chat_web_app/llm-fe/scripts/sitemap.js

/**
 * Sitemap lastmod comes from the page source, not from the build clock.
 * Changelog uses the newest entry date. Terms use the "Last Updated" line.
 * Other public pages use the last git commit that touched that file.
 */
const { execFileSync } = require('child_process');
const fs = require('fs');
const path = require('path');

const ORIGIN = 'https://hesychia.ai';
const INDEXNOW_KEY = '8d1667dd71c6ef137171989c26eeb632';
const INDEXNOW_HOST = 'hesychia.ai';

const llmFe = path.resolve(__dirname, '..');
const repoRoot = path.resolve(llmFe, '..');

/** Public pages only. Same URLs the site already lists for crawlers. */
const SITEMAP_PAGES = [
  {
    path: '/',
    file: 'llm-fe/src/llm-fe/pages/Landing/LandingPage.tsx',
    lastmod: 'git',
  },
  {
    path: '/pricing/',
    file: 'llm-fe/src/llm-fe/pages/Pricing/PricingPage.tsx',
    lastmod: 'git',
  },
  {
    path: '/privacy/',
    file: 'llm-fe/src/llm-fe/pages/Privacy/PrivacyPage.tsx',
    lastmod: 'git',
  },
  {
    path: '/changelog/',
    file: 'llm-fe/src/llm-fe/pages/Changelog/changelogEntries.ts',
    lastmod: 'changelog',
  },
  {
    path: '/terms_of_service/',
    file: 'llm-fe/src/llm-fe/components/legal/TermsOfServiceDocument.tsx',
    lastmod: 'terms',
  },
  {
    path: '/account-deletion/',
    file: 'llm-fe/src/llm-fe/pages/AccountDeletion/AccountDeletionPage.tsx',
    lastmod: 'git',
  },
];

const MONTHS = {
  January: '01',
  February: '02',
  March: '03',
  April: '04',
  May: '05',
  June: '06',
  July: '07',
  August: '08',
  September: '09',
  October: '10',
  November: '11',
  December: '12',
};

function newestChangelogDate(source) {
  const dates = [...source.matchAll(/date:\s*'(\d{4}-\d{2}-\d{2})'/g)].map((match) => match[1]);
  if (dates.length === 0) {
    throw new Error('changelog entries have no dates');
  }
  return dates.sort().at(-1);
}

function termsUpdated(source) {
  const match = source.match(/Last Updated:\s+([A-Z][a-z]+)\s+(\d{1,2}),\s+(\d{4})/);
  if (!match) {
    throw new Error('terms have no Last Updated date');
  }
  const month = MONTHS[match[1]];
  if (!month) {
    throw new Error(`unknown month in terms date: ${match[1]}`);
  }
  return `${match[3]}-${month}-${match[2].padStart(2, '0')}`;
}

function gitLastmod(file) {
  const out = execFileSync('git', ['log', '-1', '--format=%cs', '--', file], {
    cwd: repoRoot,
    encoding: 'utf8',
  }).trim();
  if (!/^\d{4}-\d{2}-\d{2}$/.test(out)) {
    throw new Error(`no git commit date for ${file}`);
  }
  return out;
}

function lastmodFor(page) {
  const abs = path.join(repoRoot, page.file);
  const source = fs.readFileSync(abs, 'utf8');
  if (page.lastmod === 'changelog') return newestChangelogDate(source);
  if (page.lastmod === 'terms') return termsUpdated(source);
  return gitLastmod(page.file);
}

function sitemapEntries() {
  return SITEMAP_PAGES.map((page) => ({
    loc: `${ORIGIN}${page.path}`,
    lastmod: lastmodFor(page),
  }));
}

function renderSitemap(entries) {
  const urls = entries
    .map(
      (entry) => `  <url>
    <loc>${entry.loc}</loc>
    <lastmod>${entry.lastmod}</lastmod>
  </url>`,
    )
    .join('\n');
  return `<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
${urls}
</urlset>
`;
}

function writeSitemap() {
  const xml = renderSitemap(sitemapEntries());
  fs.writeFileSync(path.join(llmFe, 'public', 'sitemap.xml'), xml);
  return xml;
}

function indexNowBody(key, urls) {
  return {
    host: INDEXNOW_HOST,
    key,
    keyLocation: `https://${INDEXNOW_HOST}/${key}.txt`,
    urlList: urls,
  };
}

module.exports = {
  INDEXNOW_KEY,
  SITEMAP_PAGES,
  newestChangelogDate,
  termsUpdated,
  gitLastmod,
  sitemapEntries,
  renderSitemap,
  writeSitemap,
  indexNowBody,
};