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\Database;
  30 use Friendica\Database\DBA;
  31 use Friendica\DI;
  32 use Friendica\Network\HTTPException;
  33
  34 /**
  35  * Get information about a given URL
  36  *
  37  * Class with methods for extracting certain content from an url
  38  */
  39 class ParseUrl
  40 {
  41         const DEFAULT_EXPIRATION_FAILURE = 'now + 1 day';
  42         const DEFAULT_EXPIRATION_SUCCESS = 'now + 3 months';
  43
  44         /**
  45          * Maximum number of characters for the description
  46          */
  47         const MAX_DESC_COUNT = 250;
  48
  49         /**
  50          * Minimum number of characters for the description
  51          */
  52         const MIN_DESC_COUNT = 100;
  53
  54         /**
  55          * Search for chached embeddable data of an url otherwise fetch it
  56          *
  57          * @param string $url         The url of the page which should be scraped
  58          * @param bool   $no_guessing If true the parse doens't search for
  59          *                            preview pictures
  60          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
  61          *                            to avoid endless loops
  62          *
  63          * @return array which contains needed data for embedding
  64          *    string 'url'      => The url of the parsed page
  65          *    string 'type'     => Content type
  66          *    string 'title'    => (optional) The title of the content
  67          *    string 'text'     => (optional) The description for the content
  68          *    string 'image'    => (optional) A preview image of the content (only available if $no_geuessing = false)
  69          *    array  'images'   => (optional) Array of preview pictures
  70          *    string 'keywords' => (optional) The tags which belong to the content
  71          *
  72          * @throws HTTPException\InternalServerErrorException
  73          * @see   ParseUrl::getSiteinfo() for more information about scraping
  74          * embeddable content
  75          */
  76         public static function getSiteinfoCached($url, $no_guessing = false, $do_oembed = true): array
  77         {
  78                 if (empty($url)) {
  79                         return [
  80                                 'url' => '',
  81                                 'type' => 'error',
  82                         ];
  83                 }
  84
  85                 $urlHash = hash('sha256', $url);
  86
  87                 $parsed_url = DBA::selectFirst('parsed_url', ['content'],
  88                         ['url_hash' => $urlHash, 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
  89                 );
  90                 if (!empty($parsed_url['content'])) {
  91                         $data = unserialize($parsed_url['content']);
  92                         return $data;
  93                 }
  94
  95                 $data = self::getSiteinfo($url, $no_guessing, $do_oembed);
  96
  97                 $expires = $data['expires'];
  98
  99                 unset($data['expires']);
 100
 101                 DI::dba()->insert(
 102                         'parsed_url',
 103                         [
 104                                 'url_hash' => $urlHash,
 105                                 'guessing' => !$no_guessing,
 106                                 'oembed'   => $do_oembed,
 107                                 'url'      => $url,
 108                                 'content'  => serialize($data),
 109                                 'created'  => DateTimeFormat::utcNow(),
 110                                 'expires'  => $expires,
 111                         ],
 112                         Database::INSERT_UPDATE
 113                 );
 114
 115                 return $data;
 116         }
 117
 118         /**
 119          * Parse a page for embeddable content information
 120          *
 121          * This method parses to url for meta data which can be used to embed
 122          * the content. If available it prioritizes Open Graph meta tags.
 123          * If this is not available it uses the twitter cards meta tags.
 124          * As fallback it uses standard html elements with meta informations
 125          * like \<title\>Awesome Title\</title\> or
 126          * \<meta name="description" content="An awesome description"\>
 127          *
 128          * @param string $url         The url of the page which should be scraped
 129          * @param bool   $no_guessing If true the parse doens't search for
 130          *                            preview pictures
 131          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
 132          *                            to avoid endless loops
 133          * @param int    $count       Internal counter to avoid endless loops
 134          *
 135          * @return array which contains needed data for embedding
 136          *    string 'url'      => The url of the parsed page
 137          *    string 'type'     => Content type (error, link, photo, image, audio, video)
 138          *    string 'title'    => (optional) The title of the content
 139          *    string 'text'     => (optional) The description for the content
 140          *    string 'image'    => (optional) A preview image of the content (only available if $no_guessing = false)
 141          *    array  'images'   => (optional) Array of preview pictures
 142          *    string 'keywords' => (optional) The tags which belong to the content
 143          *
 144          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 145          * @todo  https://developers.google.com/+/plugins/snippet/
 146          * @verbatim
 147          * <meta itemprop="name" content="Awesome title">
 148          * <meta itemprop="description" content="An awesome description">
 149          * <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
 150          *
 151          * <body itemscope itemtype="http://schema.org/Product">
 152          *   <h1 itemprop="name">Shiny Trinket</h1>
 153          *   <img itemprop="image" src="{image-url}" />
 154          *   <p itemprop="description">Shiny trinkets are shiny.</p>
 155          * </body>
 156          * @endverbatim
 157          */
 158         public static function getSiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1)
 159         {
 160                 if (empty($url)) {
 161                         return [
 162                                 'url' => '',
 163                                 'type' => 'error',
 164                         ];
 165                 }
 166
 167                 // Check if the URL does contain a scheme
 168                 $scheme = parse_url($url, PHP_URL_SCHEME);
 169
 170                 if ($scheme == '') {
 171                         $url = 'http://' . ltrim($url, '/');
 172                 }
 173
 174                 $url = trim($url, "'\"");
 175
 176                 $url = Network::stripTrackingQueryParams($url);
 177
 178                 $siteinfo = [
 179                         'url' => $url,
 180                         'type' => 'link',
 181                         'expires' => DateTimeFormat::utc(self::DEFAULT_EXPIRATION_FAILURE),
 182                 ];
 183
 184                 if ($count > 10) {
 185                         Logger::log('Endless loop detected for ' . $url, Logger::DEBUG);
 186                         return $siteinfo;
 187                 }
 188
 189                 $curlResult = DI::httpRequest()->get($url);
 190                 if (!$curlResult->isSuccess()) {
 191                         return $siteinfo;
 192                 }
 193
 194                 $siteinfo['expires'] = DateTimeFormat::utc(self::DEFAULT_EXPIRATION_SUCCESS);
 195
 196                 // If the file is too large then exit
 197                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 198                         return $siteinfo;
 199                 }
 200
 201                 // Native media type, no need for HTML parsing
 202                 $type = $curlResult->getHeader('Content-Type');
 203                 if ($type) {
 204                         preg_match('#(image|video|audio)/#i', $type, $matches);
 205                         if ($matches) {
 206                                 $siteinfo['type'] = array_pop($matches);
 207                                 return $siteinfo;
 208                         }
 209                 }
 210
 211                 // If it isn't a HTML file then exit
 212                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 213                         return $siteinfo;
 214                 }
 215
 216                 if ($cacheControlHeader = $curlResult->getHeader('Cache-Control')) {
 217                         if (preg_match('/max-age=([0-9]+)/i', $cacheControlHeader, $matches)) {
 218                                 $maxAge = max(86400, (int)array_pop($matches));
 219                                 $siteinfo['expires'] = DateTimeFormat::utc("now + $maxAge seconds");
 220                         }
 221                 }
 222
 223                 $header = $curlResult->getHeader();
 224                 $body = $curlResult->getBody();
 225
 226                 if ($do_oembed) {
 227                         $oembed_data = OEmbed::fetchURL($url);
 228
 229                         if (!empty($oembed_data->type)) {
 230                                 if (!in_array($oembed_data->type, ['error', 'rich', ''])) {
 231                                         $siteinfo['type'] = $oembed_data->type;
 232                                 }
 233
 234                                 // See https://github.com/friendica/friendica/pull/5763#discussion_r217913178
 235                                 if ($siteinfo['type'] != 'photo') {
 236                                         if (isset($oembed_data->title)) {
 237                                                 $siteinfo['title'] = trim($oembed_data->title);
 238                                         }
 239                                         if (isset($oembed_data->description)) {
 240                                                 $siteinfo['text'] = trim($oembed_data->description);
 241                                         }
 242                                         if (isset($oembed_data->thumbnail_url)) {
 243                                                 $siteinfo['image'] = $oembed_data->thumbnail_url;
 244                                         }
 245                                 }
 246                         }
 247                 }
 248
 249                 $charset = '';
 250                 // Look for a charset, first in headers
 251                 // Expected form: Content-Type: text/html; charset=ISO-8859-4
 252                 if (preg_match('/charset=([a-z0-9-_.\/]+)/i', $header, $matches)) {
 253                         $charset = trim(trim(trim(array_pop($matches)), ';,'));
 254                 }
 255
 256                 // Then in body that gets precedence
 257                 // Expected forms:
 258                 // - <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
 259                 // - <meta charset="utf-8">
 260                 // - <meta charset=utf-8>
 261                 // - <meta charSet="utf-8">
 262                 // We escape <style> and <script> tags since they can contain irrelevant charset information
 263                 // (see https://github.com/friendica/friendica/issues/9251#issuecomment-698636806)
 264                 Strings::performWithEscapedBlocks($body, '#<(?:style|script).*?</(?:style|script)>#ism', function ($body) use (&$charset) {
 265                         if (preg_match('/charset=["\']?([a-z0-9-_.\/]+)/i', $body, $matches)) {
 266                                 $charset = trim(trim(trim(array_pop($matches)), ';,'));
 267                         }
 268                 });
 269
 270                 $siteinfo['charset'] = $charset;
 271
 272                 if ($charset && strtoupper($charset) != 'UTF-8') {
 273                         // See https://github.com/friendica/friendica/issues/5470#issuecomment-418351211
 274                         $charset = str_ireplace('latin-1', 'latin1', $charset);
 275
 276                         Logger::log('detected charset ' . $charset, Logger::DEBUG);
 277                         $body = iconv($charset, 'UTF-8//TRANSLIT', $body);
 278                 }
 279
 280                 $body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
 281
 282                 $doc = new DOMDocument();
 283                 @$doc->loadHTML($body);
 284
 285                 XML::deleteNode($doc, 'style');
 286                 XML::deleteNode($doc, 'script');
 287                 XML::deleteNode($doc, 'option');
 288                 XML::deleteNode($doc, 'h1');
 289                 XML::deleteNode($doc, 'h2');
 290                 XML::deleteNode($doc, 'h3');
 291                 XML::deleteNode($doc, 'h4');
 292                 XML::deleteNode($doc, 'h5');
 293                 XML::deleteNode($doc, 'h6');
 294                 XML::deleteNode($doc, 'ol');
 295                 XML::deleteNode($doc, 'ul');
 296
 297                 $xpath = new DOMXPath($doc);
 298
 299                 $list = $xpath->query('//meta[@content]');
 300                 foreach ($list as $node) {
 301                         $meta_tag = [];
 302                         if ($node->attributes->length) {
 303                                 foreach ($node->attributes as $attribute) {
 304                                         $meta_tag[$attribute->name] = $attribute->value;
 305                                 }
 306                         }
 307
 308                         if (@$meta_tag['http-equiv'] == 'refresh') {
 309                                 $path = $meta_tag['content'];
 310                                 $pathinfo = explode(';', $path);
 311                                 $content = '';
 312                                 foreach ($pathinfo as $value) {
 313                                         if (substr(strtolower($value), 0, 4) == 'url=') {
 314                                                 $content = substr($value, 4);
 315                                         }
 316                                 }
 317                                 if ($content != '') {
 318                                         $siteinfo = self::getSiteinfo($content, $no_guessing, $do_oembed, ++$count);
 319                                         return $siteinfo;
 320                                 }
 321                         }
 322                 }
 323
 324                 $list = $xpath->query('//title');
 325                 if ($list->length > 0) {
 326                         $siteinfo['title'] = trim($list->item(0)->nodeValue);
 327                 }
 328
 329                 $list = $xpath->query('//meta[@name]');
 330                 foreach ($list as $node) {
 331                         $meta_tag = [];
 332                         if ($node->attributes->length) {
 333                                 foreach ($node->attributes as $attribute) {
 334                                         $meta_tag[$attribute->name] = $attribute->value;
 335                                 }
 336                         }
 337
 338                         if (empty($meta_tag['content'])) {
 339                                 continue;
 340                         }
 341
 342                         $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 343
 344                         switch (strtolower($meta_tag['name'])) {
 345                                 case 'fulltitle':
 346                                         $siteinfo['title'] = trim($meta_tag['content']);
 347                                         break;
 348                                 case 'description':
 349                                         $siteinfo['text'] = trim($meta_tag['content']);
 350                                         break;
 351                                 case 'thumbnail':
 352                                         $siteinfo['image'] = $meta_tag['content'];
 353                                         break;
 354                                 case 'twitter:image':
 355                                         $siteinfo['image'] = $meta_tag['content'];
 356                                         break;
 357                                 case 'twitter:image:src':
 358                                         $siteinfo['image'] = $meta_tag['content'];
 359                                         break;
 360                                 case 'twitter:card':
 361                                         // Detect photo pages
 362                                         if ($meta_tag['content'] == 'summary_large_image') {
 363                                                 $siteinfo['type'] = 'photo';
 364                                         }
 365                                         break;
 366                                 case 'twitter:description':
 367                                         $siteinfo['text'] = trim($meta_tag['content']);
 368                                         break;
 369                                 case 'twitter:title':
 370                                         $siteinfo['title'] = trim($meta_tag['content']);
 371                                         break;
 372                                 case 'dc.title':
 373                                         $siteinfo['title'] = trim($meta_tag['content']);
 374                                         break;
 375                                 case 'dc.description':
 376                                         $siteinfo['text'] = trim($meta_tag['content']);
 377                                         break;
 378                                 case 'keywords':
 379                                         $keywords = explode(',', $meta_tag['content']);
 380                                         break;
 381                                 case 'news_keywords':
 382                                         $keywords = explode(',', $meta_tag['content']);
 383                                         break;
 384                         }
 385                 }
 386
 387                 if (isset($keywords)) {
 388                         $siteinfo['keywords'] = [];
 389                         foreach ($keywords as $keyword) {
 390                                 if (!in_array(trim($keyword), $siteinfo['keywords'])) {
 391                                         $siteinfo['keywords'][] = trim($keyword);
 392                                 }
 393                         }
 394                 }
 395
 396                 $list = $xpath->query('//meta[@property]');
 397                 foreach ($list as $node) {
 398                         $meta_tag = [];
 399                         if ($node->attributes->length) {
 400                                 foreach ($node->attributes as $attribute) {
 401                                         $meta_tag[$attribute->name] = $attribute->value;
 402                                 }
 403                         }
 404
 405                         if (!empty($meta_tag['content'])) {
 406                                 $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 407
 408                                 switch (strtolower($meta_tag['property'])) {
 409                                         case 'og:image':
 410                                                 $siteinfo['image'] = $meta_tag['content'];
 411                                                 break;
 412                                         case 'og:title':
 413                                                 $siteinfo['title'] = trim($meta_tag['content']);
 414                                                 break;
 415                                         case 'og:description':
 416                                                 $siteinfo['text'] = trim($meta_tag['content']);
 417                                                 break;
 418                                 }
 419                         }
 420                 }
 421
 422                 // Prevent to have a photo type without an image
 423                 if ((empty($siteinfo['image']) || !empty($siteinfo['text'])) && ($siteinfo['type'] == 'photo')) {
 424                         $siteinfo['type'] = 'link';
 425                 }
 426
 427                 if (!empty($siteinfo['image'])) {
 428                         $src = self::completeUrl($siteinfo['image'], $url);
 429
 430                         unset($siteinfo['image']);
 431
 432                         $photodata = Images::getInfoFromURLCached($src);
 433
 434                         if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) {
 435                                 $siteinfo['images'][] = ['src' => $src,
 436                                         'width' => $photodata[0],
 437                                         'height' => $photodata[1]];
 438                         }
 439                 }
 440
 441                 if (!empty($siteinfo['text']) && mb_strlen($siteinfo['text']) > self::MAX_DESC_COUNT) {
 442                         $siteinfo['text'] = mb_substr($siteinfo['text'], 0, self::MAX_DESC_COUNT) . '…';
 443                         $pos = mb_strrpos($siteinfo['text'], '.');
 444                         if ($pos > self::MIN_DESC_COUNT) {
 445                                 $siteinfo['text'] = mb_substr($siteinfo['text'], 0, $pos + 1);
 446                         }
 447                 }
 448
 449                 Logger::info('Siteinfo fetched', ['url' => $url, 'siteinfo' => $siteinfo]);
 450
 451                 Hook::callAll('getsiteinfo', $siteinfo);
 452
 453                 return $siteinfo;
 454         }
 455
 456         /**
 457          * Convert tags from CSV to an array
 458          *
 459          * @param string $string Tags
 460          * @return array with formatted Hashtags
 461          */
 462         public static function convertTagsToArray($string)
 463         {
 464                 $arr_tags = str_getcsv($string);
 465                 if (count($arr_tags)) {
 466                         // add the # sign to every tag
 467                         array_walk($arr_tags, ["self", "arrAddHashes"]);
 468
 469                         return $arr_tags;
 470                 }
 471         }
 472
 473         /**
 474          * Add a hasht sign to a string
 475          *
 476          * This method is used as callback function
 477          *
 478          * @param string $tag The pure tag name
 479          * @param int    $k   Counter for internal use
 480          * @return void
 481          */
 482         private static function arrAddHashes(&$tag, $k)
 483         {
 484                 $tag = "#" . $tag;
 485         }
 486
 487         /**
 488          * Add a scheme to an url
 489          *
 490          * The src attribute of some html elements (e.g. images)
 491          * can miss the scheme so we need to add the correct
 492          * scheme
 493          *
 494          * @param string $url    The url which possibly does have
 495          *                       a missing scheme (a link to an image)
 496          * @param string $scheme The url with a correct scheme
 497          *                       (e.g. the url from the webpage which does contain the image)
 498          *
 499          * @return string The url with a scheme
 500          */
 501         private static function completeUrl($url, $scheme)
 502         {
 503                 $urlarr = parse_url($url);
 504
 505                 // If the url does allready have an scheme
 506                 // we can stop the process here
 507                 if (isset($urlarr["scheme"])) {
 508                         return($url);
 509                 }
 510
 511                 $schemearr = parse_url($scheme);
 512
 513                 $complete = $schemearr["scheme"]."://".$schemearr["host"];
 514
 515                 if (!empty($schemearr["port"])) {
 516                         $complete .= ":".$schemearr["port"];
 517                 }
 518
 519                 if (!empty($urlarr["path"])) {
 520                         if (strpos($urlarr["path"], "/") !== 0) {
 521                                 $complete .= "/";
 522                         }
 523
 524                         $complete .= $urlarr["path"];
 525                 }
 526
 527                 if (!empty($urlarr["query"])) {
 528                         $complete .= "?".$urlarr["query"];
 529                 }
 530
 531                 if (!empty($urlarr["fragment"])) {
 532                         $complete .= "#".$urlarr["fragment"];
 533                 }
 534
 535                 return($complete);
 536         }
 537 }