feat(i18n): add-locale orchestrator (config, flag, ui, docs, cli, readme, bars, site)

One command brings a new locale to every surface, in order: config/i18n.json
entry, flag SVG (lipis/flag-icons), dashboard catalog (scaffold + sync-ui-keys
--translate-markers), docs mirrors (run-translation over the core set every
existing locale carries, then the llm.txt / CHANGELOG.md stubs), CLI catalog
(generate-locales --code + batched common/program translation), README flag
link / docs index row / I18N guide row / locale counts (+ sync-llm-mirrors),
language bars, and the marketing site (lang JSON, SUPPORTED_LANGS, dropdowns).
Every phase checks presence first, so a re-run is a no-op apart from filling in
what is still missing; --dry-run prints the whole plan and touches nothing (no
files, network or child processes); --only / --skip select phases.

- scripts/i18n/lib/docs-core-set.mjs: computeDocsCoreSet — intersection of the
  existing non-docsExcluded mirrors minus llm.txt / CHANGELOG.md / I18N.md and
  paths without an English source (22 files today).
- scripts/i18n/lib/site-scaffold.mjs: addSupportedLang (re-flows the array in
  the file's own fill style) and addDropdownOption (code order, every menu).
- bin/cli/scripts/generate-locales.mjs: --code=<locale> filter.
- translate-backend parseBatchResponse: Object.hasOwn + trimmed values.
- package.json: i18n:add-locale script.
This commit is contained in:
Markus Hartung
2026-09-02 09:10:41 -03:00
parent 29ea2496a6
commit fe28438dbc
8 changed files with 1270 additions and 7 deletions

799
scripts/i18n/add-locale.mjs Normal file
View File

@@ -0,0 +1,799 @@
#!/usr/bin/env node
/**
* OmniRoute — add one locale to every surface with a single command.
*
* npm run i18n:add-locale -- --code=el --english=Greek --native=Ελληνικά --flag=🇬🇷 \
* [--aliases=el-gr] [--flag-file=gr.svg] [--rtl] [--docs=core|all] [--cli-full] [--force-cli] \
* [--site-dir=../omnirouteSite] [--batch-size=40] [--only=<phase,…>] [--skip=<phase,…>] [--dry-run]
*
* Phases, in order. Every phase checks presence first, so re-running the command
* for an already-added locale is a no-op apart from re-translating whatever is
* still missing (`__MISSING__` markers, untranslated CLI / site keys):
*
* config config/i18n.json entry (+ aliases, rtl) — the single source of truth
* flag docs/assets/flags/<cc>.svg from lipis/flag-icons (MIT) when missing
* ui src/i18n/messages/<code>.json — scaffold, then sync-ui-keys --translate-markers
* docs docs/i18n/<code>/** via run-translation — the core set every existing locale
* carries (lib/docs-core-set.mjs; --docs=all for the full source set) — then the
* llm.txt / CHANGELOG.md mirror stubs
* cli bin/cli/locales/<code>.json — generate-locales --code scaffold, then the
* common + program sections (--cli-full: every section) translated in batches
* readme README flag link, docs/i18n/README.md row, docs/guides/I18N.md row, and the
* locale counts in llm.txt (+ sync-llm-mirrors), docs/README.md and
* docs/diagrams/i18n-flow.mmd
* bars sync-language-bars — every 🌐 Languages bar gains the new locale
* site <site-dir>/lang/<code>.json (translated), js/i18n.js SUPPORTED_LANGS and the
* language dropdowns (lib/site-scaffold.mjs); node --check on the edited JS
*
* A real run ends with Prettier on the touched repo files. `--dry-run` prints every
* planned write / command — including the computed docs core set — and touches
* nothing: no files, no network, no child processes.
*
* The translating phases (ui, docs, cli, site) need OMNIROUTE_TRANSLATION_API_URL,
* _API_KEY and _MODEL (docs/guides/I18N.md → "Translation pipeline"); `.env` is loaded
* automatically. Child scripts run through execFileSync with an argument array —
* nothing is ever interpolated into a shell.
*/
import { promises as fs, existsSync, readFileSync } from "node:fs";
import path from "node:path";
import process from "node:process";
import { execFileSync } from "node:child_process";
import { fileURLToPath } from "node:url";
import { computeDocsCoreSet, docsLocaleDirs } from "./lib/docs-core-set.mjs";
import { buildMirrorBar } from "./lib/language-bar.mjs";
import {
bumpCounts,
buildMirrorStub,
flagFileFor,
insertDocsIndexRow,
insertI18nGuideRow,
insertLocaleEntry,
insertReadmeFlagLink,
} from "./lib/locale-scaffold.mjs";
import { addDropdownOption, addSupportedLang } from "./lib/site-scaffold.mjs";
import { backendConfig, translateBatch, translateString } from "./lib/translate-backend.mjs";
const SCRIPT_DIR = path.dirname(fileURLToPath(import.meta.url));
const ROOT = path.resolve(SCRIPT_DIR, "..", "..");
const PHASES = ["config", "flag", "ui", "docs", "cli", "readme", "bars", "site"];
const PHASES_READING_CONFIG_FROM_DISK = ["ui", "docs", "cli", "bars"];
const PHASES_TRANSLATING = ["ui", "docs", "cli", "site"];
const FLAG_CDN = "https://raw.githubusercontent.com/lipis/flag-icons/main/flags/4x3/";
const SITE_PAGES = ["index.html", "why/index.html", "viral/index.html"];
const CLI_DEFAULT_SECTIONS = ["common", "program"];
const MIRROR_STUBS = [
["llm.txt", "OmniRoute"],
["CHANGELOG.md", "Changelog"],
];
const PLACEHOLDER_PREFIX = "__MISSING__:";
const LOCALE_CODE = /^[a-z]{2,3}(-[A-Z][A-Za-z]{1,3})?$/;
const ALIAS = /^[a-z]{2,3}(-[a-z0-9]{2,8})*$/;
const FORBIDDEN_KEYS = new Set(["__proto__", "prototype", "constructor"]);
const DEFAULT_BATCH_SIZE = 40;
const USAGE = `Usage: node scripts/i18n/add-locale.mjs --code=<code> --english=<name> --native=<name> --flag=<emoji> [options]
--code=<code> locale code, e.g. el, pt-PT, zh-TW (required)
--english=<name> English language name, e.g. Greek (required for a new locale)
--native=<name> native language name, e.g. Ελληνικά (required for a new locale)
--flag=<emoji> flag emoji, e.g. 🇬🇷 (required for a new locale)
--aliases=<csv> browser/OS tags resolving to this locale, e.g. el-gr
--flag-file=<file> docs/assets/flags/<file> when it cannot be derived from the emoji
--rtl add the code to config/i18n.json "rtl"
--docs=core|all docs to translate: the core set every locale carries (default) or every source
--cli-full translate every CLI catalog section (default: common + program)
--force-cli retranslate CLI keys that already have a value
--site-dir=<dir> omnirouteSite checkout (relative to the repo root or absolute); skipped when absent
--batch-size=<n> strings per translation request (default ${DEFAULT_BATCH_SIZE})
--only=<phase,…> run only these phases
--skip=<phase,…> skip these phases
--dry-run print every planned write / command and touch nothing
Phases, in order: ${PHASES.join(", ")}
The translating phases (${PHASES_TRANSLATING.join(", ")}) need OMNIROUTE_TRANSLATION_API_URL / _API_KEY / _MODEL
(docs/guides/I18N.md → "Translation pipeline"); .env is loaded automatically.`;
// ----- .env loader ---------------------------------------------------------
// Same semantics as sync-ui-keys.mjs / run-translation.mjs: variables already
// set in the environment win over the file.
function loadDotEnv() {
const envPath = path.join(ROOT, ".env");
if (!existsSync(envPath)) return;
try {
for (const rawLine of readFileSync(envPath, "utf8").split(/\r?\n/)) {
const line = rawLine.trim();
if (!line || line.startsWith("#")) continue;
const eq = line.indexOf("=");
if (eq <= 0) continue;
const key = line.slice(0, eq).trim();
if (!key || process.env[key] !== undefined) continue;
let value = line.slice(eq + 1);
if (
(value.startsWith('"') && value.endsWith('"')) ||
(value.startsWith("'") && value.endsWith("'"))
) {
value = value.slice(1, -1);
}
process.env[key] = value;
}
} catch {
/* ignore — backendConfig() reports the missing variable */
}
}
// ----- CLI -----------------------------------------------------------------
function parsePhaseList(value, flag) {
const set = new Set(
value
.split(",")
.map((s) => s.trim())
.filter(Boolean)
);
for (const phase of set) {
if (!PHASES.includes(phase)) {
throw new Error(`${flag}: unknown phase "${phase}" (known: ${PHASES.join(", ")})`);
}
}
return set;
}
function parseArgs(argv) {
const o = {
code: null,
english: null,
native: null,
flag: null,
aliases: [],
flagFile: null,
rtl: false,
docs: "core",
cliFull: false,
forceCli: false,
siteDir: null,
batchSize: DEFAULT_BATCH_SIZE,
only: null,
skip: new Set(),
dryRun: false,
};
for (const arg of argv.slice(2)) {
const eq = arg.indexOf("=");
const key = eq === -1 ? arg : arg.slice(0, eq);
const value = eq === -1 ? "" : arg.slice(eq + 1);
switch (key) {
case "--help":
case "-h":
console.log(USAGE);
process.exit(0);
break;
case "--code":
o.code = value;
break;
case "--english":
o.english = value;
break;
case "--native":
o.native = value;
break;
case "--flag":
o.flag = value;
break;
case "--aliases":
o.aliases = value
.split(",")
.map((s) => s.trim().toLowerCase())
.filter(Boolean);
break;
case "--flag-file":
o.flagFile = value;
break;
case "--rtl":
o.rtl = true;
break;
case "--docs":
o.docs = value;
break;
case "--cli-full":
o.cliFull = true;
break;
case "--force-cli":
o.forceCli = true;
break;
case "--site-dir":
o.siteDir = path.resolve(ROOT, value);
break;
case "--batch-size":
// Whole numbers only; NaN / 0 / negatives fall back to the default.
o.batchSize = Math.max(1, Math.floor(Number(value)) || DEFAULT_BATCH_SIZE);
break;
case "--only":
o.only = parsePhaseList(value, "--only");
break;
case "--skip":
o.skip = parsePhaseList(value, "--skip");
break;
case "--dry-run":
o.dryRun = true;
break;
default:
throw new Error(`unknown argument ${arg}\n\n${USAGE}`);
}
}
if (!o.code) throw new Error(`--code is required\n\n${USAGE}`);
if (!LOCALE_CODE.test(o.code)) {
throw new Error(`invalid locale code "${o.code}" (expected e.g. el, pt-PT, zh-TW)`);
}
if (!["core", "all"].includes(o.docs)) {
throw new Error(`--docs must be "core" or "all" (got "${o.docs}")`);
}
for (const alias of o.aliases) {
if (!ALIAS.test(alias))
throw new Error(`invalid alias "${alias}" (lower-case BCP-47, e.g. el-gr)`);
}
return o;
}
function selectPhases(o) {
return PHASES.filter((phase) => (!o.only || o.only.has(phase)) && !o.skip.has(phase));
}
// ----- Helpers -------------------------------------------------------------
const log = (...parts) => console.log("[add-locale]", ...parts);
const warn = (...parts) => console.warn("[add-locale] WARN", ...parts);
const dryTag = (ctx) => (ctx.dry ? "[DRY] " : "");
function insideRepo(file) {
const relative = path.relative(ROOT, file);
return Boolean(relative) && !relative.startsWith("..") && !path.isAbsolute(relative);
}
/** Repo-relative POSIX path for files inside the repo, absolute otherwise (site files). */
function display(file) {
return insideRepo(file) ? path.relative(ROOT, file).split(path.sep).join("/") : file;
}
const readText = (file) => fs.readFile(file, "utf8");
const readJson = async (file) => JSON.parse(await readText(file));
async function writeFile(ctx, file, text) {
log(`${dryTag(ctx)}write ${display(file)}`);
ctx.touched.add(file);
if (ctx.dry) return;
await fs.mkdir(path.dirname(file), { recursive: true });
await fs.writeFile(file, text, "utf8");
}
function runNode(ctx, script, args = [], note = "") {
const command = ["node", display(script), ...args].join(" ");
log(`${dryTag(ctx)}${command}${note ? `${note}` : ""}`);
if (ctx.dry) return;
execFileSync(process.execPath, [script, ...args], { cwd: ROOT, stdio: "inherit" });
}
/** `{ id: "a.b.c", text }` for every string leaf of a nested catalog. */
function flattenLeaves(node, prefix = "", out = []) {
if (typeof node === "string") {
out.push({ id: prefix, text: node });
} else if (node && typeof node === "object" && !Array.isArray(node)) {
for (const [key, value] of Object.entries(node)) {
if (FORBIDDEN_KEYS.has(key)) continue;
flattenLeaves(value, prefix ? `${prefix}.${key}` : key, out);
}
}
return out;
}
function getDeep(node, id) {
let cursor = node;
for (const segment of id.split(".")) {
if (!cursor || typeof cursor !== "object") return undefined;
cursor = cursor[segment];
}
return cursor;
}
function setDeep(node, id, value) {
const segments = id.split(".");
if (segments.some((segment) => FORBIDDEN_KEYS.has(segment))) {
throw new Error(`refusing to write key ${id}`);
}
let cursor = node;
for (const segment of segments.slice(0, -1)) {
if (!cursor[segment] || typeof cursor[segment] !== "object") cursor[segment] = {};
cursor = cursor[segment];
}
cursor[segments[segments.length - 1]] = value;
}
async function walkFiles(dir, out = []) {
for (const entry of await fs.readdir(dir, { withFileTypes: true })) {
const abs = path.join(dir, entry.name);
if (entry.isDirectory()) await walkFiles(abs, out);
else out.push(abs);
}
return out;
}
/** Files under docs/ that carry a 🌐 Languages bar — what sync-language-bars rewrites. */
async function countLanguageBarFiles() {
let count = 0;
for (const file of await walkFiles(path.join(ROOT, "docs"))) {
if (!file.endsWith(".md") && path.basename(file) !== "llm.txt") continue;
const text = await readText(file);
if (text.split("\n").some((line) => line.startsWith("🌐 **Languages:**"))) count += 1;
}
return count;
}
/**
* Translates `entries` (`{ id, text }`) in batches of `--batch-size`; a batch
* that fails or cannot be parsed is retried one string at a time. Strings that
* still fail are left out of the result and recorded in `ctx.failures`, so a
* later run fills them in (the run exits 1 to flag them).
*/
async function translateEntries(ctx, entries, label) {
ctx.backend ??= backendConfig();
const out = new Map();
const size = ctx.o.batchSize;
const batches = Math.ceil(entries.length / size);
for (let i = 0; i < entries.length; i += size) {
const chunk = entries.slice(i, i + size);
try {
const translated = await translateBatch(chunk, ctx.entry, ctx.backend);
for (const { id } of chunk) out.set(id, translated.get(id));
} catch (err) {
warn(
`${label}: batch ${i / size + 1}/${batches} failed (${err.message}) — retrying one by one`
);
for (const { id, text } of chunk) {
try {
const value = await translateString(text, ctx.entry, ctx.backend);
if (!value) throw new Error("empty translation");
out.set(id, value);
} catch (inner) {
ctx.failures.push(`${label}: ${id} (${inner.message})`);
}
}
}
log(`${label}: ${Math.min(i + size, entries.length)}/${entries.length} strings translated`);
}
return out;
}
// ----- Phases --------------------------------------------------------------
async function phaseConfig(ctx) {
if (ctx.nextConfigText === ctx.configText) {
log(`config: ${ctx.code} already configured — nothing to change`);
return;
}
const changes = [];
if (!ctx.stored) changes.push(`add ${ctx.code} entry`);
if (ctx.o.rtl && !ctx.before.rtl.includes(ctx.code)) changes.push("add to rtl");
log(`config: ${changes.join(", ")} (total locales → ${ctx.total})`);
await writeFile(ctx, ctx.configPath, ctx.nextConfigText);
}
async function phaseFlag(ctx) {
const file = path.join(ROOT, "docs", "assets", "flags", ctx.flagFile);
if (existsSync(file)) {
log(`flag: ${display(file)} present`);
return;
}
const url = `${FLAG_CDN}${ctx.flagFile}`;
log(`${dryTag(ctx)}fetch ${url}${display(file)}`);
if (ctx.dry) return;
const res = await fetch(url);
if (!res.ok) {
throw new Error(
`flag download failed (${res.status}) for ${url} — place docs/assets/flags/${ctx.flagFile} manually (or pass --flag-file) and re-run`
);
}
const svg = await res.text();
if (!svg.trimStart().startsWith("<svg")) throw new Error(`flag download is not an SVG: ${url}`);
await writeFile(ctx, file, svg);
}
async function phaseUi(ctx) {
const dir = path.join(ROOT, "src", "i18n", "messages");
const file = path.join(dir, `${ctx.code}.json`);
const source = flattenLeaves(await readJson(path.join(dir, "en.json")));
const existing = existsSync(file) ? flattenLeaves(await readJson(file)) : null;
// sync-ui-keys only fills locales that already exist on disk.
if (!existing) await writeFile(ctx, file, "{}\n");
const have = new Set((existing ?? []).map((leaf) => leaf.id));
const missing = source.filter((leaf) => !have.has(leaf.id)).length;
const markers = (existing ?? []).filter((leaf) =>
leaf.text.startsWith(PLACEHOLDER_PREFIX)
).length;
log(
`ui: ${source.length} source keys — ${missing} missing, ${markers} __MISSING__ markers → ${missing + markers} to translate`
);
if (missing + markers === 0) {
log(`ui: ${display(file)} already complete`);
return;
}
runNode(
ctx,
path.join(SCRIPT_DIR, "sync-ui-keys.mjs"),
[`--locale=${ctx.code}`, "--translate-markers", `--batch-size=${ctx.o.batchSize}`],
`writes ${display(file)}`
);
ctx.touched.add(file);
}
async function phaseDocs(ctx) {
const localeDir = path.join(ROOT, "docs", "i18n", ctx.code);
let files = null;
if (ctx.o.docs === "core") {
// The target locale is left out of the intersection: a partial earlier run
// of the same locale must not shrink the set it is being caught up to.
const peers = {
...ctx.config,
locales: ctx.config.locales.filter((locale) => locale.code !== ctx.code),
};
files = computeDocsCoreSet({ root: ROOT, config: peers });
if (files.length === 0) {
throw new Error(
"docs: no existing locale mirror to derive the core set from — use --docs=all"
);
}
log(
`docs: core set = ${files.length} files (carried by every one of the ${docsLocaleDirs({ root: ROOT, config: peers }).length} existing locale mirrors)`
);
if (ctx.dry) {
for (const rel of files) {
const target = path.join(localeDir, rel);
log(`[DRY] write ${display(target)} (via run-translation)`);
ctx.touched.add(target);
}
}
} else {
log("docs: full source set (--docs=all)");
}
runNode(
ctx,
path.join(SCRIPT_DIR, "run-translation.mjs"),
[`--locale=${ctx.code}`, ...(files ? [`--files=${files.join(",")}`] : [])],
`writes docs/i18n/${ctx.code}/**`
);
if (!ctx.dry && !existsSync(localeDir)) {
warn(`docs: ${display(localeDir)} was not created — skipping the llm.txt / CHANGELOG.md stubs`);
return;
}
for (const [fileName, heading] of MIRROR_STUBS) {
const target = path.join(localeDir, fileName);
if (existsSync(target)) {
log(`docs: ${display(target)} present`);
continue;
}
const body = (await readText(path.join(ROOT, fileName))).replace(/^# .+\r?\n+/, "");
const stub = buildMirrorStub({
heading,
native: ctx.entry.native ?? ctx.entry.name,
bar: buildMirrorBar(fileName, ctx.code, ctx.config),
body,
});
await writeFile(ctx, target, stub);
}
}
async function phaseCli(ctx) {
const dir = path.join(ROOT, "bin", "cli", "locales");
const catalogPath = path.join(dir, `${ctx.code}.json`);
const en = await readJson(path.join(dir, "en.json"));
const sections = ctx.o.cliFull
? Object.keys(en)
: CLI_DEFAULT_SECTIONS.filter((section) => section in en);
if (existsSync(catalogPath)) {
log(`cli: ${display(catalogPath)} present`);
} else {
runNode(
ctx,
path.join(ROOT, "bin", "cli", "scripts", "generate-locales.mjs"),
[`--code=${ctx.code}`],
`scaffolds ${display(catalogPath)}`
);
}
const catalog = existsSync(catalogPath) ? await readJson(catalogPath) : {};
const source = flattenLeaves(
Object.fromEntries(sections.map((section) => [section, en[section]]))
);
const pending = ctx.o.forceCli
? source
: source.filter(({ id }) => typeof getDeep(catalog, id) !== "string");
const scope = ctx.o.cliFull ? `every section (${sections.length})` : sections.join(" + ");
log(
`cli: ${source.length} keys in ${scope}${pending.length} to translate${ctx.o.forceCli ? " (--force-cli)" : ""}`
);
if (pending.length === 0) {
log(`cli: ${display(catalogPath)} already translated`);
return;
}
if (ctx.dry) {
log(`[DRY] write ${display(catalogPath)}`);
ctx.touched.add(catalogPath);
return;
}
const translated = await translateEntries(ctx, pending, "cli");
for (const [id, text] of translated) setDeep(catalog, id, text);
await writeFile(ctx, catalogPath, JSON.stringify(catalog, null, 2) + "\n");
}
async function phaseReadme(ctx) {
const { entry, total, code } = ctx;
const guideRow = new RegExp(`^\\| \`${code}\` +\\|`, "m");
const edits = [
[
"README.md",
(t) =>
t.includes(`href="docs/i18n/${code}/README.md"`)
? t
: insertReadmeFlagLink(t, entry, total),
],
[
"docs/i18n/README.md",
(t) => (t.includes(`(\`${code}\`)`) ? t : insertDocsIndexRow(t, entry, total)),
],
[
"docs/guides/I18N.md",
(t) => (guideRow.test(t) ? t : insertI18nGuideRow(t, entry, total, ctx.config.rtl)),
],
["llm.txt", (t) => bumpCounts(t, total)],
[
"docs/README.md",
(t) =>
t.replace(
/in \d+ locales \(plus the English originals — \d+ languages in total\)/,
`in ${total - 1} locales (plus the English originals — ${total} languages in total)`
),
],
["docs/diagrams/i18n-flow.mmd", (t) => t.replace(/\(\d+ langs\)/, `(${total} langs)`)],
];
let llmChanged = false;
for (const [rel, transform] of edits) {
const file = path.join(ROOT, rel);
const text = await readText(file);
const next = transform(text);
if (next === text) {
log(`readme: ${rel} up to date`);
continue;
}
await writeFile(ctx, file, next);
if (rel === "llm.txt") llmChanged = true;
}
// llm.txt mirrors are strict copies of the root body (check-docs-sync).
if (llmChanged) {
runNode(ctx, path.join(SCRIPT_DIR, "sync-llm-mirrors.mjs"), [], "re-syncs docs/i18n/*/llm.txt");
}
}
async function phaseBars(ctx) {
const note = `rewrites the 🌐 Languages bar of every file under docs/ that carries one (${await countLanguageBarFiles()} today) — adds ${ctx.entry.flag} [${ctx.code}]`;
runNode(ctx, path.join(SCRIPT_DIR, "sync-language-bars.mjs"), [], note);
}
async function phaseSite(ctx) {
const siteDir = ctx.o.siteDir;
if (!siteDir) {
log("site: --site-dir not given — skipped");
return;
}
if (!existsSync(siteDir)) {
warn(`site: ${siteDir} does not exist — skipping the site phase`);
return;
}
const sourcePath = path.join(siteDir, "lang", "_source.en.json");
if (!existsSync(sourcePath)) {
throw new Error(`site: ${sourcePath} not found — is --site-dir the omnirouteSite checkout?`);
}
// lang/<code>.json — flat keys; only the ones without a translation yet.
const source = await readJson(sourcePath);
const langPath = path.join(siteDir, "lang", `${ctx.code}.json`);
const existing = existsSync(langPath) ? await readJson(langPath) : {};
const entries = Object.entries(source)
.filter(([id, text]) => typeof text === "string" && typeof existing[id] !== "string")
.map(([id, text]) => ({ id, text }));
log(
`site: ${Object.keys(source).length} keys in lang/_source.en.json — ${entries.length} to translate`
);
if (entries.length === 0) {
log(`site: ${langPath} already complete`);
} else if (ctx.dry) {
log(`[DRY] write ${langPath}`);
} else {
const translated = await translateEntries(ctx, entries, "site");
const out = { ...existing };
for (const [id, text] of translated) out[id] = text;
await writeFile(ctx, langPath, JSON.stringify(out, null, 2) + "\n");
}
// js/i18n.js — SUPPORTED_LANGS.
const jsPath = path.join(siteDir, "js", "i18n.js");
const js = await readText(jsPath);
const nextJs = addSupportedLang(js, ctx.code);
if (nextJs === js) {
log(`site: js/i18n.js already lists ${ctx.code}`);
} else {
await writeFile(ctx, jsPath, nextJs);
log(`${dryTag(ctx)}node --check ${jsPath}`);
if (!ctx.dry) execFileSync(process.execPath, ["--check", jsPath], { stdio: "inherit" });
}
// Language dropdowns.
for (const page of SITE_PAGES) {
const file = path.join(siteDir, page);
if (!existsSync(file)) {
warn(`site: ${page} not found — skipped`);
continue;
}
const html = await readText(file);
const next = addDropdownOption(html, {
code: ctx.code,
flag: ctx.entry.flag,
native: ctx.entry.native ?? ctx.entry.name,
});
if (next === html) log(`site: ${page} already lists ${ctx.code}`);
else await writeFile(ctx, file, next);
}
}
const PHASE_RUNNERS = {
config: phaseConfig,
flag: phaseFlag,
ui: phaseUi,
docs: phaseDocs,
cli: phaseCli,
readme: phaseReadme,
bars: phaseBars,
site: phaseSite,
};
async function runPrettier(ctx, phases) {
// Repo files only (the site has its own tooling) and only the kinds Prettier
// has a parser for — llm.txt and the .mmd diagram would make it exit 2.
const files = new Set(
[...ctx.touched].filter((file) => insideRepo(file) && /\.(json|md)$/.test(file))
);
const localeDir = path.join(ROOT, "docs", "i18n", ctx.code);
if (!ctx.dry && phases.includes("docs") && existsSync(localeDir)) {
for (const file of await walkFiles(localeDir)) if (file.endsWith(".md")) files.add(file);
}
if (files.size === 0) return;
const bin = path.join(ROOT, "node_modules", "prettier", "bin", "prettier.cjs");
if (!existsSync(bin)) {
warn("prettier is not installed (node_modules) — skipping the formatting pass");
return;
}
const list = [...files].sort();
log(`${dryTag(ctx)}prettier --write ${list.length} file(s): ${list.map(display).join(" ")}`);
if (ctx.dry) return;
execFileSync(process.execPath, [bin, "--write", "--log-level=warn", ...list], {
cwd: ROOT,
stdio: "inherit",
});
}
// ----- Main ----------------------------------------------------------------
async function main() {
loadDotEnv();
const o = parseArgs(process.argv);
const phases = selectPhases(o);
if (phases.length === 0) throw new Error("no phase left to run (check --only / --skip)");
const configPath = path.join(ROOT, "config", "i18n.json");
const configText = await readText(configPath);
const before = JSON.parse(configText);
const stored = before.locales.find((locale) => locale.code === o.code) ?? null;
if (!stored) {
for (const required of ["english", "native", "flag"]) {
if (!o[required]) {
throw new Error(
`--${required} is required (${o.code} is not in config/i18n.json yet)\n\n${USAGE}`
);
}
}
} else {
for (const field of ["english", "native", "flag"]) {
if (o[field] && o[field] !== stored[field]) {
warn(
`--${field}=${o[field]} differs from config/i18n.json (${stored[field]}) — the stored value wins`
);
}
}
}
// The working entry: the stored one for a known locale, else built from the
// arguments. `flagFile` is a helper key that is never written to the config.
const entry = stored
? { ...stored, ...(o.flagFile ? { flagFile: o.flagFile } : {}) }
: {
code: o.code,
label: o.code.toUpperCase(),
name: o.native,
native: o.native,
english: o.english,
flag: o.flag,
...(o.aliases.length ? { aliases: o.aliases } : {}),
...(o.flagFile ? { flagFile: o.flagFile } : {}),
};
const flagFile = flagFileFor(entry); // fails fast for a flag the file name cannot be derived from
// config/i18n.json after this run (in memory; the config phase writes it).
let nextConfigText = stored ? configText : insertLocaleEntry(configText, entry);
if (o.rtl) {
const cfg = JSON.parse(nextConfigText);
if (!cfg.rtl.includes(o.code)) {
cfg.rtl = [...cfg.rtl, o.code].sort();
nextConfigText = JSON.stringify(cfg, null, 2) + "\n";
}
}
const config = JSON.parse(nextConfigText);
const total = config.locales.length;
const ctx = {
o,
dry: o.dryRun,
code: o.code,
entry,
flagFile,
stored,
before,
config,
total,
configPath,
configText,
nextConfigText,
touched: new Set(),
failures: [],
backend: null,
};
if (!o.dryRun) {
if (!stored && !phases.includes("config")) {
const needing = phases.filter((phase) => PHASES_READING_CONFIG_FROM_DISK.includes(phase));
if (needing.length) {
throw new Error(
`config/i18n.json does not list ${o.code} yet and the ${needing.join(", ")} phase(s) read it from disk — run the config phase first`
);
}
}
// Fail before the first write when a translating phase has no backend.
if (phases.some((phase) => PHASES_TRANSLATING.includes(phase))) ctx.backend = backendConfig();
}
log(
`${dryTag(ctx)}${o.code} (${entry.english ?? entry.name}, ${entry.flag}) — phases: ${phases.join(", ")} — locales after: ${total}`
);
for (const phase of phases) {
log(`--- ${phase} ---`);
await PHASE_RUNNERS[phase](ctx);
}
log("--- prettier ---");
await runPrettier(ctx, phases);
if (ctx.failures.length) {
warn(
`${ctx.failures.length} string(s) could not be translated — re-run the command to fill them in:`
);
for (const failure of ctx.failures) console.warn(` - ${failure}`);
process.exitCode = 1;
}
log(
`${dryTag(ctx)}done: ${o.code} (${entry.english ?? entry.name}) — total locales now ${total}`
);
log(
"next: node --import tsx/esm --test tests/unit/i18n-locale-surfaces-parity.test.ts && npm run i18n:check-ui-coverage && npm run check:docs-all"
);
}
main().catch((err) => {
console.error("[add-locale] ERROR", err?.stack || err?.message || String(err));
process.exit(1);
});

View File

@@ -0,0 +1,58 @@
/**
* The docs "core set": the source documents every existing locale mirror carries.
*
* The translation pipeline (`scripts/i18n/run-translation.mjs`) knows ~150
* source files, but only a core of them is translated in every locale under
* `docs/i18n/`. A new locale should reach parity with its peers rather than
* translate the full set, so the core is derived from the tree itself — the
* intersection of the relative paths present under every existing,
* non-`docsExcluded` locale directory — instead of being hard-coded.
*
* docsLocaleDirs({ root, config }) → ["ar", "az", …] locale codes whose
* docs/i18n/<code>/ directory exists
* computeDocsCoreSet({ root, config }) → sorted repo-relative source paths
*
* Paths that are not translation targets are dropped from the result: the
* strict-mirror stubs (`llm.txt`, `CHANGELOG.md`), the operator-only
* `docs/guides/I18N.md`, and any mirror whose English source no longer exists
* at `<root>/<path>`. Synchronous and filesystem-only — no network, no writes.
*/
import { existsSync, readdirSync, statSync } from "node:fs";
import path from "node:path";
export const CORE_SET_EXCLUDED = ["llm.txt", "CHANGELOG.md", "docs/guides/I18N.md"];
function walkFiles(dir, base = dir, out = []) {
for (const entry of readdirSync(dir, { withFileTypes: true })) {
const abs = path.join(dir, entry.name);
if (entry.isDirectory()) walkFiles(abs, base, out);
else if (entry.isFile()) out.push(path.relative(base, abs).split(path.sep).join("/"));
}
return out;
}
function isDirectory(abs) {
return existsSync(abs) && statSync(abs).isDirectory();
}
function isFile(abs) {
return existsSync(abs) && statSync(abs).isFile();
}
export function docsLocaleDirs({ root, config }) {
const excluded = new Set(config.docsExcluded ?? ["en"]);
return config.locales
.map((locale) => locale.code)
.filter((code) => !excluded.has(code) && isDirectory(path.join(root, "docs", "i18n", code)));
}
export function computeDocsCoreSet({ root, config }) {
let shared = null;
for (const code of docsLocaleDirs({ root, config })) {
const files = new Set(walkFiles(path.join(root, "docs", "i18n", code)));
shared = shared === null ? files : new Set([...shared].filter((rel) => files.has(rel)));
}
if (shared === null) return [];
const dropped = new Set(CORE_SET_EXCLUDED);
return [...shared].filter((rel) => !dropped.has(rel) && isFile(path.join(root, rel))).sort();
}

View File

@@ -0,0 +1,105 @@
/**
* Pure text helpers for the marketing site (omnirouteSite — a separate repo
* mirrored next to this one; `scripts/i18n/add-locale.mjs --site-dir=…` points
* at it). Text in, text out — no filesystem.
*
* addSupportedLang(jsSource, code) js/i18n.js
* Adds `code` to the `const SUPPORTED_LANGS = [ … ];` literal, keeps the
* codes in `localeCompare(…, "en")` order and re-flows the literal in the
* file's own style: a multi-line array is filled line by line up to its
* existing width (80 columns in the site file), a single-line array stays
* on one line. Returns the input unchanged when the code is already listed.
*
* addDropdownOption(html, { code, flag, native }) index.html and friends
* Inserts
* <a href="#" class="lang-option" data-lang="<code>" role="menuitem"><flag> <native></a>
* into every language menu of the page — a menu being a contiguous run of
* `.lang-option` lines — in `data-lang` code order, with the indentation of
* its neighbours. Returns the input unchanged when every menu lists the code.
*
* Both throw when the anchor they edit cannot be found.
*/
const SUPPORTED_LANGS_RE = /(const SUPPORTED_LANGS = \[)([\s\S]*?)(\];)/;
const OPTION_LINE_RE = /^([ \t]*)<a href="#" class="lang-option(?: [^"]*)?" data-lang="([^"]+)"/;
const MIN_FILL_WIDTH = 80;
function compareCodes(a, b) {
return a.localeCompare(b, "en");
}
function fillLines(items, indent, width) {
const lines = [];
let current = "";
for (const item of items) {
const token = `${item},`;
const candidate = current ? `${current} ${token}` : `${indent}${token}`;
if (current && candidate.length > width) {
lines.push(current);
current = `${indent}${token}`;
} else {
current = candidate;
}
}
if (current) lines.push(current);
return lines;
}
export function addSupportedLang(jsSource, code) {
const match = jsSource.match(SUPPORTED_LANGS_RE);
if (!match) throw new Error("SUPPORTED_LANGS array literal not found");
const list = match[2];
const quote = list.match(/["']/)?.[0] ?? '"';
const codes = [...list.matchAll(/["']([^"']+)["']/g)].map((m) => m[1]);
if (codes.includes(code)) return jsSource;
codes.push(code);
codes.sort(compareCodes);
const items = codes.map((c) => `${quote}${c}${quote}`);
let body;
if (list.includes("\n")) {
const indent = list.match(/\n([ \t]*)\S/)?.[1] ?? " ";
const width = Math.max(MIN_FILL_WIDTH, ...list.split("\n").map((line) => line.length));
body = `\n${fillLines(items, indent, width).join("\n")}\n`;
} else {
body = items.join(", ");
}
return jsSource.replace(
SUPPORTED_LANGS_RE,
(_whole, open, _list, close) => `${open}${body}${close}`
);
}
export function addDropdownOption(html, { code, flag, native }) {
const eol = html.includes("\r\n") ? "\r\n" : "\n";
const lines = html.split(eol);
const options = [];
lines.forEach((line, index) => {
const m = line.match(OPTION_LINE_RE);
if (m) options.push({ index, indent: m[1], code: m[2] });
});
if (options.length === 0) throw new Error("no .lang-option lines found (language menu missing)");
// One menu = one contiguous run of option lines.
const menus = [];
for (const option of options) {
const menu = menus[menus.length - 1];
if (menu && option.index === menu[menu.length - 1].index + 1) menu.push(option);
else menus.push([option]);
}
// Insert bottom-up so the indexes of the menus above stay valid.
for (const menu of [...menus].reverse()) {
if (menu.some((option) => option.code === code)) continue;
const next = menu.find((option) => compareCodes(option.code, code) > 0);
const last = menu[menu.length - 1];
const at = next ? next.index : last.index + 1;
const indent = (next ?? last).indent;
lines.splice(
at,
0,
`${indent}<a href="#" class="lang-option" data-lang="${code}" role="menuitem">${flag} ${native}</a>`
);
}
return lines.join(eol);
}

View File

@@ -143,10 +143,11 @@ export const BATCH_SYSTEM = (englishName, native) =>
/**
* Pure parser for a batch answer. Accepts a bare JSON object or one wrapped in
* a ```json fence; anything else (prose around the object, arrays, invalid
* JSON) throws. Every id in `expectedIds` must be present with a non-empty
* JSON) throws. Every id in `expectedIds` must be present as an OWN property
* (an inherited name such as `constructor` counts as missing) with a non-empty
* string value — an empty translation would replace the `__MISSING__` marker
* for good, so it fails the batch instead (the per-string path rejects empty
* completions the same way).
* completions the same way). Values are trimmed, like `translateString` does.
*
* @param {string} text raw assistant content
* @param {string[]} expectedIds ids the caller sent (and expects back)
@@ -169,12 +170,13 @@ export function parseBatchResponse(text, expectedIds) {
}
const out = new Map();
for (const id of expectedIds) {
if (!(id in parsed)) throw new Error(`batch response missing id ${id}`);
if (!Object.hasOwn(parsed, id)) throw new Error(`batch response missing id ${id}`);
if (typeof parsed[id] !== "string") {
throw new Error(`batch response has non-string value for ${id}`);
}
if (!parsed[id].trim()) throw new Error(`batch response has empty value for ${id}`);
out.set(id, parsed[id]);
const value = parsed[id].trim();
if (!value) throw new Error(`batch response has empty value for ${id}`);
out.set(id, value);
}
return out;
}