SocialPub/PrivaPub/Domain/Content/ContentRenderer.cs

using AngleSharp.Dom;
using AngleSharp.Html.Parser;

using Markdig;

using PrivaPub.Federation.Actors;
using PrivaPub.Federation.Objects;
using PrivaPub.Models.Federation;

using System.Net;
using System.Text.RegularExpressions;

namespace PrivaPub.Domain.Content
{
	public sealed record ResolvedMention(string ActorUri, string UserName, string Domain, string ProfileUrl, bool IsLocal, string AccountId, string Inbox)
	{
		public string Handle => $"{UserName}@{Domain}";
	}

	public sealed record RenderedContent(string Html, IReadOnlyList<ResolvedMention> Mentions, IReadOnlyList<string> Tags);

	public interface IContentRenderer
	{
		Task<RenderedContent> Markdown(string text, CancellationToken token);
		Task<RenderedContent> PlainText(string text, CancellationToken token);
		string TagUrl(string tag);
	}

	public partial class ContentRenderer : IContentRenderer
	{
		const int MaxMentions = 20;
		static readonly MarkdownPipeline Pipeline = new MarkdownPipelineBuilder().DisableHtml().UseAutoLinks().UseEmphasisExtras().Build();
		static readonly HtmlParser Parser = new();

		readonly ILocalActorService _localActors;
		readonly IRemoteActorService _remoteActors;

		public ContentRenderer(ILocalActorService localActors, IRemoteActorService remoteActors)
		{
			_localActors = localActors;
			_remoteActors = remoteActors;
		}

		public string TagUrl(string tag) => $"{_localActors.BaseAddress}/tags/{Uri.EscapeDataString(tag)}";

		/// <summary>A text of the server's own (its description, privacy policy, terms, an announcement) from Markdown to
		/// sanitized HTML: the posts' pipeline (raw HTML off), without mentions or hashtags, so nothing is looked up.</summary>
		public static string Document(string markdown) =>
			string.IsNullOrWhiteSpace(markdown) ? string.Empty : ContentSanitizer.Html(Markdig.Markdown.ToHtml(markdown, Pipeline));

		public Task<RenderedContent> Markdown(string text, CancellationToken token) =>
			Decorate(string.IsNullOrWhiteSpace(text) ? string.Empty : Markdig.Markdown.ToHtml(text, Pipeline), token);

		public Task<RenderedContent> PlainText(string text, CancellationToken token)
		{
			if (string.IsNullOrWhiteSpace(text))
				return Decorate(string.Empty, token);
			var paragraphs = Regex.Split(text.Replace("\r\n", "\n").Trim(), "\n{2,}")
				.Select(p => "<p>" + Linkify(WebUtility.HtmlEncode(p)).Replace("\n", "<br>") + "</p>");
			return Decorate(string.Concat(paragraphs), token);
		}

		async Task<RenderedContent> Decorate(string html, CancellationToken token)
		{
			var document = Parser.ParseDocument($"<body>{html}</body>");
			var textNodes = document.Body.Descendants<IText>()
				.Where(node => !node.Ancestors<IElement>().Any(e => e.LocalName is "a" or "code" or "pre"))
				.ToList();

			var handles = textNodes.SelectMany(node => MentionPattern().Matches(node.Data).Select(m => m.Value))
				.Distinct(StringComparer.OrdinalIgnoreCase)
				.Take(MaxMentions)
				.ToList();
			var mentions = new Dictionary<string, ResolvedMention>(StringComparer.OrdinalIgnoreCase);
			foreach (var handle in handles)
			{
				var resolved = await Resolve(handle, token);
				if (resolved != default)
					mentions[handle] = resolved;
			}

			var tags = new List<string>();
			foreach (var node in textNodes)
			{
				var replaced = Replace(node.Data, mentions, tags);
				if (replaced == default)
					continue;
				var fragment = Parser.ParseFragment(replaced, node.ParentElement);
				foreach (var child in fragment.ToList())
					node.Parent.InsertBefore(child, node);
				node.Parent.RemoveChild(node);
			}

			return new RenderedContent(
				ContentSanitizer.Html(document.Body.InnerHtml),
				mentions.Values.DistinctBy(m => m.ActorUri).ToList(),
				tags.Distinct().ToList());
		}

		string Replace(string text, IReadOnlyDictionary<string, ResolvedMention> mentions, List<string> tags)
		{
			var changed = false;
			var encoded = WebUtility.HtmlEncode(text);
			encoded = MentionPattern().Replace(encoded, match =>
			{
				if (!mentions.TryGetValue(WebUtility.HtmlDecode(match.Value), out var mention))
					return match.Value;
				changed = true;
				return $"<span class=\"h-card\" translate=\"no\"><a href=\"{WebUtility.HtmlEncode(mention.ProfileUrl)}\" class=\"u-url mention\">@<span>{WebUtility.HtmlEncode(mention.UserName)}</span></a></span>";
			});
			encoded = HashtagPattern().Replace(encoded, match =>
			{
				var tag = match.Groups["tag"].Value;
				if (tag.All(char.IsDigit))
					return match.Value;
				changed = true;
				tags.Add(tag.ToLowerInvariant());
				return $"{match.Groups["lead"].Value}<a href=\"{WebUtility.HtmlEncode(TagUrl(tag.ToLowerInvariant()))}\" class=\"mention hashtag\" rel=\"tag\">#<span>{tag}</span></a>";
			});
			return changed ? encoded : default;
		}

		async Task<ResolvedMention> Resolve(string handle, CancellationToken token)
		{
			var parts = handle.TrimStart('@').Split('@');
			var localDomain = new Uri(_localActors.BaseAddress).Authority;
			if (parts.Length == 1 || parts[1].Equals(localDomain, StringComparison.OrdinalIgnoreCase))
			{
				var local = await _localActors.FindByUserName(parts[0], token);
				return local is { IsFederated: true, IsCircle: false, IsServerActor: false }
					? new ResolvedMention(local.Uri, local.UserName, local.Domain, local.Uri, true, local.Id, local.Inbox)
					: default;
			}

			var actorUri = await _remoteActors.ResolveHandle($"{parts[0]}@{parts[1]}", token);
			var actor = actorUri == default ? default : await _remoteActors.GetActor(actorUri, refresh: false, token);
			return actor == default
				? default
				: new ResolvedMention(actor.ActorURI, actor.UserName ?? parts[0], parts[1].ToLowerInvariant(), actor.Url ?? actor.ActorURI, false, actor.ID,
					actor.InboxURL);
		}

		static string Linkify(string encoded) =>
			UrlPattern().Replace(encoded, match => $"<a href=\"{match.Value}\">{match.Value}</a>");

		[GeneratedRegex(@"(?<![\w@/.])@[a-zA-Z0-9_]+(?:@[a-zA-Z0-9-]+(?:\.[a-zA-Z0-9-]+)+(?::\d+)?)?(?![\w@])")]
		private static partial Regex MentionPattern();

		[GeneratedRegex(@"(?<lead>^|[^\w&#/])#(?<tag>[\p{L}\p{N}_]{1,100})(?![\p{L}\p{N}_])")]
		private static partial Regex HashtagPattern();

		[GeneratedRegex(@"https?://[^\s<]+[^\s<.,;:!?)\]'""]")]
		private static partial Regex UrlPattern();
	}
}