SocialPub/PrivaPub/Infrastructure/Statistics/Interactions.cs

using PrivaPub.Models.Statistics;

using System.Text.RegularExpressions;

namespace PrivaPub.Infrastructure.Statistics
{
	public static partial class Interactions
	{
		public const string Receive = "recv";
		public const string In = "in";
		public const string Out = "out";
		public const string Http = "http";
		public const string Preview = "preview";
		public const string Crawl = "crawl";

		public const string Queued = "queued";
		public const string Refused = "refused";
		public const string Accepted = "accepted";
		public const string Dropped = "dropped";
		public const string Rejected = "rejected";
		public const string Ok = "ok";
		public const string Failed = "failed";
		public const string Deferred = "deferred";
		public const string Retry = "retry";
		public const string Dead = "dead";

		public const string Public = "public";
		public const string Unlisted = "unlisted";
		public const string Private = "private";
		public const string None = "none";

		public const string Unknown = "-";
		public const string Other = "other";

		static readonly HashSet<string> Channels = new(StringComparer.Ordinal) { Receive, In, Out, Http, Preview, Crawl };
		static readonly HashSet<string> Audiences = new(StringComparer.Ordinal) { Public, Unlisted, Private, None };
		static readonly HashSet<string> LocalKinds = new(StringComparer.Ordinal) { "person", "group", "application" };
		static readonly HashSet<string> Inboxes = new(StringComparer.Ordinal) { "shared", "personal" };
		static readonly int[] LatencyBuckets = { 50, 100, 250, 500, 1000, 2500, 5000, 10000, 30000 };
		static readonly int[] WaitBuckets = { 1, 10, 60, 600, 3600, 21600, 86400 };

		public static InteractionEvent Sanitize(InteractionEvent e)
		{
			e.Host = Host(e.Host);
			e.Channel = Channels.Contains(e.Channel) ? e.Channel : Other;
			e.Activity = TypeName(e.Activity);
			e.Object = TypeName(e.Object);
			e.Purpose = Word(e.Purpose);
			e.Trigger = Word(e.Trigger);
			e.Outcome = Word(e.Outcome);
			e.Reason = Word(e.Reason);
			e.Audience = e.Audience != default && Audiences.Contains(e.Audience) ? e.Audience : default;
			e.LocalKind = e.Audience is Public or Unlisted && e.LocalKind != default && LocalKinds.Contains(e.LocalKind) ? e.LocalKind : default;
			e.Inbox = e.Inbox != default && Inboxes.Contains(e.Inbox) ? e.Inbox : default;
			e.Signature = e.Signature == default ? default : SignatureName().IsMatch(e.Signature) ? e.Signature : Other;
			e.Features = e.Features?.Select(Feature).Where(f => f != default).Distinct().Take(32).ToList();
			if (e.Features?.Count == 0)
				e.Features = default;
			return e;
		}

		public static string Host(string host)
		{
			host = host?.Trim().TrimEnd('.').ToLowerInvariant();
			return string.IsNullOrEmpty(host) || host.Length > 253 || !HostName().IsMatch(host) ? Unknown : host;
		}

		public static string HostOf(string uri) =>
			Uri.TryCreate(uri, UriKind.Absolute, out var parsed) ? Host(parsed.Host) : Unknown;

		public static string Latency(int milliseconds) => Bucket("le", milliseconds, LatencyBuckets, "ms");

		public static string Wait(int milliseconds) => Bucket("le", (int)Math.Ceiling(milliseconds / 1000.0), WaitBuckets, "s");

		static string Bucket(string prefix, int value, int[] limits, string unit)
		{
			foreach (var limit in limits)
				if (value <= limit)
					return $"{prefix}{limit}{unit}";
			return "inf";
		}

		static string TypeName(string value) =>
			value == default ? default : TypeWord().IsMatch(value) ? value : Other;

		static string Word(string value) =>
			value == default ? default : KebabWord().IsMatch(value) ? value : Other;

		static string Feature(string value) =>
			value == default ? default : FeatureName().IsMatch(value) ? value : Other;

		[GeneratedRegex("^[A-Za-z][A-Za-z0-9_-]{0,31}$")]
		private static partial Regex TypeWord();

		[GeneratedRegex("^[a-z0-9][a-z0-9-]{0,39}$")]
		private static partial Regex KebabWord();

		[GeneratedRegex("^[a-z0-9][a-z0-9:-]{0,39}$")]
		private static partial Regex SignatureName();

		[GeneratedRegex("^[a-z0-9][a-z0-9:/+.-]{0,63}$")]
		private static partial Regex FeatureName();

		[GeneratedRegex("^[a-z0-9]([a-z0-9.-]*[a-z0-9])?$")]
		private static partial Regex HostName();
	}
}