Collapsing i18n keys check in CI/CD + fix existing

.github/workflows/main.yml

@@ -36,6 +36,8 @@ jobs:
         run: pnpm run knip
       - name: Check translations jsons
         run: pnpm run check-translation-jsons:no-write
+      - name: Check plural key collapse
+        run: pnpm run check-plural-collapse
       - name: Check articles
         run: pnpm run check-articles
       - name: Check test DB migrations

docs/translation.md

@@ -15,5 +15,6 @@ Things to note:
 - If you want to add a new language, ask Sendou.
 - Some lines have dynamic parts like `"articleBy": "by {{author}}"`. The `{{author}}` part should appear in the translated version unchanged - don't translate the part inside `{{}}`.
 - Another special syntax: `"project": "Sendou.ink is a project by <2>Sendou</2> with help from contributors:"`. The `<2></2>` tags should appear in the translated version, but the text inside them can change.
+- Some English keys come in plural variants like `"tournament_one"` and `"tournament_other"`. Languages that only have a single plural form (e.g. Chinese, Japanese, Korean) can only hold **one** translation for such a key, so during syncing these variants are collapsed into a single key (e.g. `"tournament"`). When collapsing, the `_other` value is kept (it's the one that includes the `{{count}}` number), and the other variants are discarded. This means for these languages you should translate the `_other` variant. 
 
 Any questions please ask Sendou!

locales/zh/badges.json

@@ -2,8 +2,7 @@
 	"patreon": "Patreon 上的 sendou.ink Supporter",
 	"patreon+": "Patreon 上的 sendou.ink Supporter+",
 	"xp": "到达 {{xpText}} 的奖励",
-	"tournament_one": "在 {{tournament}} 中获胜的奖励",
-	"tournament_other": "在 {{tournament}} 中获胜了 {{count}} 次的奖励",
+	"tournament": "在 {{tournament}} 中获胜了 {{count}} 次的奖励",
 	"forYourEvent": "在我的活动中设置徽章",
 	"managedBy": "由 <0></0> 管理",
 	"madeBy": "由 <0></0> 创作",

locales/zh/tournament.json

@@ -122,7 +122,7 @@
 	"staff.editOrganization": "编辑组织",
 	"actions.addSub": "添加替补",
 	"actions.shareLink": "分享您的邀请链接以添加成员: {{inviteLink}}",
-	"actions.sub.prompt_other": "您仍可以向阵容中添加 {{count}} 名替补",
+	"actions.sub.prompt": "您仍可以向阵容中添加 {{count}} 名替补",
 	"actions.sub.prompt_zero": "您的阵容已满，无法添加更多替补",
 	"actions.finalize": "正在结束赛事",
 	"actions.finalize.button": "结束赛事",

package.json

@@ -31,7 +31,8 @@
 		"test:e2e:flaky-detect": "playwright test --repeat-each=10 --max-failures=1",
 		"test:e2e:generate-seeds": "cross-env DB_PATH=db-test.sqlite3 pnpm run migrate up && cross-env DB_PATH=db-test.sqlite3 vite-node scripts/generate-e2e-seed-dbs.ts",
 		"check-test-db-migrations": "node --experimental-strip-types scripts/check-test-db-migrations.ts",
-		"checks": "pnpm run biome:fix && pnpm run test:unit:browser && pnpm run check-translation-jsons && pnpm run typecheck && pnpm run knip && pnpm run check-test-db-migrations",
+		"check-plural-collapse": "node --experimental-strip-types scripts/collapse-single-plural-keys.ts --check",
+		"checks": "pnpm run biome:fix && pnpm run test:unit:browser && pnpm run check-translation-jsons && pnpm run check-plural-collapse && pnpm run typecheck && pnpm run knip && pnpm run check-test-db-migrations",
 		"setup": "cross-env DB_PATH=db.sqlite3 vite-node ./scripts/setup.ts",
 		"i18n:sync": "node --experimental-strip-types scripts/collapse-single-plural-keys.ts && i18next-locales-sync -e true -p en -s da de es-ES es-US fr-CA fr-EU he it ja ko nl pl pt-BR ru zh -l locales && pnpm run biome:fix",
 		"knip": "knip"

scripts/check-translation-jsons.ts

@@ -12,6 +12,13 @@ const dontWrite = process.argv.includes(NO_WRITE_KEY);
 
 const KNOWN_SUFFIXES = ["_zero", "_one", "_two", "_few", "_many", "_other"];
 
+// `_zero` is an i18next `count === 0` override, not a CLDR plural category. It
+// may legitimately sit alongside the bare singular key that single-plural
+// languages (zh, ja, ko) use after `i18n:sync` collapses their plural forms, so
+// unlike the real plural suffixes it never counts as a with/without-suffix clash.
+const ZERO_SUFFIX = "_zero";
+const PLURAL_SUFFIXES = KNOWN_SUFFIXES.filter((sfx) => sfx !== ZERO_SUFFIX);
+
 const REPO_TRANSLATIONS_INFO_URL =
 	"https://github.com/sendou-ink/sendou.ink/blob/main/docs/translation.md";
 
@@ -187,40 +194,47 @@ function getKeysWithoutSuffix(
 	lang: string,
 	file: string,
 ): string[] {
-	const foundSuffixKeys = new Set<string>();
-	const keys = [];
+	const pluralBaseKeys = new Set<string>();
+	const bareKeys = new Set<string>();
+	const keys: string[] = [];
+
+	const pushOnce = (key: string) => {
+		if (!keys.includes(key)) keys.push(key);
+	};
 
 	for (const [key, value] of Object.entries(translations)) {
 		if (value === "") {
 			continue; // Consider key missing if untranslated
 		}
 
-		const suffix = KNOWN_SUFFIXES.find((sfx) => key.endsWith(sfx));
+		if (key.endsWith(ZERO_SUFFIX)) {
+			// `_zero` override always maps to its base key and never clashes with it
+			pushOnce(key.slice(0, -ZERO_SUFFIX.length));
+			continue;
+		}
+
+		const suffix = PLURAL_SUFFIXES.find((sfx) => key.endsWith(sfx));
 		if (!suffix) {
-			if (foundSuffixKeys.has(key)) {
+			if (pluralBaseKeys.has(key)) {
 				throw new Error(
 					`Found same key with and without suffixes in ${lang}/${file}: ${key}`,
 				);
 			}
-			keys.push(key);
+			bareKeys.add(key);
+			pushOnce(key);
 			continue;
 		}
 
-		const baseKey = key.replace(suffix, "");
+		const baseKey = key.slice(0, -suffix.length);
 
-		if (foundSuffixKeys.has(baseKey)) {
-			// Already found this key with a suffix. Duplicates are handled elsewhere.
-			continue;
-		}
-
-		if (keys.includes(baseKey)) {
+		if (bareKeys.has(baseKey)) {
 			throw new Error(
 				`Found same key with and without suffixes in ${lang}/${file}: ${baseKey}`,
 			);
 		}
 
-		keys.push(baseKey);
-		foundSuffixKeys.add(baseKey);
+		pluralBaseKeys.add(baseKey);
+		pushOnce(baseKey);
 	}
 
 	return keys;

scripts/collapse-single-plural-keys.ts

@@ -9,12 +9,24 @@ const __dirname = path.dirname(__filename);
 const LOCALES_PATH = path.join(__dirname, "..", "locales");
 const PRIMARY_LANGUAGE = "en";
 
+// When passed, no files are written. Instead the script exits non-zero if any
+// file would be collapsed, so CI can fail a PR that committed suffixed plural
+// keys in a single-plural language without running `pnpm run i18n:sync`.
+const CHECK_KEY = "--check";
+const checkOnly = process.argv.includes(CHECK_KEY);
+
 // `_zero` is intentionally excluded: it is not a CLDR plural category but an
 // i18next special-case override for `count === 0`. `i18next-locales-sync` treats
 // e.g. `foo_zero` as a plain key (English has no `_zero` plural form), so it is
 // preserved across every language and must not be collapsed into `foo`.
 const COLLAPSIBLE_PLURAL_SUFFIXES = ["_one", "_two", "_few", "_many", "_other"];
 
+// The plural form whose value we keep when collapsing. `_other` is preferred
+// because it is the form i18next resolves at runtime for these languages and it
+// typically carries the `{{count}}` interpolation. The remaining forms are only
+// used as a fallback when `_other` has no translated value.
+const PREFERRED_SUFFIX = "_other";
+
 // `i18next-locales-sync` stores plural keys for languages whose CLDR cardinal
 // rule has a single category ("other" only, e.g. zh, ja, ko) under the bare
 // singular key instead of suffixed `_one`/`_other` keys. When a translator adds
@@ -26,6 +38,8 @@ const languages = fs
 	.readdirSync(LOCALES_PATH)
 	.filter((lang) => lang !== PRIMARY_LANGUAGE && !lang.startsWith("."));
 
+const wouldCollapse: Array<{ lang: string; file: string; keys: string[] }> = [];
+
 for (const lang of languages) {
 	if (!isSinglePluralLanguage(lang)) continue;
 
@@ -41,14 +55,32 @@ for (const lang of languages) {
 			string
 		>;
 
-		const { collapsed, changed } = collapsePluralKeys(content);
-		if (!changed) continue;
+		const { collapsed, collapsedBaseKeys } = collapsePluralKeys(content);
+		if (collapsedBaseKeys.length === 0) continue;
+
+		if (checkOnly) {
+			wouldCollapse.push({ lang, file, keys: collapsedBaseKeys });
+			continue;
+		}
 
 		fs.writeFileSync(filePath, `${JSON.stringify(collapsed, null, "\t")}\n`);
 		console.info(`collapsed plural keys in ${lang}/${file}`);
 	}
 }
 
+if (checkOnly) {
+	if (wouldCollapse.length > 0) {
+		console.error(
+			"Found suffixed plural keys in single-plural languages that `pnpm run i18n:sync` would collapse. Run it and commit the result.",
+		);
+		for (const { lang, file, keys } of wouldCollapse) {
+			console.error(`  ${lang}/${file}: ${keys.join(", ")}`);
+		}
+		process.exit(1);
+	}
+	console.info("no plural keys to collapse in single-plural languages");
+}
+
 function isSinglePluralLanguage(lang: string) {
 	return (
 		new Intl.PluralRules(lang).resolvedOptions().pluralCategories.length === 1
@@ -57,7 +89,8 @@ function isSinglePluralLanguage(lang: string) {
 
 function collapsePluralKeys(content: Record<string, string>) {
 	const collapsed: Record<string, string> = {};
-	let changed = false;
+	const chosenFromPreferred = new Set<string>();
+	const collapsedBaseKeys = new Set<string>();
 
 	for (const [key, value] of Object.entries(content)) {
 		const suffix = COLLAPSIBLE_PLURAL_SUFFIXES.find((sfx) => key.endsWith(sfx));
@@ -66,14 +99,24 @@ function collapsePluralKeys(content: Record<string, string>) {
 			continue;
 		}
 
-		changed = true;
 		const baseKey = key.slice(0, -suffix.length);
+		collapsedBaseKeys.add(baseKey);
+		const isPreferred = suffix === PREFERRED_SUFFIX;
 
-		// keep the first non-empty value found across the plural forms
-		if (!(baseKey in collapsed) || (!collapsed[baseKey] && value)) {
+		// prefer the `_other` value (it carries {{count}}); otherwise keep the
+		// first non-empty value found across the remaining plural forms
+		const shouldReplace =
+			!(baseKey in collapsed) ||
+			(!collapsed[baseKey] && !!value) ||
+			(isPreferred && !!value && !chosenFromPreferred.has(baseKey));
+
+		if (shouldReplace) {
 			collapsed[baseKey] = value;
+			if (isPreferred && value) {
+				chosenFromPreferred.add(baseKey);
+			}
 		}
 	}
 
-	return { collapsed, changed };
+	return { collapsed, collapsedBaseKeys: Array.from(collapsedBaseKeys) };
 }