klz-cables.com/scripts/check-http.ts

import axios from 'axios';
import * as cheerio from 'cheerio';

const targetUrl = process.argv[2] || process.env.NEXT_PUBLIC_BASE_URL || 'http://localhost:3000';
const gatekeeperPassword = process.env.GATEKEEPER_PASSWORD || 'klz2026';

async function main() {
  console.log(`\n🚀 Starting HTTP Sitemap Validation for: ${targetUrl}\n`);

  try {
    const sitemapUrl = `${targetUrl.replace(/\/$/, '')}/sitemap.xml`;
    console.log(`📥 Fetching sitemap from ${sitemapUrl}...`);

    const response = await axios.get(sitemapUrl, {
      headers: { Cookie: `klz_gatekeeper_session=${gatekeeperPassword}` },
      validateStatus: (status) => status < 400,
    });

    const $ = cheerio.load(response.data, { xmlMode: true });
    let urls = $('url loc')
      .map((i, el) => $(el).text())
      .get();

    const urlPattern = /https?:\/\/[^\/]+/;
    urls = [...new Set(urls)]
      .filter((u) => u.startsWith('http'))
      .map((u) => u.replace(urlPattern, targetUrl.replace(/\/$/, '')))
      .sort();

    console.log(`✅ Found ${urls.length} target URLs in sitemap.`);

    if (urls.length === 0) {
      console.error('❌ No URLs found in sitemap. Is the site up?');
      process.exit(1);
    }

    console.log(`\n🔍 Verifying HTTP Status Codes (Limit: None)...`);
    let hasErrors = false;

    // Run fetches sequentially to avoid overwhelming the server during CI
    for (let i = 0; i < urls.length; i++) {
      const u = urls[i];
      try {
        const res = await axios.get(u, {
          headers: { Cookie: `klz_gatekeeper_session=${gatekeeperPassword}` },
          validateStatus: null, // Don't throw on error status
        });

        if (res.status >= 400) {
          console.error(`❌ ERROR ${res.status}: ${res.statusText} -> ${u}`);
          hasErrors = true;
        } else {
          console.log(`✅ OK ${res.status} -> ${u}`);
        }
      } catch (err: any) {
        console.error(`❌ NETWORK ERROR: ${err.message} -> ${u}`);
        hasErrors = true;
      }
    }

    if (hasErrors) {
      console.error(`\n❌ HTTP Sitemap Validation Failed. One or more pages returned an error.`);
      process.exit(1);
    } else {
      console.log(`\n✨ Success: All ${urls.length} pages are healthy! (HTTP 200)`);
      process.exit(0);
    }
  } catch (error: any) {
    console.error(`\n❌ Critical Error during Sitemap Fetch:`, error.message);
    process.exit(1);
  }
}

main();