SocialPub/PrivaPub/Infrastructure/Backup/TransferStore.cs

using System.Collections.Concurrent;
using System.Security.Cryptography;
using System.Text.Json;

namespace PrivaPub.Infrastructure.Backup
{
	// Backups leaving and arriving through the administrator's page, never through the client's memory:
	// - a download is a link with a ticket in its path, good for ten minutes for one backup (a browser downloads it, and
	//   resumes it with Range), given for the administrator's password;
	// - an upload arrives in pieces appended at the offset already received (a piece that would leave a gap or overlap is
	//   refused with what was received, so a broken upload resumes), into <backups>/.uploads, and is read into a backup
	//   once whole. One left for a day is deleted.
	public class TransferStore(Backups backups)
	{
		public static readonly TimeSpan TicketLifetime = TimeSpan.FromMinutes(10);
		public const long ChunkBytes = 32L * 1024 * 1024;
		const string Uploads = ".uploads";

		readonly ConcurrentDictionary<string, (string Backup, DateTime Expires)> _tickets = new();
		readonly ConcurrentDictionary<string, SemaphoreSlim> _appending = new();

		public (string Ticket, DateTime Expires) Ticket(string backup)
		{
			foreach (var (ticket, stale) in _tickets.Where(t => t.Value.Expires < DateTime.UtcNow).ToList())
				_tickets.TryRemove(ticket, out _);
			var expires = DateTime.UtcNow + TicketLifetime;
			var token = Convert.ToHexStringLower(RandomNumberGenerator.GetBytes(32));
			_tickets[token] = (backup, expires);
			return (token, expires);
		}

		/// <summary>The backup a ticket is for, while it is good.</summary>
		public string Redeem(string ticket) =>
			ticket != default && _tickets.TryGetValue(ticket, out var held) && held.Expires > DateTime.UtcNow ? held.Backup : default;

		string Folder => Path.Combine(backups.Root, Uploads);

		string File(string id) => Path.Combine(Folder, id + ".tar");

		static bool IsId(string id) => id is { Length: 32 } && id.All(char.IsAsciiHexDigitLower);

		public string Start(string by)
		{
			ServerBackup.MakeDirectory(backups.Root);
			ServerBackup.MakeDirectory(Folder);
			Forget(DateTime.UtcNow.AddDays(-1));
			var id = Convert.ToHexStringLower(RandomNumberGenerator.GetBytes(16));
			using (OperatingSystem.IsWindows()
				? new FileStream(File(id), FileMode.CreateNew, FileAccess.Write)
				: new FileStream(File(id), new FileStreamOptions { Mode = FileMode.CreateNew, Access = FileAccess.Write,
					UnixCreateMode = UnixFileMode.UserRead | UnixFileMode.UserWrite | UnixFileMode.GroupRead }))
			{
			}
			System.IO.File.WriteAllText(Path.Combine(Folder, id + ".json"), JsonSerializer.Serialize(new { by, at = DateTime.UtcNow }));
			return id;
		}

		/// <summary>How much of an upload has arrived; -1 when there is no such upload.</summary>
		public long Received(string id) => IsId(id) && System.IO.File.Exists(File(id)) ? new FileInfo(File(id)).Length : -1;

		/// <summary>A piece appended where the upload stands: what has arrived since, and whether it was taken.</summary>
		public async Task<(long Received, bool Taken)> Append(string id, long offset, Stream piece, CancellationToken token)
		{
			if (!IsId(id) || !System.IO.File.Exists(File(id)))
				return (-1, false);
			var gate = _appending.GetOrAdd(id, _ => new SemaphoreSlim(1, 1));
			await gate.WaitAsync(token);
			try
			{
				await using var file = new FileStream(File(id), FileMode.Open, FileAccess.Write);
				if (file.Length != offset)
					return (file.Length, false);
				file.Seek(offset, SeekOrigin.Begin);
				var buffer = new byte[81920];
				var total = 0L;
				int read;
				while ((read = await piece.ReadAsync(buffer, token)) > 0)
				{
					total += read;
					if (total > ChunkBytes)
					{
						file.SetLength(offset);
						return (offset, false);
					}
					await file.WriteAsync(buffer.AsMemory(0, read), token);
				}
				await file.FlushAsync(token);
				return (file.Length, true);
			}
			catch (IOException)
			{
				//the piece broke off: what arrived of it is dropped, so the next one starts where this one did
				await using var file = new FileStream(File(id), FileMode.Open, FileAccess.Write);
				file.SetLength(offset);
				return (offset, false);
			}
			finally
			{
				gate.Release();
			}
		}

		/// <summary>The whole upload read into a backup: the backup, or why not. The upload goes either way.</summary>
		public async Task<(BackupInfo Backup, string Error)> Finish(string id, CancellationToken token)
		{
			if (!IsId(id) || !System.IO.File.Exists(File(id)))
				return (default, "no such upload");
			try
			{
				await using var tar = new FileStream(File(id), FileMode.Open, FileAccess.Read, FileShare.Read, 81920, FileOptions.SequentialScan);
				return await BackupTar.Import(tar, backups.Root, token);
			}
			finally
			{
				Cancel(id);
			}
		}

		public bool Cancel(string id)
		{
			if (!IsId(id) || !System.IO.File.Exists(File(id)))
				return false;
			System.IO.File.Delete(File(id));
			System.IO.File.Delete(Path.Combine(Folder, id + ".json"));
			_appending.TryRemove(id, out _);
			return true;
		}

		// uploads left behind
		void Forget(DateTime before)
		{
			foreach (var file in Directory.EnumerateFiles(Folder).Where(f => System.IO.File.GetLastWriteTimeUtc(f) < before))
				System.IO.File.Delete(file);
		}
	}
}