src/Content/OEmbed.php

   1 <?php\r
   2 \r
   3 /**\r
   4  * @file src/Content/OEmbed.php\r
   5  */\r
   6 \r
   7 namespace Friendica\Content;\r
   8 \r
   9 use Friendica\Core\Cache;\r
  10 use Friendica\Core\System;\r
  11 use Friendica\Core\Config;\r
  12 use Friendica\Database\DBM;\r
  13 use Friendica\Util\ParseUrl;\r
  14 use dba;\r
  15 use DOMDocument;\r
  16 use DOMXPath;\r
  17 use DOMNode;\r
  18 use Exception;\r
  19 \r
  20 require_once 'include/dba.php';\r
  21 require_once 'mod/proxy.php';\r
  22 \r
  23 /**\r
  24  * Handles all OEmbed content fetching and replacement\r
  25  *\r
  26  * OEmbed is a standard used to allow an embedded representation of a URL on\r
  27  * third party sites\r
  28  *\r
  29  * @see https://oembed.com\r
  30  *\r
  31  * @author Hypolite Petovan <mrpetovan@gmail.com>\r
  32  */\r
  33 class OEmbed\r
  34 {\r
  35         public static function replaceCallback($matches)\r
  36         {\r
  37                 $embedurl = $matches[1];\r
  38                 $j = self::fetchURL($embedurl);\r
  39                 $s = self::formatObject($j);\r
  40 \r
  41                 return $s;\r
  42         }\r
  43 \r
  44         /**\r
  45          * @brief Get data from an URL to embed its content.\r
  46          *\r
  47          * @param string $embedurl The URL from which the data should be fetched.\r
  48          * @param bool $no_rich_type If set to true rich type content won't be fetched.\r
  49          *\r
  50          * @return bool|object Returns object with embed content or false if no embedable\r
  51          *       content exists\r
  52          */\r
  53         public static function fetchURL($embedurl, $no_rich_type = false)\r
  54         {\r
  55                 $embedurl = trim($embedurl, "'");\r
  56                 $embedurl = trim($embedurl, '"');\r
  57 \r
  58                 $a = get_app();\r
  59 \r
  60                 $condition = array('url' => normalise_link($embedurl));\r
  61                 $r = dba::select('oembed', array('content'), $condition, array('limit' => 1));\r
  62 \r
  63                 if (DBM::is_result($r)) {\r
  64                         $txt = $r["content"];\r
  65                 } else {\r
  66                         $txt = Cache::get($a->videowidth . $embedurl);\r
  67                 }\r
  68                 // These media files should now be caught in bbcode.php\r
  69                 // left here as a fallback in case this is called from another source\r
  70 \r
  71                 $noexts = array("mp3", "mp4", "ogg", "ogv", "oga", "ogm", "webm");\r
  72                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);\r
  73 \r
  74 \r
  75                 if (is_null($txt)) {\r
  76                         $txt = "";\r
  77 \r
  78                         if (!in_array($ext, $noexts)) {\r
  79                                 // try oembed autodiscovery\r
  80                                 $redirects = 0;\r
  81                                 $html_text = fetch_url($embedurl, false, $redirects, 15, "text/*");\r
  82                                 if ($html_text) {\r
  83                                         $dom = @DOMDocument::loadHTML($html_text);\r
  84                                         if ($dom) {\r
  85                                                 $xpath = new DOMXPath($dom);\r
  86                                                 $entries = $xpath->query("//link[@type='application/json+oembed']");\r
  87                                                 foreach ($entries as $e) {\r
  88                                                         $href = $e->getAttributeNode("href")->nodeValue;\r
  89                                                         $txt = fetch_url($href . '&maxwidth=' . $a->videowidth);\r
  90                                                         break;\r
  91                                                 }\r
  92                                                 $entries = $xpath->query("//link[@type='text/json+oembed']");\r
  93                                                 foreach ($entries as $e) {\r
  94                                                         $href = $e->getAttributeNode("href")->nodeValue;\r
  95                                                         $txt = fetch_url($href . '&maxwidth=' . $a->videowidth);\r
  96                                                         break;\r
  97                                                 }\r
  98                                         }\r
  99                                 }\r
 100                         }\r
 101 \r
 102                         $txt = trim($txt);\r
 103 \r
 104                         if (!$txt || $txt[0] != "{") {\r
 105                                 $txt = '{"type":"error"}';\r
 106                         } else { //save in cache\r
 107                                 $j = json_decode($txt);\r
 108                                 if ($j->type != "error") {\r
 109                                         dba::insert('oembed', array('url' => normalise_link($embedurl),\r
 110                                                 'content' => $txt, 'created' => datetime_convert()), true);\r
 111                                 }\r
 112 \r
 113                                 Cache::set($a->videowidth . $embedurl, $txt, CACHE_DAY);\r
 114                         }\r
 115                 }\r
 116 \r
 117                 $j = json_decode($txt);\r
 118 \r
 119                 if (!is_object($j)) {\r
 120                         return false;\r
 121                 }\r
 122 \r
 123                 // Always embed the SSL version\r
 124                 if (isset($j->html)) {\r
 125                         $j->html = str_replace(array("http://www.youtube.com/", "http://player.vimeo.com/"), array("https://www.youtube.com/", "https://player.vimeo.com/"), $j->html);\r
 126                 }\r
 127 \r
 128                 $j->embedurl = $embedurl;\r
 129 \r
 130                 // If fetching information doesn't work, then improve via internal functions\r
 131                 if (($j->type == "error") || ($no_rich_type && ($j->type == "rich"))) {\r
 132                         $data = ParseUrl::getSiteinfoCached($embedurl, true, false);\r
 133                         $j->type = $data["type"];\r
 134 \r
 135                         if ($j->type == "photo") {\r
 136                                 $j->url = $data["url"];\r
 137                                 //$j->width = $data["images"][0]["width"];\r
 138                                 //$j->height = $data["images"][0]["height"];\r
 139                         }\r
 140 \r
 141                         if (isset($data["title"])) {\r
 142                                 $j->title = $data["title"];\r
 143                         }\r
 144 \r
 145                         if (isset($data["text"])) {\r
 146                                 $j->description = $data["text"];\r
 147                         }\r
 148 \r
 149                         if (is_array($data["images"])) {\r
 150                                 $j->thumbnail_url = $data["images"][0]["src"];\r
 151                                 $j->thumbnail_width = $data["images"][0]["width"];\r
 152                                 $j->thumbnail_height = $data["images"][0]["height"];\r
 153                         }\r
 154                 }\r
 155 \r
 156                 call_hooks('oembed_fetch_url', $embedurl, $j);\r
 157 \r
 158                 return $j;\r
 159         }\r
 160 \r
 161         public static function formatObject($j)\r
 162         {\r
 163                 $embedurl = $j->embedurl;\r
 164                 $jhtml = $j->html;\r
 165                 $ret = '<div class="oembed ' . $j->type . '">';\r
 166                 switch ($j->type) {\r
 167                         case "video":\r
 168                                 if (isset($j->thumbnail_url)) {\r
 169                                         $tw = (isset($j->thumbnail_width) && intval($j->thumbnail_width)) ? $j->thumbnail_width : 200;\r
 170                                         $th = (isset($j->thumbnail_height) && intval($j->thumbnail_height)) ? $j->thumbnail_height : 180;\r
 171                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio\r
 172                                         $tr = (($th) ? $tw / $th : 1);\r
 173 \r
 174                                         $th = 120;\r
 175                                         $tw = $th * $tr;\r
 176                                         $tpl = get_markup_template('oembed_video.tpl');\r
 177                                         $ret .= replace_macros($tpl, array(\r
 178                                                 '$baseurl' => System::baseUrl(),\r
 179                                                 '$embedurl' => $embedurl,\r
 180                                                 '$escapedhtml' => base64_encode($jhtml),\r
 181                                                 '$tw' => $tw,\r
 182                                                 '$th' => $th,\r
 183                                                 '$turl' => $j->thumbnail_url,\r
 184                                         ));\r
 185                                 } else {\r
 186                                         $ret = $jhtml;\r
 187                                 }\r
 188                                 break;\r
 189                         case "photo":\r
 190                                 $ret .= '<img width="' . $j->width . '" src="' . proxy_url($j->url) . '">';\r
 191                                 break;\r
 192                         case "link":\r
 193                                 break;\r
 194                         case "rich":\r
 195                                 if (self::isAllowedURL($embedurl)) {\r
 196                                         $ret .= proxy_parse_html($jhtml);\r
 197                                 }\r
 198                                 break;\r
 199                 }\r
 200 \r
 201                 $ret .= '</div>';\r
 202                 // add link to source if not present in "rich" type\r
 203                 if ($j->type != 'rich' || !strpos($j->html, $embedurl)) {\r
 204                         $ret .= '<h4>';\r
 205                         if (isset($j->title)) {\r
 206                                 if (isset($j->provider_name)) {\r
 207                                         $ret .= $j->provider_name . ": ";\r
 208                                 }\r
 209 \r
 210                                 $embedlink = (isset($j->title)) ? $j->title : $embedurl;\r
 211                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedlink . '</a>';\r
 212                                 if (isset($j->author_name)) {\r
 213                                         $ret .= ' (' . $j->author_name . ')';\r
 214                                 }\r
 215                         } elseif (isset($j->provider_name) || isset($j->author_name)) {\r
 216                                 $embedlink = "";\r
 217                                 if (isset($j->provider_name)) {\r
 218                                         $embedlink .= $j->provider_name;\r
 219                                 }\r
 220 \r
 221                                 if (isset($j->author_name)) {\r
 222                                         if ($embedlink != "") {\r
 223                                                 $embedlink .= ": ";\r
 224                                         }\r
 225 \r
 226                                         $embedlink .= $j->author_name;\r
 227                                 }\r
 228                                 if (trim($embedlink) == "") {\r
 229                                         $embedlink = $embedurl;\r
 230                                 }\r
 231 \r
 232                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedlink . '</a>';\r
 233                         }\r
 234                         $ret .= "</h4>";\r
 235                 } elseif (!strpos($j->html, $embedurl)) {\r
 236                         // add <a> for html2bbcode conversion\r
 237                         $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $j->title . '</a>';\r
 238                 }\r
 239 \r
 240                 $ret = str_replace("\n", "", $ret);\r
 241                 return mb_convert_encoding($ret, 'HTML-ENTITIES', mb_detect_encoding($ret));\r
 242         }\r
 243 \r
 244         public static function BBCode2HTML($text)\r
 245         {\r
 246                 $stopoembed = Config::get("system", "no_oembed");\r
 247                 if ($stopoembed == true) {\r
 248                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);\r
 249                 }\r
 250                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);\r
 251         }\r
 252 \r
 253         /**\r
 254          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>\r
 255          * and replace it with [embed]url[/embed]\r
 256          */\r
 257         public static function HTML2BBCode($text)\r
 258         {\r
 259                 // start parser only if 'oembed' is in text\r
 260                 if (strpos($text, "oembed")) {\r
 261 \r
 262                         // convert non ascii chars to html entities\r
 263                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));\r
 264 \r
 265                         // If it doesn't parse at all, just return the text.\r
 266                         $dom = @DOMDocument::loadHTML($html_text);\r
 267                         if (!$dom) {\r
 268                                 return $text;\r
 269                         }\r
 270                         $xpath = new DOMXPath($dom);\r
 271 \r
 272                         $xattr = self::buildXPath("class", "oembed");\r
 273                         $entries = $xpath->query("//div[$xattr]");\r
 274 \r
 275                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");\r
 276                         foreach ($entries as $e) {\r
 277                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;\r
 278                                 if (!is_null($href)) {\r
 279                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);\r
 280                                 }\r
 281                         }\r
 282                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));\r
 283                 } else {\r
 284                         return $text;\r
 285                 }\r
 286         }\r
 287 \r
 288         /**\r
 289          * Determines if rich content OEmbed is allowed for the provided URL\r
 290          *\r
 291          * @brief Determines if rich content OEmbed is allowed for the provided URL\r
 292          * @param string $url\r
 293          * @return boolean\r
 294          */\r
 295         public static function isAllowedURL($url)\r
 296         {\r
 297                 if (!Config::get('system', 'no_oembed_rich_content')) {\r
 298                         return true;\r
 299                 }\r
 300 \r
 301                 $domain = parse_url($url, PHP_URL_HOST);\r
 302                 if (!x($domain)) {\r
 303                         return false;\r
 304                 }\r
 305 \r
 306                 $str_allowed = Config::get('system', 'allowed_oembed', '');\r
 307                 if (!x($str_allowed)) {\r
 308                         return false;\r
 309                 }\r
 310                 \r
 311                 $allowed = explode(',', $str_allowed);\r
 312 \r
 313                 return allowed_domain($domain, $allowed);\r
 314         }\r
 315 \r
 316         public static function getHTML($url, $title = null)\r
 317         {\r
 318                 // Always embed the SSL version\r
 319                 $url = str_replace(array("http://www.youtube.com/", "http://player.vimeo.com/"),\r
 320                                         array("https://www.youtube.com/", "https://player.vimeo.com/"), $url);\r
 321 \r
 322                 $o = OEmbed::fetchURL($url);\r
 323 \r
 324                 if (!is_object($o) || $o->type == 'error') {\r
 325                         throw new Exception('OEmbed failed for URL: ' . $url);\r
 326                 }\r
 327 \r
 328                 if (x($title)) {\r
 329                         $o->title = $title;\r
 330                 }\r
 331 \r
 332                 $html = OEmbed::formatObject($o);\r
 333 \r
 334                 return $html;\r
 335         }\r
 336 \r
 337         /**\r
 338          * @brief Generates the iframe HTML for an oembed attachment.\r
 339          *\r
 340          * Width and height are given by the remote, and are regularly too small for\r
 341          * the generated iframe.\r
 342          *\r
 343          * The width is entirely discarded for the actual width of the post, while fixed\r
 344          * height is used as a starting point before the inevitable resizing.\r
 345          *\r
 346          * Since the iframe is automatically resized on load, there are no need for ugly\r
 347          * and impractical scrollbars.\r
 348          *\r
 349          * @todo This function is currently unused until someone™ adds support for a separate OEmbed domain\r
 350          *\r
 351          * @param string $src Original remote URL to embed\r
 352          * @param string $width\r
 353          * @param string $height\r
 354          * @return string formatted HTML\r
 355          *\r
 356          * @see oembed_format_object()\r
 357          */\r
 358         private static function iframe($src, $width, $height)\r
 359         {\r
 360                 $a = get_app();\r
 361 \r
 362                 if (!$height || strstr($height, '%')) {\r
 363                         $height = '200';\r
 364                 }\r
 365                 $width = '100%';\r
 366 \r
 367                 $src = System::baseUrl() . '/oembed/' . base64url_encode($src);\r
 368                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . t('Embedded content') . '</iframe>';\r
 369         }\r
 370 \r
 371         /**\r
 372          * Generates an XPath query to select elements whose provided attribute contains\r
 373          * the provided value in a space-separated list.\r
 374          *\r
 375          * @brief Generates attribute search XPath string\r
 376          *\r
 377          * @param string $attr Name of the attribute to seach\r
 378          * @param string $value Value to search in a space-separated list\r
 379          * @return string\r
 380          */\r
 381         private static function buildXPath($attr, $value)\r
 382         {\r
 383                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath\r
 384                 return "contains(normalize-space(@$attr), ' $value ') or substring(normalize-space(@$attr), 1, string-length('$value') + 1) = '$value ' or substring(normalize-space(@$attr), string-length(@$attr) - string-length('$value')) = ' $value' or @$attr = '$value'";\r
 385         }\r
 386 \r
 387         /**\r
 388          * Returns the inner XML string of a provided DOMNode\r
 389          *\r
 390          * @brief Returns the inner XML string of a provided DOMNode\r
 391          *\r
 392          * @param DOMNode $node\r
 393          * @return string\r
 394          */\r
 395         private static function getInnerHTML(DOMNode $node)\r
 396         {\r
 397                 $innerHTML = '';\r
 398                 $children = $node->childNodes;\r
 399                 foreach ($children as $child) {\r
 400                         $innerHTML .= $child->ownerDocument->saveXML($child);\r
 401                 }\r
 402                 return $innerHTML;\r
 403         }\r
 404 \r
 405 }\r