homepage/scripts/split-i18n.ts

/**
 * Split a single-file i18n module (with an object literal whose values are
 * `Record<locale, string>`) into per-locale files under
 * src/lib/i18n/<namespace>/<locale>.ts.
 *
 * The first locale is the source of truth; others use `as const satisfies
 * Record<keyof typeof <first>, string>` so missing translations fail
 * type-checking.
 *
 * Run: pnpm exec vite-node scripts/split-i18n.ts <source> <namespace> <locales,csv> [--marker=<marker>] [--basename=<name>]
 *   e.g. ... cospendI18n.ts cospend de,en
 *        ... calendarI18n.ts calendar de,en,la --marker='export const ui = {' --basename=de
 *
 * Defaults: marker = `const translations: Translations = {`, basename = first locale.
 */

import { readFileSync, writeFileSync, mkdirSync } from 'node:fs';

const [, , srcPath, namespace, localesCsv, ...flags] = process.argv;
if (!srcPath || !namespace || !localesCsv) {
	console.error(
		'usage: split-i18n.ts <source> <namespace> <locales,csv> [--marker=...] [--basename=...]'
	);
	process.exit(1);
}
const locales = localesCsv.split(',').map((s) => s.trim()).filter(Boolean);
const markerFlag = flags.find((f) => f.startsWith('--marker='));
const startMarker = markerFlag
	? markerFlag.slice('--marker='.length)
	: 'const translations: Translations = {';
const basenameFlag = flags.find((f) => f.startsWith('--basename='));
const fileBase = basenameFlag ? basenameFlag.slice('--basename='.length) : '';

const src = readFileSync(srcPath, 'utf8');

// Slice the translations object body
const startIdx = src.indexOf(startMarker);
if (startIdx === -1) throw new Error(`marker not found in ${srcPath}: ${startMarker}`);
// Object literal can close with `};` or `} as const;` — pick the earliest match.
const candA = src.indexOf('\n};', startIdx);
const candB = src.indexOf('\n} as const', startIdx);
const endIdx =
	candA < 0 ? candB : candB < 0 ? candA : Math.min(candA, candB);
if (endIdx === -1) throw new Error('translations object end not found');
const body = src.slice(startIdx + startMarker.length, endIdx);

// Match each translation entry boundary: `key: { ...inner... },`. Each
// entry's body is then parsed independently for `loc: 'value'` pairs, so
// locale order in the source file doesn't matter.
const entryRe = /^\s*(\w+)\s*:\s*\{([\s\S]*?)\}\s*,?\s*$/gm;
// Match `loc: '...'` OR `loc: "..."` (double quotes are used when the string
// contains a literal apostrophe).
const localeRe = /(\w+)\s*:\s*(?:'([^']*)'|"((?:\\.|[^"\\])*)")/g;

function decodeJsString(raw: string, doubleQuoted: boolean): string {
	if (doubleQuoted) {
		// Already valid JSON (escapes preserved). Parse directly.
		return JSON.parse('"' + raw + '"');
	}
	// Single-quoted: convert any \' → ' and escape literal " for JSON.
	const jsonReady = '"' + raw.replace(/\\'/g, "'").replace(/"/g, '\\"') + '"';
	return JSON.parse(jsonReady);
}

interface Entry {
	key: string;
	values: Record<string, string>;
}

const entries: Entry[] = [];
let m: RegExpExecArray | null;
while ((m = entryRe.exec(body)) !== null) {
	const inner = m[2];
	const values: Record<string, string> = {};
	let lm: RegExpExecArray | null;
	while ((lm = localeRe.exec(inner)) !== null) {
		const single = lm[2];
		const double = lm[3];
		values[lm[1]] = single !== undefined
			? decodeJsString(single, false)
			: decodeJsString(double, true);
	}
	for (const loc of locales) {
		if (!(loc in values)) {
			throw new Error(`entry "${m[1]}" is missing locale "${loc}"`);
		}
	}
	entries.push({ key: m[1], values });
}
console.log(`extracted ${entries.length} entries`);

const outDir = `src/lib/i18n/${namespace}`;
mkdirSync(outDir, { recursive: true });

const sourceLocale = locales[0];

// Optional file prefix lets us split multiple tables into the same dir
// (e.g. calendar `ui` → de.ts, calendar `ui1962` → de_1962.ts).
const path = (loc: string) => `${outDir}/${fileBase ? `${loc}_${fileBase}` : loc}.ts`;

// Write the source-of-truth locale (no satisfies clause).
{
	const lines = [
		'/** Generated by scripts/split-i18n.ts. */',
		`/** ${sourceLocale.toUpperCase()} ${namespace}${fileBase ? ` (${fileBase})` : ''} UI strings — source of truth for the key set. */`,
		'',
		`export const ${sourceLocale} = {`
	];
	for (const e of entries) lines.push(`\t${e.key}: ${JSON.stringify(e.values[sourceLocale])},`);
	lines.push('} as const;', '');
	writeFileSync(path(sourceLocale), lines.join('\n'));
}

// Write the other locales with `satisfies` constraint.
const sourceFile = fileBase ? `${sourceLocale}_${fileBase}` : sourceLocale;
for (let i = 1; i < locales.length; i++) {
	const loc = locales[i];
	const lines = [
		'/** Generated by scripts/split-i18n.ts. */',
		`import type { ${sourceLocale} } from './${sourceFile}';`,
		'',
		`export const ${loc} = {`
	];
	for (const e of entries) lines.push(`\t${e.key}: ${JSON.stringify(e.values[loc])},`);
	lines.push(
		`} as const satisfies Record<keyof typeof ${sourceLocale}, string>;`,
		''
	);
	writeFileSync(path(loc), lines.join('\n'));
}

console.log(`wrote ${locales.map(path).join(', ')}`);