SocialPub/PrivaPub/Domain/Privacy/Counted.cs

using MongoDB.Driver;
using MongoDB.Entities;

using PrivaPub.Federation.Actors;
using PrivaPub.Models.Federation;
using PrivaPub.Models.Post;
using PrivaPub.Models.User;

using System.Linq.Expressions;

using PostEntity = PrivaPub.Models.Post.Post;

namespace PrivaPub.Domain.Privacy
{
	// What the numbers count, in one place, so a persona's statuses_count, its outbox's totalItems, NodeInfo and the
	// instance API agree (they used to count four different things).
	public static class Counted
	{
		// Mastodon's statuses_count: every post that is neither deleted nor a direct message, reblogs included. Circle and
		// located posts count too (owner decision 2026-10-04: "a good balance for the fediverse to work").
		public static readonly Expression<Func<PostEntity, bool>> Post = p => !p.DeletedAt.HasValue && p.Visibility != PostVisibility.Direct;

		// a persona's own posts, or every post in a group, its remote members' included
		public static Expression<Func<PostEntity, bool>> By(LocalActor actor)
		{
			var id = actor.Id;
			return actor.Kind == LocalActorKind.Group
				? p => p.GroupId == id && !p.DeletedAt.HasValue && p.Visibility != PostVisibility.Direct
				: p => p.GroupUserId == id && !p.IsFederatedCopy && !p.DeletedAt.HasValue && p.Visibility != PostVisibility.Direct;
		}

		// replies_count counts only public and unlisted replies, as Mastodon does: a count must not tell anyone that a
		// followers-only, direct or circle reply exists
		public static bool Reply(PostEntity reply) => reply.Visibility is PostVisibility.Public or PostVisibility.Unlisted;

		public static async Task<long> PostsBy(LocalActor actor, CancellationToken token) =>
			await DB.Default.CountAsync(By(actor), token);

		// Personas that do not count as this server's users, and are not found in search: those of a banned or deleted root.
		public static async Task<List<string>> BarredPersonas(CancellationToken token)
		{
			var roots = (await DB.Default.Find<RootUser>().Match(u => u.IsBanned || u.DeletedAt != null).Project(u => u.Include(x => x.ID)).ExecuteAsync(token))
				.Select(u => u.ID).ToList();
			return roots.Count == 0
				? new List<string>()
				: (await DB.Default.Find<RootToAvatar>().Match(ra => roots.Contains(ra.RootId)).ExecuteAsync(token)).Select(ra => ra.AvatarId).ToList();
		}

		public static async Task<long> Users(CancellationToken token)
		{
			var barred = await BarredPersonas(token);
			return await DB.Default.CountAsync<Avatar>(a => !a.DeletionAt.HasValue && !barred.Contains(a.ID), token);
		}

		// users who posted within the last days, as NodeInfo's activeMonth and activeHalfyear and the instance API mean it
		public static async Task<long> ActiveUsers(int days, CancellationToken token)
		{
			var barred = await BarredPersonas(token);
			var since = DateTime.UtcNow.AddDays(-days);
			var authors = await DB.Default.Find<PostEntity, string>()
				.Match(p => !p.IsFederatedCopy && p.CreationDate >= since && !barred.Contains(p.GroupUserId))
				.Project(p => p.GroupUserId)
				.ExecuteAsync(token);
			var distinct = authors.Where(a => !string.IsNullOrEmpty(a)).Distinct().ToList();
			return distinct.Count == 0 ? 0 : await DB.Default.CountAsync<Avatar>(a => distinct.Contains(a.ID) && !a.DeletionAt.HasValue, token);
		}

		// the media kept (not trashed) that which holds: their bytes and how many
		public static async Task<(long Bytes, long Files)> Media(Expression<Func<Models.Media.MediaAttachment, bool>> which, CancellationToken token)
		{
			var totals = await DB.Default.Fluent<Models.Media.MediaAttachment>()
				.Match(m => m.TrashedAt == null)
				.Match(which)
				.Group(m => true, g => new { Bytes = g.Sum(m => m.Size), Files = g.LongCount() })
				.FirstOrDefaultAsync(token);
			return totals == default ? (0, 0) : (totals.Bytes, totals.Files);
		}

		// a login's media: every persona's uploads and pictures
		public static async Task<(long Bytes, long Files)> MediaOfRoot(string rootId, CancellationToken token)
		{
			var personas = (await DB.Default.Find<Models.User.RootToAvatar>().Match(r => r.RootId == rootId).ExecuteAsync(token)).Select(r => r.AvatarId).ToList();
			return await Media(m => personas.Contains(m.OwnerAvatarId), token);
		}

		public static async Task<long> LocalPosts(CancellationToken token)
		{
			var barred = await BarredPersonas(token);
			return await DB.Default.CountAsync<PostEntity>(f => f.Where(p => !p.IsFederatedCopy && !barred.Contains(p.GroupUserId)) & f.Where(Post), token);
		}

		// Mastodon's instance activity, week by week (from Monday 00:00 UTC), this week first: the local posts made in it,
		// counted as LocalPosts counts them (by their ids, which carry when they were made, so the _id index serves it), and
		// the logins made in it (RootUser.CreatedAt). Never personas by their creation: their dates are spread on purpose.
		public static async Task<List<(DateTime Week, long Posts, long Registrations)>> Weekly(int weeks, CancellationToken token)
		{
			var barred = await BarredPersonas(token);
			var today = DateTime.UtcNow.Date;
			var monday = DateTime.SpecifyKind(today.AddDays(-(((int)today.DayOfWeek + 6) % 7)), DateTimeKind.Utc);
			var counted = new List<(DateTime, long, long)>();
			for (var i = 0; i < weeks; i++)
			{
				var start = monday.AddDays(-7 * i);
				var end = start.AddDays(7);
				var (low, high) = (IdAt(start), IdAt(end));
				var posts = await DB.Default.CountAsync<PostEntity>(f => f.Gte(p => p.ID, low) & f.Lt(p => p.ID, high)
					& f.Where(p => !p.IsFederatedCopy && !barred.Contains(p.GroupUserId)) & f.Where(Post), token);
				var registrations = await DB.Default.CountAsync<RootUser>(u => u.CreatedAt >= start && u.CreatedAt < end, token);
				counted.Add((start, posts, registrations));
			}
			return counted;
		}

		// the smallest ObjectId of a moment
		static string IdAt(DateTime at) => ((uint)new DateTimeOffset(at, TimeSpan.Zero).ToUnixTimeSeconds()).ToString("x8") + "0000000000000000";
	}
}