2014-05-06 22:00:30 +01:00
< ? php
2015-01-17 11:06:08 +00:00
if ( ! defined ( 'GNUSOCIAL' )) { exit ( 1 ); }
2014-05-06 22:00:30 +01:00
class OembedPlugin extends Plugin
{
2015-01-25 01:34:40 +00:00
// settings which can be set in config.php with addPlugin('Oembed', array('param'=>'value', ...));
2015-01-25 10:18:57 +00:00
// WARNING, these are _regexps_ (slashes added later). Always escape your dots and end your strings
2015-01-25 01:34:40 +00:00
public $domain_whitelist = array ( // hostname => service provider
2015-01-25 10:18:57 +00:00
'^i\d*\.ytimg\.com$' => 'YouTube' ,
2015-02-02 10:13:52 +00:00
'^i\d*\.vimeocdn\.com$' => 'Vimeo' ,
2015-01-25 01:34:40 +00:00
);
public $append_whitelist = array (); // fill this array as domain_whitelist to add more trusted sources
2015-04-03 20:00:19 +01:00
public $check_whitelist = false ; // security/abuse precaution
2015-01-25 01:34:40 +00:00
protected $imgData = array ();
// these should be declared protected everywhere
public function initialize ()
{
parent :: initialize ();
$this -> domain_whitelist = array_merge ( $this -> domain_whitelist , $this -> append_whitelist );
}
2014-05-06 22:00:30 +01:00
public function onCheckSchema ()
{
$schema = Schema :: get ();
$schema -> ensureTable ( 'file_oembed' , File_oembed :: schemaDef ());
return true ;
}
public function onRouterInitialized ( URLMapper $m )
{
$m -> connect ( 'main/oembed' , array ( 'action' => 'oembed' ));
}
2015-11-30 01:06:04 +00:00
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 );
2016-01-12 14:29:03 +00:00
common_log ( LOG_INFO , 'Found oEmbed API endpoint ' . $api . ' for URL ' . $url );
2015-11-30 01:06:04 +00:00
$params = array (
'maxwidth' => common_config ( 'thumbnail' , 'width' ),
'maxheight' => common_config ( 'thumbnail' , 'height' ),
);
$metadata = oEmbedHelper :: getOembedFrom ( $api , $url , $params );
2016-01-26 01:05:53 +00:00
2016-01-26 01:07:44 +00:00
// Facebook just gives us javascript in its oembed html,
// so use the content of the title element instead
2016-01-26 01:05:53 +00:00
if ( strpos ( $url , 'https://www.facebook.com/' ) === 0 ) {
2016-01-26 11:28:24 +00:00
$metadata -> html = @ $dom -> getElementsByTagName ( 'title' ) -> item ( 0 ) -> nodeValue ;
2016-01-26 01:05:53 +00:00
}
2016-01-26 01:07:44 +00:00
// Wordpress sometimes also just gives us javascript, use og:description if it is available
$xpath = new DomXpath ( $dom );
2016-01-26 11:28:24 +00:00
$generatorNode = @ $xpath -> query ( '//meta[@name="generator"][1]' ) -> item ( 0 );
2016-01-26 01:07:44 +00:00
if ( $generatorNode instanceof DomElement ) {
// when wordpress only gives us javascript, the html stripped from tags
// is the same as the title, so this helps us to identify this (common) case
if ( strpos ( $generatorNode -> getAttribute ( 'content' ), 'WordPress' ) === 0
&& trim ( strip_tags ( $metadata -> html )) == trim ( $metadata -> title )) {
2016-01-26 11:28:24 +00:00
$propertyNode = @ $xpath -> query ( '//meta[@property="og:description"][1]' ) -> item ( 0 );
2016-01-26 01:07:44 +00:00
if ( $propertyNode instanceof DomElement ) {
$metadata -> html = $propertyNode -> getAttribute ( 'content' );
}
}
}
2015-11-30 01:06:04 +00:00
} catch ( Exception $e ) {
2016-01-12 14:29:03 +00:00
common_log ( LOG_INFO , 'Could not find an oEmbed endpoint using link headers, trying OpenGraph from HTML.' );
2015-11-30 01:06:04 +00:00
// Just ignore it!
2016-01-12 14:29:03 +00:00
$metadata = OpenGraphHelper :: ogFromHtml ( $dom );
2015-11-30 01:06:04 +00:00
}
2016-01-28 15:19:29 +00:00
2016-02-25 18:46:17 +00:00
if ( isset ( $metadata -> thumbnail_url )) {
// sometimes sites serve the path, not the full URL, for images
// let's "be liberal in what you accept from others"!
// add protocol and host if the thumbnail_url starts with /
if ( substr ( $metadata -> thumbnail_url , 0 , 1 ) == '/' ) {
$thumbnail_url_parsed = parse_url ( $metadata -> url );
$metadata -> thumbnail_url = $thumbnail_url_parsed [ 'scheme' ] . " :// " . $thumbnail_url_parsed [ 'host' ] . $metadata -> thumbnail_url ;
}
2016-01-28 15:19:29 +00:00
2016-02-25 18:46:17 +00:00
// some wordpress opengraph implementations sometimes return a white blank image
// no need for us to save that!
if ( $metadata -> thumbnail_url == 'https://s0.wp.com/i/blank.jpg' ) {
unset ( $metadata -> thumbnail_url );
}
2016-01-28 15:19:29 +00:00
}
2015-11-30 01:06:04 +00:00
}
2014-05-06 22:00:30 +01:00
public function onEndShowHeadElements ( Action $action )
{
switch ( $action -> getActionName ()) {
case 'attachment' :
$action -> element ( 'link' , array ( 'rel' => 'alternate' ,
'type' => 'application/json+oembed' ,
'href' => common_local_url (
'oembed' ,
array (),
array ( 'format' => 'json' , 'url' =>
common_local_url ( 'attachment' ,
2016-07-21 00:38:48 +01:00
array ( 'attachment' => $action -> attachment -> getID ())))),
2014-05-06 22:00:30 +01:00
'title' => 'oEmbed' ), null );
$action -> element ( 'link' , array ( 'rel' => 'alternate' ,
'type' => 'text/xml+oembed' ,
'href' => common_local_url (
'oembed' ,
array (),
array ( 'format' => 'xml' , 'url' =>
common_local_url ( 'attachment' ,
2016-07-21 00:38:48 +01:00
array ( 'attachment' => $action -> attachment -> getID ())))),
2014-05-06 22:00:30 +01:00
'title' => 'oEmbed' ), null );
break ;
case 'shownotice' :
2015-12-31 00:55:18 +00:00
if ( ! $action -> notice -> isLocal ()) {
break ;
}
2014-05-10 12:06:18 +01:00
try {
$action -> element ( 'link' , array ( 'rel' => 'alternate' ,
'type' => 'application/json+oembed' ,
'href' => common_local_url (
'oembed' ,
array (),
array ( 'format' => 'json' , 'url' => $action -> notice -> getUrl ())),
'title' => 'oEmbed' ), null );
$action -> element ( 'link' , array ( 'rel' => 'alternate' ,
'type' => 'text/xml+oembed' ,
'href' => common_local_url (
'oembed' ,
array (),
array ( 'format' => 'xml' , 'url' => $action -> notice -> getUrl ())),
'title' => 'oEmbed' ), null );
} catch ( InvalidUrlException $e ) {
// The notice is probably a share or similar, which don't
// have a representational URL of their own.
}
2014-05-06 22:00:30 +01:00
break ;
}
return true ;
}
2016-03-16 23:31:45 +00:00
public function onEndShowStylesheets ( Action $action ) {
$action -> cssLink ( $this -> path ( 'css/oembed.css' ));
return true ;
}
2014-05-06 22:00:30 +01:00
/**
* Save embedding information for a File , if applicable .
*
* Normally this event is called through File :: saveNew ()
*
* @ param File $file The newly inserted File object .
*
* @ return boolean success
*/
2015-11-02 05:15:08 +00:00
public function onEndFileSaveNew ( File $file )
2014-05-06 22:00:30 +01:00
{
2016-07-21 00:38:48 +01:00
$fo = File_oembed :: getKV ( 'file_id' , $file -> getID ());
2014-05-06 22:00:30 +01:00
if ( $fo instanceof File_oembed ) {
2016-07-21 00:38:48 +01:00
common_log ( LOG_WARNING , " Strangely, a File_oembed object exists for new file { $file -> getID () } " , __FILE__ );
2015-10-01 21:18:33 +01:00
return true ;
2014-05-06 22:00:30 +01:00
}
2015-11-02 05:15:08 +00:00
if ( isset ( $file -> mimetype )
&& (( 'text/html' === substr ( $file -> mimetype , 0 , 9 )
|| 'application/xhtml+xml' === substr ( $file -> mimetype , 0 , 21 )))) {
2014-05-06 22:00:30 +01:00
try {
2015-11-02 05:15:08 +00:00
$oembed_data = File_oembed :: _getOembed ( $file -> url );
2014-05-06 22:00:30 +01:00
if ( $oembed_data === false ) {
throw new Exception ( 'Did not get oEmbed data from URL' );
}
2016-07-23 20:01:28 +01:00
$file -> setTitle ( $oembed_data -> title );
2014-05-06 22:00:30 +01:00
} catch ( Exception $e ) {
2016-08-27 23:16:31 +01:00
common_log ( LOG_WARNING , sprintf ( __METHOD__ . ': %s thrown when getting oEmbed data: %s' , get_class ( $e ), _ve ( $e -> getMessage ())));
2014-05-06 22:00:30 +01:00
return true ;
}
2016-07-21 00:38:48 +01:00
File_oembed :: saveNew ( $oembed_data , $file -> getID ());
2014-05-06 22:00:30 +01:00
}
return true ;
}
public function onEndShowAttachmentLink ( HTMLOutputter $out , File $file )
{
2016-07-21 00:38:48 +01:00
$oembed = File_oembed :: getKV ( 'file_id' , $file -> getID ());
2014-05-06 22:00:30 +01:00
if ( empty ( $oembed -> author_name ) && empty ( $oembed -> provider )) {
return true ;
}
2014-06-21 20:01:17 +01:00
$out -> elementStart ( 'div' , array ( 'id' => 'oembed_info' , 'class' => 'e-content' ));
2014-05-06 22:00:30 +01:00
if ( ! empty ( $oembed -> author_name )) {
$out -> elementStart ( 'div' , 'fn vcard author' );
if ( empty ( $oembed -> author_url )) {
$out -> text ( $oembed -> author_name );
} else {
$out -> element ( 'a' , array ( 'href' => $oembed -> author_url ,
'class' => 'url' ),
$oembed -> author_name );
}
}
if ( ! empty ( $oembed -> provider )) {
$out -> elementStart ( 'div' , 'fn vcard' );
if ( empty ( $oembed -> provider_url )) {
$out -> text ( $oembed -> provider );
} else {
$out -> element ( 'a' , array ( 'href' => $oembed -> provider_url ,
'class' => 'url' ),
$oembed -> provider );
}
}
$out -> elementEnd ( 'div' );
}
public function onFileEnclosureMetadata ( File $file , & $enclosure )
{
// Never treat generic HTML links as an enclosure type!
// But if we have oEmbed info, we'll consider it golden.
2016-07-21 00:38:48 +01:00
$oembed = File_oembed :: getKV ( 'file_id' , $file -> getID ());
2015-04-05 14:44:04 +01:00
if ( ! $oembed instanceof File_oembed || ! in_array ( $oembed -> type , array ( 'photo' , 'video' ))) {
2014-05-06 22:00:30 +01:00
return true ;
}
2016-01-14 17:29:21 +00:00
foreach ( array ( 'mimetype' , 'url' , 'title' , 'modified' , 'width' , 'height' ) as $key ) {
if ( isset ( $oembed -> { $key }) && ! empty ( $oembed -> { $key })) {
2014-05-06 22:00:30 +01:00
$enclosure -> { $key } = $oembed -> { $key };
}
}
return true ;
}
2016-03-16 23:31:45 +00:00
public function onStartShowAttachmentRepresentation ( HTMLOutputter $out , File $file )
{
try {
$oembed = File_oembed :: getByFile ( $file );
} catch ( NoResultException $e ) {
return true ;
}
2016-03-22 13:02:36 +00:00
// Show thumbnail as usual if it's a photo.
if ( $oembed -> type === 'photo' ) {
return true ;
}
2016-03-27 15:21:43 +01:00
$out -> elementStart ( 'article' , [ 'class' => 'h-entry oembed' ]);
2016-03-16 23:31:45 +00:00
$out -> elementStart ( 'header' );
try {
$thumb = $file -> getThumbnail ( 128 , 128 );
2016-03-27 15:21:43 +01:00
$out -> element ( 'img' , $thumb -> getHtmlAttrs ([ 'class' => 'u-photo oembed' ]));
2016-03-16 23:31:45 +00:00
unset ( $thumb );
} catch ( Exception $e ) {
$out -> element ( 'div' , [ 'class' => 'error' ], $e -> getMessage ());
}
2016-03-27 15:21:43 +01:00
$out -> elementStart ( 'h5' , [ 'class' => 'p-name oembed' ]);
$out -> element ( 'a' , [ 'class' => 'u-url' , 'href' => $file -> getUrl ()], common_strip_html ( $oembed -> title ));
2016-03-16 23:31:45 +00:00
$out -> elementEnd ( 'h5' );
2016-03-27 15:21:43 +01:00
$out -> elementStart ( 'div' , [ 'class' => 'p-author oembed' ]);
2016-03-16 23:31:45 +00:00
if ( ! empty ( $oembed -> author_name )) {
// TRANS: text before the author name of oEmbed attachment representation
// FIXME: The whole "By x from y" should be i18n because of different language constructions.
$out -> text ( _ ( 'By ' ));
2016-03-27 15:21:43 +01:00
$attrs = [ 'class' => 'h-card p-author' ];
2016-03-16 23:31:45 +00:00
if ( ! empty ( $oembed -> author_url )) {
$attrs [ 'href' ] = $oembed -> author_url ;
$tag = 'a' ;
} else {
$tag = 'span' ;
}
$out -> element ( $tag , $attrs , $oembed -> author_name );
}
if ( ! empty ( $oembed -> provider )) {
// TRANS: text between the oEmbed author name and provider url
// FIXME: The whole "By x from y" should be i18n because of different language constructions.
$out -> text ( _ ( ' from ' ));
$attrs = [ 'class' => 'h-card' ];
if ( ! empty ( $oembed -> provider_url )) {
$attrs [ 'href' ] = $oembed -> provider_url ;
$tag = 'a' ;
} else {
$tag = 'span' ;
}
$out -> element ( $tag , $attrs , $oembed -> provider );
}
$out -> elementEnd ( 'div' );
$out -> elementEnd ( 'header' );
2016-03-27 15:21:43 +01:00
$out -> elementStart ( 'div' , [ 'class' => 'p-summary oembed' ]);
2016-03-17 11:58:40 +00:00
$out -> raw ( common_purify ( $oembed -> html ));
$out -> elementEnd ( 'div' );
2016-03-16 23:31:45 +00:00
$out -> elementStart ( 'footer' );
$out -> elementEnd ( 'footer' );
$out -> elementEnd ( 'article' );
return false ;
}
2014-05-06 22:00:30 +01:00
2016-01-07 16:35:37 +00:00
public function onShowUnsupportedAttachmentRepresentation ( HTMLOutputter $out , File $file )
2014-05-06 22:00:30 +01:00
{
2015-10-01 21:18:33 +01:00
try {
$oembed = File_oembed :: getByFile ( $file );
} catch ( NoResultException $e ) {
2014-05-06 22:00:30 +01:00
return true ;
}
2015-10-01 21:18:33 +01:00
2016-01-07 16:35:37 +00:00
// the 'photo' type is shown through ordinary means, using StartShowAttachmentRepresentation!
2014-05-06 22:00:30 +01:00
switch ( $oembed -> type ) {
case 'video' :
case 'link' :
2014-05-18 23:54:43 +01:00
if ( ! empty ( $oembed -> html )
2015-02-27 11:44:15 +00:00
&& ( GNUsocial :: isAjax () || common_config ( 'attachments' , 'show_html' ))) {
2016-01-28 17:57:36 +00:00
require_once INSTALLDIR . '/extlib/HTMLPurifier/HTMLPurifier.auto.php' ;
$purifier = new HTMLPurifier ();
// FIXME: do we allow <object> and <embed> here? we did that when we used htmLawed, but I'm not sure anymore...
$out -> raw ( $purifier -> purify ( $oembed -> html ));
2014-05-06 22:00:30 +01:00
}
2016-01-07 16:35:37 +00:00
return false ;
2014-05-06 22:00:30 +01:00
break ;
}
2016-01-07 16:35:37 +00:00
return true ;
2014-05-06 22:00:30 +01:00
}
2015-01-25 01:34:40 +00:00
public function onCreateFileImageThumbnailSource ( File $file , & $imgPath , $media = null )
{
// If we are on a private node, we won't do any remote calls (just as a precaution until
// we can configure this from config.php for the private nodes)
if ( common_config ( 'site' , 'private' )) {
return true ;
}
// All our remote Oembed images lack a local filename property in the File object
2015-10-01 21:18:33 +01:00
if ( ! is_null ( $file -> filename )) {
2015-01-25 01:34:40 +00:00
return true ;
}
try {
// If we have proper oEmbed data, there should be an entry in the File_oembed
// and File_thumbnail tables respectively. If not, we're not going to do anything.
2015-10-01 21:18:33 +01:00
$file_oembed = File_oembed :: getByFile ( $file );
2015-01-25 01:34:40 +00:00
$thumbnail = File_thumbnail :: byFile ( $file );
2016-02-10 03:43:30 +00:00
} catch ( NoResultException $e ) {
2015-01-25 01:34:40 +00:00
// Not Oembed data, or at least nothing we either can or want to use.
return true ;
}
try {
$this -> storeRemoteFileThumbnail ( $thumbnail );
} catch ( AlreadyFulfilledException $e ) {
// aw yiss!
}
$imgPath = $thumbnail -> getPath ();
return false ;
}
/**
2015-01-25 10:18:57 +00:00
* @ return boolean false on no check made , provider name on success
2015-01-25 01:34:40 +00:00
* @ throws ServerException if check is made but fails
*/
protected function checkWhitelist ( $url )
{
if ( ! $this -> check_whitelist ) {
return false ; // indicates "no check made"
}
$host = parse_url ( $url , PHP_URL_HOST );
2015-01-25 10:18:57 +00:00
foreach ( $this -> domain_whitelist as $regex => $provider ) {
if ( preg_match ( " / $regex / " , $host )) {
return $provider ; // we trust this source, return provider name
}
2015-01-25 01:34:40 +00:00
}
2015-01-25 10:18:57 +00:00
throw new ServerException ( sprintf ( _ ( 'Domain not in remote thumbnail source whitelist: %s' ), $host ));
2015-01-25 01:34:40 +00:00
}
protected function storeRemoteFileThumbnail ( File_thumbnail $thumbnail )
{
if ( ! empty ( $thumbnail -> filename ) && file_exists ( $thumbnail -> getPath ())) {
2016-07-21 02:19:05 +01:00
throw new AlreadyFulfilledException ( sprintf ( 'A thumbnail seems to already exist for remote file with id==%u' , $thumbnail -> getFileId ()));
2015-01-25 01:34:40 +00:00
}
2016-07-21 02:19:05 +01:00
$remoteUrl = $thumbnail -> getUrl ();
$this -> checkWhitelist ( $remoteUrl );
2015-01-25 01:34:40 +00:00
2016-07-21 02:19:05 +01:00
$http = new HTTPClient ();
// First see if it's too large for us
common_debug ( __METHOD__ . ': ' . sprintf ( 'Performing HEAD request for remote file id==%u to avoid unnecessarily downloading too large files. URL: %s' , $thumbnail -> getFileId (), $remoteUrl ));
$head = $http -> head ( $remoteUrl );
2017-07-10 13:41:03 +01:00
if ( ! $head -> isOk ()) {
common_log ( LOG_WARN , 'HEAD request returned HTTP failure, so we will abort now and delete the thumbnail object.' );
$thumbnail -> delete ();
return false ;
} else {
common_debug ( 'HEAD request returned HTTP success, so we will continue.' );
}
2016-07-21 02:19:05 +01:00
$remoteUrl = $head -> getEffectiveUrl (); // to avoid going through redirects again
$headers = $head -> getHeader ();
$filesize = isset ( $headers [ 'content-length' ]) ? $headers [ 'content-length' ] : null ;
// FIXME: I just copied some checks from StoreRemoteMedia, maybe we should have other checks for thumbnails? Or at least embed into some class somewhere.
if ( empty ( $filesize )) {
// file size not specified on remote server
common_debug ( sprintf ( '%s: Ignoring remote thumbnail because we did not get a content length for thumbnail for file id==%u' , __CLASS__ , $thumbnail -> getFileId ()));
return true ;
} elseif ( $filesize > common_config ( 'attachments' , 'file_quota' )) {
// file too big according to site configuration
common_debug ( sprintf ( '%s: Skip downloading remote thumbnail because content length (%u) is larger than file_quota (%u) for file id==%u' , __CLASS__ , intval ( $filesize ), common_config ( 'attachments' , 'file_quota' ), $thumbnail -> getFileId ()));
return true ;
}
// Then we download the file to memory and test whether it's actually an image file
2015-01-25 01:34:40 +00:00
// FIXME: To support remote video/whatever files, this needs reworking.
2016-07-21 02:19:05 +01:00
common_debug ( sprintf ( 'Downloading remote thumbnail for file id==%u (should be size %u) with effective URL: %s' , $thumbnail -> getFileId (), $filesize , _ve ( $remoteUrl )));
$imgData = HTTPClient :: quickGet ( $remoteUrl );
2015-01-25 01:34:40 +00:00
$info = @ getimagesizefromstring ( $imgData );
if ( $info === false ) {
throw new UnsupportedMediaException ( _ ( 'Remote file format was not identified as an image.' ), $url );
} elseif ( ! $info [ 0 ] || ! $info [ 1 ]) {
throw new UnsupportedMediaException ( _ ( 'Image file had impossible geometry (0 width or height)' ));
}
2016-02-25 21:31:45 +00:00
$ext = File :: guessMimeExtension ( $info [ 'mime' ]);
2015-10-01 21:18:33 +01:00
// We'll trust sha256 (File::FILEHASH_ALG) not to have collision issues any time soon :)
2016-03-07 21:33:34 +00:00
$filename = 'oembed-' . hash ( File :: FILEHASH_ALG , $imgData ) . " . { $ext } " ;
2015-01-25 01:34:40 +00:00
$fullpath = File_thumbnail :: path ( $filename );
// Write the file to disk. Throw Exception on failure
if ( ! file_exists ( $fullpath ) && file_put_contents ( $fullpath , $imgData ) === false ) {
throw new ServerException ( _ ( 'Could not write downloaded file to disk.' ));
}
// Get rid of the file from memory
unset ( $imgData );
// Updated our database for the file record
$orig = clone ( $thumbnail );
$thumbnail -> filename = $filename ;
$thumbnail -> width = $info [ 0 ]; // array indexes documented on php.net:
$thumbnail -> height = $info [ 1 ]; // https://php.net/manual/en/function.getimagesize.php
2015-01-25 11:45:26 +00:00
// Throws exception on failure.
2016-01-28 15:42:59 +00:00
$thumbnail -> updateWithKeys ( $orig );
2015-01-25 01:34:40 +00:00
}
2014-05-06 22:00:30 +01:00
public function onPluginVersion ( array & $versions )
{
$versions [] = array ( 'name' => 'Oembed' ,
'version' => GNUSOCIAL_VERSION ,
'author' => 'Mikael Nordfeldth' ,
'homepage' => 'http://gnu.io/' ,
'description' =>
// TRANS: Plugin description.
_m ( 'Plugin for using and representing Oembed data.' ));
return true ;
}
}