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 stdClass;
  23
  24 require_once 'include/dba.php';
  25 require_once 'mod/proxy.php';
  26
  27 /**
  28  * Handles all OEmbed content fetching and replacement
  29  *
  30  * OEmbed is a standard used to allow an embedded representation of a URL on
  31  * third party sites
  32  *
  33  * @see https://oembed.com
  34  *
  35  * @author Hypolite Petovan <mrpetovan@gmail.com>
  36  */
  37 class OEmbed
  38 {
  39         public static function replaceCallback($matches)
  40         {
  41                 $embedurl = $matches[1];
  42                 $j = self::fetchURL($embedurl, !self::isAllowedURL($embedurl));
  43                 $s = self::formatObject($j);
  44
  45                 return $s;
  46         }
  47
  48         /**
  49          * @brief Get data from an URL to embed its content.
  50          *
  51          * @param string $embedurl The URL from which the data should be fetched.
  52          * @param bool $no_rich_type If set to true rich type content won't be fetched.
  53          *
  54          * @return bool|object Returns object with embed content or false if no embeddable
  55          *                         content exists
  56          */
  57         public static function fetchURL($embedurl, $no_rich_type = false)
  58         {
  59                 $embedurl = trim($embedurl, "'");
  60                 $embedurl = trim($embedurl, '"');
  61
  62                 $a = get_app();
  63
  64                 $condition = ['url' => normalise_link($embedurl), 'maxwidth' => $a->videowidth];
  65                 $oembed = DBA::selectFirst('oembed', ['content'], $condition);
  66                 if (DBA::is_result($oembed)) {
  67                         $txt = $oembed["content"];
  68                 } else {
  69                         $txt = Cache::get($a->videowidth . $embedurl);
  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
  74                 $noexts = ["mp3", "mp4", "ogg", "ogv", "oga", "ogm", "webm"];
  75                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  76
  77
  78                 if (is_null($txt)) {
  79                         $txt = "";
  80
  81                         if (!in_array($ext, $noexts)) {
  82                                 // try oembed autodiscovery
  83                                 $redirects = 0;
  84                                 $html_text = Network::fetchUrl($embedurl, false, $redirects, 15, "text/*");
  85                                 if ($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                                                         $txt = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
  93                                                         break;
  94                                                 }
  95                                                 $entries = $xpath->query("//link[@type='text/json+oembed']");
  96                                                 foreach ($entries as $e) {
  97                                                         $href = $e->getAttributeNode("href")->nodeValue;
  98                                                         $txt = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
  99                                                         break;
 100                                                 }
 101                                         }
 102                                 }
 103                         }
 104
 105                         $txt = trim($txt);
 106
 107                         if (!$txt || $txt[0] != "{") {
 108                                 $txt = '{"type":"error"}';
 109                         } else { //save in cache
 110                                 $j = json_decode($txt);
 111                                 if (!empty($j->type) && $j->type != "error") {
 112                                         DBA::insert('oembed', [
 113                                                 'url' => normalise_link($embedurl),
 114                                                 'maxwidth' => $a->videowidth,
 115                                                 'content' => $txt,
 116                                                 'created' => DateTimeFormat::utcNow()
 117                                         ], true);
 118                                 }
 119
 120                                 Cache::set($a->videowidth . $embedurl, $txt, CACHE_DAY);
 121                         }
 122                 }
 123
 124                 $j = json_decode($txt);
 125
 126                 if (!is_object($j)) {
 127                         return false;
 128                 }
 129
 130                 // Always embed the SSL version
 131                 if (isset($j->html)) {
 132                         $j->html = str_replace(["http://www.youtube.com/", "http://player.vimeo.com/"], ["https://www.youtube.com/", "https://player.vimeo.com/"], $j->html);
 133                 }
 134
 135                 $j->embedurl = $embedurl;
 136
 137                 // If fetching information doesn't work, then improve via internal functions
 138                 if ($no_rich_type && ($j->type == "rich")) {
 139                         $data = ParseUrl::getSiteinfoCached($embedurl, true, false);
 140                         $j->type = $data["type"];
 141
 142                         if ($j->type == "photo") {
 143                                 $j->url = $data["url"];
 144                         }
 145
 146                         if (isset($data["title"])) {
 147                                 $j->title = $data["title"];
 148                         }
 149
 150                         if (isset($data["text"])) {
 151                                 $j->description = $data["text"];
 152                         }
 153
 154                         if (is_array($data["images"])) {
 155                                 $j->thumbnail_url = $data["images"][0]["src"];
 156                                 $j->thumbnail_width = $data["images"][0]["width"];
 157                                 $j->thumbnail_height = $data["images"][0]["height"];
 158                         }
 159                 }
 160
 161                 Addon::callHooks('oembed_fetch_url', $embedurl, $j);
 162
 163                 return $j;
 164         }
 165
 166         private static function formatObject(stdClass $j)
 167         {
 168                 $embedurl = $j->embedurl;
 169                 $jhtml = $j->html;
 170                 $ret = '<div class="oembed ' . $j->type . '">';
 171
 172                 switch ($j->type) {
 173                         case "video":
 174                                 if (isset($j->thumbnail_url)) {
 175                                         $tw = (isset($j->thumbnail_width) && intval($j->thumbnail_width)) ? $j->thumbnail_width : 200;
 176                                         $th = (isset($j->thumbnail_height) && intval($j->thumbnail_height)) ? $j->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 = get_markup_template('oembed_video.tpl');
 183                                         $ret .= replace_macros($tpl, [
 184                                                 '$baseurl' => System::baseUrl(),
 185                                                 '$embedurl' => $embedurl,
 186                                                 '$escapedhtml' => base64_encode($jhtml),
 187                                                 '$tw' => $tw,
 188                                                 '$th' => $th,
 189                                                 '$turl' => $j->thumbnail_url,
 190                                         ]);
 191                                 } else {
 192                                         $ret = $jhtml;
 193                                 }
 194                                 break;
 195                         case "photo":
 196                                 $ret .= '<img width="' . $j->width . '" src="' . proxy_url($j->url) . '">';
 197                                 break;
 198                         case "link":
 199                                 break;
 200                         case "rich":
 201                                 $ret .= proxy_parse_html($jhtml);
 202                                 break;
 203                 }
 204
 205                 // add link to source if not present in "rich" type
 206                 if ($j->type != 'rich' || !strpos($j->html, $embedurl)) {
 207                         $ret .= '<h4>';
 208                         if (!empty($j->title)) {
 209                                 if (!empty($j->provider_name)) {
 210                                         $ret .= $j->provider_name . ": ";
 211                                 }
 212
 213                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $j->title . '</a>';
 214                                 if (!empty($j->author_name)) {
 215                                         $ret .= ' (' . $j->author_name . ')';
 216                                 }
 217                         } elseif (!empty($j->provider_name) || !empty($j->author_name)) {
 218                                 $embedlink = "";
 219                                 if (!empty($j->provider_name)) {
 220                                         $embedlink .= $j->provider_name;
 221                                 }
 222
 223                                 if (!empty($j->author_name)) {
 224                                         if ($embedlink != "") {
 225                                                 $embedlink .= ": ";
 226                                         }
 227
 228                                         $embedlink .= $j->author_name;
 229                                 }
 230                                 if (trim($embedlink) == "") {
 231                                         $embedlink = $embedurl;
 232                                 }
 233
 234                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedlink . '</a>';
 235                         } else {
 236                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedurl . '</a>';
 237                         }
 238                         $ret .= "</h4>";
 239                 } elseif (!strpos($j->html, $embedurl)) {
 240                         // add <a> for html2bbcode conversion
 241                         $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $j->title . '</a>';
 242                 }
 243
 244                 $ret .= '</div>';
 245
 246                 $ret = str_replace("\n", "", $ret);
 247                 return mb_convert_encoding($ret, 'HTML-ENTITIES', mb_detect_encoding($ret));
 248         }
 249
 250         public static function BBCode2HTML($text)
 251         {
 252                 $stopoembed = Config::get("system", "no_oembed");
 253                 if ($stopoembed == true) {
 254                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . L10n::t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 255                 }
 256                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 257         }
 258
 259         /**
 260          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 261          * and replace it with [embed]url[/embed]
 262          */
 263         public static function HTML2BBCode($text)
 264         {
 265                 // start parser only if 'oembed' is in text
 266                 if (strpos($text, "oembed")) {
 267
 268                         // convert non ascii chars to html entities
 269                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 270
 271                         // If it doesn't parse at all, just return the text.
 272                         $dom = @DOMDocument::loadHTML($html_text);
 273                         if (!$dom) {
 274                                 return $text;
 275                         }
 276                         $xpath = new DOMXPath($dom);
 277
 278                         $xattr = self::buildXPath("class", "oembed");
 279                         $entries = $xpath->query("//div[$xattr]");
 280
 281                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 282                         foreach ($entries as $e) {
 283                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 284                                 if (!is_null($href)) {
 285                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 286                                 }
 287                         }
 288                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 289                 } else {
 290                         return $text;
 291                 }
 292         }
 293
 294         /**
 295          * Determines if rich content OEmbed is allowed for the provided URL
 296          *
 297          * @brief Determines if rich content OEmbed is allowed for the provided URL
 298          * @param string $url
 299          * @return boolean
 300          */
 301         public static function isAllowedURL($url)
 302         {
 303                 if (!Config::get('system', 'no_oembed_rich_content')) {
 304                         return true;
 305                 }
 306
 307                 $domain = parse_url($url, PHP_URL_HOST);
 308                 if (!x($domain)) {
 309                         return false;
 310                 }
 311
 312                 $str_allowed = Config::get('system', 'allowed_oembed', '');
 313                 if (!x($str_allowed)) {
 314                         return false;
 315                 }
 316
 317                 $allowed = explode(',', $str_allowed);
 318
 319                 return Network::isDomainAllowed($domain, $allowed);
 320         }
 321
 322         public static function getHTML($url, $title = null)
 323         {
 324                 // Always embed the SSL version
 325                 $url = str_replace(["http://www.youtube.com/", "http://player.vimeo.com/"],
 326                                         ["https://www.youtube.com/", "https://player.vimeo.com/"], $url);
 327
 328                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 329
 330                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 331                         throw new Exception('OEmbed failed for URL: ' . $url);
 332                 }
 333
 334                 if (x($title)) {
 335                         $o->title = $title;
 336                 }
 337
 338                 $html = self::formatObject($o);
 339
 340                 return $html;
 341         }
 342
 343         /**
 344          * @brief Generates the iframe HTML for an oembed attachment.
 345          *
 346          * Width and height are given by the remote, and are regularly too small for
 347          * the generated iframe.
 348          *
 349          * The width is entirely discarded for the actual width of the post, while fixed
 350          * height is used as a starting point before the inevitable resizing.
 351          *
 352          * Since the iframe is automatically resized on load, there are no need for ugly
 353          * and impractical scrollbars.
 354          *
 355          * @todo This function is currently unused until someone™ adds support for a separate OEmbed domain
 356          *
 357          * @param string $src Original remote URL to embed
 358          * @param string $width
 359          * @param string $height
 360          * @return string formatted HTML
 361          *
 362          * @see oembed_format_object()
 363          */
 364         private static function iframe($src, $width, $height)
 365         {
 366                 $a = get_app();
 367
 368                 if (!$height || strstr($height, '%')) {
 369                         $height = '200';
 370                 }
 371                 $width = '100%';
 372
 373                 $src = System::baseUrl() . '/oembed/' . base64url_encode($src);
 374                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . L10n::t('Embedded content') . '</iframe>';
 375         }
 376
 377         /**
 378          * Generates an XPath query to select elements whose provided attribute contains
 379          * the provided value in a space-separated list.
 380          *
 381          * @brief Generates attribute search XPath string
 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          * @brief Returns the inner XML string of a provided DOMNode
 397          *
 398          * @param DOMNode $node
 399          * @return string
 400          */
 401         private static function getInnerHTML(DOMNode $node)
 402         {
 403                 $innerHTML = '';
 404                 $children = $node->childNodes;
 405                 foreach ($children as $child) {
 406                         $innerHTML .= $child->ownerDocument->saveXML($child);
 407                 }
 408                 return $innerHTML;
 409         }
 410
 411 }