Fix similarity (#16942)

Fix similarity

5 files changed, 57 insertions, 29 deletions

diff --git a/Emby.Server.Implementations/Library/LibraryManager.cs b/Emby.Server.Implementations/Library/LibraryManager.cs
index cc85f09d23..a826db090f 100644
--- a/Emby.Server.Implementations/Library/LibraryManager.cs
+++ b/Emby.Server.Implementations/Library/LibraryManager.cs
@@ -3395,9 +3395,9 @@ namespace Emby.Server.Implementations.Library
         }
 
         /// <inheritdoc/>
-        public IReadOnlyList<string> GetPeopleNamesByItems(IReadOnlyList<Guid> itemIds, IReadOnlyList<string> personTypes, int limit)
+        public IReadOnlyDictionary<Guid, IReadOnlyList<string>> GetPeopleNamesByItems(IReadOnlyList<Guid> itemIds, IReadOnlyList<string> personTypes)
         {
-            return _peopleRepository.GetPeopleNamesByItems(itemIds, personTypes, limit);
+            return _peopleRepository.GetPeopleNamesByItems(itemIds, personTypes);
         }
 
         public void UpdatePeople(BaseItem item, List<PersonInfo> people)
diff --git a/Emby.Server.Implementations/Library/SimilarItems/SimilarItemsManager.cs b/Emby.Server.Implementations/Library/SimilarItems/SimilarItemsManager.cs
index 358c170db2..d923cff07e 100644
--- a/Emby.Server.Implementations/Library/SimilarItems/SimilarItemsManager.cs
+++ b/Emby.Server.Implementations/Library/SimilarItems/SimilarItemsManager.cs
@@ -125,6 +125,7 @@ public class SimilarItemsManager : ISimilarItemsManager
 
         var allResults = new List<(BaseItem Item, float Score)>();
         var excludeIds = new HashSet<Guid> { item.Id };
+        var excludeKeys = new HashSet<string>(StringComparer.OrdinalIgnoreCase) { item.GetPresentationUniqueKey() };
         foreach (var (providerOrder, provider) in orderedProviders.Index())
         {
             if (allResults.Count >= requestedLimit || cancellationToken.IsCancellationRequested)
@@ -149,7 +150,9 @@ public class SimilarItemsManager : ISimilarItemsManager
 
                     foreach (var (position, resultItem) in items.Index())
                     {
-                        if (excludeIds.Add(resultItem.Id))
+                        var isNewId = excludeIds.Add(resultItem.Id);
+                        var isNewKey = excludeKeys.Add(resultItem.GetPresentationUniqueKey());
+                        if (isNewId && isNewKey)
                         {
                             var score = CalculateScore(null, providerOrder, position);
                             allResults.Add((resultItem, score));
@@ -163,7 +166,7 @@ public class SimilarItemsManager : ISimilarItemsManager
                     var cachedReferences = await TryReadSimilarItemsCacheAsync(cachePath, cancellationToken).ConfigureAwait(false);
                     if (cachedReferences is not null)
                     {
-                        var resolvedItems = ResolveRemoteReferences(cachedReferences, providerOrder, user, dtoOptions, itemKind, excludeIds);
+                        var resolvedItems = ResolveRemoteReferences(cachedReferences, providerOrder, user, dtoOptions, itemKind, excludeIds, excludeKeys);
                         allResults.AddRange(resolvedItems);
                         continue;
                     }
@@ -191,7 +194,7 @@ public class SimilarItemsManager : ISimilarItemsManager
 
                         if (pendingBatch.Count >= BatchSize)
                         {
-                            var resolvedItems = ResolveRemoteReferences(pendingBatch, providerOrder, user, dtoOptions, itemKind, excludeIds);
+                            var resolvedItems = ResolveRemoteReferences(pendingBatch, providerOrder, user, dtoOptions, itemKind, excludeIds, excludeKeys);
                             allResults.AddRange(resolvedItems);
                             remaining -= resolvedItems.Count;
                             pendingBatch.Clear();
@@ -206,7 +209,7 @@ public class SimilarItemsManager : ISimilarItemsManager
                     // Resolve any remaining references in the last partial batch
                     if (pendingBatch.Count > 0)
                     {
-                        var resolvedItems = ResolveRemoteReferences(pendingBatch, providerOrder, user, dtoOptions, itemKind, excludeIds);
+                        var resolvedItems = ResolveRemoteReferences(pendingBatch, providerOrder, user, dtoOptions, itemKind, excludeIds, excludeKeys);
                         allResults.AddRange(resolvedItems);
                     }
 
@@ -435,7 +438,11 @@ public class SimilarItemsManager : ISimilarItemsManager
     private IReadOnlyList<string> GetPeopleNames(IReadOnlyList<BaseItem> items, IReadOnlyList<string> personTypes)
     {
         var itemIds = items.Select(i => i.Id).ToArray();
-        return _libraryManager.GetPeopleNamesByItems(itemIds, personTypes, limit: 0);
+        return _libraryManager.GetPeopleNamesByItems(itemIds, personTypes)
+            .Values
+            .SelectMany(names => names)
+            .Distinct()
+            .ToArray();
     }
 
     private List<(BaseItem Item, float Score)> ResolveRemoteReferences(
@@ -444,14 +451,15 @@ public class SimilarItemsManager : ISimilarItemsManager
         User? user,
         DtoOptions dtoOptions,
         BaseItemKind itemKind,
-        HashSet<Guid> excludeIds)
+        HashSet<Guid> excludeIds,
+        HashSet<string> excludeKeys)
     {
         if (references.Count == 0)
         {
             return [];
         }
 
-        var resolvedById = new Dictionary<Guid, (BaseItem Item, float Score)>();
+        var resolvedByKey = new Dictionary<string, (BaseItem Item, float Score)>(StringComparer.OrdinalIgnoreCase);
         var providerLookup = new Dictionary<(string ProviderName, string ProviderId), (float? Score, int Position)>(StringTupleComparer.Instance);
 
         foreach (var (position, match) in references.Index())
@@ -482,7 +490,13 @@ public class SimilarItemsManager : ISimilarItemsManager
 
         foreach (var item in items)
         {
-            if (excludeIds.Contains(item.Id) || resolvedById.ContainsKey(item.Id))
+            if (excludeIds.Contains(item.Id))
+            {
+                continue;
+            }
+
+            var presentationKey = item.GetPresentationUniqueKey();
+            if (excludeKeys.Contains(presentationKey))
             {
                 continue;
             }
@@ -492,10 +506,9 @@ public class SimilarItemsManager : ISimilarItemsManager
                 if (item.TryGetProviderId(providerName, out var itemProviderId) && providerLookup.TryGetValue((providerName, itemProviderId), out var matchInfo))
                 {
                     var score = CalculateScore(matchInfo.Score, providerOrder, matchInfo.Position);
-                    if (!resolvedById.TryGetValue(item.Id, out var existing) || existing.Score < score)
+                    if (!resolvedByKey.TryGetValue(presentationKey, out var existing) || existing.Score < score)
                     {
-                        excludeIds.Add(item.Id);
-                        resolvedById[item.Id] = (item, score);
+                        resolvedByKey[presentationKey] = (item, score);
                     }
 
                     break;
@@ -503,7 +516,13 @@ public class SimilarItemsManager : ISimilarItemsManager
             }
         }
 
-        return [.. resolvedById.Values];
+        foreach (var (key, entry) in resolvedByKey)
+        {
+            excludeIds.Add(entry.Item.Id);
+            excludeKeys.Add(key);
+        }
+
+        return [.. resolvedByKey.Values];
     }
 
     private static float CalculateScore(float? matchScore, int providerOrder, int position)
diff --git a/Jellyfin.Server.Implementations/Item/PeopleRepository.cs b/Jellyfin.Server.Implementations/Item/PeopleRepository.cs
index 6062aaca2f..eb87b525fe 100644
--- a/Jellyfin.Server.Implementations/Item/PeopleRepository.cs
+++ b/Jellyfin.Server.Implementations/Item/PeopleRepository.cs
@@ -166,7 +166,7 @@ public class PeopleRepository(IDbContextFactory<JellyfinDbContext> dbProvider, I
     }
 
     /// <inheritdoc/>
-    public IReadOnlyList<string> GetPeopleNamesByItems(IReadOnlyList<Guid> itemIds, IReadOnlyList<string> personTypes, int limit)
+    public IReadOnlyDictionary<Guid, IReadOnlyList<string>> GetPeopleNamesByItems(IReadOnlyList<Guid> itemIds, IReadOnlyList<string> personTypes)
     {
         using var context = _dbProvider.CreateDbContext();
         var query = context.PeopleBaseItemMap
@@ -178,16 +178,27 @@ public class PeopleRepository(IDbContextFactory<JellyfinDbContext> dbProvider, I
             query = query.Where(m => personTypes.Contains(m.People.PersonType));
         }
 
-        var names = query
-            .Select(m => m.People.Name)
-            .Distinct();
+        var rows = query
+            .OrderBy(m => m.ListOrder)
+            .Select(m => new { m.ItemId, m.People.Name })
+            .ToList();
 
-        if (limit > 0)
+        var result = new Dictionary<Guid, IReadOnlyList<string>>();
+        foreach (var group in rows.GroupBy(r => r.ItemId))
         {
-            names = names.Take(limit);
+            var names = group
+                .Select(r => r.Name)
+                .Where(name => !string.IsNullOrEmpty(name))
+                .Distinct()
+                .ToArray();
+
+            if (names.Length > 0)
+            {
+                result[group.Key] = names;
+            }
         }
 
-        return names.ToArray();
+        return result;
     }
 
     private PersonInfo Map(People people)
diff --git a/MediaBrowser.Controller/Library/ILibraryManager.cs b/MediaBrowser.Controller/Library/ILibraryManager.cs
index c23eba75ef..0b64da291c 100644
--- a/MediaBrowser.Controller/Library/ILibraryManager.cs
+++ b/MediaBrowser.Controller/Library/ILibraryManager.cs
@@ -598,13 +598,12 @@ namespace MediaBrowser.Controller.Library
         IReadOnlyList<string> GetPeopleNames(InternalPeopleQuery query);
 
         /// <summary>
-        /// Gets distinct people names for multiple items.
+        /// Gets the distinct people names per item for multiple items.
         /// </summary>
         /// <param name="itemIds">The item IDs.</param>
         /// <param name="personTypes">The person types to include.</param>
-        /// <param name="limit">Maximum number of names.</param>
-        /// <returns>The distinct people names.</returns>
-        IReadOnlyList<string> GetPeopleNamesByItems(IReadOnlyList<Guid> itemIds, IReadOnlyList<string> personTypes, int limit);
+        /// <returns>A dictionary mapping each item ID to its distinct people names. Items with no matching people are omitted.</returns>
+        IReadOnlyDictionary<Guid, IReadOnlyList<string>> GetPeopleNamesByItems(IReadOnlyList<Guid> itemIds, IReadOnlyList<string> personTypes);
 
         /// <summary>
         /// Queries the items.
diff --git a/MediaBrowser.Controller/Persistence/IPeopleRepository.cs b/MediaBrowser.Controller/Persistence/IPeopleRepository.cs
index 7474130ec4..e2833dc722 100644
--- a/MediaBrowser.Controller/Persistence/IPeopleRepository.cs
+++ b/MediaBrowser.Controller/Persistence/IPeopleRepository.cs
@@ -34,11 +34,10 @@ public interface IPeopleRepository
     IReadOnlyList<string> GetPeopleNames(InternalPeopleQuery filter);
 
     /// <summary>
-    /// Gets distinct people names for multiple items efficiently by querying from the mapping table.
+    /// Gets the distinct people names per item for multiple items efficiently by querying from the mapping table.
     /// </summary>
     /// <param name="itemIds">The item IDs to get people for.</param>
     /// <param name="personTypes">The person types to include (e.g. "Actor", "Director").</param>
-    /// <param name="limit">Maximum number of names to return.</param>
-    /// <returns>The distinct people names.</returns>
-    IReadOnlyList<string> GetPeopleNamesByItems(IReadOnlyList<Guid> itemIds, IReadOnlyList<string> personTypes, int limit);
+    /// <returns>A dictionary mapping each item ID to its distinct people names, ordered by cast list order. Items with no matching people are omitted.</returns>
+    IReadOnlyDictionary<Guid, IReadOnlyList<string>> GetPeopleNamesByItems(IReadOnlyList<Guid> itemIds, IReadOnlyList<string> personTypes);
 }