src/Content/OEmbed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2023, 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         /**
  53          * Callback for fetching URL, checking allowance and returning formatted HTML
  54          *
  55          * @param array $matches
  56          * @return string Formatted HTML
  57          */
  58         public static function replaceCallback(array $matches): string
  59         {
  60                 $embedurl = $matches[1];
  61                 $j = self::fetchURL($embedurl, !self::isAllowedURL($embedurl));
  62                 $s = self::formatObject($j);
  63
  64                 return $s;
  65         }
  66
  67         /**
  68          * Get data from an URL to embed its content.
  69          *
  70          * @param string $embedurl     The URL from which the data should be fetched.
  71          * @param bool   $no_rich_type If set to true rich type content won't be fetched.
  72          * @param bool   $use_parseurl Use the "ParseUrl" functionality to add additional data
  73          *
  74          * @return \Friendica\Object\OEmbed
  75          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  76          */
  77         public static function fetchURL(string $embedurl, bool $no_rich_type = false, bool $use_parseurl = true): \Friendica\Object\OEmbed
  78         {
  79                 $embedurl = trim($embedurl, '\'"');
  80
  81                 $a = DI::app();
  82
  83                 $cache_key = 'oembed:' . $a->getThemeInfoValue('videowidth') . ':' . $embedurl;
  84
  85                 $condition = ['url' => Strings::normaliseLink($embedurl), 'maxwidth' => $a->getThemeInfoValue('videowidth')];
  86                 $oembed_record = DBA::selectFirst('oembed', ['content'], $condition);
  87                 if (DBA::isResult($oembed_record)) {
  88                         $json_string = $oembed_record['content'];
  89                 } else {
  90                         $json_string = DI::cache()->get($cache_key);
  91                 }
  92
  93                 // These media files should now be caught in bbcode.php
  94                 // left here as a fallback in case this is called from another source
  95                 $noexts = ['mp3', 'mp4', 'ogg', 'ogv', 'oga', 'ogm', 'webm'];
  96                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  97
  98                 $oembed = new \Friendica\Object\OEmbed($embedurl);
  99
 100                 if ($json_string) {
 101                         $oembed->parseJSON($json_string);
 102                 } else {
 103                         $json_string = '';
 104
 105                         if (!in_array($ext, $noexts)) {
 106                                 // try oembed autodiscovery
 107                                 $html_text = DI::httpClient()->fetch($embedurl, HttpClientAccept::HTML, 15);
 108                                 if (!empty($html_text)) {
 109                                         $dom = new DOMDocument();
 110                                         if (@$dom->loadHTML($html_text)) {
 111                                                 $xpath = new DOMXPath($dom);
 112                                                 foreach (
 113                                                         $xpath->query("//link[@type='application/json+oembed'] | //link[@type='text/json+oembed']")
 114                                                         as $link)
 115                                                 {
 116                                                         $href = $link->getAttributeNode('href')->nodeValue;
 117                                                         // Both Youtube and Vimeo output OEmbed endpoint URL with HTTP
 118                                                         // but their OEmbed endpoint is only accessible by HTTPS ¯\_(ツ)_/¯
 119                                                         $href = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'],
 120                                                                 ['https://www.youtube.com/', 'https://player.vimeo.com/'], $href);
 121                                                         $result = DI::httpClient()->fetchFull($href . '&maxwidth=' . $a->getThemeInfoValue('videowidth'));
 122                                                         if ($result->getReturnCode() === 200) {
 123                                                                 $json_string = $result->getBody();
 124                                                                 break;
 125                                                         }
 126                                                 }
 127                                         }
 128                                 }
 129                         }
 130
 131                         $json_string = trim($json_string);
 132
 133                         if (!$json_string || $json_string[0] != '{') {
 134                                 $json_string = '{"type":"error"}';
 135                         }
 136
 137                         $oembed->parseJSON($json_string);
 138
 139                         if (!empty($oembed->type) && $oembed->type != 'error') {
 140                                 DBA::insert('oembed', [
 141                                         'url' => Strings::normaliseLink($embedurl),
 142                                         'maxwidth' => $a->getThemeInfoValue('videowidth'),
 143                                         'content' => $json_string,
 144                                         'created' => DateTimeFormat::utcNow()
 145                                 ], Database::INSERT_UPDATE);
 146                                 $cache_ttl = Duration::DAY;
 147                         } else {
 148                                 $cache_ttl = Duration::FIVE_MINUTES;
 149                         }
 150
 151                         DI::cache()->set($cache_key, $json_string, $cache_ttl);
 152                 }
 153
 154                 // Always embed the SSL version
 155                 if (!empty($oembed->html)) {
 156                         $oembed->html = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'], ['https://www.youtube.com/', 'https://player.vimeo.com/'], $oembed->html);
 157                 }
 158
 159                 // Improve the OEmbed data with data from OpenGraph, Twitter cards and other sources
 160                 if ($use_parseurl) {
 161                         $data = ParseUrl::getSiteinfoCached($embedurl, false);
 162
 163                         if (($oembed->type == 'error') && empty($data['title']) && empty($data['text'])) {
 164                                 return $oembed;
 165                         }
 166
 167                         if ($no_rich_type || ($oembed->type == 'error')) {
 168                                 $oembed->html = '';
 169                                 $oembed->type = $data['type'];
 170
 171                                 if ($oembed->type == 'photo') {
 172                                         if (!empty($data['images'])) {
 173                                                 $oembed->url = $data['images'][0]['src'];
 174                                                 $oembed->width = $data['images'][0]['width'];
 175                                                 $oembed->height = $data['images'][0]['height'];
 176                                         } else {
 177                                                 $oembed->type = 'link';
 178                                         }
 179                                 }
 180                         }
 181
 182                         if (!empty($data['title'])) {
 183                                 $oembed->title = $data['title'];
 184                         }
 185
 186                         if (!empty($data['text'])) {
 187                                 $oembed->description = $data['text'];
 188                         }
 189
 190                         if (!empty($data['publisher_name'])) {
 191                                 $oembed->provider_name = $data['publisher_name'];
 192                         }
 193
 194                         if (!empty($data['publisher_url'])) {
 195                                 $oembed->provider_url = $data['publisher_url'];
 196                         }
 197
 198                         if (!empty($data['author_name'])) {
 199                                 $oembed->author_name = $data['author_name'];
 200                         }
 201
 202                         if (!empty($data['author_url'])) {
 203                                 $oembed->author_url = $data['author_url'];
 204                         }
 205
 206                         if (!empty($data['images']) && ($oembed->type != 'photo')) {
 207                                 $oembed->thumbnail_url = $data['images'][0]['src'];
 208                                 $oembed->thumbnail_width = $data['images'][0]['width'];
 209                                 $oembed->thumbnail_height = $data['images'][0]['height'];
 210                         }
 211                 }
 212
 213                 Hook::callAll('oembed_fetch_url', $embedurl, $oembed);
 214
 215                 return $oembed;
 216         }
 217
 218         /**
 219          * Returns a formatted string from OEmbed object
 220          *
 221          * @param \Friendica\Object\OEmbed $oembed
 222          * @return string
 223          */
 224         private static function formatObject(\Friendica\Object\OEmbed $oembed): string
 225         {
 226                 $ret = '<div class="oembed ' . $oembed->type . '">';
 227
 228                 switch ($oembed->type) {
 229                         case 'video':
 230                                 if ($oembed->thumbnail_url) {
 231                                         $tw = (isset($oembed->thumbnail_width) && intval($oembed->thumbnail_width)) ? $oembed->thumbnail_width : 200;
 232                                         $th = (isset($oembed->thumbnail_height) && intval($oembed->thumbnail_height)) ? $oembed->thumbnail_height : 180;
 233                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio
 234                                         $tr = (($th) ? $tw / $th : 1);
 235
 236                                         $th = 120;
 237                                         $tw = $th * $tr;
 238                                         $tpl = Renderer::getMarkupTemplate('oembed_video.tpl');
 239                                         $ret .= Renderer::replaceMacros($tpl, [
 240                                                 '$embedurl' => $oembed->embed_url,
 241                                                 '$escapedhtml' => base64_encode($oembed->html),
 242                                                 '$tw' => $tw,
 243                                                 '$th' => $th,
 244                                                 '$turl' => $oembed->thumbnail_url,
 245                                         ]);
 246                                 } else {
 247                                         $ret = $oembed->html;
 248                                 }
 249                                 break;
 250
 251                         case 'photo':
 252                                 $ret .= '<img width="' . $oembed->width . '" src="' . Proxy::proxifyUrl($oembed->url) . '">';
 253                                 break;
 254
 255                         case 'link':
 256                                 break;
 257
 258                         case 'rich':
 259                                 $ret .= Proxy::proxifyHtml($oembed->html);
 260                                 break;
 261                 }
 262
 263                 // add link to source if not present in "rich" type
 264                 if ($oembed->type != 'rich' || !strpos($oembed->html, $oembed->embed_url)) {
 265                         $ret .= '<h4>';
 266                         if (!empty($oembed->title)) {
 267                                 if (!empty($oembed->provider_name)) {
 268                                         $ret .= $oembed->provider_name . ": ";
 269                                 }
 270
 271                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 272                                 if (!empty($oembed->author_name)) {
 273                                         $ret .= ' (' . $oembed->author_name . ')';
 274                                 }
 275                         } elseif (!empty($oembed->provider_name) || !empty($oembed->author_name)) {
 276                                 $embedlink = "";
 277                                 if (!empty($oembed->provider_name)) {
 278                                         $embedlink .= $oembed->provider_name;
 279                                 }
 280
 281                                 if (!empty($oembed->author_name)) {
 282                                         if ($embedlink != "") {
 283                                                 $embedlink .= ": ";
 284                                         }
 285
 286                                         $embedlink .= $oembed->author_name;
 287                                 }
 288                                 if (trim($embedlink) == "") {
 289                                         $embedlink = $oembed->embed_url;
 290                                 }
 291
 292                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $embedlink . '</a>';
 293                         } else {
 294                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->embed_url . '</a>';
 295                         }
 296                         $ret .= "</h4>";
 297                 } elseif (!strpos($oembed->html, $oembed->embed_url)) {
 298                         // add <a> for html2bbcode conversion
 299                         $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 300                 }
 301
 302                 $ret .= '</div>';
 303
 304                 return str_replace("\n", "", $ret);
 305         }
 306
 307         /**
 308          * Converts BBCode to HTML code
 309          *
 310          * @param string $text
 311          * @return string
 312          */
 313         public static function BBCode2HTML(string $text): string
 314         {
 315                 $stopoembed = DI::config()->get('system', 'no_oembed');
 316                 if ($stopoembed == true) {
 317                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . DI::l10n()->t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 318                 }
 319                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", [self::class, 'replaceCallback'], $text);
 320         }
 321
 322         /**
 323          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 324          * and replace it with [embed]url[/embed]
 325          *
 326          * @param string $text
 327          * @return string
 328          */
 329         public static function HTML2BBCode(string $text): string
 330         {
 331                 // start parser only if 'oembed' is in text
 332                 if (strpos($text, 'oembed')) {
 333                         // convert non ascii chars to html entities
 334                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 335
 336                         // If it doesn't parse at all, just return the text.
 337                         $dom = @DOMDocument::loadHTML($html_text);
 338                         if (!$dom) {
 339                                 return $text;
 340                         }
 341                         $xpath = new DOMXPath($dom);
 342
 343                         $xattr = self::buildXPath('class', 'oembed');
 344                         $entries = $xpath->query("//div[$xattr]");
 345
 346                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 347                         foreach ($entries as $e) {
 348                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 349                                 if (!is_null($href)) {
 350                                         $e->parentNode->replaceChild(new DOMText('[embed]' . $href . '[/embed]'), $e);
 351                                 }
 352                         }
 353                         return self::getInnerHTML($dom->getElementsByTagName('body')->item(0));
 354                 } else {
 355                         return $text;
 356                 }
 357         }
 358
 359         /**
 360          * Determines if rich content OEmbed is allowed for the provided URL
 361          *
 362          * @param string $url
 363          * @return boolean
 364          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 365          */
 366         public static function isAllowedURL(string $url): bool
 367         {
 368                 if (!DI::config()->get('system', 'no_oembed_rich_content')) {
 369                         return true;
 370                 }
 371
 372                 $domain = parse_url($url, PHP_URL_HOST);
 373                 if (empty($domain)) {
 374                         return false;
 375                 }
 376
 377                 $str_allowed = DI::config()->get('system', 'allowed_oembed', '');
 378                 if (empty($str_allowed)) {
 379                         return false;
 380                 }
 381
 382                 $allowed = explode(',', $str_allowed);
 383
 384                 return Network::isDomainAllowed($domain, $allowed);
 385         }
 386
 387         /**
 388          * Returns a formatted HTML code from given URL and sets optional title
 389          *
 390          * @param string $url URL to fetch
 391          * @param string $title Optional title (default: what comes from OEmbed object)
 392          * @return string Formatted HTML
 393          */
 394         public static function getHTML(string $url, string $title = ''): string
 395         {
 396                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 397
 398                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 399                         throw new Exception('OEmbed failed for URL: ' . $url);
 400                 }
 401
 402                 if (!empty($title)) {
 403                         $o->title = $title;
 404                 }
 405
 406                 $html = self::formatObject($o);
 407
 408                 return $html;
 409         }
 410
 411         /**
 412          * Generates the iframe HTML for an oembed attachment.
 413          *
 414          * Width and height are given by the remote, and are regularly too small for
 415          * the generated iframe.
 416          *
 417          * The width is entirely discarded for the actual width of the post, while fixed
 418          * height is used as a starting point before the inevitable resizing.
 419          *
 420          * Since the iframe is automatically resized on load, there are no need for ugly
 421          * and impractical scrollbars.
 422          *
 423          * @todo  This function is currently unused until someone™ adds support for a separate OEmbed domain
 424          *
 425          * @param string $src Original remote URL to embed
 426          * @param string $width
 427          * @param string $height
 428          * @return string Formatted HTML
 429          *
 430          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 431          * @see   oembed_format_object()
 432          */
 433         private static function iframe(string $src, string $width, string $height): string
 434         {
 435                 if (!$height || strstr($height, '%')) {
 436                         $height = '200';
 437                 }
 438                 $width = '100%';
 439
 440                 $src = DI::baseUrl() . '/oembed/' . Strings::base64UrlEncode($src);
 441                 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>';
 442         }
 443
 444         /**
 445          * Generates attribute search XPath string
 446          *
 447          * Generates an XPath query to select elements whose provided attribute contains
 448          * the provided value in a space-separated list.
 449          *
 450          * @param string $attr Name of the attribute to seach
 451          * @param string $value Value to search in a space-separated list
 452          * @return string
 453          */
 454         private static function buildXPath(string $attr, $value): string
 455         {
 456                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 457                 return "contains(normalize-space(@$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'";
 458         }
 459
 460         /**
 461          * Returns the inner XML string of a provided DOMNode
 462          *
 463          * @param DOMNode $node
 464          * @return string
 465          */
 466         private static function getInnerHTML(DOMNode $node): string
 467         {
 468                 $innerHTML = '';
 469                 $children = $node->childNodes;
 470                 foreach ($children as $child) {
 471                         $innerHTML .= $child->ownerDocument->saveXML($child);
 472                 }
 473                 return $innerHTML;
 474         }
 475
 476 }