Refactor translation system to be reusable in non-Astro code (#1003)

10 files changed, 218 insertions, 53 deletions

diff --git a/.changeset/large-squids-wash.md b/.changeset/large-squids-wash.md
new file mode 100644
index 00000000..924b70dc
--- /dev/null
+++ b/.changeset/large-squids-wash.md
@@ -0,0 +1,5 @@
+---
+'@astrojs/starlight': patch
+---
+
+Internal: refactor translation string loading to make translations available to Starlight integration code
diff --git a/packages/starlight/__tests__/i18n/empty-src/content/i18n/.gitkeep b/packages/starlight/__tests__/i18n/empty-src/content/i18n/.gitkeep
new file mode 100644
index 00000000..e69de29b
--- /dev/null
+++ b/packages/starlight/__tests__/i18n/empty-src/content/i18n/.gitkeep
diff --git a/packages/starlight/__tests__/i18n/malformed-src/content/i18n/en.json b/packages/starlight/__tests__/i18n/malformed-src/content/i18n/en.json
new file mode 100644
index 00000000..6c916955
--- /dev/null
+++ b/packages/starlight/__tests__/i18n/malformed-src/content/i18n/en.json
@@ -0,0 +1,3 @@
+{
+,
+}
diff --git a/packages/starlight/__tests__/i18n/src/content/i18n/en.json b/packages/starlight/__tests__/i18n/src/content/i18n/en.json
new file mode 100644
index 00000000..099e4c00
--- /dev/null
+++ b/packages/starlight/__tests__/i18n/src/content/i18n/en.json
@@ -0,0 +1,3 @@
+{
+  "page.editLink": "Make this page different"
+}
diff --git a/packages/starlight/__tests__/i18n/translations-fs.test.ts b/packages/starlight/__tests__/i18n/translations-fs.test.ts
new file mode 100644
index 00000000..5b9025aa
--- /dev/null
+++ b/packages/starlight/__tests__/i18n/translations-fs.test.ts
@@ -0,0 +1,76 @@
+import { describe, expect, test } from 'vitest';
+import { createTranslationSystemFromFs } from '../../utils/translations-fs';
+
+describe('createTranslationSystemFromFs', () => {
+	test('creates a translation system that returns default strings', () => {
+		const useTranslations = createTranslationSystemFromFs(
+			{
+				locales: { en: { label: 'English', dir: 'ltr' } },
+				defaultLocale: { label: 'English', locale: 'en', dir: 'ltr' },
+			},
+			// Using non-existent `_src/` to ignore custom files in this test fixture.
+			{ srcDir: new URL('./_src/', import.meta.url) }
+		);
+		const t = useTranslations('en');
+		expect(t('page.editLink')).toMatchInlineSnapshot('"Edit page"');
+	});
+
+	test('creates a translation system that uses custom strings', () => {
+		const useTranslations = createTranslationSystemFromFs(
+			{
+				locales: { en: { label: 'English', dir: 'ltr' } },
+				defaultLocale: { label: 'English', locale: 'en', dir: 'ltr' },
+			},
+			// Using `src/` to load custom files in this test fixture.
+			{ srcDir: new URL('./src/', import.meta.url) }
+		);
+		const t = useTranslations('en');
+		expect(t('page.editLink')).toMatchInlineSnapshot('"Make this page different"');
+	});
+
+	test('supports root locale', () => {
+		const useTranslations = createTranslationSystemFromFs(
+			{
+				locales: { root: { label: 'English', dir: 'ltr', lang: 'en' } },
+				defaultLocale: { label: 'English', locale: 'root', lang: 'en', dir: 'ltr' },
+			},
+			// Using `src/` to load custom files in this test fixture.
+			{ srcDir: new URL('./src/', import.meta.url) }
+		);
+		const t = useTranslations(undefined);
+		expect(t('page.editLink')).toMatchInlineSnapshot('"Make this page different"');
+	});
+
+	test('returns translation for unknown language', () => {
+		const useTranslations = createTranslationSystemFromFs(
+			{
+				locales: { root: { label: 'English', dir: 'ltr', lang: 'en' } },
+				defaultLocale: { label: 'English', locale: undefined, dir: 'ltr' },
+			},
+			// Using `src/` to load custom files in this test fixture.
+			{ srcDir: new URL('./src/', import.meta.url) }
+		);
+		const t = useTranslations('fr');
+		expect(t('page.editLink')).toMatchInlineSnapshot('"Make this page different"');
+	});
+
+	test('handles empty i18n directory', () => {
+		const useTranslations = createTranslationSystemFromFs(
+			{ locales: {}, defaultLocale: { label: 'English', locale: 'en', dir: 'ltr' } },
+			// Using `empty-src/` to emulate empty `src/content/i18n/` directory.
+			{ srcDir: new URL('./empty-src/', import.meta.url) }
+		);
+		const t = useTranslations('en');
+		expect(t('page.editLink')).toMatchInlineSnapshot('"Edit page"');
+	});
+
+	test('throws on malformed i18n JSON', () => {
+		expect(() =>
+			createTranslationSystemFromFs(
+				{ locales: {}, defaultLocale: { label: 'English', locale: 'en', dir: 'ltr' } },
+				// Using `malformed-src/` to trigger syntax error in bad JSON file.
+				{ srcDir: new URL('./malformed-src/', import.meta.url) }
+			)
+		).toThrow(SyntaxError);
+	});
+});
diff --git a/packages/starlight/index.ts b/packages/starlight/index.ts
index a52b994f..53ffb2e2 100644
--- a/packages/starlight/index.ts
+++ b/packages/starlight/index.ts
@@ -9,6 +9,7 @@ import { vitePluginStarlightUserConfig } from './integrations/virtual-user-confi
 import { errorMap } from './utils/error-map';
 import { StarlightConfigSchema, type StarlightUserConfig } from './utils/user-config';
 import { rehypeRtlCodeSupport } from './integrations/code-rtl-support';
+import { createTranslationSystemFromFs } from './utils/translations-fs';
 
 export default function StarlightIntegration(opts: StarlightUserConfig): AstroIntegration {
 	const parsedConfig = StarlightConfigSchema.safeParse(opts, { errorMap });
@@ -26,6 +27,7 @@ export default function StarlightIntegration(opts: StarlightUserConfig): AstroIn
 		name: '@astrojs/starlight',
 		hooks: {
 			'astro:config:setup': ({ config, injectRoute, updateConfig }) => {
+				const useTranslations = createTranslationSystemFromFs(userConfig, config);
 				injectRoute({
 					pattern: '404',
 					entryPoint: '@astrojs/starlight/404.astro',
diff --git a/packages/starlight/schemas/i18n.ts b/packages/starlight/schemas/i18n.ts
index 07003412..eb26304d 100644
--- a/packages/starlight/schemas/i18n.ts
+++ b/packages/starlight/schemas/i18n.ts
@@ -3,6 +3,7 @@ import { z } from 'astro/zod';
 export function i18nSchema() {
 	return starlightI18nSchema().merge(pagefindI18nSchema());
 }
+export type i18nSchemaOutput = z.output<ReturnType<typeof i18nSchema>>;
 
 export function builtinI18nSchema() {
 	return starlightI18nSchema().required().strict().merge(pagefindI18nSchema());
diff --git a/packages/starlight/utils/createTranslationSystem.ts b/packages/starlight/utils/createTranslationSystem.ts
new file mode 100644
index 00000000..97a79e3a
--- /dev/null
+++ b/packages/starlight/utils/createTranslationSystem.ts
@@ -0,0 +1,79 @@
+import type { i18nSchemaOutput } from '../schemas/i18n';
+import builtinTranslations from '../translations';
+import type { StarlightConfig } from './user-config';
+
+export function createTranslationSystem(
+	userTranslations: Record<string, i18nSchemaOutput>,
+	config: Pick<StarlightConfig, 'defaultLocale' | 'locales'>
+) {
+	/** User-configured default locale. */
+	const defaultLocale = config.defaultLocale?.locale || 'root';
+
+	/** Default map of UI strings based on Starlight and user-configured defaults. */
+	const defaults = buildDictionary(
+		builtinTranslations.en!,
+		userTranslations.en,
+		builtinTranslations[defaultLocale] || builtinTranslations[stripLangRegion(defaultLocale)],
+		userTranslations[defaultLocale]
+	);
+
+	/**
+	 * Generate a utility function that returns UI strings for the given `locale`.
+	 * @param {string | undefined} [locale]
+	 * @example
+	 * const t = useTranslations('en');
+	 * const label = t('search.label'); // => 'Search'
+	 */
+	return function useTranslations(locale: string | undefined) {
+		const lang = localeToLang(locale, config.locales, config.defaultLocale);
+		const dictionary = buildDictionary(
+			defaults,
+			builtinTranslations[lang] || builtinTranslations[stripLangRegion(lang)],
+			userTranslations[lang]
+		);
+		const t = <K extends keyof typeof dictionary>(key: K) => dictionary[key];
+		t.pick = (startOfKey: string) =>
+			Object.fromEntries(Object.entries(dictionary).filter(([k]) => k.startsWith(startOfKey)));
+		return t;
+	};
+}
+
+/**
+ * Strips the region subtag from a BCP-47 lang string.
+ * @param {string} [lang]
+ * @example
+ * const lang = stripLangRegion('en-GB'); // => 'en'
+ */
+function stripLangRegion(lang: string) {
+	return lang.replace(/-[a-zA-Z]{2}/, '');
+}
+
+/**
+ * Get the BCP-47 language tag for the given locale.
+ * @param locale Locale string or `undefined` for the root locale.
+ */
+function localeToLang(
+	locale: string | undefined,
+	locales: StarlightConfig['locales'],
+	defaultLocale: StarlightConfig['defaultLocale']
+): string {
+	const lang = locale ? locales?.[locale]?.lang : locales?.root?.lang;
+	const defaultLang = defaultLocale?.lang || defaultLocale?.locale;
+	return lang || defaultLang || 'en';
+}
+
+/** Build a dictionary by layering preferred translation sources. */
+function buildDictionary(
+	base: (typeof builtinTranslations)[string],
+	...dictionaries: (i18nSchemaOutput | undefined)[]
+) {
+	const dictionary = { ...base };
+	// Iterate over alternate dictionaries to avoid overwriting preceding values with `undefined`.
+	for (const dict of dictionaries) {
+		for (const key in dict) {
+			const value = dict[key as keyof typeof dict];
+			if (value) dictionary[key as keyof typeof dict] = value;
+		}
+	}
+	return dictionary;
+}
diff --git a/packages/starlight/utils/translations-fs.ts b/packages/starlight/utils/translations-fs.ts
new file mode 100644
index 00000000..d218f19e
--- /dev/null
+++ b/packages/starlight/utils/translations-fs.ts
@@ -0,0 +1,44 @@
+import fs from 'node:fs';
+import type { i18nSchemaOutput } from '../schemas/i18n';
+import { createTranslationSystem } from './createTranslationSystem';
+import type { StarlightConfig } from './user-config';
+import type { AstroConfig } from 'astro';
+
+/**
+ * Loads and creates a translation system from the file system.
+ * Only for use in integration code.
+ * In modules loaded by Vite/Astro, import [`useTranslations`](./translations.ts) instead.
+ *
+ * @see [`./translations.ts`](./translations.ts)
+ */
+export function createTranslationSystemFromFs(
+	opts: Pick<StarlightConfig, 'defaultLocale' | 'locales'>,
+	{ srcDir }: Pick<AstroConfig, 'srcDir'>
+) {
+	/** All translation data from the i18n collection, keyed by `id`, which matches locale. */
+	let userTranslations: Record<string, i18nSchemaOutput> = {};
+	try {
+		const i18nDir = new URL('content/i18n/', srcDir);
+		// Load the user’s i18n directory
+		const files = fs.readdirSync(i18nDir, 'utf-8');
+		// Load the user’s i18n collection and ignore the error if it doesn’t exist.
+		userTranslations = Object.fromEntries(
+			files
+				.filter((file) => file.endsWith('.json'))
+				.map((file) => {
+					const id = file.slice(0, -5);
+					const data = JSON.parse(fs.readFileSync(new URL(file, i18nDir), 'utf-8'));
+					return [id, data] as const;
+				})
+		);
+	} catch (e: unknown) {
+		if (e instanceof Error && 'code' in e && e.code === 'ENOENT') {
+			// i18nDir doesn’t exist, so we ignore the error.
+		} else {
+			// Other errors may be meaningful, e.g. JSON syntax errors, so should be thrown.
+			throw e;
+		}
+	}
+
+	return createTranslationSystem(userTranslations, opts);
+}
diff --git a/packages/starlight/utils/translations.ts b/packages/starlight/utils/translations.ts
index b4c6da4b..1dbc8561 100644
--- a/packages/starlight/utils/translations.ts
+++ b/packages/starlight/utils/translations.ts
@@ -1,13 +1,10 @@
-import { type CollectionEntry, getCollection } from 'astro:content';
+import { getCollection } from 'astro:content';
 import config from 'virtual:starlight/user-config';
-import builtinTranslations from '../translations';
-import { localeToLang } from './slugs';
-
-/** User-configured default locale. */
-const defaultLocale = config.defaultLocale?.locale || 'root';
+import type { i18nSchemaOutput } from '../schemas/i18n';
+import { createTranslationSystem } from './createTranslationSystem';
 
 /** All translation data from the i18n collection, keyed by `id`, which matches locale. */
-let userTranslations: Record<string, CollectionEntry<'i18n'>['data']> = {};
+let userTranslations: Record<string, i18nSchemaOutput> = {};
 try {
 	// Load the user’s i18n collection and ignore the error if it doesn’t exist.
 	userTranslations = Object.fromEntries(
@@ -15,24 +12,6 @@ try {
 	);
 } catch {}
 
-/** Default map of UI strings based on Starlight and user-configured defaults. */
-const defaults = buildDictionary(
-	builtinTranslations.en!,
-	userTranslations.en,
-	builtinTranslations[defaultLocale] || builtinTranslations[stripLangRegion(defaultLocale)],
-	userTranslations[defaultLocale]
-);
-
-/**
- * Strips the region subtag from a BCP-47 lang string.
- * @param {string} [lang]
- * @example
- * const lang = stripLangRegion('en-GB'); // => 'en'
- */
-export function stripLangRegion(lang: string) {
-	return lang.replace(/-[a-zA-Z]{2}/, '');
-}
-
 /**
  * Generate a utility function that returns UI strings for the given `locale`.
  * @param {string | undefined} [locale]
@@ -40,31 +19,4 @@ export function stripLangRegion(lang: string) {
  * const t = useTranslations('en');
  * const label = t('search.label'); // => 'Search'
  */
-export function useTranslations(locale: string | undefined) {
-	const lang = localeToLang(locale);
-	const dictionary = buildDictionary(
-		defaults,
-		builtinTranslations[lang] || builtinTranslations[stripLangRegion(lang)],
-		userTranslations[lang]
-	);
-	const t = <K extends keyof typeof dictionary>(key: K) => dictionary[key];
-	t.pick = (startOfKey: string) =>
-		Object.fromEntries(Object.entries(dictionary).filter(([k]) => k.startsWith(startOfKey)));
-	return t;
-}
-
-/** Build a dictionary by layering preferred translation sources. */
-function buildDictionary(
-	base: (typeof builtinTranslations)[string],
-	...dictionaries: (CollectionEntry<'i18n'>['data'] | undefined)[]
-) {
-	const dictionary = { ...base };
-	// Iterate over alternate dictionaries to avoid overwriting preceding values with `undefined`.
-	for (const dict of dictionaries) {
-		for (const key in dict) {
-			const value = dict[key as keyof typeof dict];
-			if (value) dictionary[key as keyof typeof dict] = value;
-		}
-	}
-	return dictionary;
-}
+export const useTranslations = createTranslationSystem(userTranslations, config);