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