PLib/tests/PLib.Tests/Library/DuplicateDetectionTests.cs

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

namespace PLib.Tests.Library;

public sealed class DuplicateDetectionTests
{
    private readonly InMemoryVideoRepository _videos = new();

    [Fact]
    public void Distance_counts_the_bits_that_differ()
    {
        var left = Video("a", 0b1111);
        var right = Video("b", 0b1010);

        left.DistanceTo(right).ShouldBe(2);
    }

    [Fact]
    public void A_video_without_a_hash_has_no_distance_to_anything()
    {
        var hashed = Video("a", 1);
        var bare = new VideoItem(@"C:\videos\b.mp4", "b", 1, DateTimeOffset.UnixEpoch);

        hashed.DistanceTo(bare).ShouldBeNull();
        bare.DistanceTo(hashed).ShouldBeNull();
    }

    [Fact]
    public async Task Near_identical_videos_are_grouped_and_unrelated_ones_are_not()
    {
        _videos.Seed(
            Video("original", 0b0000_0000),
            Video("re-encode", 0b0000_0011),
            Video("unrelated", 0b1111_1111));

        var groups = await CreateService().FindDuplicatesAsync(maxDistance: 4, Token);

        groups.ShouldHaveSingleItem();
        groups[0].Select(x => x.Title).OrderBy(x => x).ShouldBe(["original", "re-encode"]);
    }

    [Fact]
    public async Task A_chain_of_near_matches_ends_up_in_one_group()
    {
        // Ends are 4 bits apart — further than the threshold — but the middle copy links
        // them, and all three are the same film.
        _videos.Seed(
            Video("a", 0b0000_0000),
            Video("b", 0b0000_0011),
            Video("c", 0b0000_1111));

        var groups = await CreateService().FindDuplicatesAsync(maxDistance: 2, Token);

        groups.ShouldHaveSingleItem();
        groups[0].Count.ShouldBe(3);
    }

    [Fact]
    public async Task Videos_that_were_never_hashed_are_left_out_entirely()
    {
        _videos.Seed(
            new VideoItem(@"C:\videos\x.mp4", "x", 1, DateTimeOffset.UnixEpoch),
            new VideoItem(@"C:\videos\y.mp4", "y", 1, DateTimeOffset.UnixEpoch));

        var groups = await CreateService().FindDuplicatesAsync(maxDistance: 64, Token);

        // Two hashless videos are not evidence of anything, however wide the threshold.
        groups.ShouldBeEmpty();
    }

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

    private static VideoItem Video(string title, ulong hash)
    {
        var item = new VideoItem($@"C:\videos\{title}.mp4", title, 1, DateTimeOffset.UnixEpoch);
        item.ApplyPerceptualHash(hash);
        return item;
    }

    private LibraryService CreateService() => new(
        _videos,
        new InMemoryLabelRepository(),
        Substitute.For<IVideoFileScanner>(),
        Substitute.For<IMediaProbe>(),
        Substitute.For<IThumbnailGenerator>(),
        Substitute.For<IAnimatedPreviewGenerator>(),
        Substitute.For<IVideoPerceptualHasher>(),
        Options.Create(new LibraryOptions()),
        NullLogger<LibraryService>.Instance);
}