phpMyFAQ\Linkverifier::openURL()

Complexity

Conditions	34
Paths	> 20000

Size

Total Lines

194

Duplication

Lines	32
Ratio	16.49 %

Importance

Changes

0

<?php
    protected function openURL($url = '', $redirect = '', $redirectCount = 0)
    {
        global $PMF_LANG;

        // If prequisites fail
        if (false === $this->isReady()) {
            return [false, $redirectCount, $PMF_LANG['ad_linkcheck_openurl_notready']];
        }

        // Recursing too much ?
        if (($redirectCount >= self::LINKVERIFIER_MAX_REDIRECT_COUNT) || ($url == $redirect)) {
            return [
                false,
                $redirectCount,
                sprintf(
                    $PMF_LANG['ad_linkcheck_openurl_maxredirect'],
                    self::LINKVERIFIER_MAX_REDIRECT_COUNT
                ),
            ];
        }

        // If destination is blank, fail.
        if ('' === trim($url)) {
            return [false, $redirectCount, $PMF_LANG['ad_linkcheck_openurl_urlisblank']];
        }

        if ('' !== $redirect) {
            $url = $this->makeAbsoluteURL($redirect, $url);
        }

        // parse URL
        $defaultParts = [
            'scheme' => 'http',
            'host' => $_SERVER['HTTP_HOST'],
            'user' => '',
            'pass' => '',
            'path' => '/',
            'query' => '',
            'fragment' => '',
        ];
        $urlParts = @parse_url($url);
        foreach ($defaultParts as $key => $value) {
            if (!(isset($urlParts[$key]))) {
                $urlParts[$key] = $value;
            }
        }

        if (!(isset($urlParts['port']))) {
            switch ($urlParts['scheme']) {
                case 'https':
                    $urlParts['port'] = 443;
                    break;
                case 'http':
                    $urlParts['port'] = 80;
                    break;
                default:
                    $urlParts['port'] = 80;
                    break;
            }
        }

        // Hack: fix any unsafe space chars in any component of the path to avoid HTTP 400 status during HEAD crawling
        if ('' !== $urlParts['path']) {
            $urlSubParts = explode('/', $urlParts['path']);
            $num = count($urlSubParts);
            for ($i = 0; $i < $num; ++$i) {
                $urlSubParts[$i] = str_replace(' ', '%20', $urlSubParts[$i]);
            }
            $urlParts['path'] = implode('/', $urlSubParts);
        }

        if ('' !== $urlParts['query']) {
            $urlParts['query'] = '?'.$urlParts['query'];
        }

        if ('' !== $urlParts['fragment']) {
            $urlParts['fragment'] = '#'.$urlParts['fragment'];
        }

        // Check whether we tried the host before
        if (isset($this->slowHosts[$urlParts['host']])) {
            return [
                false,
                $redirectCount,
                sprintf(
                    $PMF_LANG['ad_linkcheck_openurl_tooslow'],
                    Strings::htmlspecialchars($urlParts['host'])
                ),
            ];
        }

        // Check whether the hostname exists
        if (gethostbynamel($urlParts['host']) === false) {
            // mark this host too slow to verify
            $this->slowHosts[$urlParts['host']] = true;

            return [
                false,
                $redirectCount,
                sprintf(
                    $PMF_LANG['ad_linkcheck_openurl_nodns'],
                    Strings::htmlspecialchars($urlParts['host'])
                ),
            ];
        }

        $_response = '';

        // open socket for remote server with timeout (default: 5secs)
        $_host = $urlParts['host'];
        if (@extension_loaded('openssl') && ('https' == $urlParts['scheme'])) {
            $_host = 'ssl://'.$_host;
        }

        $fp = @fsockopen($_host, $urlParts['port'], $errno, $errstr, self::LINKVERIFIER_CONNECT_TIMEOUT);

        if (!$fp) {
            // mark this host too slow to verify
            $this->slowHosts[$urlParts['host']] = true;

            return [
                false,
                $redirectCount,
                sprintf(
                    $PMF_LANG['ad_linkcheck_openurl_tooslow'],
                    Strings::htmlspecialchars($urlParts['host'])
                ),
            ];
        }

        // wait for data with timeout (default: 10secs)
        stream_set_timeout($fp, self::LINKVERIFIER_RESPONSE_TIMEOUT, 0);
        $_url = $urlParts['path'].$urlParts['query'].$urlParts['fragment'];
        fputs($fp, 'HEAD '.$_url." HTTP/1.0\r\nHost: ".$urlParts['host']."\r\n");
        // Be polite: let our probe declares itself
        fputs($fp, "User-Agent: phpMyFAQ Link Checker\r\n");
        fputs($fp, "\r\n");
        while (!feof($fp)) {
            $_response .= fread($fp, 4096);
        }
        fclose($fp);

        // parse response
        $code = 0;
        $allowVerbs = 'n/a';
        $location = $url;
        $response = explode("\r\n", $_response);
        $httpStatusMsg = strip_tags($response[count($response) - 1]);

        foreach ($response as $_response) {
            if (preg_match("/^HTTP\/[^ ]+ ([01-9]+) .*$/", $_response, $matches)) {
                $code = $matches[1];
            }
            if (preg_match('/^Location: (.*)$/', $_response, $matches)) {
                $location = $matches[1];
            }
            if (preg_match('/^[a|A]llow: (.*)$/', $_response, $matches)) {
                $allowVerbs = $matches[1];
            }
        }

        // process response code
        switch ($code) {
            // TODO: Add more explicit http status management
            case '200': // OK
                $_reason = ($redirectCount > 0) ? sprintf($PMF_LANG['ad_linkcheck_openurl_redirected'], Strings::htmlspecialchars($url)) : '';

                return array(true, $redirectCount, $_reason);
                break;

break is not strictly necessary here and could be removed.

            case '301': // Moved Permanently (go recursive ?)
            case '302': // Found (go recursive ?)
                return $this->openURL($url, $location, $redirectCount + 1);
                break;
            case 400:   // Bad Request
                return array(false, $redirectCount, sprintf($PMF_LANG['ad_linkcheck_openurl_ambiguous'].'<br>'.$httpStatusMsg, $code));
                break;

break is not strictly necessary here and could be removed.

            case 404:   // Not found
                return array(false, $redirectCount, sprintf($PMF_LANG['ad_linkcheck_openurl_not_found'], $urlParts['host']));
                break;

break is not strictly necessary here and could be removed.

            case '300': // Multiple choices
            case '401': // Unauthorized (but it's there. right ?)
                return array(true, $redirectCount, sprintf($PMF_LANG['ad_linkcheck_openurl_ambiguous'], $code));
                break;
            case '405': // Method Not Allowed
                // TODO: Add a fallback to use GET method, otherwise this link should be marked as bad
                return array(true, $redirectCount, sprintf($PMF_LANG['ad_linkcheck_openurl_not_allowed'], $urlParts['host'], $allowVerbs));
                break;
            default:    // All other statuses
                return array(false, $redirectCount, sprintf($PMF_LANG['ad_linkcheck_openurl_ambiguous'], $code));
                break;

break is not strictly necessary here and could be removed.

        }

        return [false, $redirectCount, ''];
    }