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