$config['maxDepth']) return;
if (in_array($url, $visited)) return;
if ($pageCount >= $config['maxPages']) return;
$visited[] = $url;
$pageCount++;
$html = @file_get_contents($url);
if (!$html) return;
libxml_use_internal_errors(true);
$dom = new DOMDocument();
$dom->loadHTML($html);
libxml_clear_errors();
$xpath = new DOMXPath($dom);
// ===== IMAGES =====
foreach ($xpath->query("//img") as $img) {
$src = $img->getAttribute("src");
if ($src) {
$results[] = [
"type" => "image",
"url" => resolveUrl($src, $url)
];
}
}
// ===== VIDEOS =====
foreach ($xpath->query("//video | //source") as $vid) {
$src = $vid->getAttribute("src");
if ($src) {
$results[] = [
"type" => "video",
"url" => resolveUrl($src, $url)
];
}
}
// ===== LINKS (DEEPER CRAWL) =====
foreach ($xpath->query("//a") as $a) {
$href = $a->getAttribute("href");
if ($href) {
$next = resolveUrl($href, $url);
crawl($next, $depth + 1, $visited, $results, $pageCount, $config);
}
}
}
// ===== URL RESOLVER =====
function resolveUrl($relative, $base){
if (parse_url($relative, PHP_URL_SCHEME) != '') return $relative;
if ($relative[0] == '/') {
$parts = parse_url($base);
return $parts['scheme'] . "://" . $parts['host'] . $relative;
}
return rtrim($base, '/') . '/' . $relative;
}
// ===== HANDLE SCAN REQUEST =====
if (isset($_GET['scan'])) {
foreach ($config['seeds'] as $seed) {
crawl($seed, 0, $visited, $results, $pageCount, $config);
}
header('Content-Type: application/json');
echo json_encode(["media" => array_values(array_unique($results, SORT_REGULAR))]);
exit;
}
?>
PHP Media Scraper
Media Scraper