wp-includes/class-oembed.php

   1 <?php
   2 /**
   3  * API for fetching the HTML to embed remote content based on a provided URL.
   4  * Used internally by the {@link WP_Embed} class, but is designed to be generic.
   5  *
   6  * @link http://codex.wordpress.org/oEmbed oEmbed Codex Article
   7  * @link http://oembed.com/ oEmbed Homepage
   8  *
   9  * @package WordPress
  10  * @subpackage oEmbed
  11  */
  12
  13 /**
  14  * oEmbed class.
  15  *
  16  * @package WordPress
  17  * @subpackage oEmbed
  18  * @since 2.9.0
  19  */
  20 class WP_oEmbed {
  21         var $providers = array();
  22
  23         /**
  24          * Constructor
  25          *
  26          * @uses apply_filters() Filters a list of pre-defined oEmbed providers.
  27          */
  28         function __construct() {
  29                 $providers = array(
  30                         '#http://(www\.)?youtube\.com/watch.*#i'              => array( 'http://www.youtube.com/oembed',                      true  ),
  31                         '#https://(www\.)?youtube\.com/watch.*#i'             => array( 'http://www.youtube.com/oembed?scheme=https',         true  ),
  32                         '#http://youtu\.be/.*#i'                              => array( 'http://www.youtube.com/oembed',                      true  ),
  33                         '#https://youtu\.be/.*#i'                             => array( 'http://www.youtube.com/oembed?scheme=https',         true  ),
  34                         'http://blip.tv/*'                                    => array( 'http://blip.tv/oembed/',                             false ),
  35                         '#https?://(.+\.)?vimeo\.com/.*#i'                    => array( 'http://vimeo.com/api/oembed.{format}',               true  ),
  36                         '#https?://(www\.)?dailymotion\.com/.*#i'             => array( 'http://www.dailymotion.com/services/oembed',         true  ),
  37                         'http://dai.ly/*'                                     => array( 'http://www.dailymotion.com/services/oembed',         false ),
  38                         '#https?://(www\.)?flickr\.com/.*#i'                  => array( 'http://www.flickr.com/services/oembed/',             true  ),
  39                         'http://flic.kr/*'                                    => array( 'http://www.flickr.com/services/oembed/',             false ),
  40                         '#https?://(.+\.)?smugmug\.com/.*#i'                  => array( 'http://api.smugmug.com/services/oembed/',            true  ),
  41                         '#https?://(www\.)?hulu\.com/watch/.*#i'              => array( 'http://www.hulu.com/api/oembed.{format}',            true  ),
  42                         '#https?://(www\.)?viddler\.com/.*#i'                 => array( 'http://lab.viddler.com/services/oembed/',            true  ),
  43                         'http://revision3.com/*'                              => array( 'http://revision3.com/api/oembed/',                   false ),
  44                         'http://i*.photobucket.com/albums/*'                  => array( 'http://photobucket.com/oembed',                      false ),
  45                         'http://gi*.photobucket.com/groups/*'                 => array( 'http://photobucket.com/oembed',                      false ),
  46                         '#https?://(www\.)?scribd\.com/.*#i'                  => array( 'http://www.scribd.com/services/oembed',              true  ),
  47                         'http://wordpress.tv/*'                               => array( 'http://wordpress.tv/oembed/',                        false ),
  48                         '#https?://(.+\.)?polldaddy\.com/.*#i'                => array( 'http://polldaddy.com/oembed/',                       true  ),
  49                         '#https?://(www\.)?funnyordie\.com/videos/.*#i'       => array( 'http://www.funnyordie.com/oembed',                   true  ),
  50                         '#https?://(www\.)?twitter\.com/.+?/status(es)?/.*#i' => array( 'https://api.twitter.com/1/statuses/oembed.{format}', true  ),
  51                         '#https?://(www\.)?soundcloud\.com/.*#i'              => array( 'http://soundcloud.com/oembed',                       true  ),
  52                         '#https?://(www\.)?slideshare\.net/*#'                => array( 'http://www.slideshare.net/api/oembed/2',             true  ),
  53                         '#http://instagr(\.am|am\.com)/p/.*#i'                => array( 'http://api.instagram.com/oembed',                    true  ),
  54                         '#https?://(www\.)?rdio\.com/.*#i'                    => array( 'http://www.rdio.com/api/oembed/',                    true  ),
  55                         '#https?://rd\.io/x/.*#i'                             => array( 'http://www.rdio.com/api/oembed/',                    true  ),
  56                         '#https?://(open|play)\.spotify\.com/.*#i'            => array( 'https://embed.spotify.com/oembed/',                  true  ),
  57                         '#https?://(.+\.)?imgur\.com/.*#i'                    => array( 'http://api.imgur.com/oembed',                        true  ),
  58                         '#https?://(www\.)?meetu(\.ps|p\.com)/.*#i'           => array( 'http://api.meetup.com/oembed',                       true  ),
  59                 );
  60
  61                 /**
  62                  * Filter the list of oEmbed providers.
  63                  *
  64                  * Discovery is disabled for users lacking the unfiltered_html capability.
  65                  * Only providers in this array will be used for those users.
  66                  *
  67                  * @see wp_oembed_add_provider()
  68                  *
  69                  * @since 2.9.0
  70                  *
  71                  * @param array $providers An array of popular oEmbed providers.
  72                  */
  73                 $this->providers = apply_filters( 'oembed_providers', $providers );
  74
  75                 // Fix any embeds that contain new lines in the middle of the HTML which breaks wpautop().
  76                 add_filter( 'oembed_dataparse', array($this, '_strip_newlines'), 10, 3 );
  77         }
  78
  79         /**
  80          * The do-it-all function that takes a URL and attempts to return the HTML.
  81          *
  82          * @see WP_oEmbed::discover()
  83          * @see WP_oEmbed::fetch()
  84          * @see WP_oEmbed::data2html()
  85          *
  86          * @param string $url The URL to the content that should be attempted to be embedded.
  87          * @param array $args Optional arguments. Usually passed from a shortcode.
  88          * @return bool|string False on failure, otherwise the UNSANITIZED (and potentially unsafe) HTML that should be used to embed.
  89          */
  90         function get_html( $url, $args = '' ) {
  91                 $provider = false;
  92
  93                 if ( !isset($args['discover']) )
  94                         $args['discover'] = true;
  95
  96                 foreach ( $this->providers as $matchmask => $data ) {
  97                         list( $providerurl, $regex ) = $data;
  98
  99                         // Turn the asterisk-type provider URLs into regex
 100                         if ( !$regex ) {
 101                                 $matchmask = '#' . str_replace( '___wildcard___', '(.+)', preg_quote( str_replace( '*', '___wildcard___', $matchmask ), '#' ) ) . '#i';
 102                                 $matchmask = preg_replace( '|^#http\\\://|', '#https?\://', $matchmask );
 103                         }
 104
 105                         if ( preg_match( $matchmask, $url ) ) {
 106                                 $provider = str_replace( '{format}', 'json', $providerurl ); // JSON is easier to deal with than XML
 107                                 break;
 108                         }
 109                 }
 110
 111                 if ( !$provider && $args['discover'] )
 112                         $provider = $this->discover( $url );
 113
 114                 if ( !$provider || false === $data = $this->fetch( $provider, $url, $args ) )
 115                         return false;
 116
 117                 /**
 118                  * Filter the HTML returned by the oEmbed provider.
 119                  *
 120                  * @since 2.9.0
 121                  *
 122                  * @param string $data The returned oEmbed HTML.
 123                  * @param string $url  URL of the content to be embedded.
 124                  * @param array  $args Optional arguments, usually passed from a shortcode.
 125                  */
 126                 return apply_filters( 'oembed_result', $this->data2html( $data, $url ), $url, $args );
 127         }
 128
 129         /**
 130          * Attempts to find oEmbed provider discovery <link> tags at the given URL.
 131          *
 132          * @param string $url The URL that should be inspected for discovery <link> tags.
 133          * @return bool|string False on failure, otherwise the oEmbed provider URL.
 134          */
 135         function discover( $url ) {
 136                 $providers = array();
 137
 138                 // Fetch URL content
 139                 $request = wp_safe_remote_get( $url );
 140                 if ( $html = wp_remote_retrieve_body( $request ) ) {
 141
 142                         /**
 143                          * Filter the link types that contain oEmbed provider URLs.
 144                          *
 145                          * @since 2.9.0
 146                          *
 147                          * @param array $format Array of oEmbed link types. Accepts 'application/json+oembed',
 148                          *                      'text/xml+oembed', and 'application/xml+oembed' (incorrect,
 149                          *                      used by at least Vimeo).
 150                          */
 151                         $linktypes = apply_filters( 'oembed_linktypes', array(
 152                                 'application/json+oembed' => 'json',
 153                                 'text/xml+oembed' => 'xml',
 154                                 'application/xml+oembed' => 'xml',
 155                         ) );
 156
 157                         // Strip <body>
 158                         $html = substr( $html, 0, stripos( $html, '</head>' ) );
 159
 160                         // Do a quick check
 161                         $tagfound = false;
 162                         foreach ( $linktypes as $linktype => $format ) {
 163                                 if ( stripos($html, $linktype) ) {
 164                                         $tagfound = true;
 165                                         break;
 166                                 }
 167                         }
 168
 169                         if ( $tagfound && preg_match_all( '/<link([^<>]+)>/i', $html, $links ) ) {
 170                                 foreach ( $links[1] as $link ) {
 171                                         $atts = shortcode_parse_atts( $link );
 172
 173                                         if ( !empty($atts['type']) && !empty($linktypes[$atts['type']]) && !empty($atts['href']) ) {
 174                                                 $providers[$linktypes[$atts['type']]] = $atts['href'];
 175
 176                                                 // Stop here if it's JSON (that's all we need)
 177                                                 if ( 'json' == $linktypes[$atts['type']] )
 178                                                         break;
 179                                         }
 180                                 }
 181                         }
 182                 }
 183
 184                 // JSON is preferred to XML
 185                 if ( !empty($providers['json']) )
 186                         return $providers['json'];
 187                 elseif ( !empty($providers['xml']) )
 188                         return $providers['xml'];
 189                 else
 190                         return false;
 191         }
 192
 193         /**
 194          * Connects to a oEmbed provider and returns the result.
 195          *
 196          * @param string $provider The URL to the oEmbed provider.
 197          * @param string $url The URL to the content that is desired to be embedded.
 198          * @param array $args Optional arguments. Usually passed from a shortcode.
 199          * @return bool|object False on failure, otherwise the result in the form of an object.
 200          */
 201         function fetch( $provider, $url, $args = '' ) {
 202                 $args = wp_parse_args( $args, wp_embed_defaults() );
 203
 204                 $provider = add_query_arg( 'maxwidth', (int) $args['width'], $provider );
 205                 $provider = add_query_arg( 'maxheight', (int) $args['height'], $provider );
 206                 $provider = add_query_arg( 'url', urlencode($url), $provider );
 207
 208                 /**
 209                  * Filter the oEmbed URL to be fetched.
 210                  *
 211                  * @since 2.9.0
 212                  *
 213                  * @param string $provider URL of the oEmbed provider.
 214                  * @param string $url      URL of the content to be embedded.
 215                  * @param array  $args     Optional arguments, usually passed from a shortcode.
 216                  */
 217                 $provider = apply_filters( 'oembed_fetch_url', $provider, $url, $args );
 218
 219                 foreach( array( 'json', 'xml' ) as $format ) {
 220                         $result = $this->_fetch_with_format( $provider, $format );
 221                         if ( is_wp_error( $result ) && 'not-implemented' == $result->get_error_code() )
 222                                 continue;
 223                         return ( $result && ! is_wp_error( $result ) ) ? $result : false;
 224                 }
 225                 return false;
 226         }
 227
 228         /**
 229          * Fetches result from an oEmbed provider for a specific format and complete provider URL
 230          *
 231          * @since 3.0.0
 232          * @access private
 233          * @param string $provider_url_with_args URL to the provider with full arguments list (url, maxheight, etc.)
 234          * @param string $format Format to use
 235          * @return bool|object False on failure, otherwise the result in the form of an object.
 236          */
 237         function _fetch_with_format( $provider_url_with_args, $format ) {
 238                 $provider_url_with_args = add_query_arg( 'format', $format, $provider_url_with_args );
 239                 $response = wp_safe_remote_get( $provider_url_with_args );
 240                 if ( 501 == wp_remote_retrieve_response_code( $response ) )
 241                         return new WP_Error( 'not-implemented' );
 242                 if ( ! $body = wp_remote_retrieve_body( $response ) )
 243                         return false;
 244                 $parse_method = "_parse_$format";
 245                 return $this->$parse_method( $body );
 246         }
 247
 248         /**
 249          * Parses a json response body.
 250          *
 251          * @since 3.0.0
 252          * @access private
 253          */
 254         function _parse_json( $response_body ) {
 255                 return ( ( $data = json_decode( trim( $response_body ) ) ) && is_object( $data ) ) ? $data : false;
 256         }
 257
 258         /**
 259          * Parses an XML response body.
 260          *
 261          * @since 3.0.0
 262          * @access private
 263          */
 264         function _parse_xml( $response_body ) {
 265                 if ( ! function_exists( 'libxml_disable_entity_loader' ) )
 266                         return false;
 267
 268                 $loader = libxml_disable_entity_loader( true );
 269                 $errors = libxml_use_internal_errors( true );
 270
 271                 $return = $this->_parse_xml_body( $response_body );
 272
 273                 libxml_use_internal_errors( $errors );
 274                 libxml_disable_entity_loader( $loader );
 275
 276                 return $return;
 277         }
 278
 279         /**
 280          * Helper function for parsing an XML response body.
 281          *
 282          * @since 3.6.0
 283          * @access private
 284          */
 285         private function _parse_xml_body( $response_body ) {
 286                 if ( ! function_exists( 'simplexml_import_dom' ) || ! class_exists( 'DOMDocument' ) )
 287                         return false;
 288
 289                 $dom = new DOMDocument;
 290                 $success = $dom->loadXML( $response_body );
 291                 if ( ! $success )
 292                         return false;
 293
 294                 if ( isset( $dom->doctype ) )
 295                         return false;
 296
 297                 foreach ( $dom->childNodes as $child ) {
 298                         if ( XML_DOCUMENT_TYPE_NODE === $child->nodeType )
 299                                 return false;
 300                 }
 301
 302                 $xml = simplexml_import_dom( $dom );
 303                 if ( ! $xml )
 304                         return false;
 305
 306                 $return = new stdClass;
 307                 foreach ( $xml as $key => $value ) {
 308                         $return->$key = (string) $value;
 309                 }
 310
 311                 return $return;
 312         }
 313
 314         /**
 315          * Converts a data object from {@link WP_oEmbed::fetch()} and returns the HTML.
 316          *
 317          * @param object $data A data object result from an oEmbed provider.
 318          * @param string $url The URL to the content that is desired to be embedded.
 319          * @return bool|string False on error, otherwise the HTML needed to embed.
 320          */
 321         function data2html( $data, $url ) {
 322                 if ( ! is_object( $data ) || empty( $data->type ) )
 323                         return false;
 324
 325                 $return = false;
 326
 327                 switch ( $data->type ) {
 328                         case 'photo':
 329                                 if ( empty( $data->url ) || empty( $data->width ) || empty( $data->height ) )
 330                                         break;
 331                                 if ( ! is_string( $data->url ) || ! is_numeric( $data->width ) || ! is_numeric( $data->height ) )
 332                                         break;
 333
 334                                 $title = ! empty( $data->title ) && is_string( $data->title ) ? $data->title : '';
 335                                 $return = '<a href="' . esc_url( $url ) . '"><img src="' . esc_url( $data->url ) . '" alt="' . esc_attr($title) . '" width="' . esc_attr($data->width) . '" height="' . esc_attr($data->height) . '" /></a>';
 336                                 break;
 337
 338                         case 'video':
 339                         case 'rich':
 340                                 if ( ! empty( $data->html ) && is_string( $data->html ) )
 341                                         $return = $data->html;
 342                                 break;
 343
 344                         case 'link':
 345                                 if ( ! empty( $data->title ) && is_string( $data->title ) )
 346                                         $return = '<a href="' . esc_url( $url ) . '">' . esc_html( $data->title ) . '</a>';
 347                                 break;
 348
 349                         default:
 350                                 $return = false;
 351                 }
 352
 353                 /**
 354                  * Filter the returned oEmbed HTML.
 355                  *
 356                  * Use this filter to add support for custom data types, or to filter the result.
 357                  *
 358                  * @since 2.9.0
 359                  *
 360                  * @param string $return The returned oEmbed HTML.
 361                  * @param object $data   A data object result from an oEmbed provider.
 362                  * @param string $url    The URL of the content to be embedded.
 363                  */
 364                 return apply_filters( 'oembed_dataparse', $return, $data, $url );
 365         }
 366
 367         /**
 368          * Strip any new lines from the HTML.
 369          *
 370          * @access private
 371          * @param string $html Existing HTML.
 372          * @param object $data Data object from WP_oEmbed::data2html()
 373          * @param string $url The original URL passed to oEmbed.
 374          * @return string Possibly modified $html
 375          */
 376         function _strip_newlines( $html, $data, $url ) {
 377                 if ( false !== strpos( $html, "\n" ) )
 378                         $html = str_replace( array( "\r\n", "\n" ), '', $html );
 379
 380                 return $html;
 381         }
 382 }
 383
 384 /**
 385  * Returns the initialized {@link WP_oEmbed} object
 386  *
 387  * @since 2.9.0
 388  * @access private
 389  *
 390  * @see WP_oEmbed
 391  * @uses WP_oEmbed
 392  *
 393  * @return WP_oEmbed object.
 394  */
 395 function _wp_oembed_get_object() {
 396         static $wp_oembed;
 397
 398         if ( is_null($wp_oembed) )
 399                 $wp_oembed = new WP_oEmbed();
 400
 401         return $wp_oembed;
 402 }