u-claw/tests/origin.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));
const origin = JSON.parse(readFileSync(join(repoRoot, 'origin.json'), 'utf8'));

const SKIPPED_DIRS = new Set(['.git', 'node_modules', 'dist', '.download-cache', 'tests']);
// Lockfiles are full of third-party funding and repository links. They say
// nothing about where *this* build fetches from.
const SKIPPED_FILES = new Set(['package-lock.json']);
const SCANNED = new Set(['.sh', '.ps1', '.bat', '.command', '.mjs', '.js', '.yml', '.yaml', '.json', '.html', '.md']);

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('origin.json describes a coherent origin', () => {
  assert.match(origin.repo.owner, /^[\w.-]+$/);
  assert.match(origin.repo.name, /^[\w.-]+$/);
  // A template, not a base: Gitea and GitHub lay raw paths out differently.
  assert.match(origin.urls.rawTemplate, /\{ref\}.*\{path\}/, 'rawTemplate needs both placeholders');
  for (const key of ['web', 'rawTemplate', 'releases', 'issues', 'website']) {
    assert.match(origin.urls[key], /^https:\/\//, `${key} should be an https URL`);
  }
  assert.match(origin.urls.ssh, /^ssh:\/\//, 'ssh should be an ssh URL');
});

// This is a fork. Every URL the build fetches at runtime points at somebody
// else's host until we move it. The move is only safe if nothing is hiding in a
// file we forgot about, which is what this test is for.
test('no file points at an owner other than the one origin.json declares', () => {
  const declaredOwner = origin.repo.owner;
  const OWNER_PATTERN = new RegExp(
    `${origin.repo.host.replace(/\./g, '\\.')}\\/([\\w.-]+)\\/|github\\.com\\/([\\w.-]+)\\/`, 'g');
  const KNOWN_THIRD_PARTY = new Set([
    'openclaw',           // upstream runtime, a real dependency
    'electron',           // electron mirrors
    'ventoy',             // bootable USB
    'dongsheng123132',    // the project we forked — referenced, never fetched from
  ]);

  const offenders = [];
  for (const file of walk(repoRoot)) {
    if (!SCANNED.has(extname(file))) continue;
    const rel = relative(repoRoot, file);
    if (rel === 'origin.json' || SKIPPED_FILES.has(rel.split('/').pop())) continue;
    const content = readFileSync(file, 'utf8');
    content.split(/\r?\n/).forEach((line, i) => {
      for (const match of line.matchAll(OWNER_PATTERN)) {
        const owner = match[1] ?? match[2];
        if (owner === declaredOwner || KNOWN_THIRD_PARTY.has(owner)) continue;
        offenders.push(`${rel}:${i + 1} points at ${owner}`);
      }
    });
  }
  assert.deepEqual(offenders, [], `unexpected owners:\n${offenders.join('\n')}`);
});

// The one-line installers run through curl|bash with no checkout, so they cannot
// read origin.json and carry the same URLs as literals. Without this check a
// migration would update origin.json, look done, and leave curl|bash users
// fetching from the old host.
test('the standalone installers carry URLs that match origin.json', () => {
  const expected = {
    releases: origin.urls.releases,
    website: origin.urls.website,
  };

  for (const script of ['install/install.sh', 'install/install.ps1']) {
    const content = readFileSync(join(repoRoot, script), 'utf8');

    const rawPrefix = origin.urls.rawTemplate.split('{ref}')[0];
    assert.ok(
      !/raw\.githubusercontent\.com/.test(content) || content.includes(rawPrefix),
      `${script} still fetches raw content from GitHub`,
    );
    if (content.includes('install-skills.mjs')) {
      assert.ok(content.includes(rawPrefix), `${script} should fetch the installer from ${rawPrefix}`);
    }

    const releaseUrls = [...content.matchAll(new RegExp(`https://[\\w.-]+/[\\w.-]+/[\\w.-]+/releases`, 'g'))].map((m) => m[0]);
    for (const url of releaseUrls) {
      assert.equal(url, expected.releases, `${script} downloads from ${url}, origin.json says ${expected.releases}`);
    }

    const siteUrls = [...content.matchAll(/https:\/\/u-claw\.org|https:\/\/[\w.-]*u-claw[\w.-]*\.\w+/g)].map((m) => m[0]);
    for (const url of siteUrls) {
      assert.equal(url, expected.website, `${script} points at ${url}, origin.json says ${expected.website}`);
    }
  }
});

test('install-skills.mjs reads origin.json rather than hardcoding the host', () => {
  const content = readFileSync(join(repoRoot, 'lib', 'install-skills.mjs'), 'utf8');
  assert.match(content, /origin\.json/, 'it should read origin.json');
  // A literal fallback is fine and necessary — the remote installers download
  // this file on its own, with no repo around it — but it must agree.
  // A literal fallback is necessary — the remote installers download this file on
  // its own, with no repo around it — but it must agree with origin.json.
  const fallback = content.match(/\?\?\s*'(https:\/\/[^']+)'/)?.[1];
  assert.ok(fallback, 'there should be a hardcoded fallback template');
  assert.equal(fallback, origin.urls.rawTemplate, 'the fallback template disagrees with origin.json');
});

test('we do not tell users to email a domain we do not own', () => {
  // u-claw.org is the upstream project's domain. Any address on it reaches them,
  // not us — so a support line pointing there sends our users' problems to
  // someone with no reason to answer.
  const OURS = new Set(Object.values(origin.upstream?.addresses ?? {}).filter((v) => /@/.test(v)));
  const offenders = [];
  for (const file of walk(repoRoot)) {
    if (!SCANNED.has(extname(file))) continue;
    const rel = relative(repoRoot, file);
    if (rel === 'origin.json' || SKIPPED_FILES.has(rel.split('/').pop())) continue;
    readFileSync(file, 'utf8').split(/\r?\n/).forEach((line, i) => {
      for (const [found] of line.matchAll(/[\w.+-]+@u-claw\.org/g)) {
        // Upstream's own addresses may appear where the text is about upstream.
        if (OURS.has(found)) continue;
        offenders.push(`${rel}:${i + 1} offers ${found}, a domain we do not control`);
      }
    });
  }
  assert.deepEqual(offenders, [], offenders.join('\n'));
});

test('support routes somewhere we actually control', () => {
  assert.match(origin.support.issues, /^https:\/\//, 'there has to be a working support route');
  assert.ok(
    origin.support.issues.includes(origin.repo.host),
    'support should point at our own host, not the upstream project',
  );
  // Until we have an address of our own, saying so beats leaving a stale one.
  assert.ok('email' in origin.support, 'origin.json should record whether we have an address yet');
  assert.match(origin.support.note, /do not own u-claw\.org/i);
});