From 691a8955fee345d37c09da12b451d2488e9ae440 Mon Sep 17 00:00:00 2001
From: Mark McDowall <mark@mcdowall.ca>
Date: Sun, 7 Mar 2021 16:53:56 -0800
Subject: [PATCH] Fixed: Parsing similar movie titles with common words at end

---
 .../ParserTests/NormalizeTitleFixture.cs      | 28 +++++++++++++++----
 src/NzbDrone.Core/Parser/Parser.cs            |  2 +-
 2 files changed, 24 insertions(+), 6 deletions(-)

diff --git a/src/NzbDrone.Core.Test/ParserTests/NormalizeTitleFixture.cs b/src/NzbDrone.Core.Test/ParserTests/NormalizeTitleFixture.cs
index 69dde0889..23631b51d 100644
--- a/src/NzbDrone.Core.Test/ParserTests/NormalizeTitleFixture.cs
+++ b/src/NzbDrone.Core.Test/ParserTests/NormalizeTitleFixture.cs
@@ -37,16 +37,13 @@ public void should_remove_special_characters_and_casing(string dirty, string cle
         [TestCase("or")]
         [TestCase("an")]
         [TestCase("of")]
-        public void should_remove_common_words(string word)
+        public void should_remove_common_words_from_middle_of_title(string word)
         {
             var dirtyFormat = new[]
                             {
                                 "word.{0}.word",
                                 "word {0} word",
-                                "word-{0}-word",
-                                "word.word.{0}",
-                                "word-word-{0}",
-                                "word-word {0}",
+                                "word-{0}-word"
                             };
 
             foreach (var s in dirtyFormat)
@@ -56,6 +53,27 @@ public void should_remove_common_words(string word)
             }
         }
 
+        [TestCase("the")]
+        [TestCase("and")]
+        [TestCase("or")]
+        [TestCase("an")]
+        [TestCase("of")]
+        public void should_not_remove_common_words_from_end_of_title(string word)
+        {
+            var dirtyFormat = new[]
+                              {
+                                  "word.word.{0}",
+                                  "word-word-{0}",
+                                  "word-word {0}"
+                              };
+
+            foreach (var s in dirtyFormat)
+            {
+                var dirty = string.Format(s, word);
+                dirty.CleanMovieTitle().Should().Be("wordword" + word.ToLower());
+            }
+        }
+
         [Test]
         public void should_remove_a_from_middle_of_title()
         {
diff --git a/src/NzbDrone.Core/Parser/Parser.cs b/src/NzbDrone.Core/Parser/Parser.cs
index 86ec89acf..4c45bbd1e 100644
--- a/src/NzbDrone.Core/Parser/Parser.cs
+++ b/src/NzbDrone.Core/Parser/Parser.cs
@@ -100,7 +100,7 @@ public static class Parser
         // Regex to unbracket alternative titles.
         private static readonly Regex BracketedAlternativeTitleRegex = new Regex(@"(.*) \([ ]*AKA[ ]+(.*)\)", RegexOptions.IgnoreCase | RegexOptions.Compiled);
 
-        private static readonly Regex NormalizeRegex = new Regex(@"((?:\b|_)(?<!^|[^a-zA-Z0-9_']\w[^a-zA-Z0-9_'])(a(?!$|[^a-zA-Z0-9_']\w[^a-zA-Z0-9_'])|an|the|and|or|of)(?:\b|_))|\W|_",
+        private static readonly Regex NormalizeRegex = new Regex(@"((?:\b|_)(?<!^|[^a-zA-Z0-9_']\w[^a-zA-Z0-9_'])(a(?!$|[^a-zA-Z0-9_']\w[^a-zA-Z0-9_'])|an|the|and|or|of)(?!$)(?:\b|_))|\W|_",
                                                                 RegexOptions.IgnoreCase | RegexOptions.Compiled);
 
         private static readonly Regex FileExtensionRegex = new Regex(@"\.[a-z0-9]{2,4}$",