PLib/tests/PLib.Tests/Library/MetadataScanTests.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;

/// <summary>
/// The library-wide run. Its job is not to find metadata — that is the provider's — but to
/// decide what to ask about, what to write without asking, and when to stop asking a source.
/// </summary>
public sealed class MetadataScanTests
{
    private readonly InMemoryVideoRepository _videos = new();
    private readonly InMemoryLabelRepository _labels = new();
    private readonly IMetadataProvider _provider = Substitute.For<IMetadataProvider>();

    private readonly MetadataSourceOptions _source =
        new() { Name = "StashDB", Endpoint = "https://stashdb.org/graphql" };

    [Fact]
    public async Task Only_fingerprinted_videos_are_asked_about()
    {
        var hashed = Video("с отпечатком", hash: 1);
        _videos.Seed(hashed, Video("без отпечатка", hash: null));

        Answer(_source, []);

        var events = await CollectAsync();

        // Without a fingerprint there is no question to ask, so the video is not counted as
        // processed either — a total that included it would never reach itself.
        events.OfType<MetadataScanEvent.Completed>().Single().Processed.ShouldBe(1);
        await _provider.Received(1).FindByPerceptualHashAsync(_source, 1, Arg.Any<CancellationToken>());
    }

    [Fact]
    public async Task Videos_that_already_have_a_description_are_skipped_by_default()
    {
        var described = Video("описанное", hash: 1);
        described.Describe("уже есть");
        _videos.Seed(described, Video("голое", hash: 2));

        Answer(_source, []);

        (await CollectAsync()).OfType<MetadataScanEvent.Completed>().Single().Processed.ShouldBe(1);

        // Asking again for everything is what the switch is for.
        var all = await CollectAsync(new MetadataScanRequest(OnlyWithoutDescription: false));
        all.OfType<MetadataScanEvent.Completed>().Single().Processed.ShouldBe(2);
    }

    [Fact]
    public async Task A_single_candidate_is_written_only_when_the_run_was_told_it_may()
    {
        var video = Video("видео", hash: 1);
        _videos.Seed(video);

        Answer(_source, [Match("Название")]);

        await CollectAsync();
        video.Description.ShouldBeNull();

        var applied = await CollectAsync(new MetadataScanRequest(ApplyUnambiguous: true));

        video.Title.ShouldBe("Название");
        applied.OfType<MetadataScanEvent.Completed>().Single().Applied.ShouldBe(1);
        applied.OfType<MetadataScanEvent.Matched>().Single().Applied.ShouldBeTrue();
    }

    [Fact]
    public async Task Disagreeing_sources_are_never_written_without_asking()
    {
        var video = Video("видео", hash: 1);
        _videos.Seed(video);

        var second = new MetadataSourceOptions { Name = "Другой", Endpoint = "https://other/graphql" };
        Answer(_source, [Match("Одно")]);
        Answer(second, [Match("Другое")]);

        var events = await CollectAsync(new MetadataScanRequest(ApplyUnambiguous: true), _source, second);

        // Two candidates is precisely the case a human has to resolve.
        video.Title.ShouldBe("видео");
        events.OfType<MetadataScanEvent.Matched>().Single().Applied.ShouldBeFalse();
        events.OfType<MetadataScanEvent.Completed>().Single().Applied.ShouldBe(0);
    }

    [Fact]
    public async Task A_failing_source_is_reported_once_and_then_left_out_of_the_run()
    {
        _videos.Seed(Video("первое", hash: 1), Video("второе", hash: 2), Video("третье", hash: 3));

        _provider.FindByPerceptualHashAsync(_source, Arg.Any<ulong>(), Arg.Any<CancellationToken>())
            .Returns<IReadOnlyList<VideoMetadataMatch>>(_ => throw new HttpRequestException("401"));

        var events = await CollectAsync();

        // A rejected key fails on every video; three identical lines would bury the results,
        // and three hundred would be the whole page.
        events.OfType<MetadataScanEvent.SourceAbandoned>().ShouldHaveSingleItem();
        await _provider.Received(1).FindByPerceptualHashAsync(
            _source,
            Arg.Any<ulong>(),
            Arg.Any<CancellationToken>());

        // The run still finishes, having gone through every video and found nothing.
        events.OfType<MetadataScanEvent.Completed>().Single().Processed.ShouldBe(3);
    }

    private void Answer(MetadataSourceOptions source, IReadOnlyList<VideoMetadataMatch> matches) =>
        _provider.FindByPerceptualHashAsync(source, Arg.Any<ulong>(), Arg.Any<CancellationToken>())
            .Returns(matches);

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

    private static VideoItem Video(string title, ulong? hash)
    {
        var item = new VideoItem($@"C:\videos\{title}.mp4", title, 1_000, DateTimeOffset.UnixEpoch);
        item.ApplyTechnicalInfo(new VideoTechnicalInfo(TimeSpan.FromMinutes(1), 1280, 720, "h264"));
        item.ApplyPerceptualHash(hash);
        return item;
    }

    private async Task<List<MetadataScanEvent>> CollectAsync(
        MetadataScanRequest? request = null,
        params MetadataSourceOptions[] sources)
    {
        var service = new LibraryService(
            _videos,
            _labels,
            Substitute.For<IVideoFileScanner>(),
            Substitute.For<IMediaProbe>(),
            Substitute.For<IThumbnailGenerator>(),
            Substitute.For<IAnimatedPreviewGenerator>(),
            Substitute.For<IVideoPerceptualHasher>(),
            _provider,
            Options.Create(new LibraryOptions()),

            // No pause between requests: the delay exists to be kind to somebody else's
            // server, and there isn't one here.
            MetadataMonitor.With(sources.Length == 0 ? [_source] : sources, delayMilliseconds: 0),
            NullLogger<LibraryService>.Instance);

        var events = new List<MetadataScanEvent>();

        await foreach (var scanEvent in service.ScanMetadataAsync(
            request ?? new MetadataScanRequest(),
            TestContext.Current.CancellationToken))
        {
            events.Add(scanEvent);
        }

        return events;
    }
}