artist>" + artist.Name + @")\b.*\b(?<album>"+ string.Join("|",album.Select(s=>s.Title).ToList()) + @")\b",
-                    RegexOptions.IgnoreCase | RegexOptions.Compiled);
+                var escapedArtist = Regex.Escape(artist.Name);
+                var escapedAlbums = Regex.Escape(string.Join("|", album.Select(s => s.Title).ToList()));
+
+                var releaseRegex = new Regex(@"^(\W*|\b)(?<artist>" + escapedArtist + @")(\W*|\b).*(\W*|\b)(?<album>" + escapedAlbums + @")(\W*|\b)", RegexOptions.IgnoreCase);
+
 
                 var match = releaseRegex.Matches(simpleTitle);
 

src/NzbDrone.Core/Parser/QualityParser.cs

@@ -122,6 +122,7 @@ namespace NzbDrone.Core.Parser
                     else if (bitrate == BitRate.B256) { result.Quality = Quality.VORBIS_Q8; }
                     else if (bitrate == BitRate.B320) { result.Quality = Quality.VORBIS_Q9; }
                     else if (bitrate == BitRate.B500) { result.Quality = Quality.VORBIS_Q10; }
+                    else { result.Quality = Quality.Unknown; }
                     break;
                 case Codec.Unknown:
                     if (bitrate == BitRate.B192) { result.Quality = Quality.MP3_192; }