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\System;
  18 use Friendica\Database\DBA;
  19 use Friendica\Util\DateTimeFormat;
  20 use Friendica\Util\Network;
  21 use Friendica\Util\ParseUrl;
  22 use Friendica\Util\Proxy as ProxyUtils;
  23
  24 require_once 'include/dba.php';
  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   $no_rich_type If set to true rich type content won't be fetched.
  52          *
  53          * @return \Friendica\Object\OEmbed
  54          */
  55         public static function fetchURL($embedurl, $no_rich_type = false)
  56         {
  57                 $embedurl = trim($embedurl, '\'"');
  58
  59                 $a = get_app();
  60
  61                 $cache_key = 'oembed:' . $a->videowidth . ':' . $embedurl;
  62
  63                 $condition = ['url' => normalise_link($embedurl), 'maxwidth' => $a->videowidth];
  64                 $oembed_record = DBA::selectFirst('oembed', ['content'], $condition);
  65                 if (DBA::isResult($oembed_record)) {
  66                         $json_string = $oembed_record['content'];
  67                 } else {
  68                         $json_string = Cache::get($cache_key);
  69                 }
  70
  71                 // These media files should now be caught in bbcode.php
  72                 // left here as a fallback in case this is called from another source
  73                 $noexts = ['mp3', 'mp4', 'ogg', 'ogv', 'oga', 'ogm', 'webm'];
  74                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  75
  76                 $oembed = new \Friendica\Object\OEmbed($embedurl);
  77
  78                 if ($json_string) {
  79                         $oembed->parseJSON($json_string);
  80                 } else {
  81                         $json_string = '';
  82
  83                         if (!in_array($ext, $noexts)) {
  84                                 // try oembed autodiscovery
  85                                 $redirects = 0;
  86                                 $html_text = Network::fetchUrl($embedurl, false, $redirects, 15, 'text/*');
  87                                 if ($html_text) {
  88                                         $dom = @DOMDocument::loadHTML($html_text);
  89                                         if ($dom) {
  90                                                 $xpath = new DOMXPath($dom);
  91                                                 $entries = $xpath->query("//link[@type='application/json+oembed']");
  92                                                 foreach ($entries as $e) {
  93                                                         $href = $e->getAttributeNode('href')->nodeValue;
  94                                                         $json_string = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
  95                                                         break;
  96                                                 }
  97
  98                                                 $entries = $xpath->query("//link[@type='text/json+oembed']");
  99                                                 foreach ($entries as $e) {
 100                                                         $href = $e->getAttributeNode('href')->nodeValue;
 101                                                         $json_string = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
 102                                                         break;
 103                                                 }
 104                                         }
 105                                 }
 106                         }
 107
 108                         $json_string = trim($json_string);
 109
 110                         if (!$json_string || $json_string[0] != '{') {
 111                                 $json_string = '{"type":"error"}';
 112                         }
 113
 114                         $oembed->parseJSON($json_string);
 115
 116                         if (!empty($oembed->type) && $oembed->type != 'error') {
 117                                 DBA::insert('oembed', [
 118                                         'url' => normalise_link($embedurl),
 119                                         'maxwidth' => $a->videowidth,
 120                                         'content' => $json_string,
 121                                         'created' => DateTimeFormat::utcNow()
 122                                 ], true);
 123                                 $cache_ttl = CACHE_DAY;
 124                         } else {
 125                                 $cache_ttl = CACHE_FIVE_MINUTES;
 126                         }
 127
 128                         Cache::set($cache_key, $json_string, $cache_ttl);
 129                 }
 130
 131                 if ($oembed->type == 'error') {
 132                         return $oembed;
 133                 }
 134
 135                 // Always embed the SSL version
 136                 $oembed->html = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'], ['https://www.youtube.com/', 'https://player.vimeo.com/'], $oembed->html);
 137
 138                 // If fetching information doesn't work, then improve via internal functions
 139                 if ($no_rich_type && ($oembed->type == 'rich')) {
 140                         $data = ParseUrl::getSiteinfoCached($embedurl, true, false);
 141                         $oembed->type = $data['type'];
 142
 143                         if ($oembed->type == 'photo') {
 144                                 $oembed->url = $data['url'];
 145                         }
 146
 147                         if (isset($data['title'])) {
 148                                 $oembed->title = $data['title'];
 149                         }
 150
 151                         if (isset($data['text'])) {
 152                                 $oembed->description = $data['text'];
 153                         }
 154
 155                         if (!empty($data['images'])) {
 156                                 $oembed->thumbnail_url = $data['images'][0]['src'];
 157                                 $oembed->thumbnail_width = $data['images'][0]['width'];
 158                                 $oembed->thumbnail_height = $data['images'][0]['height'];
 159                         }
 160                 }
 161
 162                 Addon::callHooks('oembed_fetch_url', $embedurl, $oembed);
 163
 164                 return $oembed;
 165         }
 166
 167         private static function formatObject(\Friendica\Object\OEmbed $oembed)
 168         {
 169                 $ret = '<div class="oembed ' . $oembed->type . '">';
 170
 171                 switch ($oembed->type) {
 172                         case "video":
 173                                 if ($oembed->thumbnail_url) {
 174                                         $tw = (isset($oembed->thumbnail_width) && intval($oembed->thumbnail_width)) ? $oembed->thumbnail_width : 200;
 175                                         $th = (isset($oembed->thumbnail_height) && intval($oembed->thumbnail_height)) ? $oembed->thumbnail_height : 180;
 176                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio
 177                                         $tr = (($th) ? $tw / $th : 1);
 178
 179                                         $th = 120;
 180                                         $tw = $th * $tr;
 181                                         $tpl = get_markup_template('oembed_video.tpl');
 182                                         $ret .= replace_macros($tpl, [
 183                                                 '$baseurl' => System::baseUrl(),
 184                                                 '$embedurl' => $oembed->embed_url,
 185                                                 '$escapedhtml' => base64_encode($oembed->html),
 186                                                 '$tw' => $tw,
 187                                                 '$th' => $th,
 188                                                 '$turl' => $oembed->thumbnail_url,
 189                                         ]);
 190                                 } else {
 191                                         $ret = $oembed->html;
 192                                 }
 193                                 break;
 194
 195                         case "photo":
 196                                 $ret .= '<img width="' . $oembed->width . '" src="' . ProxyUtils::proxifyUrl($oembed->url) . '">';
 197                                 break;
 198
 199                         case "link":
 200                                 break;
 201
 202                         case "rich":
 203                                 $ret .= ProxyUtils::proxifyHtml($oembed->html);
 204                                 break;
 205                 }
 206
 207                 // add link to source if not present in "rich" type
 208                 if ($oembed->type != 'rich' || !strpos($oembed->html, $oembed->embed_url)) {
 209                         $ret .= '<h4>';
 210                         if (!empty($oembed->title)) {
 211                                 if (!empty($oembed->provider_name)) {
 212                                         $ret .= $oembed->provider_name . ": ";
 213                                 }
 214
 215                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 216                                 if (!empty($oembed->author_name)) {
 217                                         $ret .= ' (' . $oembed->author_name . ')';
 218                                 }
 219                         } elseif (!empty($oembed->provider_name) || !empty($oembed->author_name)) {
 220                                 $embedlink = "";
 221                                 if (!empty($oembed->provider_name)) {
 222                                         $embedlink .= $oembed->provider_name;
 223                                 }
 224
 225                                 if (!empty($oembed->author_name)) {
 226                                         if ($embedlink != "") {
 227                                                 $embedlink .= ": ";
 228                                         }
 229
 230                                         $embedlink .= $oembed->author_name;
 231                                 }
 232                                 if (trim($embedlink) == "") {
 233                                         $embedlink = $oembed->embed_url;
 234                                 }
 235
 236                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $embedlink . '</a>';
 237                         } else {
 238                                 $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->embed_url . '</a>';
 239                         }
 240                         $ret .= "</h4>";
 241                 } elseif (!strpos($oembed->html, $oembed->embed_url)) {
 242                         // add <a> for html2bbcode conversion
 243                         $ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
 244                 }
 245
 246                 $ret .= '</div>';
 247
 248                 $ret = str_replace("\n", "", $ret);
 249                 return mb_convert_encoding($ret, 'HTML-ENTITIES', mb_detect_encoding($ret));
 250         }
 251
 252         public static function BBCode2HTML($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'], $text);
 259         }
 260
 261         /**
 262          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 263          * and replace it with [embed]url[/embed]
 264          */
 265         public static function HTML2BBCode($text)
 266         {
 267                 // start parser only if 'oembed' is in text
 268                 if (strpos($text, "oembed")) {
 269
 270                         // convert non ascii chars to html entities
 271                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 272
 273                         // If it doesn't parse at all, just return the text.
 274                         $dom = @DOMDocument::loadHTML($html_text);
 275                         if (!$dom) {
 276                                 return $text;
 277                         }
 278                         $xpath = new DOMXPath($dom);
 279
 280                         $xattr = self::buildXPath("class", "oembed");
 281                         $entries = $xpath->query("//div[$xattr]");
 282
 283                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 284                         foreach ($entries as $e) {
 285                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 286                                 if (!is_null($href)) {
 287                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 288                                 }
 289                         }
 290                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 291                 } else {
 292                         return $text;
 293                 }
 294         }
 295
 296         /**
 297          * Determines if rich content OEmbed is allowed for the provided URL
 298          *
 299          * @brief Determines if rich content OEmbed is allowed for the provided URL
 300          * @param string $url
 301          * @return boolean
 302          */
 303         public static function isAllowedURL($url)
 304         {
 305                 if (!Config::get('system', 'no_oembed_rich_content')) {
 306                         return true;
 307                 }
 308
 309                 $domain = parse_url($url, PHP_URL_HOST);
 310                 if (!x($domain)) {
 311                         return false;
 312                 }
 313
 314                 $str_allowed = Config::get('system', 'allowed_oembed', '');
 315                 if (!x($str_allowed)) {
 316                         return false;
 317                 }
 318
 319                 $allowed = explode(',', $str_allowed);
 320
 321                 return Network::isDomainAllowed($domain, $allowed);
 322         }
 323
 324         public static function getHTML($url, $title = null)
 325         {
 326                 // Always embed the SSL version
 327                 $url = str_replace(["http://www.youtube.com/", "http://player.vimeo.com/"],
 328                                         ["https://www.youtube.com/", "https://player.vimeo.com/"], $url);
 329
 330                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 331
 332                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 333                         throw new Exception('OEmbed failed for URL: ' . $url);
 334                 }
 335
 336                 if (x($title)) {
 337                         $o->title = $title;
 338                 }
 339
 340                 $html = self::formatObject($o);
 341
 342                 return $html;
 343         }
 344
 345         /**
 346          * @brief Generates the iframe HTML for an oembed attachment.
 347          *
 348          * Width and height are given by the remote, and are regularly too small for
 349          * the generated iframe.
 350          *
 351          * The width is entirely discarded for the actual width of the post, while fixed
 352          * height is used as a starting point before the inevitable resizing.
 353          *
 354          * Since the iframe is automatically resized on load, there are no need for ugly
 355          * and impractical scrollbars.
 356          *
 357          * @todo This function is currently unused until someone™ adds support for a separate OEmbed domain
 358          *
 359          * @param string $src Original remote URL to embed
 360          * @param string $width
 361          * @param string $height
 362          * @return string formatted HTML
 363          *
 364          * @see oembed_format_object()
 365          */
 366         private static function iframe($src, $width, $height)
 367         {
 368                 $a = get_app();
 369
 370                 if (!$height || strstr($height, '%')) {
 371                         $height = '200';
 372                 }
 373                 $width = '100%';
 374
 375                 $src = System::baseUrl() . '/oembed/' . base64url_encode($src);
 376                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . L10n::t('Embedded content') . '</iframe>';
 377         }
 378
 379         /**
 380          * Generates an XPath query to select elements whose provided attribute contains
 381          * the provided value in a space-separated list.
 382          *
 383          * @brief Generates attribute search XPath string
 384          *
 385          * @param string $attr Name of the attribute to seach
 386          * @param string $value Value to search in a space-separated list
 387          * @return string
 388          */
 389         private static function buildXPath($attr, $value)
 390         {
 391                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 392                 return "contains(normalize-space(@$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'";
 393         }
 394
 395         /**
 396          * Returns the inner XML string of a provided DOMNode
 397          *
 398          * @brief Returns the inner XML string of a provided DOMNode
 399          *
 400          * @param DOMNode $node
 401          * @return string
 402          */
 403         private static function getInnerHTML(DOMNode $node)
 404         {
 405                 $innerHTML = '';
 406                 $children = $node->childNodes;
 407                 foreach ($children as $child) {
 408                         $innerHTML .= $child->ownerDocument->saveXML($child);
 409                 }
 410                 return $innerHTML;
 411         }
 412
 413 }