src/Content/OEmbed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2022, the Friendica project
   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\Enum\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\Network\HTTPClient\Client\HttpClientAccept;
  36 use Friendica\Util\DateTimeFormat;
  37 use Friendica\Util\Network;
  38 use Friendica\Util\ParseUrl;
  39 use Friendica\Util\Proxy;
  40 use Friendica\Util\Strings;
  41
  42 /**
  43  * Handles all OEmbed content fetching and replacement
  44  *
  45  * OEmbed is a standard used to allow an embedded representation of a URL on
  46  * third party sites
  47  *
  48  * @see https://oembed.com
  49  */
  50 class OEmbed
  51 {
  52         public static function replaceCallback($matches)
  53         {
  54                 $embedurl = $matches[1];
  55                 $j = self::fetchURL($embedurl, !self::isAllowedURL($embedurl));
  56                 $s = self::formatObject($j);
  57
  58                 return $s;
  59         }
  60
  61         /**
  62          * Get data from an URL to embed its content.
  63          *
  64          * @param string $embedurl     The URL from which the data should be fetched.
  65          * @param bool   $no_rich_type If set to true rich type content won't be fetched.
  66          * @param bool   $use_parseurl Use the "ParseUrl" functionality to add additional data
  67          *
  68          * @return \Friendica\Object\OEmbed
  69          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  70          */
  71         public static function fetchURL($embedurl, bool $no_rich_type = false, bool $use_parseurl = true)
  72         {
  73                 $embedurl = trim($embedurl, '\'"');
  74
  75                 $a = DI::app();
  76
  77                 $cache_key = 'oembed:' . $a->getThemeInfoValue('videowidth') . ':' . $embedurl;
  78
  79                 $condition = ['url' => Strings::normaliseLink($embedurl), 'maxwidth' => $a->getThemeInfoValue('videowidth')];
  80                 $oembed_record = DBA::selectFirst('oembed', ['content'], $condition);
  81                 if (DBA::isResult($oembed_record)) {
  82                         $json_string = $oembed_record['content'];
  83                 } else {
  84                         $json_string = DI::cache()->get($cache_key);
  85                 }
  86
  87                 // These media files should now be caught in bbcode.php
  88                 // left here as a fallback in case this is called from another source
  89                 $noexts = ['mp3', 'mp4', 'ogg', 'ogv', 'oga', 'ogm', 'webm'];
  90                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  91
  92                 $oembed = new \Friendica\Object\OEmbed($embedurl);
  93
  94                 if ($json_string) {
  95                         $oembed->parseJSON($json_string);
  96                 } else {
  97                         $json_string = '';
  98
  99                         if (!in_array($ext, $noexts)) {
 100                                 // try oembed autodiscovery
 101                                 $html_text = DI::httpClient()->fetch($embedurl, HttpClientAccept::HTML, 15);
 102                                 if (!empty($html_text)) {
 103                                         $dom = new DOMDocument();
 104                                         if (@$dom->loadHTML($html_text)) {
 105                                                 $xpath = new DOMXPath($dom);
 106                                                 foreach (
 107                                                         $xpath->query("//link[@type='application/json+oembed'] | //link[@type='text/json+oembed']")
 108                                                         as $link)
 109                                                 {
 110                                                         $href = $link->getAttributeNode('href')->nodeValue;
 111                                                         // Both Youtube and Vimeo output OEmbed endpoint URL with HTTP
 112                                                         // but their OEmbed endpoint is only accessible by HTTPS ¯\_(ツ)_/¯
 113                                                         $href = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'],
 114                                                                 ['https://www.youtube.com/', 'https://player.vimeo.com/'], $href);
 115                                                         $result = DI::httpClient()->fetchFull($href . '&maxwidth=' . $a->getThemeInfoValue('videowidth'));
 116                                                         if ($result->getReturnCode() === 200) {
 117                                                                 $json_string = $result->getBody();
 118                                                                 break;
 119                                                         }
 120                                                 }
 121                                         }
 122                                 }
 123                         }
 124
 125                         $json_string = trim($json_string);
 126
 127                         if (!$json_string || $json_string[0] != '{') {
 128                                 $json_string = '{"type":"error"}';
 129                         }
 130
 131                         $oembed->parseJSON($json_string);
 132
 133                         if (!empty($oembed->type) && $oembed->type != 'error') {
 134                                 DBA::insert('oembed', [
 135                                         'url' => Strings::normaliseLink($embedurl),
 136                                         'maxwidth' => $a->getThemeInfoValue('videowidth'),
 137                                         'content' => $json_string,
 138                                         'created' => DateTimeFormat::utcNow()
 139                                 ], Database::INSERT_UPDATE);
 140                                 $cache_ttl = Duration::DAY;
 141                         } else {
 142                                 $cache_ttl = Duration::FIVE_MINUTES;
 143                         }
 144
 145                         DI::cache()->set($cache_key, $json_string, $cache_ttl);
 146                 }
 147
 148                 // Always embed the SSL version
 149                 if (!empty($oembed->html)) {
 150                         $oembed->html = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'], ['https://www.youtube.com/', 'https://player.vimeo.com/'], $oembed->html);
 151                 }
 152
 153                 // Improve the OEmbed data with data from OpenGraph, Twitter cards and other sources
 154                 if ($use_parseurl) {
 155                         $data = ParseUrl::getSiteinfoCached($embedurl, false);
 156
 157                         if (($oembed->type == 'error') && empty($data['title']) && empty($data['text'])) {
 158                                 return $oembed;
 159                         }
 160
 161                         if ($no_rich_type || ($oembed->type == 'error')) {
 162                                 $oembed->html = '';
 163                                 $oembed->type = $data['type'];
 164
 165                                 if ($oembed->type == 'photo') {
 166                                         if (!empty($data['images'])) {
 167                                                 $oembed->url = $data['images'][0]['src'];
 168                                                 $oembed->width = $data['images'][0]['width'];
 169                                                 $oembed->height = $data['images'][0]['height'];
 170                                         } else {
 171                                                 $oembed->type = 'link';
 172                                         }
 173                                 }
 174                         }
 175
 176                         if (!empty($data['title'])) {
 177                                 $oembed->title = $data['title'];
 178                         }
 179
 180                         if (!empty($data['text'])) {
 181                                 $oembed->description = $data['text'];
 182                         }
 183
 184                         if (!empty($data['publisher_name'])) {
 185                                 $oembed->provider_name = $data['publisher_name'];
 186                         }
 187
 188                         if (!empty($data['publisher_url'])) {
 189                                 $oembed->provider_url = $data['publisher_url'];
 190                         }
 191
 192                         if (!empty($data['author_name'])) {
 193                                 $oembed->author_name = $data['author_name'];
 194                         }
 195
 196                         if (!empty($data['author_url'])) {
 197                                 $oembed->author_url = $data['author_url'];
 198                         }
 199
 200                         if (!empty($data['images']) && ($oembed->type != 'photo')) {
 201                                 $oembed->thumbnail_url = $data['images'][0]['src'];
 202                                 $oembed->thumbnail_width = $data['images'][0]['width'];
 203                                 $oembed->thumbnail_height = $data['images'][0]['height'];
 204                         }
 205                 }
 206
 207                 Hook::callAll('oembed_fetch_url', $embedurl, $oembed);
 208
 209                 return $oembed;
 210         }
 211
 212         private static function formatObject(\Friendica\Object\OEmbed $oembed)
 213         {
 214                 $ret = '<div class="oembed ' . $oembed->type . '">';
 215
 216                 switch ($oembed->type) {
 217                         case "video":
 218                                 if ($oembed->thumbnail_url) {
 219                                         $tw = (isset($oembed->thumbnail_width) && intval($oembed->thumbnail_width)) ? $oembed->thumbnail_width : 200;
 220                                         $th = (isset($oembed->thumbnail_height) && intval($oembed->thumbnail_height)) ? $oembed->thumbnail_height : 180;
 221                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio
 222                                         $tr = (($th) ? $tw / $th : 1);
 223
 224                                         $th = 120;
 225                                         $tw = $th * $tr;
 226                                         $tpl = Renderer::getMarkupTemplate('oembed_video.tpl');
 227                                         $ret .= Renderer::replaceMacros($tpl, [
 228                                                 '$embedurl' => $oembed->embed_url,
 229                                                 '$escapedhtml' => base64_encode($oembed->html),
 230                                                 '$tw' => $tw,
 231                                                 '$th' => $th,
 232                                                 '$turl' => $oembed->thumbnail_url,
 233                                         ]);
 234                                 } else {
 235                                         $ret = $oembed->html;
 236                                 }
 237                                 break;
 238
 239                         case "photo":
 240                                 $ret .= '<img width="' . $oembed->width . '" src="' . Proxy::proxifyUrl($oembed->url) . '">';
 241                                 break;
 242
 243                         case "link":
 244                                 break;
 245
 246                         case "rich":
 247                                 $ret .= Proxy::proxifyHtml($oembed->html);
 248                                 break;
 249                 }
 250
 251                 // add link to source if not present in "rich" type
 252                 if ($oembed->type != 'rich' || !strpos($oembed->html, $oembed->embed_url)) {
 253                         $ret .= '<h4>';
 254                         if (!empty($oembed->title)) {
 255                                 if (!empty($oembed->provider_name)) {
 256                                         $ret .= $oembed->provider_name . ": ";
 257                                 }
 258
 259                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 260                                 if (!empty($oembed->author_name)) {
 261                                         $ret .= ' (' . $oembed->author_name . ')';
 262                                 }
 263                         } elseif (!empty($oembed->provider_name) || !empty($oembed->author_name)) {
 264                                 $embedlink = "";
 265                                 if (!empty($oembed->provider_name)) {
 266                                         $embedlink .= $oembed->provider_name;
 267                                 }
 268
 269                                 if (!empty($oembed->author_name)) {
 270                                         if ($embedlink != "") {
 271                                                 $embedlink .= ": ";
 272                                         }
 273
 274                                         $embedlink .= $oembed->author_name;
 275                                 }
 276                                 if (trim($embedlink) == "") {
 277                                         $embedlink = $oembed->embed_url;
 278                                 }
 279
 280                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $embedlink . '</a>';
 281                         } else {
 282                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->embed_url . '</a>';
 283                         }
 284                         $ret .= "</h4>";
 285                 } elseif (!strpos($oembed->html, $oembed->embed_url)) {
 286                         // add <a> for html2bbcode conversion
 287                         $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 288                 }
 289
 290                 $ret .= '</div>';
 291
 292                 return str_replace("\n", "", $ret);
 293         }
 294
 295         public static function BBCode2HTML($text)
 296         {
 297                 $stopoembed = DI::config()->get("system", "no_oembed");
 298                 if ($stopoembed == true) {
 299                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . DI::l10n()->t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 300                 }
 301                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 302         }
 303
 304         /**
 305          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 306          * and replace it with [embed]url[/embed]
 307          *
 308          * @param $text
 309          * @return string
 310          */
 311         public static function HTML2BBCode($text)
 312         {
 313                 // start parser only if 'oembed' is in text
 314                 if (strpos($text, "oembed")) {
 315
 316                         // convert non ascii chars to html entities
 317                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 318
 319                         // If it doesn't parse at all, just return the text.
 320                         $dom = @DOMDocument::loadHTML($html_text);
 321                         if (!$dom) {
 322                                 return $text;
 323                         }
 324                         $xpath = new DOMXPath($dom);
 325
 326                         $xattr = self::buildXPath("class", "oembed");
 327                         $entries = $xpath->query("//div[$xattr]");
 328
 329                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 330                         foreach ($entries as $e) {
 331                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 332                                 if (!is_null($href)) {
 333                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 334                                 }
 335                         }
 336                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 337                 } else {
 338                         return $text;
 339                 }
 340         }
 341
 342         /**
 343          * Determines if rich content OEmbed is allowed for the provided URL
 344          *
 345          * @param string $url
 346          * @return boolean
 347          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 348          */
 349         public static function isAllowedURL($url)
 350         {
 351                 if (!DI::config()->get('system', 'no_oembed_rich_content')) {
 352                         return true;
 353                 }
 354
 355                 $domain = parse_url($url, PHP_URL_HOST);
 356                 if (empty($domain)) {
 357                         return false;
 358                 }
 359
 360                 $str_allowed = DI::config()->get('system', 'allowed_oembed', '');
 361                 if (empty($str_allowed)) {
 362                         return false;
 363                 }
 364
 365                 $allowed = explode(',', $str_allowed);
 366
 367                 return Network::isDomainAllowed($domain, $allowed);
 368         }
 369
 370         public static function getHTML($url, $title = null)
 371         {
 372                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 373
 374                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 375                         throw new Exception('OEmbed failed for URL: ' . $url);
 376                 }
 377
 378                 if (!empty($title)) {
 379                         $o->title = $title;
 380                 }
 381
 382                 $html = self::formatObject($o);
 383
 384                 return $html;
 385         }
 386
 387         /**
 388          * Generates the iframe HTML for an oembed attachment.
 389          *
 390          * Width and height are given by the remote, and are regularly too small for
 391          * the generated iframe.
 392          *
 393          * The width is entirely discarded for the actual width of the post, while fixed
 394          * height is used as a starting point before the inevitable resizing.
 395          *
 396          * Since the iframe is automatically resized on load, there are no need for ugly
 397          * and impractical scrollbars.
 398          *
 399          * @todo  This function is currently unused until someone™ adds support for a separate OEmbed domain
 400          *
 401          * @param string $src Original remote URL to embed
 402          * @param string $width
 403          * @param string $height
 404          * @return string formatted HTML
 405          *
 406          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 407          * @see   oembed_format_object()
 408          */
 409         private static function iframe($src, $width, $height)
 410         {
 411                 if (!$height || strstr($height, '%')) {
 412                         $height = '200';
 413                 }
 414                 $width = '100%';
 415
 416                 $src = DI::baseUrl() . '/oembed/' . Strings::base64UrlEncode($src);
 417                 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>';
 418         }
 419
 420         /**
 421          * Generates attribute search XPath string
 422          *
 423          * Generates an XPath query to select elements whose provided attribute contains
 424          * the provided value in a space-separated list.
 425          *
 426          * @param string $attr Name of the attribute to seach
 427          * @param string $value Value to search in a space-separated list
 428          * @return string
 429          */
 430         private static function buildXPath($attr, $value)
 431         {
 432                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 433                 return "contains(normalize-space(@$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'";
 434         }
 435
 436         /**
 437          * Returns the inner XML string of a provided DOMNode
 438          *
 439          * @param DOMNode $node
 440          * @return string
 441          */
 442         private static function getInnerHTML(DOMNode $node)
 443         {
 444                 $innerHTML = '';
 445                 $children = $node->childNodes;
 446                 foreach ($children as $child) {
 447                         $innerHTML .= $child->ownerDocument->saveXML($child);
 448                 }
 449                 return $innerHTML;
 450         }
 451
 452 }