diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/phinde/HttpRequest.php | 4 | ||||
| -rw-r--r-- | src/phinde/HubUrlExtractor.php | 226 | ||||
| -rw-r--r-- | src/phinde/Subscriptions.php | 153 |
3 files changed, 378 insertions, 5 deletions
diff --git a/src/phinde/HttpRequest.php b/src/phinde/HttpRequest.php index e68bd84..4635973 100644 --- a/src/phinde/HttpRequest.php +++ b/src/phinde/HttpRequest.php @@ -3,9 +3,9 @@ namespace phinde; class HttpRequest extends \HTTP_Request2 { - public function __construct($url) + public function __construct($url = null, $method = 'GET') { - parent::__construct($url); + parent::__construct($url, $method); $this->setConfig('follow_redirects', true); $this->setConfig('connect_timeout', 5); $this->setConfig('timeout', 10); diff --git a/src/phinde/HubUrlExtractor.php b/src/phinde/HubUrlExtractor.php new file mode 100644 index 0000000..e2d328a --- /dev/null +++ b/src/phinde/HubUrlExtractor.php @@ -0,0 +1,226 @@ +<?php +namespace phinde; + +class HubUrlExtractor +{ + /** + * HTTP request object that's used to do the requests + * + * @var \HTTP_Request2 + */ + protected $request; + + /** + * Get the hub and self/canonical URL of a given topic URL. + * Uses link headers and parses HTML link rels. + * + * @param string $url Topic URL + * + * @return array Array of URLs with keys: hub, self + */ + public function getUrls($url) + { + //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); + $res = $req->send(); + + if (intval($res->getStatus() / 100) >= 4 + && $res->getStatus() != 405 //method not supported/allowed + ) { + return null; + } + + $url = $res->getEffectiveUrl(); + $base = new \Net_URL2($url); + + $urls = $this->extractHeader($res); + if (count($urls) === 2) { + return $this->absolutifyUrls($urls, $base); + } + + list($type) = explode(';', $res->getHeader('Content-type')); + if ($type != 'text/html' && $type != 'text/xml' + && $type != 'application/xhtml+xml' + //FIXME: atom, rss + && $res->getStatus() != 405//HEAD method not allowed + ) { + //we will not be able to extract links from the content + return $urls; + } + + //HEAD failed, do a normal GET + $req->setMethod(\HTTP_Request2::METHOD_GET); + $res = $req->send(); + if (intval($res->getStatus() / 100) >= 4) { + return $urls; + } + + //yes, maybe the server does return this header now + // e.g. PHP's Phar::webPhar() does not work with HEAD + // https://bugs.php.net/bug.php?id=51918 + $urls = array_merge($this->extractHeader($res), $urls); + if (count($urls) === 2) { + return $this->absolutifyUrls($urls, $base); + } + + //FIXME: atom/rss + $body = $res->getBody(); + $doc = $this->loadHtml($body, $res); + + $xpath = new \DOMXPath($doc); + $xpath->registerNamespace('h', 'http://www.w3.org/1999/xhtml'); + + $nodeList = $xpath->query( + '/*[self::html or self::h:html]' + . '/*[self::head or self::h:head]' + . '/*[(self::link or self::h:link)' + . ' and' + . ' (' + . ' contains(concat(" ", normalize-space(@rel), " "), " hub ")' + . ' or' + . ' contains(concat(" ", normalize-space(@rel), " "), " canonical ")' + . ' or' + . ' contains(concat(" ", normalize-space(@rel), " "), " self ")' + . ' )' + . ']' + ); + + if ($nodeList->length == 0) { + //topic has no links + return $urls; + } + + foreach ($nodeList as $link) { + $uri = $link->attributes->getNamedItem('href')->nodeValue; + $types = explode( + ' ', $link->attributes->getNamedItem('rel')->nodeValue + ); + foreach ($types as $type) { + if ($type == 'canonical') { + $type = 'self'; + } + if ($type == 'hub' || $type == 'self' + && !isset($urls[$type]) + ) { + $urls[$type] = $uri; + } + } + } + + //FIXME: base href + return $this->absolutifyUrls($urls, $base); + } + + /** + * Extract hub url from the HTTP response headers. + * + * @param object $res HTTP response + * + * @return array Array with maximal two keys: hub and self + */ + protected function extractHeader(\HTTP_Request2_Response $res) + { + $http = new \HTTP2(); + + $urls = array(); + $links = $http->parseLinks($res->getHeader('Link')); + foreach ($links as $link) { + if (isset($link['_uri']) && isset($link['rel'])) { + if (!isset($urls['hub']) + && array_search('hub', $link['rel']) !== false + ) { + $urls['hub'] = $link['_uri']; + } + if (!isset($urls['self']) + && array_search('self', $link['rel']) !== false + ) { + $urls['self'] = $link['_uri']; + } + } + } + return $urls; + } + + /** + * Load a DOMDocument from the given HTML or XML + * + * @param string $sourceBody Content of $source URI + * @param object $res HTTP response from fetching $source + * + * @return \DOMDocument DOM document object with HTML/XML loaded + */ + protected static function loadHtml($sourceBody, \HTTP_Request2_Response $res) + { + $doc = new \DOMDocument(); + + libxml_clear_errors(); + $old = libxml_use_internal_errors(true); + + $typeParts = explode(';', $res->getHeader('content-type')); + $type = $typeParts[0]; + if ($type == 'application/xhtml+xml' + || $type == 'application/xml' + || $type == 'text/xml' + ) { + $doc->loadXML($sourceBody); + } else { + $doc->loadHTML($sourceBody); + } + + libxml_clear_errors(); + libxml_use_internal_errors($old); + + return $doc; + } + + /** + * Returns the HTTP request object clone that can be used + * for one HTTP request. + * + * @return HTTP_Request2 Clone of the setRequest() object + */ + public function getRequest() + { + if ($this->request === null) { + $request = new \HTTP_Request2(); + $request->setConfig('follow_redirects', true); + $this->setRequestTemplate($request); + } + + //we need to clone because previous requests could have + //set internal variables like POST data that we don't want now + return clone $this->request; + } + + /** + * Sets a custom HTTP request object that will be used to do HTTP requests + * + * @param object $request Request object + * + * @return self + */ + public function setRequestTemplate(\HTTP_Request2 $request) + { + $this->request = $request; + return $this; + } + + /** + * Make the list of urls absolute + * + * @param array $urls Array of maybe relative URLs + * @param object $base Base URL to resolve the relatives against + * + * @return array List of absolute URLs + */ + protected function absolutifyUrls($urls, \Net_URL2 $base) + { + foreach ($urls as $key => $url) { + $urls[$key] = (string) $base->resolve($url); + } + return $urls; + } +} +?> diff --git a/src/phinde/Subscriptions.php b/src/phinde/Subscriptions.php index 9db4b16..4d00ab8 100644 --- a/src/phinde/Subscriptions.php +++ b/src/phinde/Subscriptions.php @@ -1,12 +1,159 @@ <?php namespace phinde; +/** + * Database table containing information about Pubsubhubbub subscriptions + */ class Subscriptions { - public function get($url) + protected $db; + + public function __construct() + { + $this->db = new \PDO( + $GLOBALS['phinde']['db_dsn'], + $GLOBALS['phinde']['db_user'], + $GLOBALS['phinde']['db_pass'] + ); + $this->db->setAttribute(\PDO::ATTR_ERRMODE, \PDO::ERRMODE_EXCEPTION); + } + + /** + * Fetch a topic + * + * @param string $topic Topic URL + * + * @return false|object False if the row does not exist + */ + public function get($topic) + { + $stmt = $this->db->prepare( + 'SELECT * FROM subscriptions' + . ' WHERE sub_topic = :topic' + ); + $stmt->execute([':topic' => $topic]); + + //fetchObject() itself returns FALSE on failure + return $stmt->fetchObject(); + } + + /** + * Create a new subscription entry in database. + * Automatically generates secret, capkey and lease seconds. + * + * This method does NOT: + * - check for duplicates (do it yourself) + * - return the object (fetch it yourself) + * - send subscription requests to the hub + * + * @param string $topic URL to subscribe to + * + * @return void + */ + public function create($topic) + { + $stmt = $this->db->prepare( + 'INSERT INTO subscriptions' + . ' (sub_topic, sub_status, sub_lease_seconds, sub_expires' + . ', sub_secret, sub_capkey, sub_created, sub_updated' + . ', sub_pings, sub_lastping, sub_statusmessage)' + . ' VALUES ' + . ' (:topic, "subscribing", :lease_seconds, "0000-00-00 00:00:00"' + . ', :secret, :capkey, NOW(), NOW()' + . ', 0, "0000-00-00 00:00:00", "")' + ); + $stmt->execute( + [ + ':topic' => $topic, + ':lease_seconds' => 86400 * 30, + ':secret' => bin2hex(openssl_random_pseudo_bytes(16)), + ':capkey' => bin2hex(openssl_random_pseudo_bytes(16)), + ] + ); + } + + /** + * A subscription has been confirmed by the hub - mark it as active. + * + * @param integer $subId Subscription ID + * @param integer $leaseSeconds Number of seconds until subscription expires + * + * @return void + */ + public function subscribed($subId, $leaseSeconds) + { + $this->db->prepare( + 'UPDATE subscriptions' + . ' SET sub_status = "active"' + . ' , sub_lease_seconds = :leaseSeconds' + . ' , sub_expires = :expires' + . ' , sub_updated = NOW()' + . ' WHERE sub_id = :id' + )->execute( + [ + ':leaseSeconds' => $leaseSeconds, + ':expires' => gmdate('Y-m-d H:i:s', time() + $leaseSeconds), + ':id' => $subId, + ] + ); + } + + /** + * Mark a subscription as "unsubscribed" + * + * @param integer $subId Subscription ID + * + * @return void + */ + public function unsubscribed($subId) + { + $this->db->prepare( + 'UPDATE subscriptions' + . ' SET sub_status = "unsubscribed"' + . ' , sub_updated = NOW()' + . ' WHERE sub_id = :id' + )->execute([':id' => $subId]); + } + + public function denied($subId, $reason) + { + $this->db->prepare( + 'UPDATE subscriptions' + . ' SET sub_status = "denied"' + . ' , sub_statusmessage = :reason' + . ' , sub_updated = NOW()' + . ' WHERE sub_id = :id' + )->execute([':id' => $subId, ':reason' => $reason]); + } + + public function pinged($subId) + { + $this->db->prepare( + 'UPDATE subscriptions' + . ' SET sub_pings = sub_pings + 1' + . ' , sub_lastping = NOW()' + . ' , sub_updated = NOW()' + . ' WHERE sub_id = :id' + )->execute([':id' => $subId]); + } + + /** + * Detect the hub for the given topic URL + * + * @param string $url Topic URL + * + * @return array Topic URL and hub URL. Hub URL is NULL if there is none. + */ + public function detectHub($url) { - //FIXME - return false; + $hue = new HubUrlExtractor(); + $hue->setRequestTemplate(new HttpRequest()); + $urls = $hue->getUrls($url); + //we violate the spec by not requiring a self URL + $topicUrl = isset($urls['self']) ? $urls['self'] : $url; + $hubUrl = isset($urls['hub']) ? $urls['hub'] : null; + + return array($topicUrl, $hubUrl); } } ?> |