src/Util/ParseUrl.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Util;
  23
  24 use DOMDocument;
  25 use DOMXPath;
  26 use Friendica\Content\OEmbed;
  27 use Friendica\Core\Hook;
  28 use Friendica\Core\Logger;
  29 use Friendica\Database\DBA;
  30
  31 /**
  32  * Get information about a given URL
  33  *
  34  * Class with methods for extracting certain content from an url
  35  */
  36 class ParseUrl
  37 {
  38         /**
  39          * Maximum number of characters for the description
  40          */
  41         const MAX_DESC_COUNT = 250;
  42
  43         /**
  44          * Minimum number of characters for the description
  45          */
  46         const MIN_DESC_COUNT = 100;
  47
  48         /**
  49          * Search for chached embeddable data of an url otherwise fetch it
  50          *
  51          * @param string $url         The url of the page which should be scraped
  52          * @param bool   $no_guessing If true the parse doens't search for
  53          *                            preview pictures
  54          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
  55          *                            to avoid endless loops
  56          *
  57          * @return array which contains needed data for embedding
  58          *    string 'url'      => The url of the parsed page
  59          *    string 'type'     => Content type
  60          *    string 'title'    => (optional) The title of the content
  61          *    string 'text'     => (optional) The description for the content
  62          *    string 'image'    => (optional) A preview image of the content (only available if $no_geuessing = false)
  63          *    array  'images'   => (optional) Array of preview pictures
  64          *    string 'keywords' => (optional) The tags which belong to the content
  65          *
  66          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  67          * @see   ParseUrl::getSiteinfo() for more information about scraping
  68          * embeddable content
  69          */
  70         public static function getSiteinfoCached($url, $no_guessing = false, $do_oembed = true)
  71         {
  72                 if ($url == "") {
  73                         return false;
  74                 }
  75
  76                 $parsed_url = DBA::selectFirst('parsed_url', ['content'],
  77                         ['url' => Strings::normaliseLink($url), 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
  78                 );
  79                 if (!empty($parsed_url['content'])) {
  80                         $data = unserialize($parsed_url['content']);
  81                         return $data;
  82                 }
  83
  84                 $data = self::getSiteinfo($url, $no_guessing, $do_oembed);
  85
  86                 DBA::insert(
  87                         'parsed_url',
  88                         [
  89                                 'url' => substr(Strings::normaliseLink($url), 0, 255), 'guessing' => !$no_guessing,
  90                                 'oembed' => $do_oembed, 'content' => serialize($data),
  91                                 'created' => DateTimeFormat::utcNow()
  92                         ],
  93                         true
  94                 );
  95
  96                 return $data;
  97         }
  98
  99         /**
 100          * Parse a page for embeddable content information
 101          *
 102          * This method parses to url for meta data which can be used to embed
 103          * the content. If available it prioritizes Open Graph meta tags.
 104          * If this is not available it uses the twitter cards meta tags.
 105          * As fallback it uses standard html elements with meta informations
 106          * like \<title\>Awesome Title\</title\> or
 107          * \<meta name="description" content="An awesome description"\>
 108          *
 109          * @param string $url         The url of the page which should be scraped
 110          * @param bool   $no_guessing If true the parse doens't search for
 111          *                            preview pictures
 112          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
 113          *                            to avoid endless loops
 114          * @param int    $count       Internal counter to avoid endless loops
 115          *
 116          * @return array which contains needed data for embedding
 117          *    string 'url'      => The url of the parsed page
 118          *    string 'type'     => Content type
 119          *    string 'title'    => (optional) The title of the content
 120          *    string 'text'     => (optional) The description for the content
 121          *    string 'image'    => (optional) A preview image of the content (only available if $no_guessing = false)
 122          *    array  'images'   => (optional) Array of preview pictures
 123          *    string 'keywords' => (optional) The tags which belong to the content
 124          *
 125          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 126          * @todo  https://developers.google.com/+/plugins/snippet/
 127          * @verbatim
 128          * <meta itemprop="name" content="Awesome title">
 129          * <meta itemprop="description" content="An awesome description">
 130          * <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
 131          *
 132          * <body itemscope itemtype="http://schema.org/Product">
 133          *   <h1 itemprop="name">Shiny Trinket</h1>
 134          *   <img itemprop="image" src="{image-url}" />
 135          *   <p itemprop="description">Shiny trinkets are shiny.</p>
 136          * </body>
 137          * @endverbatim
 138          */
 139         public static function getSiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1)
 140         {
 141                 // Check if the URL does contain a scheme
 142                 $scheme = parse_url($url, PHP_URL_SCHEME);
 143
 144                 if ($scheme == '') {
 145                         $url = 'http://' . ltrim($url, '/');
 146                 }
 147
 148                 $url = trim($url, "'\"");
 149
 150                 $url = Network::stripTrackingQueryParams($url);
 151
 152                 $siteinfo = [
 153                         'url' => $url,
 154                         'type' => 'link',
 155                 ];
 156
 157                 if ($count > 10) {
 158                         Logger::log('Endless loop detected for ' . $url, Logger::DEBUG);
 159                         return $siteinfo;
 160                 }
 161
 162                 $curlResult = Network::curl($url);
 163                 if (!$curlResult->isSuccess()) {
 164                         return $siteinfo;
 165                 }
 166
 167                 // If the file is too large then exit
 168                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 169                         return $siteinfo;
 170                 }
 171
 172                 // If it isn't a HTML file then exit
 173                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 174                         return $siteinfo;
 175                 }
 176
 177                 $header = $curlResult->getHeader();
 178                 $body = $curlResult->getBody();
 179
 180                 if ($do_oembed) {
 181                         $oembed_data = OEmbed::fetchURL($url);
 182
 183                         if (!empty($oembed_data->type)) {
 184                                 if (!in_array($oembed_data->type, ['error', 'rich', ''])) {
 185                                         $siteinfo['type'] = $oembed_data->type;
 186                                 }
 187
 188                                 // See https://github.com/friendica/friendica/pull/5763#discussion_r217913178
 189                                 if ($siteinfo['type'] != 'photo') {
 190                                         if (isset($oembed_data->title)) {
 191                                                 $siteinfo['title'] = trim($oembed_data->title);
 192                                         }
 193                                         if (isset($oembed_data->description)) {
 194                                                 $siteinfo['text'] = trim($oembed_data->description);
 195                                         }
 196                                         if (isset($oembed_data->thumbnail_url)) {
 197                                                 $siteinfo['image'] = $oembed_data->thumbnail_url;
 198                                         }
 199                                 }
 200                         }
 201                 }
 202
 203                 // Fetch the first mentioned charset. Can be in body or header
 204                 $charset = '';
 205                 if (preg_match('/charset=(.*?)[\'"\s\n]/', $header, $matches)) {
 206                         $charset = trim(trim(trim(array_pop($matches)), ';,'));
 207                 }
 208
 209                 if ($charset && strtoupper($charset) != 'UTF-8') {
 210                         // See https://github.com/friendica/friendica/issues/5470#issuecomment-418351211
 211                         $charset = str_ireplace('latin-1', 'latin1', $charset);
 212
 213                         Logger::log('detected charset ' . $charset, Logger::DEBUG);
 214                         $body = iconv($charset, 'UTF-8//TRANSLIT', $body);
 215                 }
 216
 217                 $body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
 218
 219                 $doc = new DOMDocument();
 220                 @$doc->loadHTML($body);
 221
 222                 XML::deleteNode($doc, 'style');
 223                 XML::deleteNode($doc, 'script');
 224                 XML::deleteNode($doc, 'option');
 225                 XML::deleteNode($doc, 'h1');
 226                 XML::deleteNode($doc, 'h2');
 227                 XML::deleteNode($doc, 'h3');
 228                 XML::deleteNode($doc, 'h4');
 229                 XML::deleteNode($doc, 'h5');
 230                 XML::deleteNode($doc, 'h6');
 231                 XML::deleteNode($doc, 'ol');
 232                 XML::deleteNode($doc, 'ul');
 233
 234                 $xpath = new DOMXPath($doc);
 235
 236                 $list = $xpath->query('//meta[@content]');
 237                 foreach ($list as $node) {
 238                         $meta_tag = [];
 239                         if ($node->attributes->length) {
 240                                 foreach ($node->attributes as $attribute) {
 241                                         $meta_tag[$attribute->name] = $attribute->value;
 242                                 }
 243                         }
 244
 245                         if (@$meta_tag['http-equiv'] == 'refresh') {
 246                                 $path = $meta_tag['content'];
 247                                 $pathinfo = explode(';', $path);
 248                                 $content = '';
 249                                 foreach ($pathinfo as $value) {
 250                                         if (substr(strtolower($value), 0, 4) == 'url=') {
 251                                                 $content = substr($value, 4);
 252                                         }
 253                                 }
 254                                 if ($content != '') {
 255                                         $siteinfo = self::getSiteinfo($content, $no_guessing, $do_oembed, ++$count);
 256                                         return $siteinfo;
 257                                 }
 258                         }
 259                 }
 260
 261                 $list = $xpath->query('//title');
 262                 if ($list->length > 0) {
 263                         $siteinfo['title'] = trim($list->item(0)->nodeValue);
 264                 }
 265
 266                 $list = $xpath->query('//meta[@name]');
 267                 foreach ($list as $node) {
 268                         $meta_tag = [];
 269                         if ($node->attributes->length) {
 270                                 foreach ($node->attributes as $attribute) {
 271                                         $meta_tag[$attribute->name] = $attribute->value;
 272                                 }
 273                         }
 274
 275                         if (empty($meta_tag['content'])) {
 276                                 continue;
 277                         }
 278
 279                         $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 280
 281                         switch (strtolower($meta_tag['name'])) {
 282                                 case 'fulltitle':
 283                                         $siteinfo['title'] = trim($meta_tag['content']);
 284                                         break;
 285                                 case 'description':
 286                                         $siteinfo['text'] = trim($meta_tag['content']);
 287                                         break;
 288                                 case 'thumbnail':
 289                                         $siteinfo['image'] = $meta_tag['content'];
 290                                         break;
 291                                 case 'twitter:image':
 292                                         $siteinfo['image'] = $meta_tag['content'];
 293                                         break;
 294                                 case 'twitter:image:src':
 295                                         $siteinfo['image'] = $meta_tag['content'];
 296                                         break;
 297                                 case 'twitter:card':
 298                                         // Detect photo pages
 299                                         if ($meta_tag['content'] == 'summary_large_image') {
 300                                                 $siteinfo['type'] = 'photo';
 301                                         }
 302                                         break;
 303                                 case 'twitter:description':
 304                                         $siteinfo['text'] = trim($meta_tag['content']);
 305                                         break;
 306                                 case 'twitter:title':
 307                                         $siteinfo['title'] = trim($meta_tag['content']);
 308                                         break;
 309                                 case 'dc.title':
 310                                         $siteinfo['title'] = trim($meta_tag['content']);
 311                                         break;
 312                                 case 'dc.description':
 313                                         $siteinfo['text'] = trim($meta_tag['content']);
 314                                         break;
 315                                 case 'keywords':
 316                                         $keywords = explode(',', $meta_tag['content']);
 317                                         break;
 318                                 case 'news_keywords':
 319                                         $keywords = explode(',', $meta_tag['content']);
 320                                         break;
 321                         }
 322                 }
 323
 324                 if (isset($keywords)) {
 325                         $siteinfo['keywords'] = [];
 326                         foreach ($keywords as $keyword) {
 327                                 if (!in_array(trim($keyword), $siteinfo['keywords'])) {
 328                                         $siteinfo['keywords'][] = trim($keyword);
 329                                 }
 330                         }
 331                 }
 332
 333                 $list = $xpath->query('//meta[@property]');
 334                 foreach ($list as $node) {
 335                         $meta_tag = [];
 336                         if ($node->attributes->length) {
 337                                 foreach ($node->attributes as $attribute) {
 338                                         $meta_tag[$attribute->name] = $attribute->value;
 339                                 }
 340                         }
 341
 342                         if (!empty($meta_tag['content'])) {
 343                                 $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 344
 345                                 switch (strtolower($meta_tag['property'])) {
 346                                         case 'og:image':
 347                                                 $siteinfo['image'] = $meta_tag['content'];
 348                                                 break;
 349                                         case 'og:title':
 350                                                 $siteinfo['title'] = trim($meta_tag['content']);
 351                                                 break;
 352                                         case 'og:description':
 353                                                 $siteinfo['text'] = trim($meta_tag['content']);
 354                                                 break;
 355                                 }
 356                         }
 357                 }
 358
 359                 // Prevent to have a photo type without an image
 360                 if ((empty($siteinfo['image']) || !empty($siteinfo['text'])) && ($siteinfo['type'] == 'photo')) {
 361                         $siteinfo['type'] = 'link';
 362                 }
 363
 364                 if (!empty($siteinfo['image'])) {
 365                         $src = self::completeUrl($siteinfo['image'], $url);
 366
 367                         unset($siteinfo['image']);
 368
 369                         $photodata = Images::getInfoFromURLCached($src);
 370
 371                         if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) {
 372                                 $siteinfo['images'][] = ['src' => $src,
 373                                         'width' => $photodata[0],
 374                                         'height' => $photodata[1]];
 375                         }
 376                 }
 377
 378                 if (!empty($siteinfo['text']) && mb_strlen($siteinfo['text']) > self::MAX_DESC_COUNT) {
 379                         $siteinfo['text'] = mb_substr($siteinfo['text'], 0, self::MAX_DESC_COUNT) . '…';
 380                         $pos = mb_strrpos($siteinfo['text'], '.');
 381                         if ($pos > self::MIN_DESC_COUNT) {
 382                                 $siteinfo['text'] = mb_substr($siteinfo['text'], 0, $pos + 1);
 383                         }
 384                 }
 385
 386                 Logger::info('Siteinfo fetched', ['url' => $url, 'siteinfo' => $siteinfo]);
 387
 388                 Hook::callAll('getsiteinfo', $siteinfo);
 389
 390                 return $siteinfo;
 391         }
 392
 393         /**
 394          * Convert tags from CSV to an array
 395          *
 396          * @param string $string Tags
 397          * @return array with formatted Hashtags
 398          */
 399         public static function convertTagsToArray($string)
 400         {
 401                 $arr_tags = str_getcsv($string);
 402                 if (count($arr_tags)) {
 403                         // add the # sign to every tag
 404                         array_walk($arr_tags, ["self", "arrAddHashes"]);
 405
 406                         return $arr_tags;
 407                 }
 408         }
 409
 410         /**
 411          * Add a hasht sign to a string
 412          *
 413          * This method is used as callback function
 414          *
 415          * @param string $tag The pure tag name
 416          * @param int    $k   Counter for internal use
 417          * @return void
 418          */
 419         private static function arrAddHashes(&$tag, $k)
 420         {
 421                 $tag = "#" . $tag;
 422         }
 423
 424         /**
 425          * Add a scheme to an url
 426          *
 427          * The src attribute of some html elements (e.g. images)
 428          * can miss the scheme so we need to add the correct
 429          * scheme
 430          *
 431          * @param string $url    The url which possibly does have
 432          *                       a missing scheme (a link to an image)
 433          * @param string $scheme The url with a correct scheme
 434          *                       (e.g. the url from the webpage which does contain the image)
 435          *
 436          * @return string The url with a scheme
 437          */
 438         private static function completeUrl($url, $scheme)
 439         {
 440                 $urlarr = parse_url($url);
 441
 442                 // If the url does allready have an scheme
 443                 // we can stop the process here
 444                 if (isset($urlarr["scheme"])) {
 445                         return($url);
 446                 }
 447
 448                 $schemearr = parse_url($scheme);
 449
 450                 $complete = $schemearr["scheme"]."://".$schemearr["host"];
 451
 452                 if (!empty($schemearr["port"])) {
 453                         $complete .= ":".$schemearr["port"];
 454                 }
 455
 456                 if (!empty($urlarr["path"])) {
 457                         if (strpos($urlarr["path"], "/") !== 0) {
 458                                 $complete .= "/";
 459                         }
 460
 461                         $complete .= $urlarr["path"];
 462                 }
 463
 464                 if (!empty($urlarr["query"])) {
 465                         $complete .= "?".$urlarr["query"];
 466                 }
 467
 468                 if (!empty($urlarr["fragment"])) {
 469                         $complete .= "#".$urlarr["fragment"];
 470                 }
 471
 472                 return($complete);
 473         }
 474 }