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