Explorer-Workbench/tests/Explorer.Analysis.Tests/AnalysisTests.cs

using Explorer.Analysis;
using Explorer.Application;
using Explorer.Domain;
using Explorer.Presentation.ViewModels;
using Explorer.Storage.Sqlite;
using Microsoft.Extensions.Logging.Abstractions;

namespace Explorer.Analysis.Tests;

public class AnalysisTests
{
    [Fact]
    public async Task Largest_dirs_files_and_types()
    {
        var db = Path.Combine(Path.GetTempPath(), "ew-an", Guid.NewGuid().ToString("N"), "index.db");
        await using var store = new SqliteIndexStore(db, NullLogger<SqliteIndexStore>.Instance);
        await store.OpenAsync();
        var source = new Source { StableKey = "k", DisplayName = "Media", Kind = SourceKind.NtfsLocal, LastRootPath = @"C:\m", Status = SourceStatus.Online };
        source.Id = await store.Sources.UpsertAsync(source);
        var root = new IndexEntry { SourceId = source.Id, Name = "m", NameNorm = "m", IsDirectory = true, PathRel = "", LastSeenUtc = DateTimeOffset.UtcNow, AggregateSize = 300 };
        root.Id = await store.Entries.UpsertAsync(root);
        var movies = new IndexEntry { SourceId = source.Id, ParentId = root.Id, Name = "Movies", NameNorm = "movies", IsDirectory = true, PathRel = "Movies", LastSeenUtc = DateTimeOffset.UtcNow, AggregateSize = 200 };
        movies.Id = await store.Entries.UpsertAsync(movies);
        await store.Entries.UpsertAsync(new IndexEntry
        {
            SourceId = source.Id, ParentId = movies.Id, Name = "a.mkv", NameNorm = "a.mkv", Extension = "mkv",
            SizeBytes = 200, PathRel = @"Movies\a.mkv", LastSeenUtc = DateTimeOffset.UtcNow
        });
        await store.Entries.UpsertAsync(new IndexEntry
        {
            SourceId = source.Id, ParentId = root.Id, Name = "b.txt", NameNorm = "b.txt", Extension = "txt",
            SizeBytes = 100, PathRel = "b.txt", LastSeenUtc = DateTimeOffset.UtcNow
        });

        var analysis = new AnalysisService(store);
        var dirs = await analysis.LargestDirectoriesAsync(source.Id, root.Id);
        Assert.Equal("Movies", dirs[0].Name);
        var files = await analysis.LargestFilesAsync(source.Id, null);
        Assert.Equal("a.mkv", files[0].Name);
        var types = await analysis.UsageByExtensionAsync(source.Id, null);
        Assert.Contains(types, t => t.Extension == "mkv" && t.TotalSize == 200);
        var drill = await analysis.DrilldownAsync(root.Id);
        Assert.Contains(drill, e => e.Name == "Movies");
        Assert.Equal("Movies", dirs[0].PathRel);
        var global = await analysis.LargestDirectoriesAsync(source.Id, null);
        Assert.Contains(global, d => d.PathRel == "Movies");
        Assert.DoesNotContain(global, d => d.ParentId is null);

        var first = await analysis.GetIndexStampAsync();
        await store.Sources.UpdateIndexedAsync(source.Id, DateTimeOffset.UtcNow, source.ScanGeneration + 1);
        var second = await analysis.GetIndexStampAsync();
        Assert.NotEqual(first, second);
    }

    [Fact]
    public void Sibling_size_bars_scale_to_the_largest_child()
    {
        var siblings = new[]
        {
            new StorageNodeViewModel { Name = "A", FullPath = @"C:\A", PathRel = "A", SourceId = 1, Size = 80 },
            new StorageNodeViewModel { Name = "B", FullPath = @"C:\B", PathRel = "B", SourceId = 1, Size = 40 },
            new StorageNodeViewModel { Name = "C", FullPath = @"C:\C", PathRel = "C", SourceId = 1, Size = 0 }
        };

        AnalysisViewModel.ApplySiblingFractions(siblings);

        Assert.Equal(1, siblings[0].Fraction);
        Assert.Equal(0.5, siblings[1].Fraction);
        Assert.Equal(0, siblings[2].Fraction);
    }

    [Fact]
    public async Task Hash_worker_skips_online_only_without_opening()
    {
        var db = Path.Combine(Path.GetTempPath(), "ew-hash", Guid.NewGuid().ToString("N"), "index.db");
        var rootPath = Path.Combine(Path.GetTempPath(), "ew-hash-fs", Guid.NewGuid().ToString("N"));
        Directory.CreateDirectory(rootPath);
        File.WriteAllBytes(Path.Combine(rootPath, "local.bin"), new byte[64]);
        var missingOnline = Path.Combine(rootPath, "online-only.bin");

        await using var store = new SqliteIndexStore(db, NullLogger<SqliteIndexStore>.Instance);
        await store.OpenAsync();
        var source = new Source
        {
            StableKey = "h",
            DisplayName = "H",
            Kind = SourceKind.NtfsLocal,
            LastRootPath = rootPath,
            Status = SourceStatus.Online
        };
        source.Id = await store.Sources.UpsertAsync(source);
        var root = new IndexEntry
        {
            SourceId = source.Id,
            Name = "H",
            NameNorm = "h",
            IsDirectory = true,
            PathRel = "",
            LastSeenUtc = DateTimeOffset.UtcNow
        };
        root.Id = await store.Entries.UpsertAsync(root);
        await store.Entries.UpsertAsync(new IndexEntry
        {
            SourceId = source.Id,
            ParentId = root.Id,
            Name = "local.bin",
            NameNorm = "local.bin",
            SizeBytes = 64,
            PathRel = "local.bin",
            LastSeenUtc = DateTimeOffset.UtcNow
        });
        var online = new IndexEntry
        {
            SourceId = source.Id,
            ParentId = root.Id,
            Name = "online-only.bin",
            NameNorm = "online-only.bin",
            SizeBytes = 64,
            PathRel = "online-only.bin",
            LastSeenUtc = DateTimeOffset.UtcNow,
            Attributes = AttributeFlags.RecallOnDataAccess,
            CloudAvailability = CloudAvailability.OnlineOnly
        };
        online.Id = await store.Entries.UpsertAsync(online);

        await store.Hashes.EnqueueSizeCollisionsAsync(source.Id);
        var worker = new DuplicateHashWorker(
            store,
            new HydrationGuard(new StorageProviderRegistry([], NullLogger<StorageProviderRegistry>.Instance)),
            NullLogger<DuplicateHashWorker>.Instance);
        await worker.ProcessPendingAsync(CancellationToken.None);

        var skipped = await store.Entries.GetByPathAsync(source.Id, "online-only.bin");
        Assert.Equal(HashState.Skipped, skipped!.HashState);
        Assert.False(File.Exists(missingOnline));
        var local = await store.Entries.GetByPathAsync(source.Id, "local.bin");
        Assert.Equal(HashState.Partial, local!.HashState);
    }

    [Fact]
    public async Task Hardlinks_are_grouped_but_excluded_from_default_duplicate_list()
    {
        var (store, analysis, source) = await SeedHashedAsync(fileIdA: 42, fileIdB: 42);
        await using (store)
        {
            var raw = await store.Hashes.GetDuplicateGroupsAsync(null, null, 10);
            var group = Assert.Single(raw);
            Assert.True(group.SameFileId);

            var classified = await analysis.GetClassifiedDuplicatesAsync();
            var item = Assert.Single(classified);
            Assert.Equal(DuplicateClass.Hardlink, item.Classification);
            Assert.True(DuplicateClassifier.IsHiddenByDefault(item.Classification));
            Assert.Equal(source.Id, item.Group.Entries[0].SourceId);
        }
    }

    [Fact]
    public async Task Marked_intentional_relation_is_excluded_from_default_list()
    {
        var (store, analysis, _) = await SeedHashedAsync(fileIdA: 1, fileIdB: 2);
        await using (store)
        {
            var before = Assert.Single(await analysis.GetClassifiedDuplicatesAsync());
            Assert.Equal(DuplicateClass.Unknown, before.Classification);
            Assert.False(DuplicateClassifier.IsHiddenByDefault(before.Classification));

            await analysis.MarkDuplicateGroupAsync(before.Group.Entries, FileRelationKind.IntentionalDuplicate);
            var after = Assert.Single(await analysis.GetClassifiedDuplicatesAsync());
            Assert.Equal(DuplicateClass.Intentional, after.Classification);
            Assert.True(DuplicateClassifier.IsHiddenByDefault(after.Classification));

            await analysis.MarkDuplicateGroupAsync(after.Group.Entries, FileRelationKind.AccidentalDuplicate);
            var accidental = Assert.Single(await analysis.GetClassifiedDuplicatesAsync());
            Assert.Equal(DuplicateClass.Accidental, accidental.Classification);
            Assert.False(DuplicateClassifier.IsHiddenByDefault(accidental.Classification));
        }
    }

    private static async Task<(SqliteIndexStore Store, AnalysisService Analysis, Source Source)> SeedHashedAsync(
        long fileIdA, long fileIdB)
    {
        var db = Path.Combine(Path.GetTempPath(), "ew-dup", Guid.NewGuid().ToString("N"), "index.db");
        var store = new SqliteIndexStore(db, NullLogger<SqliteIndexStore>.Instance);
        await store.OpenAsync();
        var source = new Source
        {
            StableKey = "d",
            DisplayName = "D",
            Kind = SourceKind.NtfsLocal,
            LastRootPath = @"C:\d",
            Status = SourceStatus.Online
        };
        source.Id = await store.Sources.UpsertAsync(source);
        var hash = Enumerable.Repeat((byte)7, 32).ToArray();
        var root = new IndexEntry
        {
            SourceId = source.Id,
            Name = "d",
            NameNorm = "d",
            IsDirectory = true,
            PathRel = "",
            LastSeenUtc = DateTimeOffset.UtcNow
        };
        root.Id = await store.Entries.UpsertAsync(root);
        await store.Entries.UpsertAsync(Hashed("a.bin", fileIdA, source.Id, root.Id, hash));
        await store.Entries.UpsertAsync(Hashed("b.bin", fileIdB, source.Id, root.Id, hash));
        return (store, new AnalysisService(store), source);
    }

    private static IndexEntry Hashed(string name, long fileId, long sourceId, long parentId, byte[] hash)
        => new()
        {
            SourceId = sourceId,
            ParentId = parentId,
            Name = name,
            NameNorm = name,
            SizeBytes = 64,
            PathRel = name,
            FileId = fileId,
            ContentHash = hash,
            HashState = HashState.Full,
            LastSeenUtc = DateTimeOffset.UtcNow
        };
}