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          * @param bool   $use_parseurl Use the "ParseUrl" functionality to add additional data
  66          *
  67          * @return \Friendica\Object\OEmbed
  68          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  69          */
  70         public static function fetchURL($embedurl, bool $no_rich_type = false, bool $use_parseurl = true)
  71         {
  72                 $embedurl = trim($embedurl, '\'"');
  73
  74                 $a = DI::app();
  75
  76                 $cache_key = 'oembed:' . $a->videowidth . ':' . $embedurl;
  77
  78                 $condition = ['url' => Strings::normaliseLink($embedurl), 'maxwidth' => $a->videowidth];
  79                 $oembed_record = DBA::selectFirst('oembed', ['content'], $condition);
  80                 if (DBA::isResult($oembed_record)) {
  81                         $json_string = $oembed_record['content'];
  82                 } else {
  83                         $json_string = DI::cache()->get($cache_key);
  84                 }
  85
  86                 // These media files should now be caught in bbcode.php
  87                 // left here as a fallback in case this is called from another source
  88                 $noexts = ['mp3', 'mp4', 'ogg', 'ogv', 'oga', 'ogm', 'webm'];
  89                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  90
  91                 $oembed = new \Friendica\Object\OEmbed($embedurl);
  92
  93                 if ($json_string) {
  94                         $oembed->parseJSON($json_string);
  95                 } else {
  96                         $json_string = '';
  97
  98                         if (!in_array($ext, $noexts)) {
  99                                 // try oembed autodiscovery
 100                                 $html_text = DI::httpRequest()->fetch($embedurl, 15, 'text/*');
 101                                 if ($html_text) {
 102                                         $dom = new DOMDocument();
 103                                         if (@$dom->loadHTML($html_text)) {
 104                                                 $xpath = new DOMXPath($dom);
 105                                                 foreach (
 106                                                         $xpath->query("//link[@type='application/json+oembed'] | //link[@type='text/json+oembed']")
 107                                                         as $link)
 108                                                 {
 109                                                         $href = $link->getAttributeNode('href')->nodeValue;
 110                                                         // Both Youtube and Vimeo output OEmbed endpoint URL with HTTP
 111                                                         // but their OEmbed endpoint is only accessible by HTTPS ¯\_(ツ)_/¯
 112                                                         $href = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'],
 113                                                                 ['https://www.youtube.com/', 'https://player.vimeo.com/'], $href);
 114                                                         $result = DI::httpRequest()->fetchFull($href . '&maxwidth=' . $a->videowidth);
 115                                                         if ($result->getReturnCode() === 200) {
 116                                                                 $json_string = $result->getBody();
 117                                                                 break;
 118                                                         }
 119                                                 }
 120                                         }
 121                                 }
 122                         }
 123
 124                         $json_string = trim($json_string);
 125
 126                         if (!$json_string || $json_string[0] != '{') {
 127                                 $json_string = '{"type":"error"}';
 128                         }
 129
 130                         $oembed->parseJSON($json_string);
 131
 132                         if (!empty($oembed->type) && $oembed->type != 'error') {
 133                                 DBA::insert('oembed', [
 134                                         'url' => Strings::normaliseLink($embedurl),
 135                                         'maxwidth' => $a->videowidth,
 136                                         'content' => $json_string,
 137                                         'created' => DateTimeFormat::utcNow()
 138                                 ], Database::INSERT_UPDATE);
 139                                 $cache_ttl = Duration::DAY;
 140                         } else {
 141                                 $cache_ttl = Duration::FIVE_MINUTES;
 142                         }
 143
 144                         DI::cache()->set($cache_key, $json_string, $cache_ttl);
 145                 }
 146
 147                 // Always embed the SSL version
 148                 if (!empty($oembed->html)) {
 149                         $oembed->html = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'], ['https://www.youtube.com/', 'https://player.vimeo.com/'], $oembed->html);
 150                 }
 151
 152                 // Improve the OEmbed data with data from OpenGraph, Twitter cards and other sources
 153                 if ($use_parseurl) {
 154                         $data = ParseUrl::getSiteinfoCached($embedurl, false);
 155
 156                         if (($oembed->type == 'error') && empty($data['title']) && empty($data['text'])) {
 157                                 return $oembed;
 158                         }
 159
 160                         if ($no_rich_type || ($oembed->type == 'error')) {
 161                                 $oembed->html = '';
 162                                 $oembed->type = $data['type'];
 163
 164                                 if ($oembed->type == 'photo') {
 165                                         $oembed->url = $data['url'];
 166                                 }
 167                         }
 168
 169                         if (!empty($data['title'])) {
 170                                 $oembed->title = $data['title'];
 171                         }
 172
 173                         if (!empty($data['text'])) {
 174                                 $oembed->description = $data['text'];
 175                         }
 176
 177                         if (!empty($data['publisher_name'])) {
 178                                 $oembed->provider_name = $data['publisher_name'];
 179                         }
 180
 181                         if (!empty($data['publisher_url'])) {
 182                                 $oembed->provider_url = $data['publisher_url'];
 183                         }
 184
 185                         if (!empty($data['author_name'])) {
 186                                 $oembed->author_name = $data['author_name'];
 187                         }
 188
 189                         if (!empty($data['author_url'])) {
 190                                 $oembed->author_url = $data['author_url'];
 191                         }
 192
 193                         if (!empty($data['images'])) {
 194                                 $oembed->thumbnail_url = $data['images'][0]['src'];
 195                                 $oembed->thumbnail_width = $data['images'][0]['width'];
 196                                 $oembed->thumbnail_height = $data['images'][0]['height'];
 197                         }
 198                 }
 199
 200                 Hook::callAll('oembed_fetch_url', $embedurl, $oembed);
 201
 202                 return $oembed;
 203         }
 204
 205         private static function formatObject(\Friendica\Object\OEmbed $oembed)
 206         {
 207                 $ret = '<div class="oembed ' . $oembed->type . '">';
 208
 209                 switch ($oembed->type) {
 210                         case "video":
 211                                 if ($oembed->thumbnail_url) {
 212                                         $tw = (isset($oembed->thumbnail_width) && intval($oembed->thumbnail_width)) ? $oembed->thumbnail_width : 200;
 213                                         $th = (isset($oembed->thumbnail_height) && intval($oembed->thumbnail_height)) ? $oembed->thumbnail_height : 180;
 214                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio
 215                                         $tr = (($th) ? $tw / $th : 1);
 216
 217                                         $th = 120;
 218                                         $tw = $th * $tr;
 219                                         $tpl = Renderer::getMarkupTemplate('oembed_video.tpl');
 220                                         $ret .= Renderer::replaceMacros($tpl, [
 221                                                 '$embedurl' => $oembed->embed_url,
 222                                                 '$escapedhtml' => base64_encode($oembed->html),
 223                                                 '$tw' => $tw,
 224                                                 '$th' => $th,
 225                                                 '$turl' => $oembed->thumbnail_url,
 226                                         ]);
 227                                 } else {
 228                                         $ret = $oembed->html;
 229                                 }
 230                                 break;
 231
 232                         case "photo":
 233                                 $ret .= '<img width="' . $oembed->width . '" src="' . ProxyUtils::proxifyUrl($oembed->url) . '">';
 234                                 break;
 235
 236                         case "link":
 237                                 break;
 238
 239                         case "rich":
 240                                 $ret .= ProxyUtils::proxifyHtml($oembed->html);
 241                                 break;
 242                 }
 243
 244                 // add link to source if not present in "rich" type
 245                 if ($oembed->type != 'rich' || !strpos($oembed->html, $oembed->embed_url)) {
 246                         $ret .= '<h4>';
 247                         if (!empty($oembed->title)) {
 248                                 if (!empty($oembed->provider_name)) {
 249                                         $ret .= $oembed->provider_name . ": ";
 250                                 }
 251
 252                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 253                                 if (!empty($oembed->author_name)) {
 254                                         $ret .= ' (' . $oembed->author_name . ')';
 255                                 }
 256                         } elseif (!empty($oembed->provider_name) || !empty($oembed->author_name)) {
 257                                 $embedlink = "";
 258                                 if (!empty($oembed->provider_name)) {
 259                                         $embedlink .= $oembed->provider_name;
 260                                 }
 261
 262                                 if (!empty($oembed->author_name)) {
 263                                         if ($embedlink != "") {
 264                                                 $embedlink .= ": ";
 265                                         }
 266
 267                                         $embedlink .= $oembed->author_name;
 268                                 }
 269                                 if (trim($embedlink) == "") {
 270                                         $embedlink = $oembed->embed_url;
 271                                 }
 272
 273                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $embedlink . '</a>';
 274                         } else {
 275                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->embed_url . '</a>';
 276                         }
 277                         $ret .= "</h4>";
 278                 } elseif (!strpos($oembed->html, $oembed->embed_url)) {
 279                         // add <a> for html2bbcode conversion
 280                         $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 281                 }
 282
 283                 $ret .= '</div>';
 284
 285                 return str_replace("\n", "", $ret);
 286         }
 287
 288         public static function BBCode2HTML($text)
 289         {
 290                 $stopoembed = DI::config()->get("system", "no_oembed");
 291                 if ($stopoembed == true) {
 292                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . DI::l10n()->t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 293                 }
 294                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 295         }
 296
 297         /**
 298          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 299          * and replace it with [embed]url[/embed]
 300          *
 301          * @param $text
 302          * @return string
 303          */
 304         public static function HTML2BBCode($text)
 305         {
 306                 // start parser only if 'oembed' is in text
 307                 if (strpos($text, "oembed")) {
 308
 309                         // convert non ascii chars to html entities
 310                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 311
 312                         // If it doesn't parse at all, just return the text.
 313                         $dom = @DOMDocument::loadHTML($html_text);
 314                         if (!$dom) {
 315                                 return $text;
 316                         }
 317                         $xpath = new DOMXPath($dom);
 318
 319                         $xattr = self::buildXPath("class", "oembed");
 320                         $entries = $xpath->query("//div[$xattr]");
 321
 322                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 323                         foreach ($entries as $e) {
 324                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 325                                 if (!is_null($href)) {
 326                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 327                                 }
 328                         }
 329                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 330                 } else {
 331                         return $text;
 332                 }
 333         }
 334
 335         /**
 336          * Determines if rich content OEmbed is allowed for the provided URL
 337          *
 338          * @param string $url
 339          * @return boolean
 340          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 341          */
 342         public static function isAllowedURL($url)
 343         {
 344                 if (!DI::config()->get('system', 'no_oembed_rich_content')) {
 345                         return true;
 346                 }
 347
 348                 $domain = parse_url($url, PHP_URL_HOST);
 349                 if (empty($domain)) {
 350                         return false;
 351                 }
 352
 353                 $str_allowed = DI::config()->get('system', 'allowed_oembed', '');
 354                 if (empty($str_allowed)) {
 355                         return false;
 356                 }
 357
 358                 $allowed = explode(',', $str_allowed);
 359
 360                 return Network::isDomainAllowed($domain, $allowed);
 361         }
 362
 363         public static function getHTML($url, $title = null)
 364         {
 365                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 366
 367                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 368                         throw new Exception('OEmbed failed for URL: ' . $url);
 369                 }
 370
 371                 if (!empty($title)) {
 372                         $o->title = $title;
 373                 }
 374
 375                 $html = self::formatObject($o);
 376
 377                 return $html;
 378         }
 379
 380         /**
 381          * Generates the iframe HTML for an oembed attachment.
 382          *
 383          * Width and height are given by the remote, and are regularly too small for
 384          * the generated iframe.
 385          *
 386          * The width is entirely discarded for the actual width of the post, while fixed
 387          * height is used as a starting point before the inevitable resizing.
 388          *
 389          * Since the iframe is automatically resized on load, there are no need for ugly
 390          * and impractical scrollbars.
 391          *
 392          * @todo  This function is currently unused until someone™ adds support for a separate OEmbed domain
 393          *
 394          * @param string $src Original remote URL to embed
 395          * @param string $width
 396          * @param string $height
 397          * @return string formatted HTML
 398          *
 399          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 400          * @see   oembed_format_object()
 401          */
 402         private static function iframe($src, $width, $height)
 403         {
 404                 if (!$height || strstr($height, '%')) {
 405                         $height = '200';
 406                 }
 407                 $width = '100%';
 408
 409                 $src = DI::baseUrl() . '/oembed/' . Strings::base64UrlEncode($src);
 410                 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>';
 411         }
 412
 413         /**
 414          * Generates attribute search XPath string
 415          *
 416          * Generates an XPath query to select elements whose provided attribute contains
 417          * the provided value in a space-separated list.
 418          *
 419          * @param string $attr Name of the attribute to seach
 420          * @param string $value Value to search in a space-separated list
 421          * @return string
 422          */
 423         private static function buildXPath($attr, $value)
 424         {
 425                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 426                 return "contains(normalize-space(@$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'";
 427         }
 428
 429         /**
 430          * Returns the inner XML string of a provided DOMNode
 431          *
 432          * @param DOMNode $node
 433          * @return string
 434          */
 435         private static function getInnerHTML(DOMNode $node)
 436         {
 437                 $innerHTML = '';
 438                 $children = $node->childNodes;
 439                 foreach ($children as $child) {
 440                         $innerHTML .= $child->ownerDocument->saveXML($child);
 441                 }
 442                 return $innerHTML;
 443         }
 444
 445 }