Merge branch 'feat/backend-post-langs' into 'develop'

Add post language data to backend and API

Co-authored-by: Essem <smswessem@gmail.com>

See merge request firefish/firefish!10602

docs/api-change.md

@@ -4,6 +4,8 @@ Breaking changes are indecated by the :warning: icon.
 
 ## v1.0.5 (unreleased)
 
+- Added `lang` parameter to `notes/create` and `notes/edit`.
+
 ### dev11
 
 - :warning: `notes/translate` now requires credentials.

packages/backend/migration/1695334243217-add-post-lang.js

@@ -0,0 +1,11 @@
+export class AddPostLang1695334243217 {
+    name = 'AddPostLang1695334243217'
+
+    async up(queryRunner) {
+        await queryRunner.query(`ALTER TABLE "note" ADD "lang" character varying(10)`);
+    }
+
+    async down(queryRunner) {
+        await queryRunner.query(`ALTER TABLE "note" DROP COLUMN "lang"`);
+    }
+}

packages/backend/src/models/entities/note.ts

@@ -66,6 +66,12 @@ export class Note {
 	})
 	public text: string | null;
 
+	@Column("varchar", {
+		length: 10,
+		nullable: true,
+	})
+	public lang: string | null;
+
 	@Column("varchar", {
 		length: 256,
 		nullable: true,

packages/backend/src/models/repositories/note.ts

@@ -199,8 +199,6 @@ export const NoteRepository = db.getRepository(Note).extend({
 			host,
 		);
 
-		const lang =
-			detectLanguage(`${note.cw ?? ""}\n${note.text ?? ""}`) ?? "unknown";
 		const reactionEmoji = await populateEmojis(reactionEmojiNames, host);
 		const packed: Packed<"Note"> = await awaitAll({
 			id: note.id,
@@ -260,7 +258,7 @@ export const NoteRepository = db.getRepository(Note).extend({
 							: undefined,
 				  }
 				: {}),
-			lang: lang,
+			lang: note.lang,
 		});
 
 		if (packed.user.isCat && packed.user.speakAsCat && packed.text) {

packages/backend/src/remote/activitypub/models/note.ts

@@ -53,6 +53,7 @@ import { DB_MAX_IMAGE_COMMENT_LENGTH } from "@/misc/hard-limits.js";
 import { truncate } from "@/misc/truncate.js";
 import { type Size, getEmojiSize } from "@/misc/emoji-meta.js";
 import { fetchMeta } from "@/misc/fetch-meta.js";
+import { langmap } from "@/misc/langmap.js";
 
 const logger = apLogger;
 
@@ -305,11 +306,20 @@ export async function createNote(
 
 	// Text parsing
 	let text: string | null = null;
+	let lang: string | null = null;
 	if (
 		note.source?.mediaType === "text/x.misskeymarkdown" &&
 		typeof note.source?.content === "string"
 	) {
 		text = note.source.content;
+		if (note.contentMap != null) {
+			const key = Object.keys(note.contentMap)[0];
+			lang = Object.keys(langmap).includes(key) ? key : null;
+		}
+	} else if (note.contentMap != null) {
+		const entry = Object.entries(note.contentMap)[0];
+		lang = Object.keys(langmap).includes(entry[0]) ? entry[0] : null;
+		text = htmlToMfm(entry[1], note.tag);
 	} else if (typeof note.content === "string") {
 		text = htmlToMfm(note.content, note.tag);
 	}
@@ -378,6 +388,7 @@ export async function createNote(
 			name: note.name,
 			cw,
 			text,
+			lang,
 			localOnly: false,
 			visibility,
 			visibleUsers,
@@ -565,11 +576,20 @@ export async function updateNote(value: string | IObject, resolver?: Resolver) {
 
 	// Text parsing
 	let text: string | null = null;
+	let lang: string | null = null;
 	if (
 		post.source?.mediaType === "text/x.misskeymarkdown" &&
 		typeof post.source?.content === "string"
 	) {
 		text = post.source.content;
+		if (post.contentMap != null) {
+			const key = Object.keys(post.contentMap)[0];
+			lang = Object.keys(langmap).includes(key) ? key : null;
+		}
+	} else if (post.contentMap != null) {
+		const entry = Object.entries(post.contentMap)[0];
+		lang = Object.keys(langmap).includes(entry[0]) ? entry[0] : null;
+		text = htmlToMfm(entry[1], post.tag);
 	} else if (typeof post.content === "string") {
 		text = htmlToMfm(post.content, post.tag);
 	}
@@ -663,6 +683,9 @@ export async function updateNote(value: string | IObject, resolver?: Resolver) {
 	if (text && text !== note.text) {
 		update.text = text;
 	}
+	if (lang && lang !== note.lang) {
+		update.lang = lang;
+	}
 	if (cw !== note.cw) {
 		update.cw = cw ? cw : null;
 	}

packages/backend/src/remote/activitypub/renderer/note.ts

@@ -115,7 +115,7 @@ export default async function renderNote(
 		}),
 	);
 
-	const lang = detectLanguage(text);
+	const lang = note.lang ?? detectLanguage(text);
 	const contentMap = lang ? {
 		[lang]: content
 	} : null;

packages/backend/src/server/api/endpoints/notes/create.ts

@@ -108,6 +108,7 @@ export const paramDef = {
 			},
 		},
 		text: { type: "string", maxLength: MAX_NOTE_TEXT_LENGTH, nullable: true },
+		lang: { type: "string", nullable: true, maxLength: 10 },
 		cw: { type: "string", nullable: true, maxLength: 100 },
 		localOnly: { type: "boolean", default: false },
 		noExtractMentions: { type: "boolean", default: false },
@@ -294,6 +295,7 @@ export default define(meta, paramDef, async (ps, user) => {
 			  }
 			: undefined,
 		text: ps.text || undefined,
+		lang: ps.lang,
 		reply,
 		renote,
 		cw: ps.cw,

packages/backend/src/server/api/endpoints/notes/edit.ts

@@ -35,6 +35,8 @@ import renderUpdate from "@/remote/activitypub/renderer/update.js";
 import { deliverToRelays } from "@/services/relay.js";
 // import { deliverQuestionUpdate } from "@/services/note/polls/update.js";
 import { fetchMeta } from "@/misc/fetch-meta.js";
+import { detect as detectLanguage } from "tinyld";
+import { langmap } from "@/misc/langmap.js";
 
 export const meta = {
 	tags: ["notes"],
@@ -169,6 +171,7 @@ export const paramDef = {
 			},
 		},
 		text: { type: "string", maxLength: MAX_NOTE_TEXT_LENGTH, nullable: true },
+		lang: { type: "string", nullable: true, maxLength: 10 },
 		cw: { type: "string", nullable: true, maxLength: 250 },
 		localOnly: { type: "boolean", default: false },
 		noExtractMentions: { type: "boolean", default: false },
@@ -375,6 +378,15 @@ export default define(meta, paramDef, async (ps, user) => {
 		ps.text = null;
 	}
 
+	if (ps.lang) {
+		ps.lang = ps.lang.trim();
+		if (!Object.keys(langmap).includes(ps.lang.trim())) throw new Error("invalid param");
+	} else if (ps.text) {
+		ps.lang = detectLanguage(ps.text);
+	} else {
+		ps.lang = null;
+	}
+
 	let tags = [];
 	let emojis = [];
 	let mentionedUsers = [];
@@ -532,6 +544,9 @@ export default define(meta, paramDef, async (ps, user) => {
 	if (ps.text !== note.text) {
 		update.text = ps.text;
 	}
+	if (ps.lang !== note.lang) {
+		update.lang = ps.lang;
+	}
 	if (ps.cw !== note.cw || (ps.cw && !note.cw)) {
 		update.cw = ps.cw;
 	}

packages/backend/src/services/note/create.ts

@@ -67,6 +67,8 @@ import { shouldSilenceInstance } from "@/misc/should-block-instance.js";
 import meilisearch from "../../db/meilisearch.js";
 import { redisClient } from "@/db/redis.js";
 import { Mutex } from "redis-semaphore";
+import { detect as detectLanguage } from "tinyld";
+import { langmap } from "@/misc/langmap.js";
 
 const mutedWordsCache = new Cache<
 	{ userId: UserProfile["userId"]; mutedWords: UserProfile["mutedWords"] }[]
@@ -139,6 +141,7 @@ type Option = {
 	createdAt?: Date | null;
 	name?: string | null;
 	text?: string | null;
+	lang?: string | null;
 	reply?: Note | null;
 	renote?: Note | null;
 	files?: DriveFile[] | null;
@@ -276,6 +279,15 @@ export default async (
 			data.text = null;
 		}
 
+		if (data.lang) {
+			data.lang = data.lang.trim();
+			if (!Object.keys(langmap).includes(data.lang.trim())) throw new Error("invalid param");
+		} else if (data.text) {
+			data.lang = detectLanguage(data.text);
+		} else {
+			data.lang = null;
+		}
+
 		let tags = data.apHashtags;
 		let emojis = data.apEmojis;
 		let mentionedUsers = data.apMentions;
@@ -712,6 +724,7 @@ async function insertNote(
 			: null,
 		name: data.name,
 		text: data.text,
+		lang: data.lang,
 		hasPoll: data.poll != null,
 		cw: data.cw == null ? null : data.cw,
 		tags: tags.map((tag) => normalizeForSearch(tag)),