[backend/services] Media cleanup cron task (ISH-66, ISH-27)

Iceshrimp.Backend/Core/Configuration/Config.cs

@@ -68,14 +68,19 @@ public sealed class Config {
 	}
 
 	public sealed class StorageSection {
-		private readonly TimeSpan?         _mediaRetention;
+		public readonly TimeSpan?         MediaRetentionTimeSpan;
 		public           Enums.FileStorage Mode { get; init; } = Enums.FileStorage.Local;
 
 		public string? MediaRetention {
-			get => _mediaRetention?.ToString();
+			get => MediaRetentionTimeSpan?.ToString();
 			init {
 				if (value == null || string.IsNullOrWhiteSpace(value) || value.Trim() == "0") {
-					_mediaRetention = null;
+					MediaRetentionTimeSpan = null;
+					return;
+				}
+
+				if (value.Trim() == "-1") {
+					MediaRetentionTimeSpan = TimeSpan.MaxValue;
 					return;
 				}
 
@@ -84,7 +89,7 @@ public sealed class Config {
 
 				var suffix = value[^1];
 
-				_mediaRetention = suffix switch {
+				MediaRetentionTimeSpan = suffix switch {
 					'd' => TimeSpan.FromDays(num),
 					'w' => TimeSpan.FromDays(num * 7),
 					'm' => TimeSpan.FromDays(num * 30),
@@ -94,6 +99,9 @@ public sealed class Config {
 			}
 		}
 
+		public bool CleanAvatars = false;
+		public bool CleanBanners = false;
+
 		public LocalStorageSection?  Local         { get; init; }
 		public ObjectStorageSection? ObjectStorage { get; init; }
 	}

Iceshrimp.Backend/Core/CronTasks/MediaCleanupTask.cs

@@ -0,0 +1,44 @@
+using System.Diagnostics.CodeAnalysis;
+using Iceshrimp.Backend.Core.Configuration;
+using Iceshrimp.Backend.Core.Database;
+using Iceshrimp.Backend.Core.Queues;
+using Iceshrimp.Backend.Core.Services;
+using Microsoft.EntityFrameworkCore;
+using Microsoft.Extensions.Options;
+
+namespace Iceshrimp.Backend.Core.CronTasks;
+
+[SuppressMessage("ReSharper", "UnusedType.Global", Justification = "Instantiated at runtime by CronService")]
+public class MediaCleanupTask : ICronTask {
+	public async Task Invoke(IServiceProvider provider) {
+		var config = provider.GetRequiredService<IOptionsSnapshot<Config.StorageSection>>().Value;
+		if (config.MediaRetentionTimeSpan == TimeSpan.MaxValue) return;
+
+		var logger = provider.GetRequiredService<ILogger<MediaCleanupTask>>();
+		logger.LogInformation("Starting media cleanup task...");
+
+		var db           = provider.GetRequiredService<DatabaseContext>();
+		var queueService = provider.GetRequiredService<QueueService>();
+
+		var cutoff = DateTime.UtcNow - (config.MediaRetentionTimeSpan ?? TimeSpan.Zero);
+
+		var query = db.DriveFiles.Where(p => !p.IsLink && p.UserHost != null && p.CreatedAt < cutoff);
+
+		if (!config.CleanAvatars) query = query.Where(p => !db.Users.Any(u => u.AvatarId == p.Id));
+		if (!config.CleanBanners) query = query.Where(p => !db.Users.Any(u => u.BannerId == p.Id));
+
+		var fileIds = query.Select(p => p.Id);
+
+		logger.LogInformation("Expiring {count} files...", await fileIds.CountAsync());
+		foreach (var fileId in fileIds) {
+			await queueService.BackgroundTaskQueue.EnqueueAsync(new DriveFileDeleteJob {
+				DriveFileId = fileId,
+				Expire      = true
+			});
+		}
+	}
+
+	// Midnight
+	public TimeSpan     Trigger => TimeSpan.Zero;
+	public CronTaskType Type    => CronTaskType.Daily;
+}

Iceshrimp.Backend/Core/Extensions/ServiceExtensions.cs

@@ -52,6 +52,7 @@ public static class ServiceExtensions {
 			.AddSingleton<HttpClient>()
 			.AddSingleton<MfmConverter>()
 			.AddSingleton<HttpRequestService>()
+			.AddSingleton<CronService>()
 			.AddSingleton<QueueService>()
 			.AddSingleton<ObjectStorageService>()
 			.AddSingleton<EventService>()
@@ -62,6 +63,7 @@ public static class ServiceExtensions {
 
 		// Hosted services = long running background tasks
 		// Note: These need to be added as a singleton as well to ensure data consistency
+		services.AddHostedService<CronService>(provider => provider.GetRequiredService<CronService>());
 		services.AddHostedService<QueueService>(provider => provider.GetRequiredService<QueueService>());
 	}
 

Iceshrimp.Backend/Core/Helpers/AssemblyHelpers.cs

@@ -0,0 +1,16 @@
+using System.Reflection;
+
+namespace Iceshrimp.Backend.Core.Helpers;
+
+public static class AssemblyHelpers {
+	public static IEnumerable<Type> GetTypesWithAttribute(Type attribute, Assembly? assembly = null) {
+		assembly ??= Assembly.GetExecutingAssembly();
+		return assembly.GetTypes().Where(type => Attribute.IsDefined(type, attribute));
+	}
+
+	public static IEnumerable<Type> GetImplementationsOfInterface(Type @interface, Assembly? assembly = null) {
+		assembly ??= Assembly.GetExecutingAssembly();
+		return assembly.GetTypes().Where(type => type is { IsAbstract: false, IsClass: true } &&
+		                                         type.GetInterfaces().Contains(@interface));
+	}
+}

Iceshrimp.Backend/Core/Queues/BackgroundTaskQueue.cs

@@ -20,7 +20,10 @@ public abstract class BackgroundTaskQueue {
 		CancellationToken token
 	) {
 		if (job is DriveFileDeleteJob driveFileDeleteJob) {
-			await ProcessDriveFileDelete(driveFileDeleteJob, scope, token);
+			if (driveFileDeleteJob.Expire)
+				await ProcessDriveFileExpire(driveFileDeleteJob, scope, token);
+			else
+				await ProcessDriveFileDelete(driveFileDeleteJob, scope, token);
 		}
 	}
 
@@ -60,6 +63,53 @@ public abstract class BackgroundTaskQueue {
 			}
 		}
 	}
+
+	private static async Task ProcessDriveFileExpire(
+		DriveFileDeleteJob job,
+		IServiceProvider scope,
+		CancellationToken token
+	) {
+		var db     = scope.GetRequiredService<DatabaseContext>();
+		var logger = scope.GetRequiredService<ILogger<BackgroundTaskQueue>>();
+		logger.LogDebug("Expiring file {id}...", job.DriveFileId);
+
+		var file = await db.DriveFiles.FirstOrDefaultAsync(p => p.Id == job.DriveFileId, cancellationToken: token);
+		if (file is not { UserHost: not null, Uri: not null }) return;
+
+		file.IsLink             = true;
+		file.Url                = file.Uri;
+		file.ThumbnailUrl       = null;
+		file.WebpublicUrl       = null;
+		file.ThumbnailAccessKey = null;
+		file.WebpublicAccessKey = null;
+		file.StoredInternal     = false;
+
+		await db.Users.Where(p => p.AvatarId == file.Id)
+		        .ExecuteUpdateAsync(p => p.SetProperty(u => u.AvatarUrl, file.Uri), cancellationToken: token);
+		await db.Users.Where(p => p.BannerId == file.Id)
+		        .ExecuteUpdateAsync(p => p.SetProperty(u => u.BannerUrl, file.Uri), cancellationToken: token);
+		await db.SaveChangesAsync(token);
+
+		if (file.AccessKey == null) return;
+
+		string?[] paths = [file.AccessKey, file.ThumbnailAccessKey, file.WebpublicAccessKey];
+		if (!await db.DriveFiles.AnyAsync(p => p.Id != file.Id && p.AccessKey == file.AccessKey,
+		                                  cancellationToken: token)) {
+			if (file.StoredInternal) {
+				var pathBase = scope.GetRequiredService<IOptions<Config.StorageSection>>().Value.Local?.Path
+				               ?? throw new Exception("Cannot delete locally stored file: pathBase is null");
+
+				paths.Where(p => p != null)
+				     .Select(p => Path.Combine(pathBase, p!))
+				     .Where(File.Exists).ToList()
+				     .ForEach(File.Delete);
+			}
+			else {
+				var storageSvc = scope.GetRequiredService<ObjectStorageService>();
+				await storageSvc.RemoveFilesAsync(paths.Where(p => p != null).Select(p => p!).ToArray());
+			}
+		}
+	}
 }
 
 [ProtoContract]
@@ -69,4 +119,5 @@ public class BackgroundTaskJob : Job;
 [ProtoContract]
 public class DriveFileDeleteJob : BackgroundTaskJob {
 	[ProtoMember(1)] public required string DriveFileId;
+	[ProtoMember(2)] public required bool   Expire;
 }

Iceshrimp.Backend/Core/Services/CronService.cs

@@ -0,0 +1,97 @@
+using Iceshrimp.Backend.Core.Helpers;
+
+namespace Iceshrimp.Backend.Core.Services;
+
+public class CronService(IServiceScopeFactory serviceScopeFactory) : BackgroundService {
+	protected override Task ExecuteAsync(CancellationToken token) {
+		var tasks = AssemblyHelpers.GetImplementationsOfInterface(typeof(ICronTask))
+		                           .Select(p => Activator.CreateInstance(p) as ICronTask)
+		                           .Where(p => p != null)
+		                           .Cast<ICronTask>();
+
+		foreach (var task in tasks) {
+			ICronTrigger trigger = task.Type switch {
+				CronTaskType.Daily    => new DailyTrigger(task.Trigger, token),
+				CronTaskType.Interval => new IntervalTrigger(task.Trigger, token),
+				_                     => throw new ArgumentOutOfRangeException()
+			};
+
+			trigger.OnTrigger += async () => await task.Invoke(serviceScopeFactory.CreateScope().ServiceProvider);
+		}
+
+		return Task.CompletedTask;
+	}
+}
+
+public interface ICronTask {
+	public Task Invoke(IServiceProvider provider);
+
+	public TimeSpan     Trigger { get; }
+	public CronTaskType Type    { get; }
+}
+
+public enum CronTaskType {
+	Daily,
+	Interval
+}
+
+public interface ICronTrigger {
+	public event Action? OnTrigger;
+}
+
+file class DailyTrigger : ICronTrigger, IDisposable {
+	private TimeSpan          TriggerTime       { get; }
+	private CancellationToken CancellationToken { get; }
+	private Task              RunningTask       { get; set; }
+
+	public DailyTrigger(TimeSpan triggerTime, CancellationToken cancellationToken) {
+		TriggerTime       = triggerTime;
+		CancellationToken = cancellationToken;
+
+		RunningTask = Task.Run(async () => {
+			while (!CancellationToken.IsCancellationRequested) {
+				var nextTrigger = DateTime.Today + TriggerTime - DateTime.Now;
+				if (nextTrigger < TimeSpan.Zero)
+					nextTrigger = nextTrigger.Add(new TimeSpan(24, 0, 0));
+				await Task.Delay(nextTrigger, CancellationToken);
+				OnTrigger?.Invoke();
+			}
+		}, CancellationToken);
+	}
+
+	public void Dispose() {
+		RunningTask.Dispose();
+		RunningTask = null!;
+		GC.SuppressFinalize(this);
+	}
+
+	public event Action? OnTrigger;
+	~DailyTrigger() => Dispose();
+}
+
+file class IntervalTrigger : ICronTrigger, IDisposable {
+	private TimeSpan          TriggerInterval   { get; }
+	private CancellationToken CancellationToken { get; }
+	private Task              RunningTask       { get; set; }
+
+	public IntervalTrigger(TimeSpan triggerInterval, CancellationToken cancellationToken) {
+		TriggerInterval   = triggerInterval;
+		CancellationToken = cancellationToken;
+
+		RunningTask = Task.Run(async () => {
+			while (!CancellationToken.IsCancellationRequested) {
+				await Task.Delay(TriggerInterval, CancellationToken);
+				OnTrigger?.Invoke();
+			}
+		}, CancellationToken);
+	}
+
+	public void Dispose() {
+		RunningTask.Dispose();
+		RunningTask = null!;
+		GC.SuppressFinalize(this);
+	}
+
+	public event Action? OnTrigger;
+	~IntervalTrigger() => Dispose();
+}

Iceshrimp.Backend/Core/Services/DriveService.cs

@@ -258,7 +258,7 @@ public class DriveService(
 	}
 
 	public async Task RemoveFile(string fileId) {
-		var job = new DriveFileDeleteJob { DriveFileId = fileId };
+		var job = new DriveFileDeleteJob { DriveFileId = fileId, Expire = false };
 		await queueSvc.BackgroundTaskQueue.EnqueueAsync(job);
 	}
 

Iceshrimp.Backend/Core/Services/QueueService.cs

@@ -161,6 +161,7 @@ public class JobQueue<T>(
 [ProtoInclude(100, typeof(InboxJob))]
 [ProtoInclude(101, typeof(DeliverJob))]
 [ProtoInclude(102, typeof(PreDeliverJob))]
+[ProtoInclude(103, typeof(BackgroundTaskJob))]
 public abstract class Job {
 	public enum JobStatus {
 		Queued,

Iceshrimp.Backend/configuration.ini

@@ -58,9 +58,13 @@ Port = 6379
 ;; Options: [Local, ObjectStorage]
 Mode = Local
 
-;; Amount of time remote media is retained in the cache (0 = disabled)
+;; Amount of time remote media is retained in the cache (0 = disabled, -1 = infinite)
 MediaRetention = 30d
 
+;; Whether to cleanup avatars & banners past the media retention time
+CleanAvatars = false
+CleanBanners = false
+
 [Storage:Local]
 ;; Path where media is stored at. Must be writable for the service user.
 Path = /path/to/media/location