Oembed slimmed to only do discovery (soon we get og: discovery too)

This commit is contained in:
Mikael Nordfeldth 2015-11-30 02:06:04 +01:00
parent 8bab642cc7
commit 334a0d56e7
4 changed files with 46 additions and 110 deletions

View File

@ -276,10 +276,6 @@ $default =
'maxpeople' => 500, // maximum no. of people with the same tag by the same user 'maxpeople' => 500, // maximum no. of people with the same tag by the same user
'allow_tagging' => array('all' => true), // equivalent to array('local' => true, 'remote' => true) 'allow_tagging' => array('all' => true), // equivalent to array('local' => true, 'remote' => true)
'desclimit' => null), 'desclimit' => null),
'oembed' =>
array('endpoint' => null, // 'https://noembed.com/embed/' for proxied oEmbed data
'order' => array('built-in', 'well-known', 'service', 'discovery'),
),
'search' => 'search' =>
array('type' => 'like'), array('type' => 'like'),
'sessions' => 'sessions' =>

View File

@ -35,6 +35,24 @@ class OembedPlugin extends Plugin
$m->connect('main/oembed', array('action' => 'oembed')); $m->connect('main/oembed', array('action' => 'oembed'));
} }
public function onGetRemoteUrlMetadataFromDom($url, DOMDocument $dom, stdClass &$metadata)
{
try {
common_log(LOG_INFO, 'Trying to discover an oEmbed endpoint using link headers.');
$api = oEmbedHelper::oEmbedEndpointFromHTML($dom);
common_log(LOG_INFO, 'Found API endpoint ' . $api . ' for URL ' . $url);
$params = array(
'maxwidth' => common_config('thumbnail', 'width'),
'maxheight' => common_config('thumbnail', 'height'),
);
$metadata = oEmbedHelper::getOembedFrom($api, $url, $params);
} catch (Exception $e) {
common_log(LOG_INFO, 'Could not find an oEmbed endpoint using link headers.');
// Just ignore it!
}
}
public function onEndShowHeadElements(Action $action) public function onEndShowHeadElements(Action $action)
{ {
switch ($action->getActionName()) { switch ($action->getActionName()) {

View File

@ -68,12 +68,8 @@ class File_oembed extends Managed_DataObject
} }
static function _getOembed($url) { static function _getOembed($url) {
$parameters = array(
'maxwidth' => common_config('thumbnail', 'width'),
'maxheight' => common_config('thumbnail', 'height'),
);
try { try {
return oEmbedHelper::getObject($url, $parameters); return oEmbedHelper::getObject($url);
} catch (Exception $e) { } catch (Exception $e) {
common_log(LOG_INFO, "Error during oembed lookup for $url - " . $e->getMessage()); common_log(LOG_INFO, "Error during oembed lookup for $url - " . $e->getMessage());
return false; return false;

View File

@ -17,9 +17,7 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>. * along with this program. If not, see <http://www.gnu.org/licenses/>.
*/ */
if (!defined('STATUSNET')) { if (!defined('GNUSOCIAL')) { exit(1); }
exit(1);
}
/** /**
@ -47,8 +45,6 @@ class oEmbedHelper
'revision3.com' => 'https://revision3.com/api/oembed/', 'revision3.com' => 'https://revision3.com/api/oembed/',
'vimeo.com' => 'https://vimeo.com/api/oembed.json', 'vimeo.com' => 'https://vimeo.com/api/oembed.json',
); );
protected static $functionMap = array(
);
/** /**
* Perform or fake an oEmbed lookup for the given resource. * Perform or fake an oEmbed lookup for the given resource.
@ -71,74 +67,13 @@ class oEmbedHelper
*/ */
public static function getObject($url, $params=array()) public static function getObject($url, $params=array())
{ {
$host = parse_url($url, PHP_URL_HOST); common_log(LOG_INFO, 'Checking for remote URL metadata for ' . $url);
if (substr($host, 0, 4) == 'www.') {
$host = substr($host, 4);
}
common_log(LOG_INFO, 'Checking for oEmbed data for ' . $url); // TODO: Make this class something like UrlMetadata, or use a dataobject?
$metadata = new stdClass();
// You can fiddle with the order of discovery -- either skipping if (Event::handle('GetRemoteUrlMetadata', array($url, &$metadata))) {
// some types or re-ordering them. // If that event didn't return anything, try downloading the body and parse it
$order = common_config('oembed', 'order');
foreach ($order as $method) {
switch ($method) {
case 'built-in':
common_log(LOG_INFO, 'Considering built-in oEmbed methods...');
// Blacklist: systems with no oEmbed API of their own, which are
// either missing from or broken on noembed.com's proxy.
// we know how to look data up in another way...
if (array_key_exists($host, self::$functionMap)) {
common_log(LOG_INFO, 'We have a built-in method for ' . $host);
$func = self::$functionMap[$host];
return call_user_func($func, $url, $params);
}
break;
case 'well-known':
common_log(LOG_INFO, 'Considering well-known oEmbed endpoints...');
// Whitelist: known API endpoints for sites that don't provide discovery...
if (array_key_exists($host, self::$apiMap)) {
$api = self::$apiMap[$host];
common_log(LOG_INFO, 'Using well-known endpoint "' . $api . '" for "' . $host . '"');
break 2;
}
break;
case 'discovery':
try {
common_log(LOG_INFO, 'Trying to discover an oEmbed endpoint using link headers.');
$api = self::discover($url);
common_log(LOG_INFO, 'Found API endpoint ' . $api . ' for URL ' . $url);
break 2;
} catch (Exception $e) {
common_log(LOG_INFO, 'Could not find an oEmbed endpoint using link headers.');
// Just ignore it!
}
break;
case 'service':
$api = common_config('oembed', 'endpoint');
common_log(LOG_INFO, 'Using service API endpoint ' . $api);
break;
}
}
if (empty($api)) {
// TRANS: Server exception thrown in oEmbed action if no API endpoint is available.
throw new ServerException(_('No oEmbed API endpoint available.'));
}
return self::getObjectFrom($api, $url, $params);
}
/**
* Perform basic discovery.
* @return string
*/
static function discover($url)
{
// @fixme ideally skip this for non-HTML stuff!
$body = HTTPClient::quickGet($url); $body = HTTPClient::quickGet($url);
// DOMDocument::loadHTML may throw warnings on unrecognized elements, // DOMDocument::loadHTML may throw warnings on unrecognized elements,
@ -146,13 +81,17 @@ class oEmbedHelper
$old = error_reporting(error_reporting() & ~(E_WARNING | E_NOTICE)); $old = error_reporting(error_reporting() & ~(E_WARNING | E_NOTICE));
$dom = new DOMDocument(); $dom = new DOMDocument();
$ok = $dom->loadHTML($body); $ok = $dom->loadHTML($body);
unset($body); // storing the DOM in memory is enough...
error_reporting($old); error_reporting($old);
if (!$ok) { if (!$ok) {
throw new oEmbedHelper_BadHtmlException(); throw new oEmbedHelper_BadHtmlException();
} }
return self::discoverFromHTML($url, $dom); Event::handle('GetRemoteUrlMetadataFromDom', array($url, $dom, &$metadata));
}
return self::normalize($metadata);
} }
/** /**
@ -162,7 +101,7 @@ class oEmbedHelper
* @param string $body HTML body text * @param string $body HTML body text
* @return mixed string with URL or false if no target found * @return mixed string with URL or false if no target found
*/ */
static function discoverFromHTML($url, DOMDocument $dom) static function oEmbedEndpointFromHTML(DOMDocument $dom)
{ {
// Ok... now on to the links! // Ok... now on to the links!
$feeds = array( $feeds = array(
@ -207,16 +146,19 @@ class oEmbedHelper
* @param array $params * @param array $params
* @return object * @return object
*/ */
static function getObjectFrom($api, $url, $params=array()) static function getOembedFrom($api, $url, $params=array())
{ {
if (empty($api)) {
// TRANS: Server exception thrown in oEmbed action if no API endpoint is available.
throw new ServerException(_('No oEmbed API endpoint available.'));
}
$params['url'] = $url; $params['url'] = $url;
$params['format'] = 'json'; $params['format'] = 'json';
$key=common_config('oembed','apikey'); $key=common_config('oembed','apikey');
if(isset($key)) { if(isset($key)) {
$params['key'] = common_config('oembed','apikey'); $params['key'] = common_config('oembed','apikey');
} }
$data = self::json($api, $params); return HTTPClient::quickGetJson($api, $params);
return self::normalize($data);
} }
/** /**
@ -225,14 +167,11 @@ class oEmbedHelper
* @param object $orig * @param object $orig
* @return object * @return object
*/ */
static function normalize($orig) static function normalize(stdClass $data)
{ {
$data = clone($orig);
if (empty($data->type)) { if (empty($data->type)) {
throw new Exception('Invalid oEmbed data: no type field.'); throw new Exception('Invalid oEmbed data: no type field.');
} }
if ($data->type == 'image') { if ($data->type == 'image') {
// YFrog does this. // YFrog does this.
$data->type = 'photo'; $data->type = 'photo';
@ -248,19 +187,6 @@ class oEmbedHelper
return $data; return $data;
} }
/**
* Fetch some URL and return JSON data.
*
* @param string $url
* @param array $params query-string params
* @return object
*/
static protected function json($url, $params=array())
{
$data = HTTPClient::quickGet($url, null, $params);
return json_decode($data);
}
} }
class oEmbedHelper_Exception extends Exception class oEmbedHelper_Exception extends Exception