PHP: Extended ressource/file/url link check with redirect handling (Linkchecker)

gistfile1.aw

/**
 * Checks HTTP Links.
 * Based on function phpLinkCheck from Johannes Froemter <[email protected]>, 2001-04-14.
 *
 * Usage:
 * <code>
 * $url = 'http://www.google.de/?q=foobar';
 * $res = mp_linkCheck($url);
 *
 * // key 'Status-Code' will contain the HTTP status code (e.g. 200 or 404).
 * echo $res['Status-Code'] . "\n";
 *
 * // key 'Location-Status-Code' will contain the status code of the new loaction, in case of a 3xx code (redirection).
 * echo $res['Location-Status-Code'] . "\n";
 *
 * // dump variable to see the complete result
 * print_r($res);
 * </code>
 *
 * @param  string  $url
 * @param  array   $options  Options array with following feasible values:
 *                           - $options['max_redirects']   (int)     Number of maximal redirects
 *                           - $options['port']            (int)     The port for the connection, default value is 80.
 *                                                                   Url containing a port definition overwrites the port in options!
 *                           - $options['socket_timeout']  (int)     Socket timeout in seconds, default value is 30.
 *                           - $options['user_agent']      (string)  User agent string.
 */
function mp_linkCheck($url, array $options = array()) {
    $url = trim($url);

    if (strpos($url, '://') === false) {
        $url = "http://$url";
    }

    // parse url
    $comp = parse_url($url);
    if (strtolower($comp['scheme']) != 'http') {
        return false;
    }

    // options check
    if (!isset($options['max_redirects']) || (int) $options['max_redirects'] < 0) {
        $options['max_redirects'] = 0;
    }
    if (!isset($options['port']) || (int) $options['port'] < 0) {
        $options['port'] = 80;
    }
    $options['this.counter'] = (isset($options['this.counter'])) ? ++$$options['this.counter'] : 1;
    if (!isset($options['socket_timeout']) || (int) $options['socket_timeout'] < 1) {
        $options['socket_timeout'] = 30;
    }
    if (!isset($options['user_agent'])) {
        $options['user_agent'] = null;
    } else {
        $options['user_agent'] = trim($options['user_agent']);
    }


    $host = $comp['host'];
    $port = (isset($comp['port'])) ? $comp['port'] : $options['port'];
    $path = $comp['path'];
    if (isset($comp['query'])) {
        $path .= '?' . $comp['query'];
    }
    if (isset($comp['fragment'])) {
        $path .= '#' . $comp['fragment'];
    }

    // open connection
    if (!$fp = fsockopen($host, $port, $errno, $errstr, $options['socket_timeout'])) {
        return false;
    }

    // compose request
    $httpRequest = "HEAD ". $path ." HTTP/1.1\r\n"
                 . "Host: ". $host ."\r\n";
    if ($options['user_agent']) {
        $httpRequest .= "User-Agent: " . $options['user_agent'] . "\r\n";
    }
    $httpRequest .= "Connection: close\r\n\r\n";

    // send request and read response
    $httpResponse = '';
    fputs($fp, $httpRequest);
    while (!feof($fp)) {
        $httpResponse .= fgets($fp, 1024);
    }
    fclose($fp);

    // extract
    if (!preg_match('=^(HTTP/\d+\.\d+) (\d{3}) ([^\r\n]*)=', $httpResponse, $matches)) {
        return false;
    }
    $http['Status-Line']   = $matches[0];
    $http['HTTP-Version']  = $matches[1];
    $http['Status-Code']   = $matches[2];
    $http['Reason-Phrase'] = $matches[3];

    if ($options['this.counter'] >= $options['max_redirects']) {
        return $http;
    }

    $rclasses = array('Informational', 'Success', 'Redirection', 'Client Error', 'Server Error');
    $http['Response-Class'] = $rclasses[$http['Status-Code'][0] - 1];

    if (preg_match_all('=^(.+): ([^\r\n]*)=m', $head, $matches, PREG_SET_ORDER)) {
        foreach ($matches as $line) {
            $http[$line[1]] = $line[2];
        }

        if ($http['Status-Code'][0] == 3) {
            $http['Location-Status-Code'] = mp_linkCheck($http['Location'], $options);
        }
    }

    return $http;
}