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\ParseUrl;\r
  14 use dba;\r
  15 use DOMDocument;\r
  16 use DOMXPath;\r
  17 use DOMNode;\r
  18 \r
  19 require_once 'include/dba.php';\r
  20 require_once 'mod/proxy.php';\r
  21 \r
  22 /**\r
  23  * Handles all OEmbed content fetching and replacement\r
  24  *\r
  25  * OEmbed is a standard used to allow an embedded representation of a URL on\r
  26  * third party sites\r
  27  *\r
  28  * @see https://oembed.com\r
  29  *\r
  30  * @author Hypolite Petovan <mrpetovan@gmail.com>\r
  31  */\r
  32 class OEmbed\r
  33 {\r
  34         public static function replaceCallback($matches)\r
  35         {\r
  36                 $embedurl = $matches[1];\r
  37                 $j = self::fetchURL($embedurl);\r
  38                 $s = self::formatObject($j);\r
  39 \r
  40                 return $s;\r
  41         }\r
  42 \r
  43         /**\r
  44          * @brief Get data from an URL to embed its content.\r
  45          *\r
  46          * @param string $embedurl The URL from which the data should be fetched.\r
  47          * @param bool $no_rich_type If set to true rich type content won't be fetched.\r
  48          *\r
  49          * @return bool|object Returns object with embed content or false if no embedable\r
  50          *       content exists\r
  51          */\r
  52         public static function fetchURL($embedurl, $no_rich_type = false)\r
  53         {\r
  54                 $embedurl = trim($embedurl, "'");\r
  55                 $embedurl = trim($embedurl, '"');\r
  56 \r
  57                 $a = get_app();\r
  58 \r
  59                 $condition = array('url' => normalise_link($embedurl));\r
  60                 $r = dba::select('oembed', array('content'), $condition, array('limit' => 1));\r
  61 \r
  62                 if (DBM::is_result($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                 $jhtml = self::iframe($j->embedurl, (isset($j->width) ? $j->width : null), (isset($j->height) ? $j->height : null));\r
 164                 $ret = "<span 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 = $jhtml;\r
 186                                 }\r
 187                                 //$ret.="<br>";\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                                 // not so safe..\r
 196                                 if (self::isAllowedURL($embedurl)) {\r
 197                                         $ret .= proxy_parse_html($jhtml);\r
 198                                 }\r
 199                                 break;\r
 200                 }\r
 201 \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'>$embedlink</a>";\r
 233                         }\r
 234                         //if (isset($j->author_name)) $ret.=" by ".$j->author_name;\r
 235                         //if (isset($j->provider_name)) $ret.=" on ".$j->provider_name;\r
 236                         $ret .= "</h4>";\r
 237                 } else {\r
 238                         // add <a> for html2bbcode conversion\r
 239                         $ret .= "<a href='$embedurl' rel='oembed'>$embedurl</a>";\r
 240                 }\r
 241                 $ret.="</span>";\r
 242                 $ret = str_replace("\n", "", $ret);\r
 243                 return mb_convert_encoding($ret, 'HTML-ENTITIES', mb_detect_encoding($ret));\r
 244         }\r
 245 \r
 246         public static function BBCode2HTML($text)\r
 247         {\r
 248                 $stopoembed = Config::get("system", "no_oembed");\r
 249                 if ($stopoembed == true) {\r
 250                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);\r
 251                 }\r
 252                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);\r
 253         }\r
 254 \r
 255         /**\r
 256          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>\r
 257          * and replace it with [embed]url[/embed]\r
 258          */\r
 259         public static function HTML2BBCode($text)\r
 260         {\r
 261                 // start parser only if 'oembed' is in text\r
 262                 if (strpos($text, "oembed")) {\r
 263 \r
 264                         // convert non ascii chars to html entities\r
 265                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));\r
 266 \r
 267                         // If it doesn't parse at all, just return the text.\r
 268                         $dom = @DOMDocument::loadHTML($html_text);\r
 269                         if (!$dom) {\r
 270                                 return $text;\r
 271                         }\r
 272                         $xpath = new DOMXPath($dom);\r
 273 \r
 274                         $xattr = self::buildXPath("class", "oembed");\r
 275                         $entries = $xpath->query("//span[$xattr]");\r
 276 \r
 277                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");\r
 278                         foreach ($entries as $e) {\r
 279                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;\r
 280                                 if (!is_null($href)) {\r
 281                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);\r
 282                                 }\r
 283                         }\r
 284                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));\r
 285                 } else {\r
 286                         return $text;\r
 287                 }\r
 288         }\r
 289 \r
 290         /**\r
 291          * @brief Generates the iframe HTML for an oembed attachment.\r
 292          *\r
 293          * Width and height are given by the remote, and are regularly too small for\r
 294          * the generated iframe.\r
 295          *\r
 296          * The width is entirely discarded for the actual width of the post, while fixed\r
 297          * height is used as a starting point before the inevitable resizing.\r
 298          *\r
 299          * Since the iframe is automatically resized on load, there are no need for ugly\r
 300          * and impractical scrollbars.\r
 301          *\r
 302          * @param string $src Original remote URL to embed\r
 303          * @param string $width\r
 304          * @param string $height\r
 305          * @return string formatted HTML\r
 306          *\r
 307          * @see oembed_format_object()\r
 308          */\r
 309         private static function iframe($src, $width, $height)\r
 310         {\r
 311                 $a = get_app();\r
 312 \r
 313                 if (!$height || strstr($height, '%')) {\r
 314                         $height = '200';\r
 315                 }\r
 316                 $width = '100%';\r
 317 \r
 318                 // Only proxy OEmbed URLs to avoid mixed-content errors\r
 319                 if (Config::get('system', 'ssl_policy') == SSL_POLICY_FULL && parse_url($src, PHP_URL_SCHEME) !== 'https') {\r
 320                         $src = System::baseUrl() . '/oembed/' . base64url_encode($src);\r
 321                 }\r
 322                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $s . '" allowfullscreen scrolling="no" frameborder="no">' . t('Embedded content') . '</iframe>';\r
 323         }\r
 324 \r
 325         /**\r
 326          * Generates an XPath query to select elements whose provided attribute contains\r
 327          * the provided value in a space-separated list.\r
 328          *\r
 329          * @brief Generates attribute search XPath string\r
 330          *\r
 331          * @param string $attr Name of the attribute to seach\r
 332          * @param string $value Value to search in a space-separated list\r
 333          * @return string\r
 334          */\r
 335         private static function buildXPath($attr, $value)\r
 336         {\r
 337                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath\r
 338                 return "contains(normalize-space(@$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
 339         }\r
 340 \r
 341         /**\r
 342          * Returns the inner XML string of a provided DOMNode\r
 343          *\r
 344          * @brief Returns the inner XML string of a provided DOMNode\r
 345          *\r
 346          * @param DOMNode $node\r
 347          * @return string\r
 348          */\r
 349         private static function getInnerHTML(DOMNode $node)\r
 350         {\r
 351                 $innerHTML = '';\r
 352                 $children = $node->childNodes;\r
 353                 foreach ($children as $child) {\r
 354                         $innerHTML .= $child->ownerDocument->saveXML($child);\r
 355                 }\r
 356                 return $innerHTML;\r
 357         }\r
 358 \r
 359         /**\r
 360          * Determines if rich content OEmbed is allowed for the provided URL\r
 361          *\r
 362          * @brief Determines if rich content OEmbed is allowed for the provided URL\r
 363          * @param string $url\r
 364          * @return boolean\r
 365          */\r
 366         private static function isAllowedURL($url)\r
 367         {\r
 368                 if (!Config::get('system', 'no_oembed_rich_content')) {\r
 369                         return true;\r
 370                 }\r
 371 \r
 372                 $domain = parse_url($url, PHP_URL_HOST);\r
 373 \r
 374                 $str_allowed = Config::get('system', 'allowed_oembed', '');\r
 375                 $allowed = explode(',', $str_allowed);\r
 376 \r
 377                 return allowed_domain($domain, $allowed, true);\r
 378         }\r
 379 }\r