Persist file categories on the index and classify archives and unknown types during idle maintenance.

Co-authored-by: Cursor <cursoragent@cursor.com>

tests/Explorer.Storage.Tests/StorageTests.cs

@@ -114,6 +114,135 @@ public class EntryAndSearchTests
         Assert.Contains(byDate, e => e.Name == "Movie.mkv");
     }
 
+    [Fact]
+    public async Task Upsert_assigns_and_preserves_categories()
+    {
+        await using var store = await Stores.Open();
+        var source = await store.AddSourceAsync(@"C:\media");
+        var root = new IndexEntry
+        {
+            SourceId = source.Id,
+            Name = "media",
+            NameNorm = "media",
+            IsDirectory = true,
+            PathRel = "",
+            LastSeenUtc = DateTimeOffset.UtcNow
+        };
+        root.Id = await store.Entries.UpsertAsync(root);
+        var photo = new IndexEntry
+        {
+            SourceId = source.Id,
+            ParentId = root.Id,
+            Name = "shot.jpg",
+            NameNorm = "shot.jpg",
+            Extension = "jpg",
+            PathRel = "shot.jpg",
+            LastSeenUtc = DateTimeOffset.UtcNow
+        };
+        photo.Id = await store.Entries.UpsertAsync(photo);
+        var loaded = await store.Entries.GetAsync(photo.Id);
+        Assert.Equal(FileCategory.Photos, loaded!.Category);
+        Assert.Equal(CategorySources.Extension, loaded.CategorySource);
+
+        await store.Entries.SetUserCategoryAsync(photo.Id, FileCategory.Documents, "Manual");
+        photo.SizeBytes = 100;
+        await store.Entries.UpsertAsync(photo);
+        loaded = await store.Entries.GetAsync(photo.Id);
+        Assert.Equal(FileCategory.Documents, loaded!.Category);
+        Assert.Equal(CategorySources.User, loaded.CategorySource);
+
+        var other = new IndexEntry
+        {
+            SourceId = source.Id,
+            ParentId = root.Id,
+            Name = "clip.mkv",
+            NameNorm = "clip.mkv",
+            Extension = "mkv",
+            PathRel = "clip.mkv",
+            LastSeenUtc = DateTimeOffset.UtcNow
+        };
+        other.Id = await store.Entries.UpsertAsync(other);
+        await store.Entries.RefreshCategoryFromChildrenAsync(root.Id);
+        var folder = await store.Entries.GetAsync(root.Id);
+        Assert.Equal(FileCategory.Unknown, folder!.Category);
+    }
+
+    [Fact]
+    public async Task Archive_contents_and_signature_queues()
+    {
+        await using var store = await Stores.Open();
+        var source = await store.AddSourceAsync(@"C:\media");
+        var zip = new IndexEntry
+        {
+            SourceId = source.Id,
+            Name = "photos.zip",
+            NameNorm = "photos.zip",
+            Extension = "zip",
+            PathRel = "photos.zip",
+            LastSeenUtc = DateTimeOffset.UtcNow
+        };
+        zip.Id = await store.Entries.UpsertAsync(zip);
+        Assert.Equal(FileCategory.Archive, (await store.Entries.GetAsync(zip.Id))!.Category);
+
+        for (var i = 0; i < 3; i++)
+        {
+            await store.Entries.UpsertAsync(new IndexEntry
+            {
+                SourceId = source.Id,
+                ParentId = zip.Id,
+                Name = $"shot{i}.jpg",
+                NameNorm = $"shot{i}.jpg",
+                Extension = "jpg",
+                PathRel = $"photos.zip/shot{i}.jpg",
+                LastSeenUtc = DateTimeOffset.UtcNow
+            });
+        }
+
+        var needing = await store.Entries.GetArchiveIdsNeedingContentClassifyAsync(10);
+        Assert.Contains(zip.Id, needing);
+        await store.Entries.RefreshCategoryFromChildrenAsync(zip.Id);
+        var loaded = await store.Entries.GetAsync(zip.Id);
+        Assert.Equal(FileCategory.Photos, loaded!.Category);
+        Assert.Equal(CategorySources.ArchiveContents, loaded.CategorySource);
+        Assert.Empty(await store.Entries.GetArchiveIdsNeedingContentClassifyAsync(10));
+
+        var mystery = new IndexEntry
+        {
+            SourceId = source.Id,
+            Name = "mystery.bin",
+            NameNorm = "mystery.bin",
+            Extension = "bin",
+            PathRel = "mystery.bin",
+            LastSeenUtc = DateTimeOffset.UtcNow
+        };
+        mystery.Id = await store.Entries.UpsertAsync(mystery);
+        var unknowns = await store.Entries.GetUnknownFilesForSignatureAsync(10);
+        Assert.Contains(unknowns, e => e.Id == mystery.Id);
+        Assert.Equal(0, await store.Entries.BackfillCheapCategoriesAsync(50));
+    }
+
+    [Fact]
+    public async Task User_category_by_path_is_preserved()
+    {
+        await using var store = await Stores.Open();
+        var source = await store.AddSourceAsync(@"C:\media");
+        var photo = new IndexEntry
+        {
+            SourceId = source.Id,
+            Name = "shot.jpg",
+            NameNorm = "shot.jpg",
+            Extension = "jpg",
+            PathRel = "shot.jpg",
+            LastSeenUtc = DateTimeOffset.UtcNow
+        };
+        photo.Id = await store.Entries.UpsertAsync(photo);
+        var mutations = new Explorer.Application.LocalIndexMutations(store);
+        await mutations.SetUserCategoryByPathAsync(@"C:\media\shot.jpg", FileCategory.Installer, "Manual");
+        var loaded = await store.Entries.GetAsync(photo.Id);
+        Assert.Equal(FileCategory.Installer, loaded!.Category);
+        Assert.Equal(CategorySources.User, loaded.CategorySource);
+    }
+
     [Fact]
     public async Task Folder_aggregates_and_tombstones()
     {