refactor: improve matcher algorithm for some edge cases

- Add direct return for single video and subtitle match - Implement merging of items with the same filename - Add unit test for MergeSameFilenameItems function
qwqcode · Nov 14, 2024 · d229f38 · d229f38
1 parent 5eb0512
commit d229f38
Show file tree

Hide file tree

Showing 3 changed files with 94 additions and 2 deletions.
diff --git a/SubRenamer.Core/Matcher.cs b/SubRenamer.Core/Matcher.cs
@@ -27,14 +27,27 @@ public static List<MatchItem> Execute(IReadOnlyList<MatchItem> inputItems, Match
             subtitleFiles.Add(item.Subtitle);
         });
 
+        // Return directly if only 1 video and 1 subtitle
+        if (videoFiles.Count == 1 && subtitleFiles.Count == 1)
+            return [new MatchItem("1", videoFiles[0], subtitleFiles[0])];
+
         // Get file keys
         var video2Keys = CalculateFileKeys(videoFiles, customRegex: options.VideoRegex);
         var subtitle2Keys = CalculateFileKeys(subtitleFiles, customRegex: options.SubtitleRegex);
 
+        // Merge items with same filename
+        result = MatcherHelper.MergeSameFilenameItems(result);
+
         // Apply keys
         List<MatchItem> keyedItems = [];
         foreach (var item in result)
         {
+            if (item.Key != "")
+            {
+                keyedItems.Add(item);
+                continue;
+            }
+
             string? k = null;
 
             if (!string.IsNullOrEmpty(item.Video)) video2Keys.TryGetValue(item.Video, out k);
@@ -77,15 +90,17 @@ private static Dictionary<string, string> CalculateFileKeys(IReadOnlyList<string
             // Extract Match keys
             foreach (var f in files)
             {
-                result[f] = MatcherHelper.PatchKey(MatcherDiff.ExtractMatchKeyByDiff(diff, Path.GetFileNameWithoutExtension(f)));
+                result[f] = MatcherHelper.PatchKey(
+                    MatcherDiff.ExtractMatchKeyByDiff(diff, Path.GetFileNameWithoutExtension(f)));
             }
         }
         else
         {
             // Method 2. Custom Regex
             foreach (var f in files)
             {
-                result[f] = MatcherHelper.PatchKey(MatcherHelper.ExtractMatchKeyRegex(customRegex, Path.GetFileName(f)));
+                result[f] = MatcherHelper.PatchKey(
+                    MatcherHelper.ExtractMatchKeyRegex(customRegex, Path.GetFileName(f)));
             }
         }
 

diff --git a/SubRenamer.Core/MatcherHelper.cs b/SubRenamer.Core/MatcherHelper.cs
@@ -65,6 +65,43 @@ public static List<MatchItem> MergeSameKeysItems(IReadOnlyList<MatchItem> items)
         return result;
     }
 
+    public static List<MatchItem> MergeSameFilenameItems(IReadOnlyList<MatchItem> items)
+    {
+        var groupBNotNullCNull = items
+            .Where(item => string.IsNullOrEmpty(item.Key) && !string.IsNullOrEmpty(item.Video) && string.IsNullOrEmpty(item.Subtitle))
+            .ToList();
+
+        var groupBNullCNotNull = items
+            .Where(item => string.IsNullOrEmpty(item.Key) && string.IsNullOrEmpty(item.Video) && !string.IsNullOrEmpty(item.Subtitle))
+            .ToList();
+
+        var result = new List<MatchItem>();
+
+        var mergedItems = groupBNotNullCNull
+            .Join(groupBNullCNotNull,
+                item1 => Path.GetFileNameWithoutExtension(item1.Video),           
+                item2 => Path.GetFileNameWithoutExtension(item2.Subtitle),           
+                (item1, item2) => new MatchItem(Path.GetFileNameWithoutExtension(item1.Video), item1.Video, item2.Subtitle))
+            .ToList();
+
+        result.AddRange(mergedItems);
+
+        var mergedVideos = new HashSet<string>(mergedItems.Select(x => x.Video));
+        var mergedSubtitles = new HashSet<string>(mergedItems.Select(x => x.Subtitle));
+
+        var itemsToRemove = items.Where(item =>
+        {
+            if (item.Key != "") return false;
+            if (item.Video != "" && item.Subtitle != "") return false;
+            if (item.Video != "" && mergedVideos.Contains(item.Video)) return true;
+            if (item.Subtitle != "" && mergedSubtitles.Contains(item.Subtitle)) return true;
+            return false;
+        });
+        result.AddRange(items.Where(item => !itemsToRemove.Contains(item)));
+
+        return result;
+    }
+
     public static List<MatchItem> MoveEmptyKeyItemsToLast(IReadOnlyList<MatchItem> items)
     {
         var keyedItems = items.Where(x => !string.IsNullOrEmpty(x.Key));

diff --git a/SubRenamer.Tests/MatcherTests/MergeSameFilenameItemsTest.cs b/SubRenamer.Tests/MatcherTests/MergeSameFilenameItemsTest.cs
@@ -0,0 +1,40 @@
+using SubRenamer.Core;
+using static SubRenamer.Core.MatcherHelper;
+
+namespace SubRenamer.Tests.MatcherTests;
+
+[TestFixture]
+public class MergeSameFilenameItemsTests
+{
+    [Test]
+    public void Basic()
+    {
+        var input = new List<MatchItem>
+        {
+            new("", "a.mp4", ""),
+            new("", "", "b.srt"),
+
+            new("", "", "a.srt"),
+            new("", "b.mp4", ""),
+
+
+            new("", "v1.mp4", ""),
+            new("", "", "s1.srt"),
+            new("2", "v2.mp4", ""),
+            new("2", "", "s2.srt"),
+        };
+        var output = new List<MatchItem>
+        {
+            new("a", "a.mp4", "a.srt"),
+            new("b", "b.mp4", "b.srt"),
+
+            new("", "v1.mp4", ""),
+            new("", "", "s1.srt"),
+            new("2", "v2.mp4", ""),
+            new("2", "", "s2.srt"),
+        };
+
+        var result = MergeSameFilenameItems(input);
+        Assert.That(result, Is.EqualTo(output));
+    }
+}