src/Content/OEmbed.php

   1 <?php
   2
   3 /**
   4  * @file src/Content/OEmbed.php
   5  */
   6
   7 namespace Friendica\Content;
   8
   9 use Friendica\Core\Cache;
  10 use Friendica\Core\System;
  11 use Friendica\Core\Config;
  12 use Friendica\Database\DBM;
  13 use Friendica\Util\ParseUrl;
  14 use dba;
  15 use DOMDocument;
  16 use DOMXPath;
  17 use DOMNode;
  18 use Exception;
  19
  20 require_once 'include/dba.php';
  21 require_once 'mod/proxy.php';
  22
  23 /**
  24  * Handles all OEmbed content fetching and replacement
  25  *
  26  * OEmbed is a standard used to allow an embedded representation of a URL on
  27  * third party sites
  28  *
  29  * @see https://oembed.com
  30  *
  31  * @author Hypolite Petovan <mrpetovan@gmail.com>
  32  */
  33 class OEmbed
  34 {
  35         public static function replaceCallback($matches)
  36         {
  37                 $embedurl = $matches[1];
  38                 $j = self::fetchURL($embedurl, !self::isAllowedURL($embedurl));
  39                 $s = self::formatObject($j);
  40
  41                 return $s;
  42         }
  43
  44         /**
  45          * @brief Get data from an URL to embed its content.
  46          *
  47          * @param string $embedurl The URL from which the data should be fetched.
  48          * @param bool $no_rich_type If set to true rich type content won't be fetched.
  49          *
  50          * @return bool|object Returns object with embed content or false if no embeddable
  51          *                         content exists
  52          */
  53         public static function fetchURL($embedurl, $no_rich_type = false)
  54         {
  55                 $embedurl = trim($embedurl, "'");
  56                 $embedurl = trim($embedurl, '"');
  57
  58                 $a = get_app();
  59
  60                 $condition = ['url' => normalise_link($embedurl), 'maxwidth' => $a->videowidth];
  61                 $oembed = dba::selectFirst('oembed', ['content'], $condition);
  62                 if (DBM::is_result($oembed)) {
  63                         $txt = $oembed["content"];
  64                 } else {
  65                         $txt = Cache::get($a->videowidth . $embedurl);
  66                 }
  67                 // These media files should now be caught in bbcode.php
  68                 // left here as a fallback in case this is called from another source
  69
  70                 $noexts = array("mp3", "mp4", "ogg", "ogv", "oga", "ogm", "webm");
  71                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  72
  73
  74                 if (is_null($txt)) {
  75                         $txt = "";
  76
  77                         if (!in_array($ext, $noexts)) {
  78                                 // try oembed autodiscovery
  79                                 $redirects = 0;
  80                                 $html_text = fetch_url($embedurl, false, $redirects, 15, "text/*");
  81                                 if ($html_text) {
  82                                         $dom = @DOMDocument::loadHTML($html_text);
  83                                         if ($dom) {
  84                                                 $xpath = new DOMXPath($dom);
  85                                                 $entries = $xpath->query("//link[@type='application/json+oembed']");
  86                                                 foreach ($entries as $e) {
  87                                                         $href = $e->getAttributeNode("href")->nodeValue;
  88                                                         $txt = fetch_url($href . '&maxwidth=' . $a->videowidth);
  89                                                         break;
  90                                                 }
  91                                                 $entries = $xpath->query("//link[@type='text/json+oembed']");
  92                                                 foreach ($entries as $e) {
  93                                                         $href = $e->getAttributeNode("href")->nodeValue;
  94                                                         $txt = fetch_url($href . '&maxwidth=' . $a->videowidth);
  95                                                         break;
  96                                                 }
  97                                         }
  98                                 }
  99                         }
 100
 101                         $txt = trim($txt);
 102
 103                         if (!$txt || $txt[0] != "{") {
 104                                 $txt = '{"type":"error"}';
 105                         } else { //save in cache
 106                                 $j = json_decode($txt);
 107                                 if ($j->type != "error") {
 108                                         dba::insert('oembed', [
 109                                                 'url' => normalise_link($embedurl),
 110                                                 'maxwidth' => $a->videowidth,
 111                                                 'content' => $txt,
 112                                                 'created' => datetime_convert()
 113                                         ], true);
 114                                 }
 115
 116                                 Cache::set($a->videowidth . $embedurl, $txt, CACHE_DAY);
 117                         }
 118                 }
 119
 120                 $j = json_decode($txt);
 121
 122                 if (!is_object($j)) {
 123                         return false;
 124                 }
 125
 126                 // Always embed the SSL version
 127                 if (isset($j->html)) {
 128                         $j->html = str_replace(array("http://www.youtube.com/", "http://player.vimeo.com/"), array("https://www.youtube.com/", "https://player.vimeo.com/"), $j->html);
 129                 }
 130
 131                 $j->embedurl = $embedurl;
 132
 133                 // If fetching information doesn't work, then improve via internal functions
 134                 if (($j->type == "error") || ($no_rich_type && ($j->type == "rich"))) {
 135                         $data = ParseUrl::getSiteinfoCached($embedurl, true, false);
 136                         $j->type = $data["type"];
 137
 138                         if ($j->type == "photo") {
 139                                 $j->url = $data["url"];
 140                                 //$j->width = $data["images"][0]["width"];
 141                                 //$j->height = $data["images"][0]["height"];
 142                         }
 143
 144                         if (isset($data["title"])) {
 145                                 $j->title = $data["title"];
 146                         }
 147
 148                         if (isset($data["text"])) {
 149                                 $j->description = $data["text"];
 150                         }
 151
 152                         if (is_array($data["images"])) {
 153                                 $j->thumbnail_url = $data["images"][0]["src"];
 154                                 $j->thumbnail_width = $data["images"][0]["width"];
 155                                 $j->thumbnail_height = $data["images"][0]["height"];
 156                         }
 157                 }
 158
 159                 call_hooks('oembed_fetch_url', $embedurl, $j);
 160
 161                 return $j;
 162         }
 163
 164         private static function formatObject($j)
 165         {
 166                 $embedurl = $j->embedurl;
 167                 $jhtml = $j->html;
 168                 $ret = '<div class="oembed ' . $j->type . '">';
 169                 switch ($j->type) {
 170                         case "video":
 171                                 if (isset($j->thumbnail_url)) {
 172                                         $tw = (isset($j->thumbnail_width) && intval($j->thumbnail_width)) ? $j->thumbnail_width : 200;
 173                                         $th = (isset($j->thumbnail_height) && intval($j->thumbnail_height)) ? $j->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 = get_markup_template('oembed_video.tpl');
 180                                         $ret .= replace_macros($tpl, array(
 181                                                 '$baseurl' => System::baseUrl(),
 182                                                 '$embedurl' => $embedurl,
 183                                                 '$escapedhtml' => base64_encode($jhtml),
 184                                                 '$tw' => $tw,
 185                                                 '$th' => $th,
 186                                                 '$turl' => $j->thumbnail_url,
 187                                         ));
 188                                 } else {
 189                                         $ret = $jhtml;
 190                                 }
 191                                 break;
 192                         case "photo":
 193                                 $ret .= '<img width="' . $j->width . '" src="' . proxy_url($j->url) . '">';
 194                                 break;
 195                         case "link":
 196                                 break;
 197                         case "rich":
 198                                 $ret .= proxy_parse_html($jhtml);
 199                                 break;
 200                 }
 201
 202                 // add link to source if not present in "rich" type
 203                 if ($j->type != 'rich' || !strpos($j->html, $embedurl)) {
 204                         $ret .= '<h4>';
 205                         if (isset($j->title)) {
 206                                 if (isset($j->provider_name)) {
 207                                         $ret .= $j->provider_name . ": ";
 208                                 }
 209
 210                                 $embedlink = (isset($j->title)) ? $j->title : $embedurl;
 211                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedlink . '</a>';
 212                                 if (isset($j->author_name)) {
 213                                         $ret .= ' (' . $j->author_name . ')';
 214                                 }
 215                         } elseif (isset($j->provider_name) || isset($j->author_name)) {
 216                                 $embedlink = "";
 217                                 if (isset($j->provider_name)) {
 218                                         $embedlink .= $j->provider_name;
 219                                 }
 220
 221                                 if (isset($j->author_name)) {
 222                                         if ($embedlink != "") {
 223                                                 $embedlink .= ": ";
 224                                         }
 225
 226                                         $embedlink .= $j->author_name;
 227                                 }
 228                                 if (trim($embedlink) == "") {
 229                                         $embedlink = $embedurl;
 230                                 }
 231
 232                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedlink . '</a>';
 233                         }
 234                         $ret .= "</h4>";
 235                 } elseif (!strpos($j->html, $embedurl)) {
 236                         // add <a> for html2bbcode conversion
 237                         $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $j->title . '</a>';
 238                 }
 239
 240                 $ret .= '</div>';
 241
 242                 $ret = str_replace("\n", "", $ret);
 243                 return mb_convert_encoding($ret, 'HTML-ENTITIES', mb_detect_encoding($ret));
 244         }
 245
 246         public static function BBCode2HTML($text)
 247         {
 248                 $stopoembed = Config::get("system", "no_oembed");
 249                 if ($stopoembed == true) {
 250                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 251                 }
 252                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 253         }
 254
 255         /**
 256          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 257          * and replace it with [embed]url[/embed]
 258          */
 259         public static function HTML2BBCode($text)
 260         {
 261                 // start parser only if 'oembed' is in text
 262                 if (strpos($text, "oembed")) {
 263
 264                         // convert non ascii chars to html entities
 265                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 266
 267                         // If it doesn't parse at all, just return the text.
 268                         $dom = @DOMDocument::loadHTML($html_text);
 269                         if (!$dom) {
 270                                 return $text;
 271                         }
 272                         $xpath = new DOMXPath($dom);
 273
 274                         $xattr = self::buildXPath("class", "oembed");
 275                         $entries = $xpath->query("//div[$xattr]");
 276
 277                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 278                         foreach ($entries as $e) {
 279                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 280                                 if (!is_null($href)) {
 281                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 282                                 }
 283                         }
 284                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 285                 } else {
 286                         return $text;
 287                 }
 288         }
 289
 290         /**
 291          * Determines if rich content OEmbed is allowed for the provided URL
 292          *
 293          * @brief Determines if rich content OEmbed is allowed for the provided URL
 294          * @param string $url
 295          * @return boolean
 296          */
 297         public static function isAllowedURL($url)
 298         {
 299                 if (!Config::get('system', 'no_oembed_rich_content')) {
 300                         return true;
 301                 }
 302
 303                 $domain = parse_url($url, PHP_URL_HOST);
 304                 if (!x($domain)) {
 305                         return false;
 306                 }
 307
 308                 $str_allowed = Config::get('system', 'allowed_oembed', '');
 309                 if (!x($str_allowed)) {
 310                         return false;
 311                 }
 312
 313                 $allowed = explode(',', $str_allowed);
 314
 315                 return allowed_domain($domain, $allowed);
 316         }
 317
 318         public static function getHTML($url, $title = null)
 319         {
 320                 // Always embed the SSL version
 321                 $url = str_replace(array("http://www.youtube.com/", "http://player.vimeo.com/"),
 322                                         array("https://www.youtube.com/", "https://player.vimeo.com/"), $url);
 323
 324                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 325
 326                 if (!is_object($o) || $o->type == 'error') {
 327                         throw new Exception('OEmbed failed for URL: ' . $url);
 328                 }
 329
 330                 if (x($title)) {
 331                         $o->title = $title;
 332                 }
 333
 334                 $html = self::formatObject($o);
 335
 336                 return $html;
 337         }
 338
 339         /**
 340          * @brief Generates the iframe HTML for an oembed attachment.
 341          *
 342          * Width and height are given by the remote, and are regularly too small for
 343          * the generated iframe.
 344          *
 345          * The width is entirely discarded for the actual width of the post, while fixed
 346          * height is used as a starting point before the inevitable resizing.
 347          *
 348          * Since the iframe is automatically resized on load, there are no need for ugly
 349          * and impractical scrollbars.
 350          *
 351          * @todo This function is currently unused until someone™ adds support for a separate OEmbed domain
 352          *
 353          * @param string $src Original remote URL to embed
 354          * @param string $width
 355          * @param string $height
 356          * @return string formatted HTML
 357          *
 358          * @see oembed_format_object()
 359          */
 360         private static function iframe($src, $width, $height)
 361         {
 362                 $a = get_app();
 363
 364                 if (!$height || strstr($height, '%')) {
 365                         $height = '200';
 366                 }
 367                 $width = '100%';
 368
 369                 $src = System::baseUrl() . '/oembed/' . base64url_encode($src);
 370                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . t('Embedded content') . '</iframe>';
 371         }
 372
 373         /**
 374          * Generates an XPath query to select elements whose provided attribute contains
 375          * the provided value in a space-separated list.
 376          *
 377          * @brief Generates attribute search XPath string
 378          *
 379          * @param string $attr Name of the attribute to seach
 380          * @param string $value Value to search in a space-separated list
 381          * @return string
 382          */
 383         private static function buildXPath($attr, $value)
 384         {
 385                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 386                 return "contains(normalize-space(@$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'";
 387         }
 388
 389         /**
 390          * Returns the inner XML string of a provided DOMNode
 391          *
 392          * @brief Returns the inner XML string of a provided DOMNode
 393          *
 394          * @param DOMNode $node
 395          * @return string
 396          */
 397         private static function getInnerHTML(DOMNode $node)
 398         {
 399                 $innerHTML = '';
 400                 $children = $node->childNodes;
 401                 foreach ($children as $child) {
 402                         $innerHTML .= $child->ownerDocument->saveXML($child);
 403                 }
 404                 return $innerHTML;
 405         }
 406
 407 }