Improve NoteService.CreateNote

Iceshrimp.Backend/Core/Extensions/StringExtensions.cs

@@ -1,7 +1,17 @@
+using System.Globalization;
+
 namespace Iceshrimp.Backend.Core.Extensions;
 
 public static class StringExtensions {
 	public static string Truncate(this string target, int maxLength) {
 		return target[..Math.Min(target.Length, maxLength)];
 	}
+
+	public static string ToPunycode(this string target) {
+		return new IdnMapping().GetAscii(target);
+	}
+	
+	public static string FromPunycode(this string target) {
+		return new IdnMapping().GetUnicode(target);
+	}
 }

Iceshrimp.Backend/Core/Federation/ActivityStreams/Types/ASNote.cs

@@ -14,8 +14,8 @@ public class ASNote : ASObject {
 	public string? Content { get; set; }
 
 	[J("https://www.w3.org/ns/activitystreams#url")]
-	[JC(typeof(LDIdObjectConverter))]
-	public LDIdObject? Url { get; set; }
+	[JC(typeof(ASLinkConverter))]
+	public ASLink? Url { get; set; }
 
 	[J("https://www.w3.org/ns/activitystreams#sensitive")]
 	[JC(typeof(VC))]

Iceshrimp.Backend/Core/Helpers/MfmHelpers.cs

@@ -0,0 +1,98 @@
+using System.Text;
+using System.Text.RegularExpressions;
+using AngleSharp.Dom;
+using AngleSharp.Html.Parser;
+
+namespace Iceshrimp.Backend.Core.Helpers;
+
+public static class MfmHelpers {
+	public static async Task<string> FromHtml(string? html) {
+		if (html == null) return "";
+		
+		// Ensure compatibility with AP servers that send both <br> as well as newlines
+		var regex = new Regex(@"<br\s?\/?>\r?\n", RegexOptions.IgnoreCase);
+		html = regex.Replace(html, "\n");
+
+		var dom = await new HtmlParser().ParseDocumentAsync(html);
+		if (dom.Body == null) return "";
+
+		var sb = new StringBuilder();
+		dom.Body.ChildNodes.Select(ParseNode).ToList().ForEach(s => sb.Append(s));
+		return sb.ToString().Trim();
+	}
+
+	private static string? ParseNode(INode node) {
+		if (node.NodeType is NodeType.Text)
+			return node.TextContent;
+		if (node.NodeType is NodeType.Comment or NodeType.Document)
+			return null;
+
+		switch (node.NodeName) {
+			case "BR": {
+				return "\n";
+			}
+			case "A": {
+				//TODO: implement parsing of links & mentions (automatically correct split domain mentions for the latter)
+				return null;
+			}
+			case "H1": {
+				return $"【{ParseChildren(node)}】\n";
+			}
+			case "B":
+			case "STRONG": {
+				return $"**{ParseChildren(node)}**";
+			}
+			case "SMALL": {
+				return $"<small>{ParseChildren(node)}</small>";
+			}
+			case "S":
+			case "DEL": {
+				return $"~~{ParseChildren(node)}~~";
+			}
+			case "I":
+			case "EM": {
+				return $"<i>{ParseChildren(node)}</i>";
+			}
+			case "PRE": {
+				return node.ChildNodes is [{ NodeName: "CODE" }]
+					? $"\n```\n{string.Join(null, node.ChildNodes[0].TextContent)}\n```\n"
+					: ParseChildren(node);
+			}
+			case "CODE": {
+				return $"`{ParseChildren(node)}`";
+			}
+			case "BLOCKQUOTE": {
+				return node.TextContent.Length > 0
+					? $"\n> {string.Join("\n> ", node.TextContent.Split("\n"))}"
+					: null;
+			}
+
+			case "P":
+			case "H2":
+			case "H3":
+			case "H4":
+			case "H5":
+			case "H6": {
+				return $"\n\n{ParseChildren(node)}";
+			}
+
+			case "DIV":
+			case "HEADER":
+			case "FOOTER":
+			case "ARTICLE":
+			case "LI":
+			case "DT":
+			case "DD": {
+				return $"\n{ParseChildren(node)}";
+			}
+
+			default: {
+				return ParseChildren(node);
+			}
+		}
+	}
+
+	private static string ParseChildren(INode node) {
+		return string.Join(null, node.ChildNodes.Select(ParseNode));
+	}
+}

Iceshrimp.Backend/Core/Middleware/ErrorHandlerMiddleware.cs

@@ -69,6 +69,14 @@ public class GracefulException(HttpStatusCode statusCode, string error, string m
 
 	public GracefulException(string message, string? details = null) :
 		this(HttpStatusCode.InternalServerError, HttpStatusCode.InternalServerError.ToString(), message, details) { }
+
+	public static GracefulException UnprocessableEntity(string message, string? details = null) {
+		return new GracefulException(HttpStatusCode.UnprocessableEntity, message, details);
+	}
+
+	public static GracefulException Forbidden(string message, string? details = null) {
+		return new GracefulException(HttpStatusCode.Forbidden, message, details);
+	}
 }
 
 public enum ExceptionVerbosity {

Iceshrimp.Backend/Core/Services/NoteService.cs

@@ -3,6 +3,7 @@ using Iceshrimp.Backend.Core.Database.Tables;
 using Iceshrimp.Backend.Core.Federation.ActivityPub;
 using Iceshrimp.Backend.Core.Federation.ActivityStreams.Types;
 using Iceshrimp.Backend.Core.Helpers;
+using Iceshrimp.Backend.Core.Middleware;
 using Microsoft.EntityFrameworkCore;
 
 namespace Iceshrimp.Backend.Core.Services;
@@ -19,16 +20,34 @@ public class NoteService(ILogger<NoteService> logger, DatabaseContext db, UserRe
 		var user = await userResolver.Resolve(actor.Id);
 		logger.LogDebug("Resolved user to {userId}", user.Id);
 
+		// Validate note
+		if (note.AttributedTo is not { Count: 1 } || note.AttributedTo[0].Id != user.Uri)
+			throw GracefulException.UnprocessableEntity("User.Uri doesn't match Note.AttributedTo");
+		if (user.Uri == null)
+			throw GracefulException.UnprocessableEntity("User.Uri is null");
+		if (new Uri(note.Id).IdnHost != new Uri(user.Uri).IdnHost)
+			throw GracefulException.UnprocessableEntity("User.Uri host doesn't match Note.Id host");
+		if (!note.Id.StartsWith("https://"))
+			throw GracefulException.UnprocessableEntity("Note.Id schema is invalid");
+		if (note.Url?.Link != null && !note.Url.Link.StartsWith("https://"))
+			throw GracefulException.UnprocessableEntity("Note.Url schema is invalid");
+		if (note.PublishedAt is null or { Year: < 2007 } || note.PublishedAt > DateTime.Now + TimeSpan.FromDays(3))
+			throw GracefulException.UnprocessableEntity("Note.PublishedAt is nonsensical");
+		if (user.IsSuspended)
+			throw GracefulException.Forbidden("User is suspended");
+
+		//TODO: validate AP object type
+		//TODO: parse note visibility
 		//TODO: resolve anything related to the note as well (reply thread, attachments, emoji, etc)
 
 		var dbNote = new Note {
 			Id     = IdHelpers.GenerateSlowflakeId(),
 			Uri    = note.Id,
-			Url    = note.Url?.Id,                   //FIXME: this doesn't seem to work yet
-			Text   = note.MkContent ?? note.Content, //TODO: html-to-mfm
+			Url    = note.Url?.Id,                                      //FIXME: this doesn't seem to work yet
+			Text   = note.MkContent ?? await MfmHelpers.FromHtml(note.Content),
 			UserId = user.Id,
 			CreatedAt = note.PublishedAt?.ToUniversalTime() ??
-			            throw new Exception("Missing or invalid PublishedAt field"),
+			            throw GracefulException.UnprocessableEntity("Missing or invalid PublishedAt field"),
 			UserHost   = user.Host,
 			Visibility = Note.NoteVisibility.Public //TODO: parse to & cc fields
 		};

Iceshrimp.Backend/Core/Services/UserService.cs

@@ -1,7 +1,9 @@
+using System.Globalization;
 using System.Net;
 using System.Security.Cryptography;
 using Iceshrimp.Backend.Core.Database;
 using Iceshrimp.Backend.Core.Database.Tables;
+using Iceshrimp.Backend.Core.Extensions;
 using Iceshrimp.Backend.Core.Federation.ActivityPub;
 using Iceshrimp.Backend.Core.Helpers;
 using Iceshrimp.Backend.Core.Middleware;
@@ -16,7 +18,7 @@ public class UserService(ILogger<UserService> logger, DatabaseContext db, APFetc
 		var split = acct[5..].Split('@');
 		if (split.Length != 2) throw new GracefulException(HttpStatusCode.BadRequest, "Invalid query");
 
-		return (split[0], split[1]);
+		return (split[0], split[1].ToPunycode());
 	}
 
 	public Task<User?> GetUserFromQuery(string query) {

Iceshrimp.Backend/Iceshrimp.Backend.csproj

@@ -14,6 +14,7 @@
     </PropertyGroup>
 
     <ItemGroup>
+        <PackageReference Include="AngleSharp" Version="1.1.0" />
         <PackageReference Include="Asp.Versioning.Http" Version="8.0.0"/>
         <PackageReference Include="cuid.net" Version="5.0.2"/>
         <PackageReference Include="dotNetRdf.Core" Version="3.2.1-dev"/>