u-claw/tests/no-china-sources.test.mjs

import { readFileSync, readdirSync, statSync } from 'node:fs';
import { fileURLToPath } from 'node:url';
import { join, relative, extname } from 'node:path';
import test from 'node:test';
import assert from 'node:assert/strict';

const repoRoot = fileURLToPath(new URL('..', import.meta.url));

// Overseas build: every download source must be reachable without routing through
// China. These patterns are slower from Singapore, and the GitHub proxies put an
// unaccountable third party in the middle of the supply chain.
const FORBIDDEN = [
  { pattern: /npmmirror\.com/, use: 'registry.npmjs.org / nodejs.org' },
  { pattern: /mirrors\.(tuna\.tsinghua|ustc|aliyun|cloud\.tencent|huaweicloud)/, use: 'the upstream project mirror' },
  { pattern: /ghfast\.top|ghproxy\.net|gh\.idayer\.com/, use: 'github.com directly' },
];

// Executable surfaces only. Docs and the Config model list are handled by the
// content pass (they need rewriting, not a URL swap).
const SCANNED_EXTENSIONS = new Set(['.sh', '.ps1', '.bat', '.command', '.yml', '.yaml', '.mjs', '.js']);
const SKIPPED_DIRS = new Set(['.git', 'node_modules', 'dist', '.download-cache']);

function* walk(dir) {
  for (const entry of readdirSync(dir)) {
    if (SKIPPED_DIRS.has(entry)) continue;
    const full = join(dir, entry);
    if (statSync(full).isDirectory()) yield* walk(full);
    else yield full;
  }
}

test('no China-routed download sources in scripts, launchers or CI', () => {
  const offenders = [];
  for (const file of walk(repoRoot)) {
    if (!SCANNED_EXTENSIONS.has(extname(file))) continue;
    const lines = readFileSync(file, 'utf8').split(/\r?\n/);
    lines.forEach((line, i) => {
      for (const { pattern, use } of FORBIDDEN) {
        if (pattern.test(line)) {
          offenders.push(`${relative(repoRoot, file)}:${i + 1} matches ${pattern} — use ${use}`);
        }
      }
    });
  }
  assert.deepEqual(offenders, [], `China-routed sources found:\n${offenders.join('\n')}`);
});

test('u-claw-app lockfile resolves every package from the official npm registry', () => {
  const lock = JSON.parse(readFileSync(join(repoRoot, 'u-claw-app', 'package-lock.json'), 'utf8'));
  const offenders = Object.entries(lock.packages ?? {})
    .filter(([, meta]) => meta.resolved && !meta.resolved.startsWith('https://registry.npmjs.org/'))
    .map(([name, meta]) => `${name} -> ${meta.resolved}`);
  // A lockfile bakes the mirror host into every `resolved` field, so `npm ci`
  // keeps hitting it no matter what --registry the scripts pass. Regenerating
  // requires deleting the lockfile first: npm will not rewrite existing entries.
  assert.deepEqual(offenders, [], `lockfile pins non-official sources:\n${offenders.join('\n')}`);
});

test('lockfile openclaw version matches OPENCLAW_VERSION', () => {
  const declared = readFileSync(join(repoRoot, 'OPENCLAW_VERSION'), 'utf8').trim();
  const lock = JSON.parse(readFileSync(join(repoRoot, 'u-claw-app', 'package-lock.json'), 'utf8'));
  const locked = lock.packages?.['node_modules/openclaw']?.version;
  assert.equal(locked, declared, `OPENCLAW_VERSION says ${declared} but the lockfile pins ${locked}`);
});

test('no user-facing Chinese is left in the English build surfaces', () => {
  // Chinese belongs in lib/messages/zh-CN.json and in the two deliberately
  // bilingual pages — not hardcoded in a script an English-speaking user sees.
  // Comments are fine: they are for whoever maintains this next.
  const CJK = /[一-鿿]/;
  const ALLOWED = new Set([
    'portable/lib/messages/zh-CN.json',   // the Chinese catalogue itself
    'portable/lib/i18n/messages.js',      // generated from it
    'portable/lib/language.html',         // bilingual by design
    'portable/advanced/Config.html',      // bilingual by design
  ]);

  // Strip comments before looking, so a Chinese note at the end of a code line
  // or inside a multi-line block does not read as a user-facing string.
  function stripComments(text, file) {
    let out = text
      .replace(/\/\*[\s\S]*?\*\//g, '')       // /* ... */
      .replace(/<!--[\s\S]*?-->/g, '');        // <!-- ... -->
    return out
      .split(/\r?\n/)
      .map((line) => {
        if (/\.(sh|command)$/.test(file)) return line.replace(/(^|\s)#.*$/, '');
        if (/\.(bat|ps1)$/.test(file)) return line.replace(/(^|\s)(REM\b|::|#).*$/i, '');
        return line.replace(/(^|[^:])\/\/.*$/, '$1');
      })
      .join('\n');
  }

  const offenders = [];
  for (const file of walk(join(repoRoot, 'portable'))) {
    const rel = relative(repoRoot, file);
    if (ALLOWED.has(rel)) continue;
    if (!/\.(sh|ps1|bat|command|mjs|js|html)$/.test(file)) continue;
    stripComments(readFileSync(file, 'utf8'), file)
      .split(/\r?\n/)
      .forEach((line, i) => { if (CJK.test(line)) offenders.push(`${rel}:${i + 1}  ${line.trim().slice(0, 70)}`); });
  }
  assert.deepEqual(offenders, [], `Chinese left in user-facing strings:\n${offenders.join('\n')}`);
});