decePubClient/Services/PortabilityService.cs

using collAnon.Client.Services;
using System.Text;

using decePubClient.Extensions;
using decePubClient.Models;
using decePubClient.Models.Mastodon;
using decePubClient.Models.Types;

using PrivaPub.ClientModels;

namespace decePubClient.Services
{
	/// <summary>What a persona carries from one server to another, each in the CSV file Mastodon exports and imports.</summary>
	public enum PortableData
	{
		Follows = 1,
		Lists,
		Blocks,
		Mutes,
		ServerBlocks,
		Bookmarks
	}

	/// <summary>How an import went: how many rows were done, and the rows that could not be, with why.</summary>
	public sealed class ImportReport
	{
		public int Total { get; set; }
		public int Done { get; set; }
		public List<string> Failed { get; set; } = [];
	}

	/// <summary>
	/// The persona's follows, lists, blocks, mutes, blocked servers and bookmarks as Mastodon's CSV exports, written in
	/// the browser from the Mastodon API, and the same files imported: each row resolved through PrivaPub (a remote
	/// account or post is fetched by PrivaPub, never by the browser) and acted on one at a time. Nothing throws; an import
	/// reports the rows it could not do.
	/// </summary>
	public class PortabilityService(AccountService accounts, ListService lists, StatusService statuses, TimelineService timelines,
		InstanceService instances, CoalescingStringLocalizer localizer)
	{
		/// <summary>The most rows read from an export or an import file.</summary>
		public const int MaxRows = 5000;

		const string AccountHeader = "Account address";

		public static string FileName(PortableData data) => data switch
		{
			PortableData.Follows => "following_accounts.csv",
			PortableData.Lists => "lists.csv",
			PortableData.Blocks => "blocked_accounts.csv",
			PortableData.Mutes => "muted_accounts.csv",
			PortableData.ServerBlocks => "blocked_domains.csv",
			_ => "bookmarks.csv"
		};

		/// <summary>The file Mastodon would export, as UTF-8 bytes.</summary>
		// a page that could not be read stops the export, which says so, instead of a file silently missing the rest
		sealed class Stop
		{
			public WebResult Failed { get; set; }
		}

		/// <summary>The file, as Data (bytes); invalid when a page of it could not be read.</summary>
		public async Task<WebResult> Export(PortableData data, string personaId, CancellationToken cancellationToken = default)
		{
			var stop = new Stop();
			var rows = data switch
			{
				PortableData.Follows => await Follows(personaId, stop, cancellationToken),
				PortableData.Lists => await Lists(cancellationToken),
				PortableData.Blocks => [.. (await Pages(accounts.Blocks, stop, cancellationToken)).Select(a => Csv(Full(a.Acct)))],
				PortableData.Mutes => await Mutes(stop, cancellationToken),
				PortableData.ServerBlocks => [.. (await accounts.DomainBlocks(cancellationToken)).Select(d => Csv(d))],
				_ => await Bookmarks(personaId, stop, cancellationToken)
			};
			if (stop.Failed is { } failed)
				return failed;
			return new WebResult { Data = Encoding.UTF8.GetBytes(string.Concat(rows.Select(r => r + "\n"))) };
		}

		async Task<List<string>> Follows(string personaId, Stop stop, CancellationToken cancellationToken)
		{
			var followed = await Pages((maxId, token) => accounts.Following(personaId, maxId, token), stop, cancellationToken);
			var relationships = (await accounts.Relationships([.. followed.Select(a => a.Id)], cancellationToken)).ToDictionary(r => r.Id);
			return
			[
				Csv(AccountHeader, "Show boosts", "Notify on new posts", "Languages"),
				.. followed.Select(a => relationships.GetValueOrDefault(a.Id) is { } r
					? Csv(Full(a.Acct), Flag(r.ShowingReblogs), Flag(r.Notifying), null)
					: Csv(Full(a.Acct), "true", "false", null))
			];
		}

		async Task<List<string>> Mutes(Stop stop, CancellationToken cancellationToken)
		{
			var muted = await Pages(accounts.Mutes, stop, cancellationToken);
			var relationships = (await accounts.Relationships([.. muted.Select(a => a.Id)], cancellationToken)).ToDictionary(r => r.Id);
			return
			[
				Csv(AccountHeader, "Hide notifications"),
				.. muted.Select(a => Csv(Full(a.Acct), Flag(relationships.GetValueOrDefault(a.Id)?.MutingNotifications ?? true)))
			];
		}

		async Task<List<string>> Lists(CancellationToken cancellationToken)
		{
			var rows = new List<string>();
			foreach (var list in await lists.All(cancellationToken))
				rows.AddRange((await lists.Members(list.Id, cancellationToken)).Select(member => Csv(list.Title, Full(member.Acct))));
			return rows;
		}

		async Task<List<string>> Bookmarks(string personaId, Stop stop, CancellationToken cancellationToken)
		{
			var rows = new List<string>();
			var maxId = default(string);
			do
			{
				var result = await timelines.Load(new FeedQuery(TimelineType.Bookmarks, personaId), maxId, cancellationToken: cancellationToken);
				if (result.IsInvalid || result.Data is not FeedPage page)
				{
					stop.Failed = result.IsInvalid ? result : new WebResult().Invalidate(localizer["The server sent an invalid response."]);
					return rows;
				}
				rows.AddRange(page.Posts.Select(p => Csv(p.Status.Uri)));
				maxId = page.Posts.Count == 0 ? null : page.MaxId;
			}
			while (maxId is not null && rows.Count < MaxRows);
			return rows;
		}

		// every page of an account list, up to MaxRows
		async Task<List<Account>> Pages(Func<string, CancellationToken, Task<WebResult>> page, Stop stop, CancellationToken cancellationToken)
		{
			var all = new List<Account>();
			var maxId = default(string);
			do
			{
				var result = await page(maxId, cancellationToken);
				if (result.IsInvalid || result.Data is not AccountPage accountPage)
				{
					stop.Failed = result.IsInvalid ? result : new WebResult().Invalidate(localizer["The server sent an invalid response."]);
					return all;
				}
				all.AddRange(accountPage.Accounts);
				maxId = accountPage.MaxId;
			}
			while (maxId is not null && all.Count < MaxRows);
			return all;
		}

		/// <summary>Does what each row of an exported file says, one row at a time; progress counts the rows done.</summary>
		public async Task<ImportReport> Import(PortableData data, string csv, IProgress<ImportReport> progress = default,
			CancellationToken cancellationToken = default)
		{
			var rows = Parse(csv.TrimStart('\uFEFF')).Where(r => r.Count > 0 && !string.IsNullOrWhiteSpace(r[0])).ToList();
			if (rows.Count > 0 && rows[0][0].Equals(AccountHeader, StringComparison.OrdinalIgnoreCase))
				rows.RemoveAt(0);
			var report = new ImportReport { Total = Math.Min(rows.Count, MaxRows) };
			var known = data == PortableData.Lists ? await lists.All(cancellationToken) : [];
			foreach (var row in rows.Take(MaxRows))
			{
				if (cancellationToken.IsCancellationRequested)
					break;
				var failure = data switch
				{
					PortableData.Follows => await FollowRow(row, cancellationToken),
					PortableData.Lists => await ListRow(row, known, cancellationToken),
					PortableData.Blocks => await AccountRow(row[0], AccountAction.Block, cancellationToken),
					PortableData.Mutes => await AccountRow(row[0], AccountAction.Mute, cancellationToken),
					PortableData.ServerBlocks => Failure(row[0], await accounts.BlockDomain(row[0], cancellationToken)),
					_ => await BookmarkRow(row[0].Trim(), cancellationToken)
				};
				if (failure is null)
					report.Done++;
				else
					report.Failed.Add(failure);
				progress?.Report(report);
			}
			return report;
		}

		async Task<string> FollowRow(List<string> row, CancellationToken cancellationToken)
		{
			if (await accounts.Lookup(row[0], cancellationToken) is not { } account)
				return Unknown(row[0]);
			var showReblogs = row.Count < 2 || !row[1].Trim().Equals("false", StringComparison.OrdinalIgnoreCase);
			return Failure(row[0], await accounts.Follow(account.Id, showReblogs, cancellationToken));
		}

		async Task<string> AccountRow(string acct, AccountAction action, CancellationToken cancellationToken)
		{
			if (await accounts.Lookup(acct, cancellationToken) is not { } account)
				return Unknown(acct);
			return Failure(acct, await accounts.Act(account.Id, action, cancellationToken));
		}

		// a list row names the list and an account; the list is made when missing, and an account is followed first, since
		// a list holds followed accounts only
		async Task<string> ListRow(List<string> row, List<MastodonList> known, CancellationToken cancellationToken)
		{
			if (row.Count < 2 || string.IsNullOrWhiteSpace(row[1]))
				return string.Join(',', row);
			var title = row[0].Trim();
			var list = known.FirstOrDefault(l => string.Equals(l.Title, title, StringComparison.Ordinal));
			if (list is null)
			{
				var made = await lists.Save(new MastodonList { Title = title }, cancellationToken);
				if (made.IsInvalid || made.Data is not MastodonList created)
					return Failure(title, made) ?? title;
				known.Add(list = created);
			}
			if (await accounts.Lookup(row[1], cancellationToken) is not { } account)
				return Unknown(row[1]);
			var relationship = await accounts.Relationship(account.Id, cancellationToken);
			if (relationship is not { Following: true })
			{
				var followed = await accounts.Follow(account.Id, true, cancellationToken);
				if (followed.IsInvalid)
					return Failure(row[1], followed);
			}
			return Failure($"{title}: {row[1]}", await lists.SetMember(list.Id, account.Id, true, cancellationToken));
		}

		async Task<string> BookmarkRow(string uri, CancellationToken cancellationToken)
		{
			var found = await accounts.Search(uri, resolve: true, cancellationToken);
			if ((found.Statuses.FirstOrDefault(s => s.Uri == uri || s.Url == uri) ?? found.Statuses.FirstOrDefault()) is not { } status)
				return Unknown(uri);
			return Failure(uri, await statuses.Act(status.Id, StatusAction.Bookmark, cancellationToken));
		}

		static string Unknown(string what) => $"{what}: ?";

		static string Failure(string what, WebResult result) => result.IsInvalid ? $"{what}: {result.ErrorMessage}" : null;

		// a local account's acct has no host; an export names every account in full
		string Full(string acct) => acct.Contains('@') ? acct : $"{acct}@{instances.SelfHost}";

		static string Flag(bool value) => value ? "true" : "false";

		static string Csv(params string[] cells) => string.Join(',', cells.Select(c => c is null ? "" :
			c.IndexOfAny([',', '"', '\n', '\r']) < 0 ? c : $"\"{c.Replace("\"", "\"\"")}\""));

		// RFC 4180: quoted cells may hold commas, quotes (doubled) and line breaks
		static IEnumerable<List<string>> Parse(string csv)
		{
			var row = new List<string>();
			var cell = new StringBuilder();
			var quoted = false;
			for (var i = 0; i < csv.Length; i++)
			{
				var c = csv[i];
				if (quoted)
				{
					if (c == '"' && i + 1 < csv.Length && csv[i + 1] == '"')
					{
						cell.Append('"');
						i++;
					}
					else if (c == '"')
						quoted = false;
					else
						cell.Append(c);
					continue;
				}
				switch (c)
				{
					case '"':
						quoted = true;
						break;
					case ',':
						row.Add(cell.ToString());
						cell.Clear();
						break;
					case '\r':
						break;
					case '\n':
						row.Add(cell.ToString());
						cell.Clear();
						yield return row;
						row = [];
						break;
					default:
						cell.Append(c);
						break;
				}
			}
			if (cell.Length > 0 || row.Count > 0)
			{
				row.Add(cell.ToString());
				yield return row;
			}
		}
	}
}