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\Database\DBM;
  20 use Friendica\Util\DateTimeFormat;
  21 use Friendica\Util\Network;
  22 use Friendica\Util\ParseUrl;
  23 use stdClass;
  24
  25 require_once 'include/dba.php';
  26 require_once 'mod/proxy.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 <mrpetovan@gmail.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 $no_rich_type If set to true rich type content won't be fetched.
  54          *
  55          * @return bool|object Returns object with embed content or false if no embeddable
  56          *                         content exists
  57          */
  58         public static function fetchURL($embedurl, $no_rich_type = false)
  59         {
  60                 $embedurl = trim($embedurl, "'");
  61                 $embedurl = trim($embedurl, '"');
  62
  63                 $a = get_app();
  64
  65                 $condition = ['url' => normalise_link($embedurl), 'maxwidth' => $a->videowidth];
  66                 $oembed = dba::selectFirst('oembed', ['content'], $condition);
  67                 if (DBM::is_result($oembed)) {
  68                         $txt = $oembed["content"];
  69                 } else {
  70                         $txt = Cache::get($a->videowidth . $embedurl);
  71                 }
  72                 // These media files should now be caught in bbcode.php
  73                 // left here as a fallback in case this is called from another source
  74
  75                 $noexts = ["mp3", "mp4", "ogg", "ogv", "oga", "ogm", "webm"];
  76                 $ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
  77
  78
  79                 if (is_null($txt)) {
  80                         $txt = "";
  81
  82                         if (!in_array($ext, $noexts)) {
  83                                 // try oembed autodiscovery
  84                                 $redirects = 0;
  85                                 $html_text = Network::fetchUrl($embedurl, false, $redirects, 15, "text/*");
  86                                 if ($html_text) {
  87                                         $dom = @DOMDocument::loadHTML($html_text);
  88                                         if ($dom) {
  89                                                 $xpath = new DOMXPath($dom);
  90                                                 $entries = $xpath->query("//link[@type='application/json+oembed']");
  91                                                 foreach ($entries as $e) {
  92                                                         $href = $e->getAttributeNode("href")->nodeValue;
  93                                                         $txt = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
  94                                                         break;
  95                                                 }
  96                                                 $entries = $xpath->query("//link[@type='text/json+oembed']");
  97                                                 foreach ($entries as $e) {
  98                                                         $href = $e->getAttributeNode("href")->nodeValue;
  99                                                         $txt = Network::fetchUrl($href . '&maxwidth=' . $a->videowidth);
 100                                                         break;
 101                                                 }
 102                                         }
 103                                 }
 104                         }
 105
 106                         $txt = trim($txt);
 107
 108                         if (!$txt || $txt[0] != "{") {
 109                                 $txt = '{"type":"error"}';
 110                         } else { //save in cache
 111                                 $j = json_decode($txt);
 112                                 if (!empty($j->type) && $j->type != "error") {
 113                                         dba::insert('oembed', [
 114                                                 'url' => normalise_link($embedurl),
 115                                                 'maxwidth' => $a->videowidth,
 116                                                 'content' => $txt,
 117                                                 'created' => DateTimeFormat::utcNow()
 118                                         ], true);
 119                                 }
 120
 121                                 Cache::set($a->videowidth . $embedurl, $txt, CACHE_DAY);
 122                         }
 123                 }
 124
 125                 $j = json_decode($txt);
 126
 127                 if (!is_object($j)) {
 128                         return false;
 129                 }
 130
 131                 // Always embed the SSL version
 132                 if (isset($j->html)) {
 133                         $j->html = str_replace(["http://www.youtube.com/", "http://player.vimeo.com/"], ["https://www.youtube.com/", "https://player.vimeo.com/"], $j->html);
 134                 }
 135
 136                 $j->embedurl = $embedurl;
 137
 138                 // If fetching information doesn't work, then improve via internal functions
 139                 if ($no_rich_type && ($j->type == "rich")) {
 140                         $data = ParseUrl::getSiteinfoCached($embedurl, true, false);
 141                         $j->type = $data["type"];
 142
 143                         if ($j->type == "photo") {
 144                                 $j->url = $data["url"];
 145                         }
 146
 147                         if (isset($data["title"])) {
 148                                 $j->title = $data["title"];
 149                         }
 150
 151                         if (isset($data["text"])) {
 152                                 $j->description = $data["text"];
 153                         }
 154
 155                         if (is_array($data["images"])) {
 156                                 $j->thumbnail_url = $data["images"][0]["src"];
 157                                 $j->thumbnail_width = $data["images"][0]["width"];
 158                                 $j->thumbnail_height = $data["images"][0]["height"];
 159                         }
 160                 }
 161
 162                 Addon::callHooks('oembed_fetch_url', $embedurl, $j);
 163
 164                 return $j;
 165         }
 166
 167         private static function formatObject(stdClass $j)
 168         {
 169                 $embedurl = $j->embedurl;
 170                 $jhtml = $j->html;
 171                 $ret = '<div class="oembed ' . $j->type . '">';
 172
 173                 switch ($j->type) {
 174                         case "video":
 175                                 if (isset($j->thumbnail_url)) {
 176                                         $tw = (isset($j->thumbnail_width) && intval($j->thumbnail_width)) ? $j->thumbnail_width : 200;
 177                                         $th = (isset($j->thumbnail_height) && intval($j->thumbnail_height)) ? $j->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 = get_markup_template('oembed_video.tpl');
 184                                         $ret .= replace_macros($tpl, [
 185                                                 '$baseurl' => System::baseUrl(),
 186                                                 '$embedurl' => $embedurl,
 187                                                 '$escapedhtml' => base64_encode($jhtml),
 188                                                 '$tw' => $tw,
 189                                                 '$th' => $th,
 190                                                 '$turl' => $j->thumbnail_url,
 191                                         ]);
 192                                 } else {
 193                                         $ret = $jhtml;
 194                                 }
 195                                 break;
 196                         case "photo":
 197                                 $ret .= '<img width="' . $j->width . '" src="' . proxy_url($j->url) . '">';
 198                                 break;
 199                         case "link":
 200                                 break;
 201                         case "rich":
 202                                 $ret .= proxy_parse_html($jhtml);
 203                                 break;
 204                 }
 205
 206                 // add link to source if not present in "rich" type
 207                 if ($j->type != 'rich' || !strpos($j->html, $embedurl)) {
 208                         $ret .= '<h4>';
 209                         if (!empty($j->title)) {
 210                                 if (!empty($j->provider_name)) {
 211                                         $ret .= $j->provider_name . ": ";
 212                                 }
 213
 214                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $j->title . '</a>';
 215                                 if (!empty($j->author_name)) {
 216                                         $ret .= ' (' . $j->author_name . ')';
 217                                 }
 218                         } elseif (!empty($j->provider_name) || !empty($j->author_name)) {
 219                                 $embedlink = "";
 220                                 if (!empty($j->provider_name)) {
 221                                         $embedlink .= $j->provider_name;
 222                                 }
 223
 224                                 if (!empty($j->author_name)) {
 225                                         if ($embedlink != "") {
 226                                                 $embedlink .= ": ";
 227                                         }
 228
 229                                         $embedlink .= $j->author_name;
 230                                 }
 231                                 if (trim($embedlink) == "") {
 232                                         $embedlink = $embedurl;
 233                                 }
 234
 235                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedlink . '</a>';
 236                         } else {
 237                                 $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $embedurl . '</a>';
 238                         }
 239                         $ret .= "</h4>";
 240                 } elseif (!strpos($j->html, $embedurl)) {
 241                         // add <a> for html2bbcode conversion
 242                         $ret .= '<a href="' . $embedurl . '" rel="oembed">' . $j->title . '</a>';
 243                 }
 244
 245                 $ret .= '</div>';
 246
 247                 $ret = str_replace("\n", "", $ret);
 248                 return mb_convert_encoding($ret, 'HTML-ENTITIES', mb_detect_encoding($ret));
 249         }
 250
 251         public static function BBCode2HTML($text)
 252         {
 253                 $stopoembed = Config::get("system", "no_oembed");
 254                 if ($stopoembed == true) {
 255                         return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . L10n::t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
 256                 }
 257                 return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
 258         }
 259
 260         /**
 261          * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
 262          * and replace it with [embed]url[/embed]
 263          */
 264         public static function HTML2BBCode($text)
 265         {
 266                 // start parser only if 'oembed' is in text
 267                 if (strpos($text, "oembed")) {
 268
 269                         // convert non ascii chars to html entities
 270                         $html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
 271
 272                         // If it doesn't parse at all, just return the text.
 273                         $dom = @DOMDocument::loadHTML($html_text);
 274                         if (!$dom) {
 275                                 return $text;
 276                         }
 277                         $xpath = new DOMXPath($dom);
 278
 279                         $xattr = self::buildXPath("class", "oembed");
 280                         $entries = $xpath->query("//div[$xattr]");
 281
 282                         $xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
 283                         foreach ($entries as $e) {
 284                                 $href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
 285                                 if (!is_null($href)) {
 286                                         $e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
 287                                 }
 288                         }
 289                         return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
 290                 } else {
 291                         return $text;
 292                 }
 293         }
 294
 295         /**
 296          * Determines if rich content OEmbed is allowed for the provided URL
 297          *
 298          * @brief Determines if rich content OEmbed is allowed for the provided URL
 299          * @param string $url
 300          * @return boolean
 301          */
 302         public static function isAllowedURL($url)
 303         {
 304                 if (!Config::get('system', 'no_oembed_rich_content')) {
 305                         return true;
 306                 }
 307
 308                 $domain = parse_url($url, PHP_URL_HOST);
 309                 if (!x($domain)) {
 310                         return false;
 311                 }
 312
 313                 $str_allowed = Config::get('system', 'allowed_oembed', '');
 314                 if (!x($str_allowed)) {
 315                         return false;
 316                 }
 317
 318                 $allowed = explode(',', $str_allowed);
 319
 320                 return Network::isDomainAllowed($domain, $allowed);
 321         }
 322
 323         public static function getHTML($url, $title = null)
 324         {
 325                 // Always embed the SSL version
 326                 $url = str_replace(["http://www.youtube.com/", "http://player.vimeo.com/"],
 327                                         ["https://www.youtube.com/", "https://player.vimeo.com/"], $url);
 328
 329                 $o = self::fetchURL($url, !self::isAllowedURL($url));
 330
 331                 if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
 332                         throw new Exception('OEmbed failed for URL: ' . $url);
 333                 }
 334
 335                 if (x($title)) {
 336                         $o->title = $title;
 337                 }
 338
 339                 $html = self::formatObject($o);
 340
 341                 return $html;
 342         }
 343
 344         /**
 345          * @brief Generates the iframe HTML for an oembed attachment.
 346          *
 347          * Width and height are given by the remote, and are regularly too small for
 348          * the generated iframe.
 349          *
 350          * The width is entirely discarded for the actual width of the post, while fixed
 351          * height is used as a starting point before the inevitable resizing.
 352          *
 353          * Since the iframe is automatically resized on load, there are no need for ugly
 354          * and impractical scrollbars.
 355          *
 356          * @todo This function is currently unused until someone™ adds support for a separate OEmbed domain
 357          *
 358          * @param string $src Original remote URL to embed
 359          * @param string $width
 360          * @param string $height
 361          * @return string formatted HTML
 362          *
 363          * @see oembed_format_object()
 364          */
 365         private static function iframe($src, $width, $height)
 366         {
 367                 $a = get_app();
 368
 369                 if (!$height || strstr($height, '%')) {
 370                         $height = '200';
 371                 }
 372                 $width = '100%';
 373
 374                 $src = System::baseUrl() . '/oembed/' . base64url_encode($src);
 375                 return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . L10n::t('Embedded content') . '</iframe>';
 376         }
 377
 378         /**
 379          * Generates an XPath query to select elements whose provided attribute contains
 380          * the provided value in a space-separated list.
 381          *
 382          * @brief Generates attribute search XPath string
 383          *
 384          * @param string $attr Name of the attribute to seach
 385          * @param string $value Value to search in a space-separated list
 386          * @return string
 387          */
 388         private static function buildXPath($attr, $value)
 389         {
 390                 // https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
 391                 return "contains(normalize-space(@$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'";
 392         }
 393
 394         /**
 395          * Returns the inner XML string of a provided DOMNode
 396          *
 397          * @brief Returns the inner XML string of a provided DOMNode
 398          *
 399          * @param DOMNode $node
 400          * @return string
 401          */
 402         private static function getInnerHTML(DOMNode $node)
 403         {
 404                 $innerHTML = '';
 405                 $children = $node->childNodes;
 406                 foreach ($children as $child) {
 407                         $innerHTML .= $child->ownerDocument->saveXML($child);
 408                 }
 409                 return $innerHTML;
 410         }
 411
 412 }