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 = ["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(["http://www.youtube.com/", "http://player.vimeo.com/"], ["https://www.youtube.com/", "https://player.vimeo.com/"], $j->html);
 129                 }
 130
 131                 $j->embedurl = $embedurl;
 132
 133                 // If fetching information doesn't work, then improve via internal functions
 134                 if ($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 = $data["url"];
 140                         }
 141
 142                         if (isset($data["title"])) {
 143                                 $j->title = $data["title"];
 144                         }
 145
 146                         if (isset($data["text"])) {
 147                                 $j->description = $data["text"];
 148                         }
 149
 150                         if (is_array($data["images"])) {
 151                                 $j->thumbnail_url = $data["images"][0]["src"];
 152                                 $j->thumbnail_width = $data["images"][0]["width"];
 153                                 $j->thumbnail_height = $data["images"][0]["height"];
 154                         }
 155                 }
 156
 157                 call_hooks('oembed_fetch_url', $embedurl, $j);
 158
 159                 return $j;
 160         }
 161
 162         private static function formatObject($j)
 163         {
 164                 $embedurl = $j->embedurl;
 165                 $jhtml = $j->html;
 166                 $ret = '<div class="oembed ' . $j->type . '">';
 167                 switch ($j->type) {
 168                         case "video":
 169                                 if (isset($j->thumbnail_url)) {
 170                                         $tw = (isset($j->thumbnail_width) && intval($j->thumbnail_width)) ? $j->thumbnail_width : 200;
 171                                         $th = (isset($j->thumbnail_height) && intval($j->thumbnail_height)) ? $j->thumbnail_height : 180;
 172                                         // make sure we don't attempt divide by zero, fallback is a 1:1 ratio
 173                                         $tr = (($th) ? $tw / $th : 1);
 174
 175                                         $th = 120;
 176                                         $tw = $th * $tr;
 177                                         $tpl = get_markup_template('oembed_video.tpl');
 178                                         $ret .= replace_macros($tpl, [
 179                                                 '$baseurl' => System::baseUrl(),
 180                                                 '$embedurl' => $embedurl,
 181                                                 '$escapedhtml' => base64_encode($jhtml),
 182                                                 '$tw' => $tw,
 183                                                 '$th' => $th,
 184                                                 '$turl' => $j->thumbnail_url,
 185                                         ]);
 186                                 } else {
 187                                         $ret = $jhtml;
 188                                 }
 189                                 break;
 190                         case "photo":
 191                                 $ret .= '<img width="' . $j->width . '" src="' . proxy_url($j->url) . '">';
 192                                 break;
 193                         case "link":
 194                                 break;
 195                         case "rich":
 196                                 $ret .= proxy_parse_html($jhtml);
 197                                 break;
 198                 }
 199
 200                 // add link to source if not present in "rich" type
 201                 if ($j->type != 'rich' || !strpos($j->html, $embedurl)) {
 202                         $ret .= '<h4>';
 203                         if (!empty($j->title)) {
 204                                 if (!empty($j->provider_name)) {
 205                                         $ret .= $j->provider_name . ": ";
 206                                 }
 207
 208                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $j->title . '</a>';
 209                                 if (!empty($j->author_name)) {
 210                                         $ret .= ' (' . $j->author_name . ')';
 211                                 }
 212                         } elseif (!empty($j->provider_name) || !empty($j->author_name)) {
 213                                 $embedlink = "";
 214                                 if (!empty($j->provider_name)) {
 215                                         $embedlink .= $j->provider_name;
 216                                 }
 217
 218                                 if (!empty($j->author_name)) {
 219                                         if ($embedlink != "") {
 220                                                 $embedlink .= ": ";
 221                                         }
 222
 223                                         $embedlink .= $j->author_name;
 224                                 }
 225                                 if (trim($embedlink) == "") {
 226                                         $embedlink = $embedurl;
 227                                 }
 228
 229                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedlink . '</a>';
 230                         } else {
 231                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedurl . '</a>';
 232                         }
 233                         $ret .= "</h4>";
 234                 } elseif (!strpos($j->html, $embedurl)) {
 235                         // add <a> for html2bbcode conversion
 236                         $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $j->title . '</a>';
 237                 }
 238
 239                 $ret .= '</div>';
 240
 241                 $ret = str_replace("\n", "", $ret);
 242                 return mb_convert_encoding($ret, 'HTML-ENTITIES', mb_detect_encoding($ret));
 243         }
 244
 245         public static function BBCode2HTML($text)
 246         {
 247                 $stopoembed = Config::get("system", "no_oembed");
 248                 if ($stopoembed == true) {
 249                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 250                 }
 251                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 252         }
 253
 254         /**
 255          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 256          * and replace it with [embed]url[/embed]
 257          */
 258         public static function HTML2BBCode($text)
 259         {
 260                 // start parser only if 'oembed' is in text
 261                 if (strpos($text, "oembed")) {
 262
 263                         // convert non ascii chars to html entities
 264                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 265
 266                         // If it doesn't parse at all, just return the text.
 267                         $dom = @DOMDocument::loadHTML($html_text);
 268                         if (!$dom) {
 269                                 return $text;
 270                         }
 271                         $xpath = new DOMXPath($dom);
 272
 273                         $xattr = self::buildXPath("class", "oembed");
 274                         $entries = $xpath->query("//div[$xattr]");
 275
 276                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 277                         foreach ($entries as $e) {
 278                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 279                                 if (!is_null($href)) {
 280                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 281                                 }
 282                         }
 283                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 284                 } else {
 285                         return $text;
 286                 }
 287         }
 288
 289         /**
 290          * Determines if rich content OEmbed is allowed for the provided URL
 291          *
 292          * @brief Determines if rich content OEmbed is allowed for the provided URL
 293          * @param string $url
 294          * @return boolean
 295          */
 296         public static function isAllowedURL($url)
 297         {
 298                 if (!Config::get('system', 'no_oembed_rich_content')) {
 299                         return true;
 300                 }
 301
 302                 $domain = parse_url($url, PHP_URL_HOST);
 303                 if (!x($domain)) {
 304                         return false;
 305                 }
 306
 307                 $str_allowed = Config::get('system', 'allowed_oembed', '');
 308                 if (!x($str_allowed)) {
 309                         return false;
 310                 }
 311
 312                 $allowed = explode(',', $str_allowed);
 313
 314                 return allowed_domain($domain, $allowed);
 315         }
 316
 317         public static function getHTML($url, $title = null)
 318         {
 319                 // Always embed the SSL version
 320                 $url = str_replace(["http://www.youtube.com/", "http://player.vimeo.com/"],
 321                                         ["https://www.youtube.com/", "https://player.vimeo.com/"], $url);
 322
 323                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 324
 325                 if (!is_object($o) || $o->type == 'error') {
 326                         throw new Exception('OEmbed failed for URL: ' . $url);
 327                 }
 328
 329                 if (x($title)) {
 330                         $o->title = $title;
 331                 }
 332
 333                 $html = self::formatObject($o);
 334
 335                 return $html;
 336         }
 337
 338         /**
 339          * @brief Generates the iframe HTML for an oembed attachment.
 340          *
 341          * Width and height are given by the remote, and are regularly too small for
 342          * the generated iframe.
 343          *
 344          * The width is entirely discarded for the actual width of the post, while fixed
 345          * height is used as a starting point before the inevitable resizing.
 346          *
 347          * Since the iframe is automatically resized on load, there are no need for ugly
 348          * and impractical scrollbars.
 349          *
 350          * @todo This function is currently unused until someone™ adds support for a separate OEmbed domain
 351          *
 352          * @param string $src Original remote URL to embed
 353          * @param string $width
 354          * @param string $height
 355          * @return string formatted HTML
 356          *
 357          * @see oembed_format_object()
 358          */
 359         private static function iframe($src, $width, $height)
 360         {
 361                 $a = get_app();
 362
 363                 if (!$height || strstr($height, '%')) {
 364                         $height = '200';
 365                 }
 366                 $width = '100%';
 367
 368                 $src = System::baseUrl() . '/oembed/' . base64url_encode($src);
 369                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . t('Embedded content') . '</iframe>';
 370         }
 371
 372         /**
 373          * Generates an XPath query to select elements whose provided attribute contains
 374          * the provided value in a space-separated list.
 375          *
 376          * @brief Generates attribute search XPath string
 377          *
 378          * @param string $attr Name of the attribute to seach
 379          * @param string $value Value to search in a space-separated list
 380          * @return string
 381          */
 382         private static function buildXPath($attr, $value)
 383         {
 384                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 385                 return "contains(normalize-space(@$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'";
 386         }
 387
 388         /**
 389          * Returns the inner XML string of a provided DOMNode
 390          *
 391          * @brief Returns the inner XML string of a provided DOMNode
 392          *
 393          * @param DOMNode $node
 394          * @return string
 395          */
 396         private static function getInnerHTML(DOMNode $node)
 397         {
 398                 $innerHTML = '';
 399                 $children = $node->childNodes;
 400                 foreach ($children as $child) {
 401                         $innerHTML .= $child->ownerDocument->saveXML($child);
 402                 }
 403                 return $innerHTML;
 404         }
 405
 406 }