Function siteChecker

apps/api/src/controllers/tools.controller.ts:491–660

(
  request: FastifyRequest<{ Querystring: { url?: string } }>,
  reply: FastifyReply,
)

}

export async function siteChecker(
  request: FastifyRequest<{ Querystring: { url?: string } }>,
  reply: FastifyReply,
) {
  const urlParam = request.query.url;

  if (!urlParam) {
    return reply.status(400).send({ error: 'URL parameter is required' });
  }

  const { ip } = getClientIpFromHeaders(request.headers);
  if (ip && !checkRateLimit(`site:${ip}`, SITE_CHECK_WINDOW, SITE_CHECK_MAX)) {
    return reply
      .status(429)
      .send({ error: 'Rate limit exceeded. Please try again later.' });
  }

  let url: URL;
  try {
    url = new URL(urlParam);
  } catch {
    return reply.status(400).send({ error: 'Invalid URL' });
  }

  if (!url.protocol || !url.protocol.startsWith('http')) {
    url = new URL(`https://${urlParam}`);
  }

  try {
    const { finalUrl, redirectChain, html, headers, statusCode, timing } =
      await fetchWithRedirects(url.toString());

    const finalUrlObj = new URL(finalUrl);
    const $ = cheerio.load(html);

    const title = $('title').first().text().trim();
    const description =
      $('meta[name="description"]').attr('content') ||
      $('meta[property="og:description"]').attr('content') ||
      '';
    const canonical =
      $('link[rel="canonical"]').attr('href') ||
      $('meta[property="og:url"]').attr('content') ||
      null;
    const h1Tags = $('h1')
      .map((_, el) => $(el).text().trim())
      .get()
      .filter(Boolean);
    const robotsMeta =
      $('meta[name="robots"]').attr('content') ||
      $('meta[name="googlebot"]').attr('content') ||
      null;

    const ogTitle = $('meta[property="og:title"]').attr('content') || null;
    const ogDescription =
      $('meta[property="og:description"]').attr('content') || null;
    const ogImage = $('meta[property="og:image"]').attr('content') || null;
    const ogUrl = $('meta[property="og:url"]').attr('content') || null;