PLib/tests/PLib.Tests/Library/LibraryServiceTests.cs

using Microsoft.Extensions.Logging.Abstractions;
using Microsoft.Extensions.Options;
using NSubstitute;
using PLib.Application.Abstractions;
using PLib.Application.Library;
using PLib.Application.Metadata;
using PLib.Domain.Videos;
using Shouldly;

namespace PLib.Tests.Library;

public sealed class LibraryServiceTests
{
    private const string Root = @"C:\videos";

    private readonly InMemoryVideoRepository _repository = new();
    private readonly InMemoryLabelRepository _labels = new();
    private readonly IVideoFileScanner _scanner = Substitute.For<IVideoFileScanner>();
    private readonly IMediaProbe _probe = Substitute.For<IMediaProbe>();
    private readonly IThumbnailGenerator _thumbnails = Substitute.For<IThumbnailGenerator>();
    private readonly IAnimatedPreviewGenerator _previews = Substitute.For<IAnimatedPreviewGenerator>();
    private readonly IVideoPerceptualHasher _hasher = Substitute.For<IVideoPerceptualHasher>();
    private readonly IMetadataProvider _metadata = Substitute.For<IMetadataProvider>();

    public LibraryServiceTests()
    {
        _probe.ProbeAsync(Arg.Any<string>(), Arg.Any<CancellationToken>())
            .Returns(new VideoTechnicalInfo(TimeSpan.FromMinutes(2), 1920, 1080, "h264"));

        _thumbnails.GetOrCreateAsync(Arg.Any<string>(), Arg.Any<TimeSpan?>(), Arg.Any<CancellationToken>())
            .Returns(callInfo => $@"C:\cache\{Path.GetFileNameWithoutExtension(callInfo.Arg<string>())}.jpg");

        _previews.GetOrCreateAsync(Arg.Any<string>(), Arg.Any<TimeSpan?>(), Arg.Any<CancellationToken>())
            .Returns(callInfo => new AnimatedPreview(
                $@"C:\cache\{Path.GetFileNameWithoutExtension(callInfo.Arg<string>())}.strip.jpg",
                12));

        // By default every remembered image is still on disk.
        _thumbnails.IsAvailable(Arg.Any<string?>()).Returns(true);
        _previews.IsAvailable(Arg.Any<string?>()).Returns(true);

        _hasher.ComputeAsync(Arg.Any<string>(), Arg.Any<TimeSpan?>(), Arg.Any<CancellationToken>())
            .Returns(0xDEADBEEFUL);
    }

    [Fact]
    public async Task Files_that_are_new_on_disk_are_added_probed_and_given_a_thumbnail()
    {
        GivenFilesOnDisk(File(@"C:\videos\a.mp4"), File(@"C:\videos\b.mkv"));

        var events = await CollectAsync(CreateService());

        _repository.Items.Count.ShouldBe(2);
        _repository.Items.ShouldAllBe(x => x.IsIndexed);

        events.OfType<LibraryScanEvent.ItemAdded>().Count().ShouldBe(2);

        // Three passes touch every file: the poster frame, the animated preview, the hash.
        events.OfType<LibraryScanEvent.ItemUpdated>().Count().ShouldBe(6);
        events.OfType<LibraryScanEvent.Completed>().Single().LibrarySize.ShouldBe(2);
    }

    [Fact]
    public async Task Each_pass_finishes_for_every_file_before_the_next_one_starts()
    {
        GivenFilesOnDisk(File(@"C:\videos\a.mp4"), File(@"C:\videos\b.mp4"), File(@"C:\videos\c.mp4"));

        var events = await CollectAsync(CreateService());

        // Cheapest and most visible first: interleaving would hold the poster frame of every
        // file behind the frame grabs of the one before it.
        var lastThumbnail = events.FindLastIndex(e => e is LibraryScanEvent.IndexingProgress);
        var firstPreview = events.FindIndex(e => e is LibraryScanEvent.PreviewProgress);
        var lastPreview = events.FindLastIndex(e => e is LibraryScanEvent.PreviewProgress);
        var firstHash = events.FindIndex(e => e is LibraryScanEvent.HashingProgress);

        firstPreview.ShouldBeGreaterThan(lastThumbnail);
        firstHash.ShouldBeGreaterThan(lastPreview);
    }

    [Fact]
    public async Task A_file_that_already_has_a_poster_frame_still_gets_a_preview_and_a_hash()
    {
        var indexed = new VideoItem(@"C:\videos\a.mp4", "a", 5_000, DateTimeOffset.UnixEpoch);
        indexed.ApplyTechnicalInfo(new VideoTechnicalInfo(TimeSpan.FromMinutes(1), 1280, 720, "h264"));
        indexed.AttachThumbnail(@"C:\cache\a.jpg");
        _repository.Seed(indexed);

        GivenFilesOnDisk(File(@"C:\videos\a.mp4", sizeInBytes: 5_000));

        await CollectAsync(CreateService());

        // Nothing to re-probe, but the later passes have their own work left to do.
        await _probe.DidNotReceive().ProbeAsync(Arg.Any<string>(), Arg.Any<CancellationToken>());
        _repository.Items.Single().PreviewPath.ShouldNotBeNull();
        _repository.Items.Single().PerceptualHash.ShouldNotBeNull();
    }

    [Fact]
    public async Task An_animated_preview_that_disappeared_from_the_cache_is_rendered_again()
    {
        var indexed = new VideoItem(@"C:\videos\a.mp4", "a", 5_000, DateTimeOffset.UnixEpoch);
        indexed.ApplyTechnicalInfo(new VideoTechnicalInfo(TimeSpan.FromMinutes(1), 1280, 720, "h264"));
        indexed.AttachThumbnail(@"C:\cache\a.jpg");
        indexed.AttachPreview(@"C:\cache\deleted.strip.jpg", 12);
        indexed.ApplyPerceptualHash(1);
        _repository.Seed(indexed);

        _previews.IsAvailable(@"C:\cache\deleted.strip.jpg").Returns(false);
        GivenFilesOnDisk(File(@"C:\videos\a.mp4", sizeInBytes: 5_000));

        await CollectAsync(CreateService());

        _repository.Items.Single().PreviewPath.ShouldBe(@"C:\cache\a.strip.jpg");
    }

    [Fact]
    public async Task A_file_whose_preview_could_not_be_rendered_keeps_the_rest_of_its_indexing()
    {
        _previews.GetOrCreateAsync(Arg.Any<string>(), Arg.Any<TimeSpan?>(), Arg.Any<CancellationToken>())
            .Returns((AnimatedPreview?)null);

        GivenFilesOnDisk(File(@"C:\videos\a.mp4"));

        await CollectAsync(CreateService());

        // A missing preview only costs the hover animation; the card itself is unaffected.
        var item = _repository.Items.Single();
        item.PreviewPath.ShouldBeNull();
        item.IsIndexed.ShouldBeTrue();
        item.PerceptualHash.ShouldNotBeNull();
    }

    [Fact]
    public async Task Entries_whose_file_is_gone_are_dropped_from_the_library()
    {
        _repository.Seed(new VideoItem(@"C:\videos\stale.mp4", "stale", 5_000, DateTimeOffset.UnixEpoch));
        GivenFilesOnDisk(File(@"C:\videos\a.mp4"));

        var events = await CollectAsync(CreateService());

        _repository.Items.Select(x => x.FullPath).ShouldBe([@"C:\videos\a.mp4"]);
        events.OfType<LibraryScanEvent.ItemRemoved>().Count().ShouldBe(1);
    }

    [Fact]
    public async Task Files_below_the_minimum_size_are_not_part_of_the_library()
    {
        GivenFilesOnDisk(File(@"C:\videos\tiny.mp4", sizeInBytes: 128), File(@"C:\videos\real.mp4"));

        await CollectAsync(CreateService(new LibraryOptions { MinimumFileSizeInBytes = 1_024 }));

        _repository.Items.Select(x => x.FullPath).ShouldBe([@"C:\videos\real.mp4"]);
    }

    [Fact]
    public async Task An_item_that_is_already_indexed_is_not_probed_again()
    {
        var indexed = new VideoItem(@"C:\videos\a.mp4", "a", 5_000, DateTimeOffset.UnixEpoch);
        indexed.ApplyTechnicalInfo(new VideoTechnicalInfo(TimeSpan.FromMinutes(1), 1280, 720, "h264"));
        indexed.AttachThumbnail(@"C:\cache\a.jpg");
        indexed.AttachPreview(@"C:\cache\a.strip.jpg", 12);
        indexed.ApplyPerceptualHash(1);
        _repository.Seed(indexed);

        GivenFilesOnDisk(File(@"C:\videos\a.mp4", sizeInBytes: 5_000));

        await CollectAsync(CreateService());

        await _probe.DidNotReceive().ProbeAsync(Arg.Any<string>(), Arg.Any<CancellationToken>());
        await _previews.DidNotReceive()
            .GetOrCreateAsync(Arg.Any<string>(), Arg.Any<TimeSpan?>(), Arg.Any<CancellationToken>());
    }

    [Fact]
    public async Task An_item_whose_file_changed_on_disk_is_indexed_again()
    {
        var indexed = new VideoItem(@"C:\videos\a.mp4", "a", 5_000, DateTimeOffset.UnixEpoch);
        indexed.ApplyTechnicalInfo(new VideoTechnicalInfo(TimeSpan.FromMinutes(1), 1280, 720, "h264"));
        indexed.AttachThumbnail(@"C:\cache\old.jpg");
        _repository.Seed(indexed);

        GivenFilesOnDisk(File(@"C:\videos\a.mp4", sizeInBytes: 9_999));

        await CollectAsync(CreateService());

        await _probe.Received(1).ProbeAsync(@"C:\videos\a.mp4", Arg.Any<CancellationToken>());
        _repository.Items.Single().ThumbnailPath.ShouldBe(@"C:\cache\a.jpg");
    }

    [Fact]
    public async Task A_poster_frame_that_disappeared_from_the_cache_is_generated_again()
    {
        var indexed = new VideoItem(@"C:\videos\a.mp4", "a", 5_000, DateTimeOffset.UnixEpoch);
        indexed.ApplyTechnicalInfo(new VideoTechnicalInfo(TimeSpan.FromMinutes(1), 1280, 720, "h264"));
        indexed.AttachThumbnail(@"C:\cache\deleted.jpg");
        _repository.Seed(indexed);

        _thumbnails.IsAvailable(@"C:\cache\deleted.jpg").Returns(false);
        GivenFilesOnDisk(File(@"C:\videos\a.mp4", sizeInBytes: 5_000));

        await CollectAsync(CreateService());

        await _thumbnails.Received(1)
            .GetOrCreateAsync(@"C:\videos\a.mp4", Arg.Any<TimeSpan?>(), Arg.Any<CancellationToken>());
        _repository.Items.Single().ThumbnailPath.ShouldBe(@"C:\cache\a.jpg");
    }

    [Fact]
    public async Task Cached_frames_nothing_points_at_are_purged_once_the_scan_is_whole()
    {
        GivenFilesOnDisk(File(@"C:\videos\a.mp4"));

        await CollectAsync(CreateService());

        await _thumbnails.Received(1).PurgeUnusedAsync(
            Arg.Is<IReadOnlyCollection<string>>(paths => paths != null && paths.SequenceEqual(new[] { @"C:\cache\a.jpg" })),
            Arg.Any<CancellationToken>());

        // Every cache is swept, not just the poster frames: previews outlive their videos too.
        await _previews.Received(1).PurgeUnusedAsync(
            Arg.Is<IReadOnlyCollection<string>>(paths => paths != null && paths.SequenceEqual(new[] { @"C:\cache\a.strip.jpg" })),
            Arg.Any<CancellationToken>());
    }

    [Fact]
    public async Task The_same_file_reached_through_two_overlapping_roots_is_only_added_once()
    {
        GivenFilesOnDisk(File(@"C:\videos\a.mp4"));

        var events = await CollectAsync(CreateService(), Root, Root);

        _repository.Items.Count.ShouldBe(1);
        events.OfType<LibraryScanEvent.DiscoveryCompleted>().Single().FilesFound.ShouldBe(1);
    }

    [Fact]
    public async Task Clearing_one_kind_of_derived_data_leaves_every_other_kind_alone()
    {
        var item = FullyIndexed();
        _repository.Seed(item);

        await CreateService().ResetAsync(LibraryDataKind.PerceptualHashes, Token);

        item.PerceptualHash.ShouldBeNull();

        // Clearing the hash is a database edit; nothing on disk has anything to do with it.
        item.ThumbnailPath.ShouldNotBeNull();
        item.PreviewPath.ShouldNotBeNull();
        item.Duration.ShouldNotBeNull();
        await _thumbnails.DidNotReceive().ClearAsync(Arg.Any<CancellationToken>());
        await _previews.DidNotReceive().ClearAsync(Arg.Any<CancellationToken>());
    }

    [Fact]
    public async Task Clearing_an_image_cache_deletes_the_files_as_well_as_the_references()
    {
        var item = FullyIndexed();
        _repository.Seed(item);

        await CreateService().ResetAsync(LibraryDataKind.AnimatedPreviews, Token);

        // A path kept after the file was deleted would leave the card pointing at nothing.
        item.PreviewPath.ShouldBeNull();
        await _previews.Received(1).ClearAsync(Arg.Any<CancellationToken>());
        await _thumbnails.DidNotReceive().ClearAsync(Arg.Any<CancellationToken>());
        item.ThumbnailPath.ShouldNotBeNull();
    }

    [Fact]
    public async Task Clearing_everything_takes_all_four_kinds()
    {
        var item = FullyIndexed();
        _repository.Seed(item);

        await CreateService().ResetAsync(LibraryDataKind.All, Token);

        item.ThumbnailPath.ShouldBeNull();
        item.PreviewPath.ShouldBeNull();
        item.PerceptualHash.ShouldBeNull();
        item.Duration.ShouldBeNull();
        item.IsIndexed.ShouldBeFalse();
    }

    [Fact]
    public async Task Usage_says_how_far_each_kind_has_got_through_the_library()
    {
        _repository.Seed(FullyIndexed());
        _repository.Seed(new VideoItem(@"C:\videos\b.mp4", "b", 5_000, DateTimeOffset.UnixEpoch));

        _previews.GetCacheSizeInBytesAsync(Arg.Any<CancellationToken>()).Returns(4_096L);

        var usage = (await CreateService().GetDataUsageAsync(Token)).ToDictionary(entry => entry.Kind);

        usage[LibraryDataKind.PerceptualHashes].ShouldSatisfyAllConditions(
            entry => entry.Videos.ShouldBe(2),
            entry => entry.Present.ShouldBe(1));

        // Only the caches that are files on disk have a size to report.
        usage[LibraryDataKind.AnimatedPreviews].Bytes.ShouldBe(4_096L);
        usage[LibraryDataKind.PerceptualHashes].Bytes.ShouldBe(0);
    }

    [Fact]
    public async Task A_video_without_a_fingerprint_cannot_be_looked_up()
    {
        var item = new VideoItem(@"C:\videos\a.mp4", "a", 5_000, DateTimeOffset.UnixEpoch);
        _repository.Seed(item);

        var result = await CreateService(sources: MetadataMonitor.With(Source("StashDB")))
            .FindMetadataAsync(item.Id, Token);

        // The answer is "wait for the scan", not "check your sources", so no source is called.
        result.HasPerceptualHash.ShouldBeFalse();
        await _metadata.DidNotReceive().FindByPerceptualHashAsync(
            Arg.Any<MetadataSourceOptions>(),
            Arg.Any<ulong>(),
            Arg.Any<CancellationToken>());
    }

    [Fact]
    public async Task A_source_that_fails_does_not_hide_what_the_others_found()
    {
        var item = FullyIndexed();
        _repository.Seed(item);

        var good = Source("Хороший");
        var bad = Source("Сломанный");

        _metadata.FindByPerceptualHashAsync(good, Arg.Any<ulong>(), Arg.Any<CancellationToken>())
            .Returns([Match("Сцена", good.Name)]);
        _metadata.FindByPerceptualHashAsync(bad, Arg.Any<ulong>(), Arg.Any<CancellationToken>())
            .Returns<IReadOnlyList<VideoMetadataMatch>>(_ => throw new HttpRequestException("401"));

        var result = await CreateService(sources: MetadataMonitor.With(bad, good)).FindMetadataAsync(item.Id, Token);

        result.Matches.Single().Title.ShouldBe("Сцена");
        result.Failures.ShouldHaveSingleItem().ShouldContain("Сломанный");
    }

    [Fact]
    public async Task Sources_that_are_switched_off_or_half_filled_are_not_called()
    {
        var item = FullyIndexed();
        _repository.Seed(item);

        var disabled = new MetadataSourceOptions { Name = "Выключен", Endpoint = "https://a/graphql", IsEnabled = false };
        var blank = new MetadataSourceOptions { Name = "Недописан", Endpoint = "   " };

        await CreateService(sources: MetadataMonitor.With(disabled, blank)).FindMetadataAsync(item.Id, Token);

        await _metadata.DidNotReceive().FindByPerceptualHashAsync(
            Arg.Any<MetadataSourceOptions>(),
            Arg.Any<ulong>(),
            Arg.Any<CancellationToken>());
    }

    [Fact]
    public async Task Applying_a_match_writes_the_text_and_a_label_of_the_right_kind_for_each_name()
    {
        var item = FullyIndexed();
        _repository.Seed(item);

        var match = new VideoMetadataMatch(
            "StashDB",
            "scene-1",
            "Настоящее название",
            "Описание",
            Tags: ["Драма", "драма"],
            Performers: ["Актёр Один"],
            Studios: ["Студия"]);

        await CreateService().ApplyMetadataAsync(item.Id, match, Token);

        item.Title.ShouldBe("Настоящее название");
        item.Description.ShouldBe("Описание");

        // The two spellings of the tag are one label: names are matched case-insensitively.
        item.Labels.Count(label => label.Kind == LabelKind.Tag).ShouldBe(1);
        item.Labels.Single(label => label.Kind == LabelKind.Performer).Name.ShouldBe("Актёр Один");
        item.Labels.Single(label => label.Kind == LabelKind.Studio).Name.ShouldBe("Студия");
    }

    [Fact]
    public async Task Applying_a_match_adds_to_the_labels_already_on_the_video()
    {
        var item = FullyIndexed();
        _repository.Seed(item);

        var service = CreateService();
        await service.AttachLabelAsync(item.Id, "Моё", LabelKind.Tag, Token);

        await service.ApplyMetadataAsync(item.Id, Match("Название", "StashDB") with { Tags = ["Их"] }, Token);

        // A match is a proposal, not a replacement: what the user put there stays.
        item.Labels.Select(label => label.Name).ShouldBe(["Моё", "Их"], ignoreOrder: true);
    }

    private static MetadataSourceOptions Source(string name) =>
        new() { Name = name, Endpoint = $"https://{name}.example/graphql", ApiKey = "key" };

    private static VideoMetadataMatch Match(string title, string sourceName) =>
        new(sourceName, "id", title, null, [], [], []);

    private static CancellationToken Token => TestContext.Current.CancellationToken;

    private static VideoItem FullyIndexed()
    {
        var item = new VideoItem(@"C:\videos\a.mp4", "a", 5_000, DateTimeOffset.UnixEpoch);
        item.ApplyTechnicalInfo(new VideoTechnicalInfo(TimeSpan.FromMinutes(1), 1280, 720, "h264"));
        item.AttachThumbnail(@"C:\cache\a.jpg");
        item.AttachPreview(@"C:\cache\a.strip.jpg", 12);
        item.ApplyPerceptualHash(1);
        return item;
    }

    private static DiscoveredVideoFile File(string path, long sizeInBytes = 10_000) =>
        new(path, sizeInBytes, DateTimeOffset.UnixEpoch);

    private void GivenFilesOnDisk(params DiscoveredVideoFile[] files) =>
        _scanner.ScanAsync(Arg.Any<string>(), Arg.Any<CancellationToken>())
            .Returns(_ => files.ToAsyncEnumerable());

    private LibraryService CreateService(
        LibraryOptions? options = null,
        MetadataMonitor? sources = null) => new(
        _repository,
        _labels,
        _scanner,
        _probe,
        _thumbnails,
        _previews,
        _hasher,
        _metadata,
        Options.Create(options ?? new LibraryOptions { MinimumFileSizeInBytes = 0 }),
        sources ?? MetadataMonitor.Empty,
        NullLogger<LibraryService>.Instance);

    private static async Task<List<LibraryScanEvent>> CollectAsync(
        LibraryService service,
        params string[] folders)
    {
        var events = new List<LibraryScanEvent>();

        await foreach (var scanEvent in service.ScanAsync(folders.Length == 0 ? [Root] : folders))
        {
            events.Add(scanEvent);
        }

        return events;
    }
}

internal static class AsyncEnumerableExtensions
{
    public static async IAsyncEnumerable<T> ToAsyncEnumerable<T>(this IEnumerable<T> source)
    {
        foreach (var item in source)
        {
            yield return item;
        }

        await Task.CompletedTask;
    }
}