src/Content/OEmbed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Content;
  23
  24 use DOMDocument;
  25 use DOMNode;
  26 use DOMText;
  27 use DOMXPath;
  28 use Exception;
  29 use Friendica\Core\Cache\Duration;
  30 use Friendica\Core\Hook;
  31 use Friendica\Core\Renderer;
  32 use Friendica\Database\Database;
  33 use Friendica\Database\DBA;
  34 use Friendica\DI;
  35 use Friendica\Util\DateTimeFormat;
  36 use Friendica\Util\Network;
  37 use Friendica\Util\ParseUrl;
  38 use Friendica\Util\Proxy as ProxyUtils;
  39 use Friendica\Util\Strings;
  40
  41 /**
  42  * Handles all OEmbed content fetching and replacement
  43  *
  44  * OEmbed is a standard used to allow an embedded representation of a URL on
  45  * third party sites
  46  *
  47  * @see https://oembed.com
  48  */
  49 class OEmbed
  50 {
  51         public static function replaceCallback($matches)
  52         {
  53                 $embedurl = $matches[1];
  54                 $j = self::fetchURL($embedurl, !self::isAllowedURL($embedurl));
  55                 $s = self::formatObject($j);
  56
  57                 return $s;
  58         }
  59
  60         /**
  61          * Get data from an URL to embed its content.
  62          *
  63          * @param string $embedurl     The URL from which the data should be fetched.
  64          * @param bool   $no_rich_type If set to true rich type content won't be fetched.
  65          *
  66          * @return \Friendica\Object\OEmbed
  67          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  68          */
  69         public static function fetchURL($embedurl, $no_rich_type = false)
  70         {
  71                 $embedurl = trim($embedurl, '\'"');
  72
  73                 $a = DI::app();
  74
  75                 $cache_key = 'oembed:' . $a->videowidth . ':' . $embedurl;
  76
  77                 $condition = ['url' => Strings::normaliseLink($embedurl), 'maxwidth' => $a->videowidth];
  78                 $oembed_record = DBA::selectFirst('oembed', ['content'], $condition);
  79                 if (DBA::isResult($oembed_record)) {
  80                         $json_string = $oembed_record['content'];
  81                 } else {
  82                         $json_string = DI::cache()->get($cache_key);
  83                 }
  84
  85                 // These media files should now be caught in bbcode.php
  86                 // left here as a fallback in case this is called from another source
  87                 $noexts = ['mp3', 'mp4', 'ogg', 'ogv', 'oga', 'ogm', 'webm'];
  88                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  89
  90                 $oembed = new \Friendica\Object\OEmbed($embedurl);
  91
  92                 if ($json_string) {
  93                         $oembed->parseJSON($json_string);
  94                 } else {
  95                         $json_string = '';
  96
  97                         if (!in_array($ext, $noexts)) {
  98                                 // try oembed autodiscovery
  99                                 $html_text = DI::httpRequest()->fetch($embedurl, 15, 'text/*');
 100                                 if ($html_text) {
 101                                         $dom = new DOMDocument();
 102                                         if (@$dom->loadHTML($html_text)) {
 103                                                 $xpath = new DOMXPath($dom);
 104                                                 foreach (
 105                                                         $xpath->query("//link[@type='application/json+oembed'] | //link[@type='text/json+oembed']")
 106                                                         as $link)
 107                                                 {
 108                                                         $href = $link->getAttributeNode('href')->nodeValue;
 109                                                         // Both Youtube and Vimeo output OEmbed endpoint URL with HTTP
 110                                                         // but their OEmbed endpoint is only accessible by HTTPS ¯\_(ツ)_/¯
 111                                                         $href = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'],
 112                                                                 ['https://www.youtube.com/', 'https://player.vimeo.com/'], $href);
 113                                                         $result = DI::httpRequest()->fetchFull($href . '&maxwidth=' . $a->videowidth);
 114                                                         if ($result->getReturnCode() === 200) {
 115                                                                 $json_string = $result->getBody();
 116                                                                 break;
 117                                                         }
 118                                                 }
 119                                         }
 120                                 }
 121                         }
 122
 123                         $json_string = trim($json_string);
 124
 125                         if (!$json_string || $json_string[0] != '{') {
 126                                 $json_string = '{"type":"error"}';
 127                         }
 128
 129                         $oembed->parseJSON($json_string);
 130
 131                         if (!empty($oembed->type) && $oembed->type != 'error') {
 132                                 DBA::insert('oembed', [
 133                                         'url' => Strings::normaliseLink($embedurl),
 134                                         'maxwidth' => $a->videowidth,
 135                                         'content' => $json_string,
 136                                         'created' => DateTimeFormat::utcNow()
 137                                 ], Database::INSERT_UPDATE);
 138                                 $cache_ttl = Duration::DAY;
 139                         } else {
 140                                 $cache_ttl = Duration::FIVE_MINUTES;
 141                         }
 142
 143                         DI::cache()->set($cache_key, $json_string, $cache_ttl);
 144                 }
 145
 146                 // Always embed the SSL version
 147                 if (!empty($oembed->html)) {
 148                         $oembed->html = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'], ['https://www.youtube.com/', 'https://player.vimeo.com/'], $oembed->html);
 149                 }
 150
 151                 // Improve the OEmbed data with data from OpenGraph, Twitter cards and other sources
 152                 $data = ParseUrl::getSiteinfoCached($embedurl, true, false);
 153                 if (($oembed->type == 'error') && empty($data['title']) && empty($data['text'])) {
 154                         return $oembed;
 155                 }
 156
 157                 if ($no_rich_type || ($oembed->type == 'error')) {
 158                         $oembed->html = '';
 159                         $oembed->type = $data['type'];
 160
 161                         if ($oembed->type == 'photo') {
 162                                 $oembed->url = $data['url'];
 163                         }
 164                 }
 165
 166                 if (!empty($data['title']) && empty($oembed->title)) {
 167                         $oembed->title = $data['title'];
 168                 }
 169
 170                 if (!empty($data['text']) && empty($oembed->description)) {
 171                         $oembed->description = $data['text'];
 172                 }
 173
 174                 if (!empty($data['publisher']) && empty($oembed->provider_name)) {
 175                         $oembed->provider_name = $data['publisher'];
 176                 }
 177
 178                 if (!empty($data['publisher_url']) && empty($oembed->provider_url)) {
 179                         $oembed->provider_url = $data['publisher_url'];
 180                 }
 181
 182                 if (!empty($data['author']) && empty($oembed->author_name)) {
 183                         $oembed->author_name = $data['author'];
 184                 }
 185
 186                 if (!empty($data['author_url']) && empty($oembed->author_url)) {
 187                         $oembed->author_url = $data['author_url'];
 188                 }
 189
 190                 if (!empty($data['images']) && empty($oembed->thumbnail_url)) {
 191                         $oembed->thumbnail_url = $data['images'][0]['src'];
 192                         $oembed->thumbnail_width = $data['images'][0]['width'];
 193                         $oembed->thumbnail_height = $data['images'][0]['height'];
 194                 }
 195
 196                 Hook::callAll('oembed_fetch_url', $embedurl, $oembed);
 197
 198                 return $oembed;
 199         }
 200
 201         private static function formatObject(\Friendica\Object\OEmbed $oembed)
 202         {
 203                 $ret = '<div class="oembed ' . $oembed->type . '">';
 204
 205                 switch ($oembed->type) {
 206                         case "video":
 207                                 if ($oembed->thumbnail_url) {
 208                                         $tw = (isset($oembed->thumbnail_width) && intval($oembed->thumbnail_width)) ? $oembed->thumbnail_width : 200;
 209                                         $th = (isset($oembed->thumbnail_height) && intval($oembed->thumbnail_height)) ? $oembed->thumbnail_height : 180;
 210                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio
 211                                         $tr = (($th) ? $tw / $th : 1);
 212
 213                                         $th = 120;
 214                                         $tw = $th * $tr;
 215                                         $tpl = Renderer::getMarkupTemplate('oembed_video.tpl');
 216                                         $ret .= Renderer::replaceMacros($tpl, [
 217                                                 '$embedurl' => $oembed->embed_url,
 218                                                 '$escapedhtml' => base64_encode($oembed->html),
 219                                                 '$tw' => $tw,
 220                                                 '$th' => $th,
 221                                                 '$turl' => $oembed->thumbnail_url,
 222                                         ]);
 223                                 } else {
 224                                         $ret = $oembed->html;
 225                                 }
 226                                 break;
 227
 228                         case "photo":
 229                                 $ret .= '<img width="' . $oembed->width . '" src="' . ProxyUtils::proxifyUrl($oembed->url) . '">';
 230                                 break;
 231
 232                         case "link":
 233                                 break;
 234
 235                         case "rich":
 236                                 $ret .= ProxyUtils::proxifyHtml($oembed->html);
 237                                 break;
 238                 }
 239
 240                 // add link to source if not present in "rich" type
 241                 if ($oembed->type != 'rich' || !strpos($oembed->html, $oembed->embed_url)) {
 242                         $ret .= '<h4>';
 243                         if (!empty($oembed->title)) {
 244                                 if (!empty($oembed->provider_name)) {
 245                                         $ret .= $oembed->provider_name . ": ";
 246                                 }
 247
 248                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 249                                 if (!empty($oembed->author_name)) {
 250                                         $ret .= ' (' . $oembed->author_name . ')';
 251                                 }
 252                         } elseif (!empty($oembed->provider_name) || !empty($oembed->author_name)) {
 253                                 $embedlink = "";
 254                                 if (!empty($oembed->provider_name)) {
 255                                         $embedlink .= $oembed->provider_name;
 256                                 }
 257
 258                                 if (!empty($oembed->author_name)) {
 259                                         if ($embedlink != "") {
 260                                                 $embedlink .= ": ";
 261                                         }
 262
 263                                         $embedlink .= $oembed->author_name;
 264                                 }
 265                                 if (trim($embedlink) == "") {
 266                                         $embedlink = $oembed->embed_url;
 267                                 }
 268
 269                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $embedlink . '</a>';
 270                         } else {
 271                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->embed_url . '</a>';
 272                         }
 273                         $ret .= "</h4>";
 274                 } elseif (!strpos($oembed->html, $oembed->embed_url)) {
 275                         // add <a> for html2bbcode conversion
 276                         $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 277                 }
 278
 279                 $ret .= '</div>';
 280
 281                 return str_replace("\n", "", $ret);
 282         }
 283
 284         public static function BBCode2HTML($text)
 285         {
 286                 $stopoembed = DI::config()->get("system", "no_oembed");
 287                 if ($stopoembed == true) {
 288                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . DI::l10n()->t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 289                 }
 290                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 291         }
 292
 293         /**
 294          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 295          * and replace it with [embed]url[/embed]
 296          *
 297          * @param $text
 298          * @return string
 299          */
 300         public static function HTML2BBCode($text)
 301         {
 302                 // start parser only if 'oembed' is in text
 303                 if (strpos($text, "oembed")) {
 304
 305                         // convert non ascii chars to html entities
 306                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 307
 308                         // If it doesn't parse at all, just return the text.
 309                         $dom = @DOMDocument::loadHTML($html_text);
 310                         if (!$dom) {
 311                                 return $text;
 312                         }
 313                         $xpath = new DOMXPath($dom);
 314
 315                         $xattr = self::buildXPath("class", "oembed");
 316                         $entries = $xpath->query("//div[$xattr]");
 317
 318                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 319                         foreach ($entries as $e) {
 320                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 321                                 if (!is_null($href)) {
 322                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 323                                 }
 324                         }
 325                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 326                 } else {
 327                         return $text;
 328                 }
 329         }
 330
 331         /**
 332          * Determines if rich content OEmbed is allowed for the provided URL
 333          *
 334          * @param string $url
 335          * @return boolean
 336          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 337          */
 338         public static function isAllowedURL($url)
 339         {
 340                 if (!DI::config()->get('system', 'no_oembed_rich_content')) {
 341                         return true;
 342                 }
 343
 344                 $domain = parse_url($url, PHP_URL_HOST);
 345                 if (empty($domain)) {
 346                         return false;
 347                 }
 348
 349                 $str_allowed = DI::config()->get('system', 'allowed_oembed', '');
 350                 if (empty($str_allowed)) {
 351                         return false;
 352                 }
 353
 354                 $allowed = explode(',', $str_allowed);
 355
 356                 return Network::isDomainAllowed($domain, $allowed);
 357         }
 358
 359         public static function getHTML($url, $title = null)
 360         {
 361                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 362
 363                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 364                         throw new Exception('OEmbed failed for URL: ' . $url);
 365                 }
 366
 367                 if (!empty($title)) {
 368                         $o->title = $title;
 369                 }
 370
 371                 $html = self::formatObject($o);
 372
 373                 return $html;
 374         }
 375
 376         /**
 377          * Generates the iframe HTML for an oembed attachment.
 378          *
 379          * Width and height are given by the remote, and are regularly too small for
 380          * the generated iframe.
 381          *
 382          * The width is entirely discarded for the actual width of the post, while fixed
 383          * height is used as a starting point before the inevitable resizing.
 384          *
 385          * Since the iframe is automatically resized on load, there are no need for ugly
 386          * and impractical scrollbars.
 387          *
 388          * @todo  This function is currently unused until someone™ adds support for a separate OEmbed domain
 389          *
 390          * @param string $src Original remote URL to embed
 391          * @param string $width
 392          * @param string $height
 393          * @return string formatted HTML
 394          *
 395          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 396          * @see   oembed_format_object()
 397          */
 398         private static function iframe($src, $width, $height)
 399         {
 400                 if (!$height || strstr($height, '%')) {
 401                         $height = '200';
 402                 }
 403                 $width = '100%';
 404
 405                 $src = DI::baseUrl() . '/oembed/' . Strings::base64UrlEncode($src);
 406                 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>';
 407         }
 408
 409         /**
 410          * Generates attribute search XPath string
 411          *
 412          * Generates an XPath query to select elements whose provided attribute contains
 413          * the provided value in a space-separated list.
 414          *
 415          * @param string $attr Name of the attribute to seach
 416          * @param string $value Value to search in a space-separated list
 417          * @return string
 418          */
 419         private static function buildXPath($attr, $value)
 420         {
 421                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 422                 return "contains(normalize-space(@$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'";
 423         }
 424
 425         /**
 426          * Returns the inner XML string of a provided DOMNode
 427          *
 428          * @param DOMNode $node
 429          * @return string
 430          */
 431         private static function getInnerHTML(DOMNode $node)
 432         {
 433                 $innerHTML = '';
 434                 $children = $node->childNodes;
 435                 foreach ($children as $child) {
 436                         $innerHTML .= $child->ownerDocument->saveXML($child);
 437                 }
 438                 return $innerHTML;
 439         }
 440
 441 }