Improve the fuzzy matching (#522)

* Fixed: improve track matching

* Deal with tracks sequentially numbered across discs

src/NzbDrone.Common.Test/ExtensionTests/FuzzyContainsFixture.cs

@@ -0,0 +1,61 @@
+using FluentAssertions;
+using NUnit.Framework;
+using NzbDrone.Common.Extensions;
+using NzbDrone.Test.Common;
+
+namespace NzbDrone.Common.Test
+{
+    [TestFixture]
+    public class FuzzyContainsFixture : TestBase
+    {
+        [TestCase("abcdef", "abcdef", 0.5, 0)]
+        [TestCase("", "abcdef", 0.5, -1)]
+        [TestCase("abcdef", "", 0.5, -1)]
+        [TestCase("", "", 0.5, -1)]
+        [TestCase("abcdef", "de", 0.5, 3)]
+        [TestCase("abcdef", "defy", 0.5, 3)]
+        [TestCase("abcdef", "abcdefy", 0.5, 0)]
+        [TestCase("I am the very model of a modern major general.", " that berry ", 0.3, 4)]
+        [TestCase("abcdefghijk", "fgh", 0.5, 5)]
+        [TestCase("abcdefghijk", "fgh", 0.5, 5)]
+        [TestCase("abcdefghijk", "efxhi", 0.5, 4)]
+        [TestCase("abcdefghijk", "cdefxyhijk", 0.5, 2)]
+        [TestCase("abcdefghijk", "bxy", 0.5, -1)]
+        [TestCase("123456789xx0", "3456789x0", 0.5, 2)]
+        [TestCase("abcdef", "xxabc", 0.5, 0)]
+        [TestCase("abcdef", "defyy", 0.5, 3)]
+        [TestCase("abcdef", "xabcdefy", 0.5, 0)]
+        [TestCase("abcdefghijk", "efxyhi", 0.6, 4)]
+        [TestCase("abcdefghijk", "efxyhi", 0.7, -1)]
+        [TestCase("abcdefghijk", "bcdef", 0.0, 1)]
+        [TestCase("abcdexyzabcde", "abccde", 0.5, 0)]
+        [TestCase("abcdefghijklmnopqrstuvwxyz", "abcdxxefg", 0.5, 0)]
+        [TestCase("abcdefghijklmnopqrstuvwxyz", "abcdefg", 0.5, 0)]
+        [TestCase("The quick brown fox jumps over the lazy dog", "The quick brown fox jumps over the lazy d", 0.5, 0)]
+        [TestCase("The quick brown fox jumps over the lazy dog", "The quick brown fox jumps over the lazy g", 0.5, 0)]
+        [TestCase("The quick brown fox jumps over the lazy dog", "quikc brown fox jumps over the lazy dog", 0.5, 4)]
+        [TestCase("The quick brown fox jumps over the lazy dog", "qui jumps over the lazy dog", 0.5, 16)]
+        [TestCase("The quick brown fox jumps over the lazy dog", "quikc brown fox jumps over the lazy dog", 0.5, 4)]
+        [TestCase("u6IEytQiYpzAccsbjQ5ISuE4smDQ1ZiU42cFBrTeKB2XrVLEqAvgIiKlDP75iApy07jzmK", "xEytQiYpzAccsbjQ5ISuE4smDQ1ZiU42cFBrTeKB2XrVLEqAvgIiKlDP75iApy07jzmK", 0.5, 2)]
+        [TestCase("plusifeelneedforredundantinformationintitlefield", "anthology", 0.5, -1)]
+        public void FuzzyFind(string text, string pattern, double threshold, int expected)
+        {
+            text.FuzzyFind(pattern, threshold).Should().Be(expected);
+        }
+
+        [TestCase("abcdef", "abcdef", 1)]
+        [TestCase("", "abcdef", 0)]
+        [TestCase("abcdef", "", 0)]
+        [TestCase("", "", 0)]
+        [TestCase("abcdef", "de", 1)]
+        [TestCase("abcdef", "defy", 0.75)]
+        [TestCase("abcdef", "abcdefghk", 6.0/9)]
+        [TestCase("abcdef", "zabcdefz", 6.0/8)]
+        [TestCase("plusifeelneedforredundantinformationintitlefield", "anthology", 4.0/9)]
+        [TestCase("+ (Plus) - I feel the need for redundant information in the title field", "+", 1)]
+        public void FuzzyContains(string text, string pattern, double expectedScore)
+        {
+            text.FuzzyContains(pattern).Should().BeApproximately(expectedScore, 1e-9);
+        }
+    }
+}

src/NzbDrone.Common.Test/LevenshteinDistanceFixture.cs

@@ -42,5 +42,21 @@ namespace NzbDrone.Common.Test
         {
             text.ToLower().LevenshteinDistanceClean(other.ToLower()).Should().Be(expected);
         }
+
+        [TestCase("hello", "hello")]
+        [TestCase("hello", "bye")]
+        [TestCase("a longer string", "a different long string")]
+        public void FuzzyMatchSymmetric(string a, string b)
+        {
+            a.FuzzyMatch(b).Should().Be(b.FuzzyMatch(a));
+        }
+
+        [TestCase("", "", 0)]
+        [TestCase("a", "", 0)]
+        [TestCase("", "a", 0)]
+        public void FuzzyMatchEmptyValuesReturnZero(string a, string b, double expected)
+        {
+            a.FuzzyMatch(b).Should().Be(expected);
+        }
     }
 }

src/NzbDrone.Common.Test/NzbDrone.Common.Test.csproj

@@ -84,6 +84,7 @@
     <Compile Include="ExtensionTests\IEnumerableExtensionTests\IntersectByFixture.cs" />
     <Compile Include="ExtensionTests\Int64ExtensionFixture.cs" />
     <Compile Include="ExtensionTests\UrlExtensionsFixture.cs" />
+    <Compile Include="ExtensionTests\FuzzyContainsFixture.cs" />
     <Compile Include="HashUtilFixture.cs" />
     <Compile Include="Http\HttpClientFixture.cs" />
     <Compile Include="Http\HttpHeaderFixture.cs" />