= $latestPageUpdate && (time() - $cacheTime) < SITEMAP_MAX_AGE) {
$shouldRegenerate = false;
}
}
if ($shouldRegenerate) {
$xml = generate_sitemap_xml();
@file_put_contents($cacheFile, $xml);
@chmod($cacheFile, 0644);
}
header('Content-Type: application/xml; charset=utf-8');
header('X-Robots-Tag: noindex'); // Don't index the sitemap itself
header('Cache-Control: public, max-age=3600'); // CDN caches for 1h
if (is_file($cacheFile)) {
readfile($cacheFile);
} else {
echo generate_sitemap_xml(); // Fallback if file write failed
}
exit;
/* ------------------------------------------------------------------ */
function generate_sitemap_xml(): string {
$base = rtrim(canonical_base_url(), '/');
$now = gmdate('Y-m-d');
$xml = '' . "\n";
$xml .= '' . "\n";
// Home — high priority
$xml .= sitemap_url("$base/", $now, '1.0', 'daily');
// Static landing pages — medium priority, change rarely
$xml .= sitemap_url("$base/pages", $now, '0.6', 'daily');
$xml .= sitemap_url("$base/recent", $now, '0.6', 'hourly');
// Categories
$cats = DB::all(
"SELECT slug, updated_at FROM categories ORDER BY sort_order ASC, name ASC"
);
foreach ($cats as $c) {
$lastmod = sitemap_format_date($c['updated_at'] ?? null);
$xml .= sitemap_url(
"$base/category/" . rawurlencode($c['slug']),
$lastmod,
'0.7',
'daily'
);
}
// All published pages
$pages = DB::all(
"SELECT slug, updated_at, view_count
FROM pages
WHERE status = 'published' AND is_deleted = 0
ORDER BY view_count DESC"
);
foreach ($pages as $p) {
// Heavily-viewed pages get higher priority — they're the ones users
// care about most. Mapping: views > 1000 -> 0.9, > 100 -> 0.8, else 0.6
$views = (int) ($p['view_count'] ?? 0);
$priority = $views > 1000 ? '0.9' : ($views > 100 ? '0.8' : '0.6');
$xml .= sitemap_url(
"$base/wiki/" . rawurlencode($p['slug']),
sitemap_format_date($p['updated_at'] ?? null),
$priority,
'weekly'
);
}
$xml .= '' . "\n";
return $xml;
}
function sitemap_url(string $loc, string $lastmod, string $priority, string $changefreq): string {
return " \n"
. " " . htmlspecialchars($loc, ENT_XML1 | ENT_QUOTES, 'UTF-8') . "\n"
. " " . $lastmod . "\n"
. " " . $changefreq . "\n"
. " " . $priority . "\n"
. " \n";
}
function sitemap_format_date(?string $sqlDate): string {
if (!$sqlDate) return gmdate('Y-m-d');
$ts = strtotime($sqlDate);
return $ts ? gmdate('Y-m-d', $ts) : gmdate('Y-m-d');
}
/**
* Build the canonical https://host base URL using the SITE_URL config
* if defined, otherwise reconstruct from the request.
*/
function canonical_base_url(): string {
if (defined('SITE_URL') && SITE_URL !== '') {
return rtrim(SITE_URL, '/');
}
$scheme = (!empty($_SERVER['HTTPS']) && $_SERVER['HTTPS'] !== 'off') ? 'https' : 'http';
$host = $_SERVER['HTTP_HOST'] ?? 'localhost';
return $scheme . '://' . $host;
}