src/Content/OEmbed.php

   1 <?php\r
   2 \r
   3 /*\r
   4  * To change this license header, choose License Headers in Project Properties.\r
   5  * To change this template file, choose Tools | Templates\r
   6  * and open the template in the editor.\r
   7  */\r
   8 \r
   9 namespace Friendica\Content;\r
  10 \r
  11 use Friendica\Core\Cache;\r
  12 use Friendica\Core\System;\r
  13 use Friendica\ParseUrl;\r
  14 use Friendica\Core\Config;\r
  15 use Friendica\Database\DBM;\r
  16 use dba;\r
  17 use DOMDocument;\r
  18 use DOMXPath;\r
  19 use DOMNode;\r
  20 \r
  21 require_once 'include/dba.php';\r
  22 require_once 'mod/proxy.php';\r
  23 \r
  24 /**\r
  25  * Description of OEmbed\r
  26  *\r
  27  * @author benlo\r
  28  */\r
  29 class OEmbed\r
  30 {\r
  31         public static function replaceCallback($matches)\r
  32         {\r
  33                 $embedurl = $matches[1];\r
  34                 $j = OEmbed::fetchURL($embedurl);\r
  35                 $s = OEmbed::formatObject($j);\r
  36 \r
  37                 return $s;\r
  38         }\r
  39 \r
  40         /**\r
  41          * @brief Get data from an URL to embed its content.\r
  42          *\r
  43          * @param string $embedurl The URL from which the data should be fetched.\r
  44          * @param bool $no_rich_type If set to true rich type content won't be fetched.\r
  45          *\r
  46          * @return bool|object Returns object with embed content or false if no embedable\r
  47          *       content exists\r
  48          */\r
  49         public static function fetchURL($embedurl, $no_rich_type = false)\r
  50         {\r
  51                 $embedurl = trim($embedurl, "'");\r
  52                 $embedurl = trim($embedurl, '"');\r
  53 \r
  54                 $a = get_app();\r
  55 \r
  56                 $condition = array('url' => normalise_link($embedurl));\r
  57                 $r = dba::select('oembed', array('content'), $condition, array('limit' => 1));\r
  58 \r
  59                 if (DBM::is_result($r)) {\r
  60                         $txt = $r["content"];\r
  61                 } else {\r
  62                         $txt = Cache::get($a->videowidth . $embedurl);\r
  63                 }\r
  64                 // These media files should now be caught in bbcode.php\r
  65                 // left here as a fallback in case this is called from another source\r
  66 \r
  67                 $noexts = array("mp3", "mp4", "ogg", "ogv", "oga", "ogm", "webm");\r
  68                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);\r
  69 \r
  70 \r
  71                 if (is_null($txt)) {\r
  72                         $txt = "";\r
  73 \r
  74                         if (!in_array($ext, $noexts)) {\r
  75                                 // try oembed autodiscovery\r
  76                                 $redirects = 0;\r
  77                                 $html_text = fetch_url($embedurl, false, $redirects, 15, "text/*");\r
  78                                 if ($html_text) {\r
  79                                         $dom = @DOMDocument::loadHTML($html_text);\r
  80                                         if ($dom) {\r
  81                                                 $xpath = new DOMXPath($dom);\r
  82                                                 $entries = $xpath->query("//link[@type='application/json+oembed']");\r
  83                                                 foreach ($entries as $e) {\r
  84                                                         $href = $e->getAttributeNode("href")->nodeValue;\r
  85                                                         $txt = fetch_url($href . '&maxwidth=' . $a->videowidth);\r
  86                                                         break;\r
  87                                                 }\r
  88                                                 $entries = $xpath->query("//link[@type='text/json+oembed']");\r
  89                                                 foreach ($entries as $e) {\r
  90                                                         $href = $e->getAttributeNode("href")->nodeValue;\r
  91                                                         $txt = fetch_url($href . '&maxwidth=' . $a->videowidth);\r
  92                                                         break;\r
  93                                                 }\r
  94                                         }\r
  95                                 }\r
  96                         }\r
  97 \r
  98                         $txt = trim($txt);\r
  99 \r
 100                         if ($txt[0] != "{") {\r
 101                                 $txt = '{"type":"error"}';\r
 102                         } else { //save in cache\r
 103                                 $j = json_decode($txt);\r
 104                                 if ($j->type != "error") {\r
 105                                         dba::insert('oembed', array('url' => normalise_link($embedurl),\r
 106                                                 'content' => $txt, 'created' => datetime_convert()), true);\r
 107                                 }\r
 108 \r
 109                                 Cache::set($a->videowidth . $embedurl, $txt, CACHE_DAY);\r
 110                         }\r
 111                 }\r
 112 \r
 113                 $j = json_decode($txt);\r
 114 \r
 115                 if (!is_object($j)) {\r
 116                         return false;\r
 117                 }\r
 118 \r
 119                 // Always embed the SSL version\r
 120                 if (isset($j->html)) {\r
 121                         $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
 122                 }\r
 123 \r
 124                 $j->embedurl = $embedurl;\r
 125 \r
 126                 // If fetching information doesn't work, then improve via internal functions\r
 127                 if (($j->type == "error") || ($no_rich_type && ($j->type == "rich"))) {\r
 128                         $data = ParseUrl::getSiteinfoCached($embedurl, true, false);\r
 129                         $j->type = $data["type"];\r
 130 \r
 131                         if ($j->type == "photo") {\r
 132                                 $j->url = $data["url"];\r
 133                                 //$j->width = $data["images"][0]["width"];\r
 134                                 //$j->height = $data["images"][0]["height"];\r
 135                         }\r
 136 \r
 137                         if (isset($data["title"])) {\r
 138                                 $j->title = $data["title"];\r
 139                         }\r
 140 \r
 141                         if (isset($data["text"])) {\r
 142                                 $j->description = $data["text"];\r
 143                         }\r
 144 \r
 145                         if (is_array($data["images"])) {\r
 146                                 $j->thumbnail_url = $data["images"][0]["src"];\r
 147                                 $j->thumbnail_width = $data["images"][0]["width"];\r
 148                                 $j->thumbnail_height = $data["images"][0]["height"];\r
 149                         }\r
 150                 }\r
 151 \r
 152                 call_hooks('oembed_fetch_url', $embedurl, $j);\r
 153 \r
 154                 return $j;\r
 155         }\r
 156 \r
 157         public static function formatObject($j)\r
 158         {\r
 159                 $embedurl = $j->embedurl;\r
 160                 $jhtml = OEmbed::iframe($j->embedurl, (isset($j->width) ? $j->width : null), (isset($j->height) ? $j->height : null));\r
 161                 $ret = "<span class='oembed " . $j->type . "'>";\r
 162                 switch ($j->type) {\r
 163                         case "video":\r
 164                                 if (isset($j->thumbnail_url)) {\r
 165                                         $tw = (isset($j->thumbnail_width) && intval($j->thumbnail_width)) ? $j->thumbnail_width : 200;\r
 166                                         $th = (isset($j->thumbnail_height) && intval($j->thumbnail_height)) ? $j->thumbnail_height : 180;\r
 167                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio\r
 168                                         $tr = (($th) ? $tw / $th : 1);\r
 169 \r
 170                                         $th = 120;\r
 171                                         $tw = $th * $tr;\r
 172                                         $tpl = get_markup_template('oembed_video.tpl');\r
 173                                         $ret.=replace_macros($tpl, array(\r
 174                                                 '$baseurl' => System::baseUrl(),\r
 175                                                 '$embedurl' => $embedurl,\r
 176                                                 '$escapedhtml' => base64_encode($jhtml),\r
 177                                                 '$tw' => $tw,\r
 178                                                 '$th' => $th,\r
 179                                                 '$turl' => $j->thumbnail_url,\r
 180                                         ));\r
 181                                 } else {\r
 182                                         $ret = $jhtml;\r
 183                                 }\r
 184                                 //$ret.="<br>";\r
 185                                 break;\r
 186                         case "photo":\r
 187                                 $ret.= "<img width='" . $j->width . "' src='" . proxy_url($j->url) . "'>";\r
 188                                 break;\r
 189                         case "link":\r
 190                                 break;\r
 191                         case "rich":\r
 192                                 // not so safe..\r
 193                                 if (!Config::get("system", "no_oembed_rich_content")) {\r
 194                                         $ret.= proxy_parse_html($jhtml);\r
 195                                 }\r
 196                                 break;\r
 197                 }\r
 198 \r
 199                 // add link to source if not present in "rich" type\r
 200                 if ($j->type != 'rich' || !strpos($j->html, $embedurl)) {\r
 201                         $ret .= "<h4>";\r
 202                         if (isset($j->title)) {\r
 203                                 if (isset($j->provider_name)) {\r
 204                                         $ret .= $j->provider_name . ": ";\r
 205                                 }\r
 206 \r
 207                                 $embedlink = (isset($j->title)) ? $j->title : $embedurl;\r
 208                                 $ret .= "<a href='$embedurl' rel='oembed'>$embedlink</a>";\r
 209                                 if (isset($j->author_name)) {\r
 210                                         $ret.=" (" . $j->author_name . ")";\r
 211                                 }\r
 212                         } elseif (isset($j->provider_name) || isset($j->author_name)) {\r
 213                                 $embedlink = "";\r
 214                                 if (isset($j->provider_name)) {\r
 215                                         $embedlink .= $j->provider_name;\r
 216                                 }\r
 217 \r
 218                                 if (isset($j->author_name)) {\r
 219                                         if ($embedlink != "") {\r
 220                                                 $embedlink .= ": ";\r
 221                                         }\r
 222 \r
 223                                         $embedlink .= $j->author_name;\r
 224                                 }\r
 225                                 if (trim($embedlink) == "") {\r
 226                                         $embedlink = $embedurl;\r
 227                                 }\r
 228 \r
 229                                 $ret .= "<a href='$embedurl' rel='oembed'>$embedlink</a>";\r
 230                         }\r
 231                         //if (isset($j->author_name)) $ret.=" by ".$j->author_name;\r
 232                         //if (isset($j->provider_name)) $ret.=" on ".$j->provider_name;\r
 233                         $ret .= "</h4>";\r
 234                 } else {\r
 235                         // add <a> for html2bbcode conversion\r
 236                         $ret .= "<a href='$embedurl' rel='oembed'>$embedurl</a>";\r
 237                 }\r
 238                 $ret.="</span>";\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 = OEmbed::buildXPath("class", "oembed");\r
 272                         $entries = $xpath->query("//span[$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]"), $e);\r
 279                                 }\r
 280                         }\r
 281                         return OEmbed::getInnerHTML($dom->getElementsByTagName("body")->item(0));\r
 282                 } else {\r
 283                         return $text;\r
 284                 }\r
 285         }\r
 286 \r
 287         /**\r
 288          * @brief Generates the iframe HTML for an oembed attachment.\r
 289          *\r
 290          * Width and height are given by the remote, and are regularly too small for\r
 291          * the generated iframe.\r
 292          *\r
 293          * The width is entirely discarded for the actual width of the post, while fixed\r
 294          * height is used as a starting point before the inevitable resizing.\r
 295          *\r
 296          * Since the iframe is automatically resized on load, there are no need for ugly\r
 297          * and impractical scrollbars.\r
 298          *\r
 299          * @param string $src Original remote URL to embed\r
 300          * @param string $width\r
 301          * @param string $height\r
 302          * @return string formatted HTML\r
 303          *\r
 304          * @see oembed_format_object()\r
 305          */\r
 306         private static function iframe($src, $width, $height)\r
 307         {\r
 308                 $a = get_app();\r
 309 \r
 310                 if (!$height || strstr($height, '%')) {\r
 311                         $height = '200';\r
 312                 }\r
 313                 $width = '100%';\r
 314 \r
 315                 $s = System::baseUrl() . '/oembed/' . base64url_encode($src);\r
 316                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $s . '" allowfullscreen scrolling="no" frameborder="no">' . t('Embedded content') . '</iframe>';\r
 317         }\r
 318 \r
 319         /**\r
 320          * Generates an XPath query to select elements whose provided attribute contains\r
 321          * the provided value in a space-separated list.\r
 322          *\r
 323          * @brief Generates attribute search XPath string\r
 324          *\r
 325          * @param string $attr Name of the attribute to seach\r
 326          * @param string $value Value to search in a space-separated list\r
 327          * @return string\r
 328          */\r
 329         private static function buildXPath($attr, $value)\r
 330         {\r
 331                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath\r
 332                 return "contains( normalize-space( @$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
 333         }\r
 334 \r
 335         /**\r
 336          * Returns the inner XML string of a provided DOMNode\r
 337          *\r
 338          * @brief Returns the inner XML string of a provided DOMNode\r
 339          *\r
 340          * @param DOMNode $node\r
 341          * @return string\r
 342          */\r
 343         private static function getInnerHTML(DOMNode $node)\r
 344         {\r
 345                 $innerHTML = '';\r
 346                 $children = $node->childNodes;\r
 347                 foreach ($children as $child) {\r
 348                         $innerHTML .= $child->ownerDocument->saveXML($child);\r
 349                 }\r
 350                 return $innerHTML;\r
 351         }\r
 352 }\r