// Bama crawl trigger — fires the server's /api/crawl every 15 min via PM2
// cron_restart. The web app itself never crawls on page load.
const fs = require("fs");
const path = require("path");

// Load .env.production (KEY=VALUE lines) so the cron process has the real
// secret + port.
const envFile = path.join(__dirname, "..", ".env.production");
try {
  for (const line of fs.readFileSync(envFile, "utf8").split("\n")) {
    const m = line.match(/^([A-Z0-9_]+)=(.*)$/);
    if (m && !process.env[m[1]]) process.env[m[1]] = m[2];
  }
} catch (e) {}

const SECRET = process.env.CRON_SECRET || "bama-cron";
const PORT = process.env.PORT || "3000";

let run = async () => {
  const res = await fetch(`http://127.0.0.1:${PORT}/api/crawl`, {
    method: "POST",
    headers: { "x-bama-cron": SECRET },
    signal: AbortSignal.timeout(280000),
  });
  const j = await res.json().catch(() => ({}));
  const c = (j && j.crawl) || {};
  console.log(
    `[bama-crawl] ${res.status} added=${c.added} named=${c.named} fallback=${c.usedFallback ? "yes" : "no"} gated=${c.gatedOut} unverified=${c.skippedUnverified} imgs=${c.withImage} articles=${c.articles ? c.articles.ok + "/" + c.articles.fetched : "?"} ${c.errors && c.errors.length ? "errs=" + c.errors.length : ""}`
  );
  process.exit(res.ok ? 0 : 1);
};

run().catch((e) => {
  console.error("[bama-crawl] FAILED", e && e.message);
  process.exit(1);
});