src/Content/OEmbed.php

   1 <?php
   2
   3 /**
   4  * @file src/Content/OEmbed.php
   5  */
   6 namespace Friendica\Content;
   7
   8 use DOMDocument;
   9 use DOMNode;
  10 use DOMText;
  11 use DOMXPath;
  12 use Exception;
  13 use Friendica\Core\Cache\Duration;
  14 use Friendica\Core\Config;
  15 use Friendica\Core\Hook;
  16 use Friendica\Core\Renderer;
  17 use Friendica\Database\DBA;
  18 use Friendica\DI;
  19 use Friendica\Util\DateTimeFormat;
  20 use Friendica\Util\Network;
  21 use Friendica\Util\ParseUrl;
  22 use Friendica\Util\Proxy as ProxyUtils;
  23 use Friendica\Util\Strings;
  24
  25 /**
  26  * Handles all OEmbed content fetching and replacement
  27  *
  28  * OEmbed is a standard used to allow an embedded representation of a URL on
  29  * third party sites
  30  *
  31  * @see https://oembed.com
  32  *
  33  * @author Hypolite Petovan <hypolite@mrpetovan.com>
  34  */
  35 class OEmbed
  36 {
  37         public static function replaceCallback($matches)
  38         {
  39                 $embedurl = $matches[1];
  40                 $j = self::fetchURL($embedurl, !self::isAllowedURL($embedurl));
  41                 $s = self::formatObject($j);
  42
  43                 return $s;
  44         }
  45
  46         /**
  47          * Get data from an URL to embed its content.
  48          *
  49          * @param string $embedurl     The URL from which the data should be fetched.
  50          * @param bool   $no_rich_type If set to true rich type content won't be fetched.
  51          *
  52          * @return \Friendica\Object\OEmbed
  53          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  54          */
  55         public static function fetchURL($embedurl, $no_rich_type = false)
  56         {
  57                 $embedurl = trim($embedurl, '\'"');
  58
  59                 $a = DI::app();
  60
  61                 $cache_key = 'oembed:' . $a->videowidth . ':' . $embedurl;
  62
  63                 $condition = ['url' => Strings::normaliseLink($embedurl), 'maxwidth' => $a->videowidth];
  64                 $oembed_record = DBA::selectFirst('oembed', ['content'], $condition);
  65                 if (DBA::isResult($oembed_record)) {
  66                         $json_string = $oembed_record['content'];
  67                 } else {
  68                         $json_string = DI::cache()->get($cache_key);
  69                 }
  70
  71                 // These media files should now be caught in bbcode.php
  72                 // left here as a fallback in case this is called from another source
  73                 $noexts = ['mp3', 'mp4', 'ogg', 'ogv', 'oga', 'ogm', 'webm'];
  74                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  75
  76                 $oembed = new \Friendica\Object\OEmbed($embedurl);
  77
  78                 if ($json_string) {
  79                         $oembed->parseJSON($json_string);
  80                 } else {
  81                         $json_string = '';
  82
  83                         if (!in_array($ext, $noexts)) {
  84                                 // try oembed autodiscovery
  85                                 $html_text = Network::fetchUrl($embedurl, false, 15, 'text/*');
  86                                 if ($html_text) {
  87                                         $dom = @DOMDocument::loadHTML($html_text);
  88                                         if ($dom) {
  89                                                 $xpath = new DOMXPath($dom);
  90                                                 $entries = $xpath->query("//link[@type='application/json+oembed']");
  91                                                 foreach ($entries as $e) {
  92                                                         $href = $e->getAttributeNode('href')->nodeValue;
  93                                                         $json_string = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
  94                                                         break;
  95                                                 }
  96
  97                                                 $entries = $xpath->query("//link[@type='text/json+oembed']");
  98                                                 foreach ($entries as $e) {
  99                                                         $href = $e->getAttributeNode('href')->nodeValue;
 100                                                         $json_string = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
 101                                                         break;
 102                                                 }
 103                                         }
 104                                 }
 105                         }
 106
 107                         $json_string = trim($json_string);
 108
 109                         if (!$json_string || $json_string[0] != '{') {
 110                                 $json_string = '{"type":"error"}';
 111                         }
 112
 113                         $oembed->parseJSON($json_string);
 114
 115                         if (!empty($oembed->type) && $oembed->type != 'error') {
 116                                 DBA::insert('oembed', [
 117                                         'url' => Strings::normaliseLink($embedurl),
 118                                         'maxwidth' => $a->videowidth,
 119                                         'content' => $json_string,
 120                                         'created' => DateTimeFormat::utcNow()
 121                                 ], true);
 122                                 $cache_ttl = Duration::DAY;
 123                         } else {
 124                                 $cache_ttl = Duration::FIVE_MINUTES;
 125                         }
 126
 127                         DI::cache()->set($cache_key, $json_string, $cache_ttl);
 128                 }
 129
 130                 if ($oembed->type == 'error') {
 131                         return $oembed;
 132                 }
 133
 134                 // Always embed the SSL version
 135                 $oembed->html = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'], ['https://www.youtube.com/', 'https://player.vimeo.com/'], $oembed->html);
 136
 137                 // If fetching information doesn't work, then improve via internal functions
 138                 if ($no_rich_type && ($oembed->type == 'rich')) {
 139                         $data = ParseUrl::getSiteinfoCached($embedurl, true, false);
 140                         $oembed->type = $data['type'];
 141
 142                         if ($oembed->type == 'photo') {
 143                                 $oembed->url = $data['url'];
 144                         }
 145
 146                         if (isset($data['title'])) {
 147                                 $oembed->title = $data['title'];
 148                         }
 149
 150                         if (isset($data['text'])) {
 151                                 $oembed->description = $data['text'];
 152                         }
 153
 154                         if (!empty($data['images'])) {
 155                                 $oembed->thumbnail_url = $data['images'][0]['src'];
 156                                 $oembed->thumbnail_width = $data['images'][0]['width'];
 157                                 $oembed->thumbnail_height = $data['images'][0]['height'];
 158                         }
 159                 }
 160
 161                 Hook::callAll('oembed_fetch_url', $embedurl, $oembed);
 162
 163                 return $oembed;
 164         }
 165
 166         private static function formatObject(\Friendica\Object\OEmbed $oembed)
 167         {
 168                 $ret = '<div class="oembed ' . $oembed->type . '">';
 169
 170                 switch ($oembed->type) {
 171                         case "video":
 172                                 if ($oembed->thumbnail_url) {
 173                                         $tw = (isset($oembed->thumbnail_width) && intval($oembed->thumbnail_width)) ? $oembed->thumbnail_width : 200;
 174                                         $th = (isset($oembed->thumbnail_height) && intval($oembed->thumbnail_height)) ? $oembed->thumbnail_height : 180;
 175                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio
 176                                         $tr = (($th) ? $tw / $th : 1);
 177
 178                                         $th = 120;
 179                                         $tw = $th * $tr;
 180                                         $tpl = Renderer::getMarkupTemplate('oembed_video.tpl');
 181                                         $ret .= Renderer::replaceMacros($tpl, [
 182                                                 '$embedurl' => $oembed->embed_url,
 183                                                 '$escapedhtml' => base64_encode($oembed->html),
 184                                                 '$tw' => $tw,
 185                                                 '$th' => $th,
 186                                                 '$turl' => $oembed->thumbnail_url,
 187                                         ]);
 188                                 } else {
 189                                         $ret = $oembed->html;
 190                                 }
 191                                 break;
 192
 193                         case "photo":
 194                                 $ret .= '<img width="' . $oembed->width . '" src="' . ProxyUtils::proxifyUrl($oembed->url) . '">';
 195                                 break;
 196
 197                         case "link":
 198                                 break;
 199
 200                         case "rich":
 201                                 $ret .= ProxyUtils::proxifyHtml($oembed->html);
 202                                 break;
 203                 }
 204
 205                 // add link to source if not present in "rich" type
 206                 if ($oembed->type != 'rich' || !strpos($oembed->html, $oembed->embed_url)) {
 207                         $ret .= '<h4>';
 208                         if (!empty($oembed->title)) {
 209                                 if (!empty($oembed->provider_name)) {
 210                                         $ret .= $oembed->provider_name . ": ";
 211                                 }
 212
 213                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 214                                 if (!empty($oembed->author_name)) {
 215                                         $ret .= ' (' . $oembed->author_name . ')';
 216                                 }
 217                         } elseif (!empty($oembed->provider_name) || !empty($oembed->author_name)) {
 218                                 $embedlink = "";
 219                                 if (!empty($oembed->provider_name)) {
 220                                         $embedlink .= $oembed->provider_name;
 221                                 }
 222
 223                                 if (!empty($oembed->author_name)) {
 224                                         if ($embedlink != "") {
 225                                                 $embedlink .= ": ";
 226                                         }
 227
 228                                         $embedlink .= $oembed->author_name;
 229                                 }
 230                                 if (trim($embedlink) == "") {
 231                                         $embedlink = $oembed->embed_url;
 232                                 }
 233
 234                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $embedlink . '</a>';
 235                         } else {
 236                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->embed_url . '</a>';
 237                         }
 238                         $ret .= "</h4>";
 239                 } elseif (!strpos($oembed->html, $oembed->embed_url)) {
 240                         // add <a> for html2bbcode conversion
 241                         $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 242                 }
 243
 244                 $ret .= '</div>';
 245
 246                 return str_replace("\n", "", $ret);
 247         }
 248
 249         public static function BBCode2HTML($text)
 250         {
 251                 $stopoembed = Config::get("system", "no_oembed");
 252                 if ($stopoembed == true) {
 253                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . DI::l10n()->t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 254                 }
 255                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 256         }
 257
 258         /**
 259          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 260          * and replace it with [embed]url[/embed]
 261          *
 262          * @param $text
 263          * @return string
 264          */
 265         public static function HTML2BBCode($text)
 266         {
 267                 // start parser only if 'oembed' is in text
 268                 if (strpos($text, "oembed")) {
 269
 270                         // convert non ascii chars to html entities
 271                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 272
 273                         // If it doesn't parse at all, just return the text.
 274                         $dom = @DOMDocument::loadHTML($html_text);
 275                         if (!$dom) {
 276                                 return $text;
 277                         }
 278                         $xpath = new DOMXPath($dom);
 279
 280                         $xattr = self::buildXPath("class", "oembed");
 281                         $entries = $xpath->query("//div[$xattr]");
 282
 283                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 284                         foreach ($entries as $e) {
 285                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 286                                 if (!is_null($href)) {
 287                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 288                                 }
 289                         }
 290                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 291                 } else {
 292                         return $text;
 293                 }
 294         }
 295
 296         /**
 297          * Determines if rich content OEmbed is allowed for the provided URL
 298          *
 299          * @param string $url
 300          * @return boolean
 301          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 302          */
 303         public static function isAllowedURL($url)
 304         {
 305                 if (!Config::get('system', 'no_oembed_rich_content')) {
 306                         return true;
 307                 }
 308
 309                 $domain = parse_url($url, PHP_URL_HOST);
 310                 if (empty($domain)) {
 311                         return false;
 312                 }
 313
 314                 $str_allowed = Config::get('system', 'allowed_oembed', '');
 315                 if (empty($str_allowed)) {
 316                         return false;
 317                 }
 318
 319                 $allowed = explode(',', $str_allowed);
 320
 321                 return Network::isDomainAllowed($domain, $allowed);
 322         }
 323
 324         public static function getHTML($url, $title = null)
 325         {
 326                 // Always embed the SSL version
 327                 $url = str_replace(["http://www.youtube.com/", "http://player.vimeo.com/"],
 328                                         ["https://www.youtube.com/", "https://player.vimeo.com/"], $url);
 329
 330                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 331
 332                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 333                         throw new Exception('OEmbed failed for URL: ' . $url);
 334                 }
 335
 336                 if (!empty($title)) {
 337                         $o->title = $title;
 338                 }
 339
 340                 $html = self::formatObject($o);
 341
 342                 return $html;
 343         }
 344
 345         /**
 346          * Generates the iframe HTML for an oembed attachment.
 347          *
 348          * Width and height are given by the remote, and are regularly too small for
 349          * the generated iframe.
 350          *
 351          * The width is entirely discarded for the actual width of the post, while fixed
 352          * height is used as a starting point before the inevitable resizing.
 353          *
 354          * Since the iframe is automatically resized on load, there are no need for ugly
 355          * and impractical scrollbars.
 356          *
 357          * @todo  This function is currently unused until someone™ adds support for a separate OEmbed domain
 358          *
 359          * @param string $src Original remote URL to embed
 360          * @param string $width
 361          * @param string $height
 362          * @return string formatted HTML
 363          *
 364          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 365          * @see   oembed_format_object()
 366          */
 367         private static function iframe($src, $width, $height)
 368         {
 369                 if (!$height || strstr($height, '%')) {
 370                         $height = '200';
 371                 }
 372                 $width = '100%';
 373
 374                 $src = DI::baseUrl() . '/oembed/' . Strings::base64UrlEncode($src);
 375                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . DI::l10n()->t('Embedded content') . '</iframe>';
 376         }
 377
 378         /**
 379          * Generates attribute search XPath string
 380          *
 381          * Generates an XPath query to select elements whose provided attribute contains
 382          * the provided value in a space-separated list.
 383          *
 384          * @param string $attr Name of the attribute to seach
 385          * @param string $value Value to search in a space-separated list
 386          * @return string
 387          */
 388         private static function buildXPath($attr, $value)
 389         {
 390                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 391                 return "contains(normalize-space(@$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'";
 392         }
 393
 394         /**
 395          * Returns the inner XML string of a provided DOMNode
 396          *
 397          * @param DOMNode $node
 398          * @return string
 399          */
 400         private static function getInnerHTML(DOMNode $node)
 401         {
 402                 $innerHTML = '';
 403                 $children = $node->childNodes;
 404                 foreach ($children as $child) {
 405                         $innerHTML .= $child->ownerDocument->saveXML($child);
 406                 }
 407                 return $innerHTML;
 408         }
 409
 410 }