Preserve remote post languages from contentMap

CHANGES.md

@@ -6,6 +6,20 @@ Version 0.9.22
 
 To be released.
 
+ -  Fixed the language of remote posts not being stored when they were
+    received from Mastodon, Hollo, or any other software that sends `content`
+    together with `contentMap`.  [[#647], [#659]]
+
+ -  Upgraded Fedify to 2.2.16, which normalizes BCP 47 extended language
+    subtags before parsing language tags.  Language-tagged strings with tags
+    like `zh-YUE`, which Mastodon uses for Cantonese, were previously dropped
+    because `Intl.Locale` rejects them; they are now kept with the canonical
+    tag (e.g., `yue`).  [[#647], [fedify-dev/fedify#1229], [#659]]
+
+[fedify-dev/fedify#1229]: https://github.com/fedify-dev/fedify/issues/1229
+[#647]: https://github.com/fedify-dev/hollo/issues/647
+[#659]: https://github.com/fedify-dev/hollo/pull/659
+
 
 Version 0.9.21
 --------------

package.json

@@ -32,14 +32,14 @@
   },
   "dependencies": {
     "@aws-sdk/credential-providers": "^3.1037.0",
-    "@fedify/debugger": "~2.2.15",
-    "@fedify/fedify": "~2.2.15",
-    "@fedify/hono": "~2.2.15",
+    "@fedify/debugger": "~2.2.16",
+    "@fedify/fedify": "~2.2.16",
+    "@fedify/hono": "~2.2.16",
     "@fedify/markdown-it-hashtag": "~0.3.0",
     "@fedify/markdown-it-mention": "~0.3.0",
-    "@fedify/postgres": "~2.2.15",
-    "@fedify/vocab": "~2.2.15",
-    "@fedify/webfinger": "~2.2.15",
+    "@fedify/postgres": "~2.2.16",
+    "@fedify/vocab": "~2.2.16",
+    "@fedify/webfinger": "~2.2.16",
     "@hexagon/base64": "^2.0.4",
     "@hono/node-server": "^1.19.13",
     "@hono/zod-validator": "^0.7.6",
@@ -88,7 +88,7 @@
   },
   "devDependencies": {
     "@dotenvx/dotenvx": "^1.52.0",
-    "@fedify/lint": "~2.2.15",
+    "@fedify/lint": "~2.2.16",
     "@iconify-json/lucide": "^1.2.104",
     "@reporters/github": "^1.13.1",
     "@types/fluent-ffmpeg": "^2.1.28",

src/federation/post.test.ts

@@ -755,6 +755,99 @@ describe("persistPost", () => {
       { accountId: mentioned.id, postId: result.id },
     ]);
   });
+
+  it("stores the language when content is accompanied by contentMap", async () => {
+    expect.assertions(2);
+    const author = await seedRemoteAccount("author");
+    const object = await Note.fromJsonLd({
+      "@context": "https://www.w3.org/ns/activitystreams",
+      type: "Note",
+      id: "https://remote.test/@author/posts/content-map",
+      content: "<p>こんにちは</p>",
+      contentMap: { ja: "<p>こんにちは</p>" },
+      to: PUBLIC_COLLECTION.href,
+    });
+
+    const result = await persistPost(
+      db,
+      object.clone({ attribution: createPerson(author) }),
+      "https://hollo.test",
+      { account: author },
+    );
+
+    expect(result?.contentHtml).toBe("<p>こんにちは</p>");
+    expect(result?.language).toBe("ja");
+  });
+
+  it("stores the language from summaryMap when content has none", async () => {
+    expect.assertions(1);
+    const author = await seedRemoteAccount("author");
+    const object = await Note.fromJsonLd({
+      "@context": "https://www.w3.org/ns/activitystreams",
+      type: "Note",
+      id: "https://remote.test/@author/posts/summary-map",
+      summary: "CW",
+      summaryMap: { "zh-TW": "CW" },
+      content: "<p>內容</p>",
+      to: PUBLIC_COLLECTION.href,
+    });
+
+    const result = await persistPost(
+      db,
+      object.clone({ attribution: createPerson(author) }),
+      "https://hollo.test",
+      { account: author },
+    );
+
+    expect(result?.language).toBe("zh-TW");
+  });
+
+  it("normalizes extended language tags such as zh-YUE", async () => {
+    expect.assertions(1);
+    const author = await seedRemoteAccount("author");
+    const object = await Note.fromJsonLd({
+      "@context": "https://www.w3.org/ns/activitystreams",
+      type: "Note",
+      id: "https://remote.test/@author/posts/extlang",
+      content: "<p>你好</p>",
+      contentMap: { "zh-YUE": "<p>你好</p>" },
+      to: PUBLIC_COLLECTION.href,
+    });
+
+    const result = await persistPost(
+      db,
+      object.clone({ attribution: createPerson(author) }),
+      "https://hollo.test",
+      { account: author },
+    );
+
+    expect(result?.language).toBe("yue");
+  });
+
+  it("ignores undetermined languages from a default @language", async () => {
+    expect.assertions(1);
+    const author = await seedRemoteAccount("author");
+    const object = await Note.fromJsonLd({
+      "@context": [
+        "https://www.w3.org/ns/activitystreams",
+        { "@language": "und" },
+      ],
+      type: "Note",
+      id: "https://remote.test/@author/posts/default-language",
+      content: "<p>こんにちは</p>",
+      contentMap: { ja: "<p>こんにちは</p>" },
+      to: PUBLIC_COLLECTION.href,
+    });
+
+    const result = await persistPost(
+      db,
+      object.clone({ attribution: createPerson(author) }),
+      "https://hollo.test",
+      { account: author },
+    );
+
+    expect(result?.language).toBe("ja");
+  });
 });
 
 describe("toObject", () => {