New: Added Icelandic language and improved Chinese language detection

src/NzbDrone.Core.Test/Languages/LanguageFixture.cs

@@ -21,8 +21,8 @@ namespace NzbDrone.Core.Test.Languages
                 new object[] {6, Language.Danish},
                 new object[] {7, Language.Dutch},
                 new object[] {8, Language.Japanese},
-                new object[] {9, Language.Cantonese},
+                new object[] {9, Language.Icelandic},
-                new object[] {10, Language.Mandarin},
+                new object[] {10, Language.Chinese},
                 new object[] {11, Language.Russian},
                 new object[] {12, Language.Polish},
                 new object[] {13, Language.Vietnamese},
@@ -47,8 +47,8 @@ namespace NzbDrone.Core.Test.Languages
                 new object[] {Language.Danish, 6},
                 new object[] {Language.Dutch, 7},
                 new object[] {Language.Japanese, 8},
-                new object[] {Language.Cantonese, 9},
+                new object[] {Language.Icelandic, 9},
-                new object[] {Language.Mandarin, 10},
+                new object[] {Language.Chinese, 10},
                 new object[] {Language.Russian, 11},
                 new object[] {Language.Polish, 12},
                 new object[] {Language.Vietnamese, 13},

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

@@ -88,18 +88,30 @@ namespace NzbDrone.Core.Test.ParserTests
             result.Language.Id.Should().Be(Language.Japanese.Id);
         }
 
-        [TestCase("Castle.2009.S01E14.Cantonese.HDTV.XviD-LOL")]
+        [TestCase("Castle.2009.S01E14.Icelandic.HDTV.XviD-LOL")]
-        public void should_parse_language_cantonese(string postTitle)
+        [TestCase("S.B.S01E03.1080p.WEB-DL.DD5.1.H.264-SbR Icelandic")]
+        public void should_parse_language_icelandic(string postTitle)
         {
             var result = Parser.Parser.ParseTitle(postTitle);
-            result.Language.Id.Should().Be(Language.Cantonese.Id);
+            result.Language.Id.Should().Be(Language.Icelandic.Id);
         }
 
+        [TestCase("Castle.2009.S01E14.Chinese.HDTV.XviD-LOL")]
+        [TestCase("Castle.2009.S01E14.Cantonese.HDTV.XviD-LOL")]
         [TestCase("Castle.2009.S01E14.Mandarin.HDTV.XviD-LOL")]
-        public void should_parse_language_mandarin(string postTitle)
+        [TestCase("[abc] My Series - 01 [CHS]")]
+        [TestCase("[abc] My Series - 01 [CHT]")]
+        [TestCase("[abc] My Series - 01 [BIG5]")]
+        [TestCase("[abc] My Series - 01 [GB]")]
+        [TestCase("[abc] My Series - 01 [繁中]")]
+        [TestCase("[abc] My Series - 01 [繁体]")]
+        [TestCase("[abc] My Series - 01 [简繁外挂]")]
+        [TestCase("[abc] My Series - 01 [简繁内封字幕]")]
+        [TestCase("[ZERO字幕组]My Series/My Series[01][HDTV]")]
+        public void should_parse_language_chinese(string postTitle)
         {
             var result = Parser.Parser.ParseTitle(postTitle);
-            result.Language.Id.Should().Be(Language.Mandarin.Id);
+            result.Language.Id.Should().Be(Language.Chinese.Id);
         }
 
         [TestCase("Castle.2009.S01E14.Korean.HDTV.XviD-LOL")]

src/NzbDrone.Core/Languages/Language.cs

@@ -64,8 +64,8 @@ namespace NzbDrone.Core.Languages
         public static Language Danish       { get { return new Language(6, "Danish"); } }
         public static Language Dutch        { get { return new Language(7, "Dutch"); } }
         public static Language Japanese     { get { return new Language(8, "Japanese"); } }
-        public static Language Cantonese    { get { return new Language(9, "Cantonese"); } }
+        public static Language Icelandic    { get { return new Language(9, "Icelandic"); } }
-        public static Language Mandarin     { get { return new Language(10, "Mandarin"); } }
+        public static Language Chinese     { get { return new Language(10, "Chinese"); } }
         public static Language Russian      { get { return new Language(11, "Russian"); } }
         public static Language Polish       { get { return new Language(12, "Polish"); } }
         public static Language Vietnamese   { get { return new Language(13, "Vietnamese"); } }
@@ -98,8 +98,8 @@ namespace NzbDrone.Core.Languages
                     Danish,
                     Dutch,
                     Japanese,
-                    Cantonese,
+                    Icelandic,
-                    Mandarin,
+                    Chinese,
                     Russian,
                     Polish,
                     Vietnamese,

src/NzbDrone.Core/Parser/IsoLanguages.cs

@@ -16,8 +16,8 @@ namespace NzbDrone.Core.Parser
                                                                new IsoLanguage("da", "dan", Language.Danish),
                                                                new IsoLanguage("nl", "nld", Language.Dutch),
                                                                new IsoLanguage("ja", "jpn", Language.Japanese),
-//                                                             new IsoLanguage("", "", Language.Cantonese),
+                                                               new IsoLanguage("is", "isl", Language.Icelandic),
-//                                                             new IsoLanguage("", "", Language.Mandarin),
+                                                               new IsoLanguage("zh", "zho", Language.Chinese),
                                                                new IsoLanguage("ru", "rus", Language.Russian),
                                                                new IsoLanguage("pl", "pol", Language.Polish),
                                                                new IsoLanguage("vi", "vie", Language.Vietnamese),

src/NzbDrone.Core/Parser/LanguageParser.cs

@@ -17,7 +17,7 @@ namespace NzbDrone.Core.Parser
                 new RegexReplace(@".*?\.(S\d{2}E\d{2,4}\..*)", "$1", RegexOptions.Compiled | RegexOptions.IgnoreCase)
             };
 
-        private static readonly Regex LanguageRegex = new Regex(@"(?:\W|_)(?<italian>\b(?:ita|italian)\b)|(?<german>german\b|videomann)|(?<flemish>flemish)|(?<greek>greek)|(?<french>(?:\W|_)(?:FR|VOSTFR)(?:\W|_))|(?<russian>\brus\b)|(?<dutch>nl\W?subs?)|(?<hungarian>\b(?:HUNDUB|HUN)\b)|(?<hebrew>\bHebDub\b)",
+        private static readonly Regex LanguageRegex = new Regex(@"(?:\W|_)(?<italian>\b(?:ita|italian)\b)|(?<german>german\b|videomann)|(?<flemish>flemish)|(?<greek>greek)|(?<french>(?:\W|_)(?:FR|VOSTFR)(?:\W|_))|(?<russian>\brus\b)|(?<dutch>nl\W?subs?)|(?<hungarian>\b(?:HUNDUB|HUN)\b)|(?<hebrew>\bHebDub\b)|(?<chinese>\[(?:CH[ST]|BIG5|GB)\]|简|繁|字幕组)",
                                                                 RegexOptions.IgnoreCase | RegexOptions.Compiled);
 
         private static readonly Regex CaseSensitiveLanguageRegex = new Regex(@"(?<lithuanian>\bLT\b)|(?<czech>\bCZ\b)",
@@ -54,11 +54,11 @@ namespace NzbDrone.Core.Parser
             if (lowerTitle.Contains("japanese"))
                 return Language.Japanese;
 
-            if (lowerTitle.Contains("cantonese"))
+            if (lowerTitle.Contains("icelandic"))
-                return Language.Cantonese;
+                return Language.Icelandic;
 
-            if (lowerTitle.Contains("mandarin"))
+            if (lowerTitle.Contains("mandarin") || lowerTitle.Contains("cantonese") || lowerTitle.Contains("chinese"))
-                return Language.Mandarin;
+                return Language.Chinese;
 
             if (lowerTitle.Contains("korean"))
                 return Language.Korean;
@@ -182,6 +182,9 @@ namespace NzbDrone.Core.Parser
             if (match.Groups["hebrew"].Success)
                 return Language.Hebrew;
 
+            if (match.Groups["chinese"].Success)
+                return Language.Chinese;
+
             return Language.Unknown;
         }
     }

src/NzbDrone.Core/Parser/Parser.cs

@@ -600,100 +600,7 @@ namespace NzbDrone.Core.Parser
 
             return title;
         }
-
+        
-        public static Language ParseLanguage(string title)
-        {
-            var lowerTitle = title.ToLower();
-
-            if (lowerTitle.Contains("english"))
-                return Language.English;
-
-            if (lowerTitle.Contains("french"))
-                return Language.French;
-
-            if (lowerTitle.Contains("spanish"))
-                return Language.Spanish;
-
-            if (lowerTitle.Contains("danish"))
-                return Language.Danish;
-
-            if (lowerTitle.Contains("dutch"))
-                return Language.Dutch;
-
-            if (lowerTitle.Contains("japanese"))
-                return Language.Japanese;
-
-            if (lowerTitle.Contains("cantonese"))
-                return Language.Cantonese;
-
-            if (lowerTitle.Contains("mandarin"))
-                return Language.Mandarin;
-
-            if (lowerTitle.Contains("korean"))
-                return Language.Korean;
-
-            if (lowerTitle.Contains("russian"))
-                return Language.Russian;
-
-            if (lowerTitle.Contains("polish"))
-                return Language.Polish;
-
-            if (lowerTitle.Contains("vietnamese"))
-                return Language.Vietnamese;
-
-            if (lowerTitle.Contains("swedish"))
-                return Language.Swedish;
-
-            if (lowerTitle.Contains("norwegian"))
-                return Language.Norwegian;
-
-            if (lowerTitle.Contains("nordic"))
-                return Language.Norwegian;
-
-            if (lowerTitle.Contains("finnish"))
-                return Language.Finnish;
-
-            if (lowerTitle.Contains("turkish"))
-                return Language.Turkish;
-
-            if (lowerTitle.Contains("portuguese"))
-                return Language.Portuguese;
-
-            if (lowerTitle.Contains("hungarian"))
-                return Language.Hungarian;
-                
-            var match = LanguageRegex.Match(title);
-
-            if (match.Groups["italian"].Captures.Cast<Capture>().Any())
-                return Language.Italian;
-
-            if (match.Groups["german"].Captures.Cast<Capture>().Any())
-                return Language.German;
-
-            if (match.Groups["flemish"].Captures.Cast<Capture>().Any())
-                return Language.Flemish;
-
-            if (match.Groups["greek"].Captures.Cast<Capture>().Any())
-                return Language.Greek;
-
-            if (match.Groups["spanish"].Captures.Cast<Capture>().Any())
-                return Language.Spanish;
-
-            if (match.Groups["french"].Success)
-                return Language.French;
-
-            if (match.Groups["russian"].Success)
-                return Language.Russian;
-
-            if (match.Groups["dutch"].Success)
-                return Language.Dutch;
-
-            if (match.Groups["hungarian"].Success)
-                return Language.Hungarian;
-
-            return Language.English;
-        }
-
         private static SeriesTitleInfo GetSeriesTitleInfo(string title)
         {
             var seriesTitleInfo = new SeriesTitleInfo();