diff options
Diffstat (limited to 'src/phinde/HubUrlExtractor.php')
| -rw-r--r-- | src/phinde/HubUrlExtractor.php | 15 |
1 files changed, 13 insertions, 2 deletions
diff --git a/src/phinde/HubUrlExtractor.php b/src/phinde/HubUrlExtractor.php index 4f1baa3..da29650 100644 --- a/src/phinde/HubUrlExtractor.php +++ b/src/phinde/HubUrlExtractor.php @@ -19,7 +19,8 @@ class HubUrlExtractor * Get the hub and self/canonical URL of a given topic URL. * Uses link headers and parses HTML link rels. * - * @param string $url Topic URL + * @param string $url Topic URL + * @param int $redirects Number of redirects that were followed * * @return array Array of URLs with keys: hub, self. * - "self" value is the URL @@ -27,12 +28,13 @@ class HubUrlExtractor * Keys may be there but most not if the URL * does not advertise them. */ - public function getUrls($url) + public function getUrls($url, $redirects = 0) { //at first, try a HEAD request that does not transfer so much data $req = $this->getRequest(); $req->setUrl($url); $req->setMethod(\HTTP_Request2::METHOD_HEAD); + $req->setConfig('follow_redirects', false); $res = $req->send(); if (intval($res->getStatus() / 100) >= 4 @@ -49,6 +51,15 @@ class HubUrlExtractor return $this->absolutifyUrls($urls, $base); } + if ($res->isRedirect()) { + //we tried header links and that failed, now follow the redirect + if ($redirects > 5) { + return []; + } + $redirectUrl = (string) $base->resolve($res->getHeader('location')); + return $this->getUrls($redirectUrl, $redirects + 1); + } + list($type) = explode(';', $res->getHeader('Content-type')); if ($type != 'text/html' && $type != 'text/xml' && $type != 'application/xhtml+xml' |