SocialPub/PrivaPub/Federation/Objects/NoteParser.cs

using PrivaPub.Models.Post;

using System.Globalization;
using System.Text.Json.Nodes;

using static PrivaPub.Federation.Objects.ActivityJson;

namespace PrivaPub.Federation.Objects
{
	public sealed class NoteDocument
	{
		public string Id { get; init; }
		public string Type { get; init; }
		public JsonObject Raw { get; init; }
		public string AttributedTo { get; init; }
		public string ContentHtml { get; init; }
		public string Title { get; init; }
		public string SpoilerText { get; init; }
		public string Excerpt { get; init; }
		public bool Sensitive { get; init; }
		public string Language { get; init; }
		public string Url { get; init; }
		public string InReplyTo { get; init; }
		public string Context { get; init; }
		public string Audience { get; init; }
		public string QuoteUri { get; init; }
		public DateTime Published { get; init; }
		public DateTime? Updated { get; init; }
		public IReadOnlyList<string> To { get; init; } = Array.Empty<string>();
		public IReadOnlyList<string> Cc { get; init; } = Array.Empty<string>();
		public IReadOnlyList<PostMention> Mentions { get; init; } = Array.Empty<PostMention>();
		public IReadOnlyList<string> Tags { get; init; } = Array.Empty<string>();
		public IReadOnlyList<PostMedia> Attachments { get; init; } = Array.Empty<PostMedia>();
	}

	public static class NoteParser
	{
		static readonly string[] NoteTypes = { "Note", "Article", "Page", "Question", "Event", "Video", "Audio", "Image" };
		const int MaxAttachments = 16;
		const int MaxTags = 64;

		public static NoteDocument Parse(JsonNode node)
		{
			if (node is not JsonObject note)
				return default;
			var type = Value(note, "type");
			var id = Id(note);
			if (!NoteTypes.Contains(type) || Origin.Of(id) == default)
				return default;

			var (contentHtml, language) = Content(note);
			var summary = Value(note, "summary");
			var sensitive = note["sensitive"] is JsonValue flag && flag.TryGetValue<bool>(out var marked) && marked;
			var summaryWarns = type is "Note" or "Question" || sensitive;
			return new NoteDocument
			{
				Id = id,
				Type = type,
				Raw = note,
				AttributedTo = Attribution(note["attributedTo"]),
				ContentHtml = ContentSanitizer.Html(contentHtml),
				Title = Plain(Value(note, "name"), 500),
				SpoilerText = summaryWarns ? Plain(summary, 500) : default,
				Excerpt = summaryWarns ? default : Plain(summary, 1000),
				Sensitive = sensitive || summaryWarns && !string.IsNullOrWhiteSpace(summary),
				Language = language,
				Url = UrlOf(note["url"]) ?? id,
				InReplyTo = Id(note["inReplyTo"]),
				Context = Id(note["context"]) ?? Value(note, "conversation"),
				Audience = Id(note["audience"]),
				QuoteUri = Value(note, "quote") ?? Value(note, "quoteUrl") ?? Value(note, "quoteUri") ?? Value(note, "_misskey_quote"),
				Published = Time(Value(note, "published")) ?? DateTime.UtcNow,
				Updated = Time(Value(note, "updated")),
				To = List(note["to"]),
				Cc = List(note["cc"]),
				Mentions = Tags(note, "Mention").Select(t => new PostMention { ActorURI = Value(t, "href"), Handle = Value(t, "name") })
					.Where(m => Origin.Of(m.ActorURI) != default)
					.DistinctBy(m => m.ActorURI)
					.Take(MaxTags)
					.ToList(),
				Tags = Tags(note, "Hashtag").Select(t => Value(t, "name")?.TrimStart('#').ToLowerInvariant())
					.Where(t => !string.IsNullOrEmpty(t))
					.Distinct()
					.Take(MaxTags)
					.ToList(),
				Attachments = Attachments(note)
			};
		}

		static (string Html, string Language) Content(JsonObject note)
		{
			var content = Value(note, "content");
			string language = default;
			if (note["contentMap"] is JsonObject map && map.Count > 0)
			{
				var first = map.First();
				language = first.Key;
				if (string.IsNullOrEmpty(content) && first.Value is JsonValue value && value.TryGetValue<string>(out var mapped))
					content = mapped;
			}
			if (string.IsNullOrEmpty(content) && Value(note, "_misskey_content") is { } misskey)
				content = System.Net.WebUtility.HtmlEncode(misskey).Replace("\n", "<br>");
			return (content, language);
		}

		static string Attribution(JsonNode node) => node switch
		{
			JsonArray array => array.OfType<JsonObject>().Where(o => Value(o, "type") is "Person" or "Service" or "Application" or "Organization").Select(Id)
				.Concat(array.OfType<JsonValue>().Select(Id))
				.Concat(array.OfType<JsonObject>().Where(o => Value(o, "type") == "Group").Select(Id))
				.FirstOrDefault(id => id != default),
			_ => Id(node)
		};

		static IEnumerable<JsonObject> Tags(JsonObject note, string type) =>
			(note["tag"] switch
			{
				JsonArray array => array.OfType<JsonObject>(),
				JsonObject single => new[] { single },
				_ => Enumerable.Empty<JsonObject>()
			}).Where(t => Value(t, "type") == type);

		static List<PostMedia> Attachments(JsonObject note)
		{
			var items = note["attachment"] switch
			{
				JsonArray array => array.OfType<JsonObject>(),
				JsonObject single => new[] { single },
				_ => Enumerable.Empty<JsonObject>()
			};
			return items
				.Where(a => Value(a, "type") is "Document" or "Image" or "Video" or "Audio")
				.Select(a => new PostMedia
				{
					RemoteURL = UrlOf(a["url"]),
					ContentType = Value(a, "mediaType"),
					Description = Plain(Value(a, "name") ?? Value(a, "summary"), 1500),
					Blurhash = Value(a, "blurhash"),
					Width = Int(a["width"]),
					Height = Int(a["height"]),
					Focus = a["focalPoint"] is JsonArray focus && focus.Count == 2
						? focus.Select(f => f is JsonValue v && v.TryGetValue<double>(out var d) ? (float)Math.Clamp(d, -1, 1) : 0f).ToArray()
						: default
				})
				.Where(m => Origin.Of(m.RemoteURL) != default)
				.Take(MaxAttachments)
				.ToList();
		}

		static string UrlOf(JsonNode node) => node switch
		{
			JsonValue value when value.TryGetValue<string>(out var text) => text,
			JsonObject obj => Value(obj, "href") ?? Value(obj, "id"),
			JsonArray array => array.Select(item => item is JsonObject link && Value(link, "mediaType") is { } mediaType && !mediaType.StartsWith("text/html")
					? default
					: UrlOf(item))
				.FirstOrDefault(url => url != default) ?? array.Select(UrlOf).FirstOrDefault(url => url != default),
			_ => default
		};

		static IReadOnlyList<string> List(JsonNode node) => node switch
		{
			JsonArray array => array.Select(Id).Where(id => id != default).Distinct(StringComparer.Ordinal).ToList(),
			JsonNode single when Id(single) is { } id => new[] { id },
			_ => Array.Empty<string>()
		};

		static DateTime? Time(string value) =>
			DateTimeOffset.TryParse(value, CultureInfo.InvariantCulture, DateTimeStyles.AssumeUniversal, out var parsed)
				? parsed.UtcDateTime
				: (DateTime?)null;

		static int? Int(JsonNode node) =>
			node is JsonValue value && value.TryGetValue<int>(out var number) && number is > 0 and < 100_000 ? number : (int?)null;

		static string Plain(string value, int max)
		{
			if (string.IsNullOrWhiteSpace(value))
				return default;
			value = value.Trim();
			return value.Length <= max ? value : value[..max];
		}
	}
}