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;
  14 use Friendica\Core\Config;
  15 use Friendica\Core\Hook;
  16 use Friendica\Core\L10n;
  17 use Friendica\Core\Renderer;
  18 use Friendica\Core\System;
  19 use Friendica\Database\DBA;
  20 use Friendica\Util\DateTimeFormat;
  21 use Friendica\Util\Network;
  22 use Friendica\Util\ParseUrl;
  23 use Friendica\Util\Proxy as ProxyUtils;
  24 use Friendica\Util\Strings;
  25
  26 /**
  27  * Handles all OEmbed content fetching and replacement
  28  *
  29  * OEmbed is a standard used to allow an embedded representation of a URL on
  30  * third party sites
  31  *
  32  * @see https://oembed.com
  33  *
  34  * @author Hypolite Petovan <hypolite@mrpetovan.com>
  35  */
  36 class OEmbed
  37 {
  38         public static function replaceCallback($matches)
  39         {
  40                 $embedurl = $matches[1];
  41                 $j = self::fetchURL($embedurl, !self::isAllowedURL($embedurl));
  42                 $s = self::formatObject($j);
  43
  44                 return $s;
  45         }
  46
  47         /**
  48          * @brief Get data from an URL to embed its content.
  49          *
  50          * @param string $embedurl     The URL from which the data should be fetched.
  51          * @param bool   $no_rich_type If set to true rich type content won't be fetched.
  52          *
  53          * @return \Friendica\Object\OEmbed
  54          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  55          */
  56         public static function fetchURL($embedurl, $no_rich_type = false)
  57         {
  58                 $embedurl = trim($embedurl, '\'"');
  59
  60                 $a = \get_app();
  61
  62                 $cache_key = 'oembed:' . $a->videowidth . ':' . $embedurl;
  63
  64                 $condition = ['url' => Strings::normaliseLink($embedurl), 'maxwidth' => $a->videowidth];
  65                 $oembed_record = DBA::selectFirst('oembed', ['content'], $condition);
  66                 if (DBA::isResult($oembed_record)) {
  67                         $json_string = $oembed_record['content'];
  68                 } else {
  69                         $json_string = Cache::get($cache_key);
  70                 }
  71
  72                 // These media files should now be caught in bbcode.php
  73                 // left here as a fallback in case this is called from another source
  74                 $noexts = ['mp3', 'mp4', 'ogg', 'ogv', 'oga', 'ogm', 'webm'];
  75                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  76
  77                 $oembed = new \Friendica\Object\OEmbed($embedurl);
  78
  79                 if ($json_string) {
  80                         $oembed->parseJSON($json_string);
  81                 } else {
  82                         $json_string = '';
  83
  84                         if (!in_array($ext, $noexts)) {
  85                                 // try oembed autodiscovery
  86                                 $redirects = 0;
  87                                 $html_text = Network::fetchUrl($embedurl, false, $redirects, 15, 'text/*');
  88                                 if ($html_text) {
  89                                         $dom = @DOMDocument::loadHTML($html_text);
  90                                         if ($dom) {
  91                                                 $xpath = new DOMXPath($dom);
  92                                                 $entries = $xpath->query("//link[@type='application/json+oembed']");
  93                                                 foreach ($entries as $e) {
  94                                                         $href = $e->getAttributeNode('href')->nodeValue;
  95                                                         $json_string = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
  96                                                         break;
  97                                                 }
  98
  99                                                 $entries = $xpath->query("//link[@type='text/json+oembed']");
 100                                                 foreach ($entries as $e) {
 101                                                         $href = $e->getAttributeNode('href')->nodeValue;
 102                                                         $json_string = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
 103                                                         break;
 104                                                 }
 105                                         }
 106                                 }
 107                         }
 108
 109                         $json_string = trim($json_string);
 110
 111                         if (!$json_string || $json_string[0] != '{') {
 112                                 $json_string = '{"type":"error"}';
 113                         }
 114
 115                         $oembed->parseJSON($json_string);
 116
 117                         if (!empty($oembed->type) && $oembed->type != 'error') {
 118                                 DBA::insert('oembed', [
 119                                         'url' => Strings::normaliseLink($embedurl),
 120                                         'maxwidth' => $a->videowidth,
 121                                         'content' => $json_string,
 122                                         'created' => DateTimeFormat::utcNow()
 123                                 ], true);
 124                                 $cache_ttl = Cache::DAY;
 125                         } else {
 126                                 $cache_ttl = Cache::FIVE_MINUTES;
 127                         }
 128
 129                         Cache::set($cache_key, $json_string, $cache_ttl);
 130                 }
 131
 132                 if ($oembed->type == 'error') {
 133                         return $oembed;
 134                 }
 135
 136                 // Always embed the SSL version
 137                 $oembed->html = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'], ['https://www.youtube.com/', 'https://player.vimeo.com/'], $oembed->html);
 138
 139                 // If fetching information doesn't work, then improve via internal functions
 140                 if ($no_rich_type && ($oembed->type == 'rich')) {
 141                         $data = ParseUrl::getSiteinfoCached($embedurl, true, false);
 142                         $oembed->type = $data['type'];
 143
 144                         if ($oembed->type == 'photo') {
 145                                 $oembed->url = $data['url'];
 146                         }
 147
 148                         if (isset($data['title'])) {
 149                                 $oembed->title = $data['title'];
 150                         }
 151
 152                         if (isset($data['text'])) {
 153                                 $oembed->description = $data['text'];
 154                         }
 155
 156                         if (!empty($data['images'])) {
 157                                 $oembed->thumbnail_url = $data['images'][0]['src'];
 158                                 $oembed->thumbnail_width = $data['images'][0]['width'];
 159                                 $oembed->thumbnail_height = $data['images'][0]['height'];
 160                         }
 161                 }
 162
 163                 Hook::callAll('oembed_fetch_url', $embedurl, $oembed);
 164
 165                 return $oembed;
 166         }
 167
 168         private static function formatObject(\Friendica\Object\OEmbed $oembed)
 169         {
 170                 $ret = '<div class="oembed ' . $oembed->type . '">';
 171
 172                 switch ($oembed->type) {
 173                         case "video":
 174                                 if ($oembed->thumbnail_url) {
 175                                         $tw = (isset($oembed->thumbnail_width) && intval($oembed->thumbnail_width)) ? $oembed->thumbnail_width : 200;
 176                                         $th = (isset($oembed->thumbnail_height) && intval($oembed->thumbnail_height)) ? $oembed->thumbnail_height : 180;
 177                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio
 178                                         $tr = (($th) ? $tw / $th : 1);
 179
 180                                         $th = 120;
 181                                         $tw = $th * $tr;
 182                                         $tpl = Renderer::getMarkupTemplate('oembed_video.tpl');
 183                                         $ret .= Renderer::replaceMacros($tpl, [
 184                                                 '$baseurl' => System::baseUrl(),
 185                                                 '$embedurl' => $oembed->embed_url,
 186                                                 '$escapedhtml' => base64_encode($oembed->html),
 187                                                 '$tw' => $tw,
 188                                                 '$th' => $th,
 189                                                 '$turl' => $oembed->thumbnail_url,
 190                                         ]);
 191                                 } else {
 192                                         $ret = $oembed->html;
 193                                 }
 194                                 break;
 195
 196                         case "photo":
 197                                 $ret .= '<img width="' . $oembed->width . '" src="' . ProxyUtils::proxifyUrl($oembed->url) . '">';
 198                                 break;
 199
 200                         case "link":
 201                                 break;
 202
 203                         case "rich":
 204                                 $ret .= ProxyUtils::proxifyHtml($oembed->html);
 205                                 break;
 206                 }
 207
 208                 // add link to source if not present in "rich" type
 209                 if ($oembed->type != 'rich' || !strpos($oembed->html, $oembed->embed_url)) {
 210                         $ret .= '<h4>';
 211                         if (!empty($oembed->title)) {
 212                                 if (!empty($oembed->provider_name)) {
 213                                         $ret .= $oembed->provider_name . ": ";
 214                                 }
 215
 216                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 217                                 if (!empty($oembed->author_name)) {
 218                                         $ret .= ' (' . $oembed->author_name . ')';
 219                                 }
 220                         } elseif (!empty($oembed->provider_name) || !empty($oembed->author_name)) {
 221                                 $embedlink = "";
 222                                 if (!empty($oembed->provider_name)) {
 223                                         $embedlink .= $oembed->provider_name;
 224                                 }
 225
 226                                 if (!empty($oembed->author_name)) {
 227                                         if ($embedlink != "") {
 228                                                 $embedlink .= ": ";
 229                                         }
 230
 231                                         $embedlink .= $oembed->author_name;
 232                                 }
 233                                 if (trim($embedlink) == "") {
 234                                         $embedlink = $oembed->embed_url;
 235                                 }
 236
 237                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $embedlink . '</a>';
 238                         } else {
 239                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->embed_url . '</a>';
 240                         }
 241                         $ret .= "</h4>";
 242                 } elseif (!strpos($oembed->html, $oembed->embed_url)) {
 243                         // add <a> for html2bbcode conversion
 244                         $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 245                 }
 246
 247                 $ret .= '</div>';
 248
 249                 return str_replace("\n", "", $ret);
 250         }
 251
 252         public static function BBCode2HTML($text)
 253         {
 254                 $stopoembed = Config::get("system", "no_oembed");
 255                 if ($stopoembed == true) {
 256                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . L10n::t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 257                 }
 258                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 259         }
 260
 261         /**
 262          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 263          * and replace it with [embed]url[/embed]
 264          *
 265          * @param $text
 266          * @return string
 267          */
 268         public static function HTML2BBCode($text)
 269         {
 270                 // start parser only if 'oembed' is in text
 271                 if (strpos($text, "oembed")) {
 272
 273                         // convert non ascii chars to html entities
 274                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 275
 276                         // If it doesn't parse at all, just return the text.
 277                         $dom = @DOMDocument::loadHTML($html_text);
 278                         if (!$dom) {
 279                                 return $text;
 280                         }
 281                         $xpath = new DOMXPath($dom);
 282
 283                         $xattr = self::buildXPath("class", "oembed");
 284                         $entries = $xpath->query("//div[$xattr]");
 285
 286                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 287                         foreach ($entries as $e) {
 288                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 289                                 if (!is_null($href)) {
 290                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 291                                 }
 292                         }
 293                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 294                 } else {
 295                         return $text;
 296                 }
 297         }
 298
 299         /**
 300          * Determines if rich content OEmbed is allowed for the provided URL
 301          *
 302          * @brief Determines if rich content OEmbed is allowed for the provided URL
 303          * @param string $url
 304          * @return boolean
 305          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 306          */
 307         public static function isAllowedURL($url)
 308         {
 309                 if (!Config::get('system', 'no_oembed_rich_content')) {
 310                         return true;
 311                 }
 312
 313                 $domain = parse_url($url, PHP_URL_HOST);
 314                 if (empty($domain)) {
 315                         return false;
 316                 }
 317
 318                 $str_allowed = Config::get('system', 'allowed_oembed', '');
 319                 if (empty($str_allowed)) {
 320                         return false;
 321                 }
 322
 323                 $allowed = explode(',', $str_allowed);
 324
 325                 return Network::isDomainAllowed($domain, $allowed);
 326         }
 327
 328         public static function getHTML($url, $title = null)
 329         {
 330                 // Always embed the SSL version
 331                 $url = str_replace(["http://www.youtube.com/", "http://player.vimeo.com/"],
 332                                         ["https://www.youtube.com/", "https://player.vimeo.com/"], $url);
 333
 334                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 335
 336                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 337                         throw new Exception('OEmbed failed for URL: ' . $url);
 338                 }
 339
 340                 if (!empty($title)) {
 341                         $o->title = $title;
 342                 }
 343
 344                 $html = self::formatObject($o);
 345
 346                 return $html;
 347         }
 348
 349         /**
 350          * @brief Generates the iframe HTML for an oembed attachment.
 351          *
 352          * Width and height are given by the remote, and are regularly too small for
 353          * the generated iframe.
 354          *
 355          * The width is entirely discarded for the actual width of the post, while fixed
 356          * height is used as a starting point before the inevitable resizing.
 357          *
 358          * Since the iframe is automatically resized on load, there are no need for ugly
 359          * and impractical scrollbars.
 360          *
 361          * @todo  This function is currently unused until someone™ adds support for a separate OEmbed domain
 362          *
 363          * @param string $src Original remote URL to embed
 364          * @param string $width
 365          * @param string $height
 366          * @return string formatted HTML
 367          *
 368          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 369          * @see   oembed_format_object()
 370          */
 371         private static function iframe($src, $width, $height)
 372         {
 373                 $a = \get_app();
 374
 375                 if (!$height || strstr($height, '%')) {
 376                         $height = '200';
 377                 }
 378                 $width = '100%';
 379
 380                 $src = System::baseUrl() . '/oembed/' . Strings::base64UrlEncode($src);
 381                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . L10n::t('Embedded content') . '</iframe>';
 382         }
 383
 384         /**
 385          * Generates an XPath query to select elements whose provided attribute contains
 386          * the provided value in a space-separated list.
 387          *
 388          * @brief Generates attribute search XPath string
 389          *
 390          * @param string $attr Name of the attribute to seach
 391          * @param string $value Value to search in a space-separated list
 392          * @return string
 393          */
 394         private static function buildXPath($attr, $value)
 395         {
 396                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 397                 return "contains(normalize-space(@$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'";
 398         }
 399
 400         /**
 401          * Returns the inner XML string of a provided DOMNode
 402          *
 403          * @brief Returns the inner XML string of a provided DOMNode
 404          *
 405          * @param DOMNode $node
 406          * @return string
 407          */
 408         private static function getInnerHTML(DOMNode $node)
 409         {
 410                 $innerHTML = '';
 411                 $children = $node->childNodes;
 412                 foreach ($children as $child) {
 413                         $innerHTML .= $child->ownerDocument->saveXML($child);
 414                 }
 415                 return $innerHTML;
 416         }
 417
 418 }