Explorer-Workbench/src/Explorer.Analysis/AnalysisService.cs

using Explorer.Domain;
using Explorer.Domain.Abstractions;

namespace Explorer.Analysis;

public sealed class AnalysisService
{
    private readonly IIndexStore _store;
    private readonly AnalysisResultCache _cache = new();
    private readonly SemaphoreSlim _ready = new(1, 1);
    private bool _readyDone;

    public AnalysisService(IIndexStore store) => _store = store;

    public Task EnsureReadyAsync(CancellationToken cancellationToken = default)
        => RunOffUiAsync(async ct =>
        {
            if (_readyDone)
            {
                return 0;
            }

            await _ready.WaitAsync(ct).ConfigureAwait(false);
            try
            {
                if (!_readyDone)
                {
                    await _store.Analysis.EnsureReadyAsync(CancellationToken.None).ConfigureAwait(false);
                    _readyDone = true;
                }
            }
            finally
            {
                _ready.Release();
            }

            return 0;
        }, cancellationToken);

    public Task<long> GetIndexStampAsync(CancellationToken cancellationToken = default)
        => RunOffUiAsync(ct => _store.Analysis.GetIndexStampAsync(ct), cancellationToken);

    public Task<IReadOnlyList<IndexEntry>> GetDirectoryRootsAsync(CancellationToken cancellationToken = default)
        => CachedAsync("roots", ct => _store.Analysis.GetDirectoryRootsAsync(ct), cancellationToken);

    public Task<IReadOnlyList<IndexEntry>> LargestDirectoriesAsync(
        long? sourceId,
        long? parentId,
        int take = AppConstants.AnalysisTopN,
        CancellationToken cancellationToken = default)
        => CachedAsync(
            $"dirs:{sourceId}:{parentId}:{take}",
            ct => _store.Analysis.LargestDirectoriesAsync(sourceId, parentId, take, ct),
            cancellationToken);

    public Task<IReadOnlyList<IndexEntry>> LargestFilesAsync(
        long? sourceId,
        string? pathRelPrefix,
        int take = AppConstants.AnalysisTopN,
        CancellationToken cancellationToken = default)
        => CachedAsync(
            $"files:{sourceId}:{pathRelPrefix}:{take}",
            ct => _store.Analysis.LargestFilesAsync(sourceId, pathRelPrefix, take, ct),
            cancellationToken);

    public Task<IReadOnlyList<ExtensionUsage>> UsageByExtensionAsync(
        long? sourceId,
        string? pathRelPrefix,
        int take = AppConstants.AnalysisTopN,
        CancellationToken cancellationToken = default)
        => CachedAsync(
            $"types:{sourceId}:{pathRelPrefix}:{take}",
            ct => _store.Analysis.UsageByExtensionAsync(sourceId, pathRelPrefix, take, ct),
            cancellationToken);

    public Task<IReadOnlyList<SourceUsage>> UsageBySourceAsync(CancellationToken cancellationToken = default)
        => CachedAsync("sources", ct => _store.Analysis.UsageBySourceAsync(ct), cancellationToken);

    public Task<IReadOnlyList<IndexEntry>> DrilldownAsync(
        long parentId,
        int take = AppConstants.AnalysisTopN,
        CancellationToken cancellationToken = default)
        => CachedAsync(
            $"children:{parentId}:{take}",
            ct => _store.Analysis.ChildrenBySizeAsync(parentId, take, ct),
            cancellationToken);

    public Task<IReadOnlyList<Source>> GetKnownSourcesAsync(CancellationToken cancellationToken = default)
        => RunOffUiAsync(ct => _store.Sources.GetAllAsync(ct), cancellationToken);

    public Task<IReadOnlyList<ClassifiedDuplicateGroup>> GetClassifiedDuplicatesAsync(
        int take = 200,
        CancellationToken cancellationToken = default)
        => RunOffUiAsync(async ct =>
        {
            var raw = await _store.Hashes.GetDuplicateGroupsAsync(null, null, Math.Max(take * 8, 400), ct)
                .ConfigureAwait(false);
            var ids = raw.SelectMany(g => g.Entries.Select(e => e.Id)).Distinct().ToList();
            var relations = await _store.Relations.GetAmongAsync(ids, ct).ConfigureAwait(false);
            return (IReadOnlyList<ClassifiedDuplicateGroup>)raw
                .Select(g => DuplicateClassifier.ClassifyGroup(g, DuplicateClassifier.RelationsFor(g.Entries, relations)))
                .ToList();
        }, cancellationToken);

    public Task MarkDuplicateGroupAsync(
        IReadOnlyList<IndexEntry> entries,
        FileRelationKind kind,
        CancellationToken cancellationToken = default)
        => RunOffUiAsync(async ct =>
        {
            var ids = entries.Select(e => e.Id).Distinct().ToList();
            await _store.Relations.DeleteAmongAsync(
                    ids,
                    [FileRelationKind.AccidentalDuplicate, FileRelationKind.IntentionalDuplicate],
                    ct)
                .ConfigureAwait(false);
            var utc = DateTimeOffset.UtcNow;
            foreach (var (left, right) in DuplicateClassifier.Pairs(entries))
            {
                await _store.Relations.UpsertAsync(new FileRelation
                {
                    LeftEntryId = left,
                    RightEntryId = right,
                    Kind = kind,
                    Origin = FileRelationOrigin.User,
                    CreatedUtc = utc
                }, ct).ConfigureAwait(false);
            }

            return 0;
        }, cancellationToken);

    private async Task<T> CachedAsync<T>(string key, Func<CancellationToken, Task<T>> query, CancellationToken cancellationToken)
    {
        await EnsureReadyAsync(cancellationToken).ConfigureAwait(false);
        var stamp = await GetIndexStampAsync(cancellationToken).ConfigureAwait(false);
        if (_cache.TryGet<T>(key, stamp, out var hit))
        {
            return hit;
        }

        var value = await RunOffUiAsync(query, cancellationToken).ConfigureAwait(false);
        cancellationToken.ThrowIfCancellationRequested();
        _cache.Set(key, stamp, value);
        return value;
    }

    public static async Task<T> RunOffUiAsync<T>(Func<CancellationToken, Task<T>> work, CancellationToken cancellationToken)
    {
        cancellationToken.ThrowIfCancellationRequested();
        if (SynchronizationContext.Current is null)
        {
            return await work(cancellationToken).ConfigureAwait(false);
        }

        return await Task.Run(async () => await work(cancellationToken).ConfigureAwait(false), cancellationToken)
            .ConfigureAwait(false);
    }
}