diff --git a/plugins/OStatus/lib/discovery.php b/plugins/OStatus/lib/discovery.php new file mode 100644 index 0000000000..1159f21510 --- /dev/null +++ b/plugins/OStatus/lib/discovery.php @@ -0,0 +1,303 @@ +. + * + * @package StatusNet + * @author James Walker + * @copyright 2010 StatusNet, Inc. + * @license http://www.fsf.org/licensing/licenses/agpl-3.0.html AGPL 3.0 + * @link http://status.net/ + */ + +/** + * This class implements LRDD-based service discovery based on the "Hammer Draft" + * (including webfinger) + * + * @see http://groups.google.com/group/webfinger/browse_thread/thread/9f3d93a479e91bbf + */ +class Discovery +{ + + const LRDD_REL = 'lrdd'; + const PROFILEPAGE = 'http://webfinger.net/rel/profile-page'; + const UPDATESFROM = 'http://schemas.google.com/g/2010#updates-from'; + + public $methods = array(); + + public function __construct() + { + $this->registerMethod('Discovery_LRDD_Host_Meta'); + $this->registerMethod('Discovery_LRDD_Link_Header'); + $this->registerMethod('Discovery_LRDD_Link_HTML'); + } + + + public function registerMethod($class) + { + $this->methods[] = $class; + } + + /** + * Given a "user id" make sure it's normalized to either a webfinger + * acct: uri or a profile HTTP URL. + */ + public static function normalize($user_id) + { + if (substr($user_id, 0, 5) == 'http:' || + substr($user_id, 0, 6) == 'https:' || + substr($user_id, 0, 5) == 'acct:') { + return $user_id; + } + + if (strpos($user_id, '@') !== FALSE) { + return 'acct:' . $user_id; + } + + return 'http://' . $user_id; + } + + public static function isWebfinger($user_id) + { + $uri = Discovery::normalize($user_id); + + return (substr($uri, 0, 5) == 'acct:'); + } + + /** + * This implements the actual lookup procedure + */ + public function lookup($id) + { + // Normalize the incoming $id to make sure we have a uri + $uri = $this->normalize($id); + + foreach ($this->methods as $class) { + $links = call_user_func(array($class, 'discover'), $uri); + + if ($link = Discovery::getService($links, Discovery::LRDD_REL)) { + // Load the LRDD XRD + if ($link['template']) { + $xrd_uri = Discovery::applyTemplate($link['template'], $uri); + } else { + $xrd_uri = $link['href']; + } + + $xrd = $this->fetchXrd($xrd_uri); + if ($xrd) { + return $xrd; + } + } + } + + throw new Exception('Unable to find services for '. $id); + } + + public static function getService($links, $service) { + foreach ($links as $link) { + if ($link['rel'] == $service) { + return $link; + } + } + } + + + public static function applyTemplate($template, $id) + { + $template = str_replace('{uri}', urlencode($id), $template); + + return $template; + } + + + public static function fetchXrd($url) + { + try { + $client = new HTTPClient(); + $response = $client->get($url); + } catch (HTTP_Request2_Exception $e) { + return false; + } + + if ($response->getStatus() != 200) { + return false; + } + + return XRD::parse($response->getBody()); + } +} + +interface Discovery_LRDD +{ + public function discovery($uri); +} + +class Discovery_LRDD_Host_Meta implements Discovery_LRDD +{ + function discover($uri) + { + if (Discovery::isWebfinger($uri)) { + // We have a webfinger acct: - start with host-meta + list($name, $domain) = explode('@', $id); + } else { + $domain = @parse_url($uri, PHP_URL_HOST); + } + + $url = 'http://'. $domain .'/.well-known/host-meta'; + + $xrd = Discovery::fetchXrd($url); + + if ($xrd) { + if ($xrd->host != $domain) { + return false; + } + + return $xrd->links; + } + } +} + +class Discovery_LRDD_Link_Header implements Discovery_LRDD +{ + public function discover($uri) + { + try { + $client = new HTTPClient(); + $response = $client->get($url); + } catch (HTTP_Request2_Exception $e) { + return false; + } + + if ($response->getStatus() != 200) { + return false; + } + + $link_header = $response->getHeader('Link'); + if (!$link_header) { + return false; + } + + return Discovery_LRDD_Link_Header::parseHeader($header); + } + + protected static function parseHeader($header) + { + preg_match('/^<[^>]+>/', $header, $uri_reference); + if (empty($uri_reference)) return; + + $link_uri = trim($uri_reference[0], '<>'); + $link_rel = array(); + $link_type = null; + + // remove uri-reference from header + $header = substr($header, strlen($uri_reference[0])); + + // parse link-params + $params = explode($header, ';'); + + foreach ($params as $param) { + if (empty($param)) continue; + list($param_name, $param_value) = explode('=', $param, 2); + $param_name = trim($param_name); + $param_value = preg_replace('(^"|"$)', '', trim($param_value)); + + // for now we only care about 'rel' and 'type' link params + // TODO do something with the other links-params + switch ($param_name) { + case 'rel': + $link_rel = trim($param_value); + break; + + case 'type': + $link_type = trim($param_value); + } + } + + return array( + 'href' => $link_uri, + 'rel' => $link_rel, + 'type' => $link_type); + } +} + +class Discovery_LRDD_Link_HTML implements Discovery_LRDD +{ + public function discover($uri) + { + try { + $client = new HTTPClient(); + $response = $client->get($url); + } catch (HTTP_Request2_Exception $e) { + return false; + } + + if ($response->getStatus() != 200) { + return false; + } + + return Discovery_LRDD_Link_HTML::parse($response->getBody()); + } + + + public function parse($html) + { + $links = array(); + + preg_match('/]*)?>(.*?)<\/head>/is', $html, $head_matches); + $head_html = $head_matches[2]; + + preg_match_all('/]*>/i', $head_html, $link_matches); + + foreach ($link_matches[0] as $link_html) { + $link_url = null; + $link_rel = null; + $link_type = null; + + preg_match('/\srel=(("|\')([^\\2]*?)\\2|[^"\'\s]+)/i', $link_html, $rel_matches); + if ( isset($rel_matches[3]) ) { + $link_rel = $rel_matches[3]; + } else if ( isset($rel_matches[1]) ) { + $link_rel = $rel_matches[1]; + } + + preg_match('/\shref=(("|\')([^\\2]*?)\\2|[^"\'\s]+)/i', $link_html, $href_matches); + if ( isset($href_matches[3]) ) { + $link_uri = $href_matches[3]; + } else if ( isset($href_matches[1]) ) { + $link_uri = $href_matches[1]; + } + + preg_match('/\stype=(("|\')([^\\2]*?)\\2|[^"\'\s]+)/i', $link_html, $type_matches); + if ( isset($type_matches[3]) ) { + $link_type = $type_matches[3]; + } else if ( isset($type_matches[1]) ) { + $link_type = $type_matches[1]; + } + + $links[] = array( + 'href' => $link_url, + 'rel' => $link_rel, + 'type' => $link_type, + ); + } + + return $links; + } +}