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