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          * Fetch the content type of the given url
  56          * @param string $url URL of the page
  57          * @return string content type
  58          */
  59         public static function getContentType(string $url)
  60         {
  61                 $curlResult = DI::httpRequest()->head($url);
  62                 if (!$curlResult->isSuccess()) {
  63                         return '';
  64                 }
  65
  66                 $contenttype =  $curlResult->getHeader('Content-Type');
  67                 if (empty($contenttype)) {
  68                         return '';
  69                 }
  70
  71                 if (!preg_match('#(image|video|audio)/#i', $contenttype, $matches)) {
  72                         return '';
  73                 }
  74
  75                 return array_pop($matches);
  76         }
  77
  78         /**
  79          * Search for chached embeddable data of an url otherwise fetch it
  80          *
  81          * @param string $url         The url of the page which should be scraped
  82          * @param bool   $no_guessing If true the parse doens't search for
  83          *                            preview pictures
  84          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
  85          *                            to avoid endless loops
  86          *
  87          * @return array which contains needed data for embedding
  88          *    string 'url'      => The url of the parsed page
  89          *    string 'type'     => Content type
  90          *    string 'title'    => (optional) The title of the content
  91          *    string 'text'     => (optional) The description for the content
  92          *    string 'image'    => (optional) A preview image of the content (only available if $no_geuessing = false)
  93          *    array  'images'   => (optional) Array of preview pictures
  94          *    string 'keywords' => (optional) The tags which belong to the content
  95          *
  96          * @throws HTTPException\InternalServerErrorException
  97          * @see   ParseUrl::getSiteinfo() for more information about scraping
  98          * embeddable content
  99          */
 100         public static function getSiteinfoCached($url, $no_guessing = false, $do_oembed = true): array
 101         {
 102                 if (empty($url)) {
 103                         return [
 104                                 'url' => '',
 105                                 'type' => 'error',
 106                         ];
 107                 }
 108
 109                 $urlHash = hash('sha256', $url);
 110
 111                 $parsed_url = DBA::selectFirst('parsed_url', ['content'],
 112                         ['url_hash' => $urlHash, 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
 113                 );
 114                 if (!empty($parsed_url['content'])) {
 115                         $data = unserialize($parsed_url['content']);
 116                         return $data;
 117                 }
 118
 119                 $data = self::getSiteinfo($url, $no_guessing, $do_oembed);
 120
 121                 $expires = $data['expires'];
 122
 123                 unset($data['expires']);
 124
 125                 DI::dba()->insert(
 126                         'parsed_url',
 127                         [
 128                                 'url_hash' => $urlHash,
 129                                 'guessing' => !$no_guessing,
 130                                 'oembed'   => $do_oembed,
 131                                 'url'      => $url,
 132                                 'content'  => serialize($data),
 133                                 'created'  => DateTimeFormat::utcNow(),
 134                                 'expires'  => $expires,
 135                         ],
 136                         Database::INSERT_UPDATE
 137                 );
 138
 139                 return $data;
 140         }
 141
 142         /**
 143          * Parse a page for embeddable content information
 144          *
 145          * This method parses to url for meta data which can be used to embed
 146          * the content. If available it prioritizes Open Graph meta tags.
 147          * If this is not available it uses the twitter cards meta tags.
 148          * As fallback it uses standard html elements with meta informations
 149          * like \<title\>Awesome Title\</title\> or
 150          * \<meta name="description" content="An awesome description"\>
 151          *
 152          * @param string $url         The url of the page which should be scraped
 153          * @param bool   $no_guessing If true the parse doens't search for
 154          *                            preview pictures
 155          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
 156          *                            to avoid endless loops
 157          * @param int    $count       Internal counter to avoid endless loops
 158          *
 159          * @return array which contains needed data for embedding
 160          *    string 'url'      => The url of the parsed page
 161          *    string 'type'     => Content type (error, link, photo, image, audio, video)
 162          *    string 'title'    => (optional) The title of the content
 163          *    string 'text'     => (optional) The description for the content
 164          *    string 'image'    => (optional) A preview image of the content (only available if $no_guessing = false)
 165          *    array  'images'   => (optional) Array of preview pictures
 166          *    string 'keywords' => (optional) The tags which belong to the content
 167          *
 168          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 169          * @todo  https://developers.google.com/+/plugins/snippet/
 170          * @verbatim
 171          * <meta itemprop="name" content="Awesome title">
 172          * <meta itemprop="description" content="An awesome description">
 173          * <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
 174          *
 175          * <body itemscope itemtype="http://schema.org/Product">
 176          *   <h1 itemprop="name">Shiny Trinket</h1>
 177          *   <img itemprop="image" src="{image-url}" />
 178          *   <p itemprop="description">Shiny trinkets are shiny.</p>
 179          * </body>
 180          * @endverbatim
 181          */
 182         public static function getSiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1)
 183         {
 184                 if (empty($url)) {
 185                         return [
 186                                 'url' => '',
 187                                 'type' => 'error',
 188                         ];
 189                 }
 190
 191                 // Check if the URL does contain a scheme
 192                 $scheme = parse_url($url, PHP_URL_SCHEME);
 193
 194                 if ($scheme == '') {
 195                         $url = 'http://' . ltrim($url, '/');
 196                 }
 197
 198                 $url = trim($url, "'\"");
 199
 200                 $url = Network::stripTrackingQueryParams($url);
 201
 202                 $siteinfo = [
 203                         'url' => $url,
 204                         'type' => 'link',
 205                         'expires' => DateTimeFormat::utc(self::DEFAULT_EXPIRATION_FAILURE),
 206                 ];
 207
 208                 if ($count > 10) {
 209                         Logger::log('Endless loop detected for ' . $url, Logger::DEBUG);
 210                         return $siteinfo;
 211                 }
 212
 213                 $curlResult = DI::httpRequest()->get($url);
 214                 if (!$curlResult->isSuccess()) {
 215                         return $siteinfo;
 216                 }
 217
 218                 $siteinfo['expires'] = DateTimeFormat::utc(self::DEFAULT_EXPIRATION_SUCCESS);
 219
 220                 // If the file is too large then exit
 221                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 222                         return $siteinfo;
 223                 }
 224
 225                 // Native media type, no need for HTML parsing
 226                 $type = $curlResult->getHeader('Content-Type');
 227                 if ($type) {
 228                         preg_match('#(image|video|audio)/#i', $type, $matches);
 229                         if ($matches) {
 230                                 $siteinfo['type'] = array_pop($matches);
 231                                 return $siteinfo;
 232                         }
 233                 }
 234
 235                 // If it isn't a HTML file then exit
 236                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 237                         return $siteinfo;
 238                 }
 239
 240                 if ($cacheControlHeader = $curlResult->getHeader('Cache-Control')) {
 241                         if (preg_match('/max-age=([0-9]+)/i', $cacheControlHeader, $matches)) {
 242                                 $maxAge = max(86400, (int)array_pop($matches));
 243                                 $siteinfo['expires'] = DateTimeFormat::utc("now + $maxAge seconds");
 244                         }
 245                 }
 246
 247                 $header = $curlResult->getHeader();
 248                 $body = $curlResult->getBody();
 249
 250                 if ($do_oembed) {
 251                         $oembed_data = OEmbed::fetchURL($url);
 252
 253                         if (!empty($oembed_data->type)) {
 254                                 if (!in_array($oembed_data->type, ['error', 'rich', ''])) {
 255                                         $siteinfo['type'] = $oembed_data->type;
 256                                 }
 257
 258                                 // See https://github.com/friendica/friendica/pull/5763#discussion_r217913178
 259                                 if ($siteinfo['type'] != 'photo') {
 260                                         if (isset($oembed_data->title)) {
 261                                                 $siteinfo['title'] = trim($oembed_data->title);
 262                                         }
 263                                         if (isset($oembed_data->description)) {
 264                                                 $siteinfo['text'] = trim($oembed_data->description);
 265                                         }
 266                                         if (isset($oembed_data->thumbnail_url)) {
 267                                                 $siteinfo['image'] = $oembed_data->thumbnail_url;
 268                                         }
 269                                 }
 270                         }
 271                 }
 272
 273                 $charset = '';
 274                 // Look for a charset, first in headers
 275                 // Expected form: Content-Type: text/html; charset=ISO-8859-4
 276                 if (preg_match('/charset=([a-z0-9-_.\/]+)/i', $header, $matches)) {
 277                         $charset = trim(trim(trim(array_pop($matches)), ';,'));
 278                 }
 279
 280                 // Then in body that gets precedence
 281                 // Expected forms:
 282                 // - <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
 283                 // - <meta charset="utf-8">
 284                 // - <meta charset=utf-8>
 285                 // - <meta charSet="utf-8">
 286                 // We escape <style> and <script> tags since they can contain irrelevant charset information
 287                 // (see https://github.com/friendica/friendica/issues/9251#issuecomment-698636806)
 288                 Strings::performWithEscapedBlocks($body, '#<(?:style|script).*?</(?:style|script)>#ism', function ($body) use (&$charset) {
 289                         if (preg_match('/charset=["\']?([a-z0-9-_.\/]+)/i', $body, $matches)) {
 290                                 $charset = trim(trim(trim(array_pop($matches)), ';,'));
 291                         }
 292                 });
 293
 294                 $siteinfo['charset'] = $charset;
 295
 296                 if ($charset && strtoupper($charset) != 'UTF-8') {
 297                         // See https://github.com/friendica/friendica/issues/5470#issuecomment-418351211
 298                         $charset = str_ireplace('latin-1', 'latin1', $charset);
 299
 300                         Logger::log('detected charset ' . $charset, Logger::DEBUG);
 301                         $body = iconv($charset, 'UTF-8//TRANSLIT', $body);
 302                 }
 303
 304                 $body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
 305
 306                 $doc = new DOMDocument();
 307                 @$doc->loadHTML($body);
 308
 309                 XML::deleteNode($doc, 'style');
 310                 XML::deleteNode($doc, 'script');
 311                 XML::deleteNode($doc, 'option');
 312                 XML::deleteNode($doc, 'h1');
 313                 XML::deleteNode($doc, 'h2');
 314                 XML::deleteNode($doc, 'h3');
 315                 XML::deleteNode($doc, 'h4');
 316                 XML::deleteNode($doc, 'h5');
 317                 XML::deleteNode($doc, 'h6');
 318                 XML::deleteNode($doc, 'ol');
 319                 XML::deleteNode($doc, 'ul');
 320
 321                 $xpath = new DOMXPath($doc);
 322
 323                 $list = $xpath->query('//meta[@content]');
 324                 foreach ($list as $node) {
 325                         $meta_tag = [];
 326                         if ($node->attributes->length) {
 327                                 foreach ($node->attributes as $attribute) {
 328                                         $meta_tag[$attribute->name] = $attribute->value;
 329                                 }
 330                         }
 331
 332                         if (@$meta_tag['http-equiv'] == 'refresh') {
 333                                 $path = $meta_tag['content'];
 334                                 $pathinfo = explode(';', $path);
 335                                 $content = '';
 336                                 foreach ($pathinfo as $value) {
 337                                         if (substr(strtolower($value), 0, 4) == 'url=') {
 338                                                 $content = substr($value, 4);
 339                                         }
 340                                 }
 341                                 if ($content != '') {
 342                                         $siteinfo = self::getSiteinfo($content, $no_guessing, $do_oembed, ++$count);
 343                                         return $siteinfo;
 344                                 }
 345                         }
 346                 }
 347
 348                 $list = $xpath->query('//title');
 349                 if ($list->length > 0) {
 350                         $siteinfo['title'] = trim($list->item(0)->nodeValue);
 351                 }
 352
 353                 $list = $xpath->query('//meta[@name]');
 354                 foreach ($list as $node) {
 355                         $meta_tag = [];
 356                         if ($node->attributes->length) {
 357                                 foreach ($node->attributes as $attribute) {
 358                                         $meta_tag[$attribute->name] = $attribute->value;
 359                                 }
 360                         }
 361
 362                         if (empty($meta_tag['content'])) {
 363                                 continue;
 364                         }
 365
 366                         $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 367
 368                         switch (strtolower($meta_tag['name'])) {
 369                                 case 'fulltitle':
 370                                         $siteinfo['title'] = trim($meta_tag['content']);
 371                                         break;
 372                                 case 'description':
 373                                         $siteinfo['text'] = trim($meta_tag['content']);
 374                                         break;
 375                                 case 'thumbnail':
 376                                         $siteinfo['image'] = $meta_tag['content'];
 377                                         break;
 378                                 case 'twitter:image':
 379                                         $siteinfo['image'] = $meta_tag['content'];
 380                                         break;
 381                                 case 'twitter:image:src':
 382                                         $siteinfo['image'] = $meta_tag['content'];
 383                                         break;
 384                                 case 'twitter:card':
 385                                         // Detect photo pages
 386                                         if ($meta_tag['content'] == 'summary_large_image') {
 387                                                 $siteinfo['type'] = 'photo';
 388                                         }
 389                                         break;
 390                                 case 'twitter:description':
 391                                         $siteinfo['text'] = trim($meta_tag['content']);
 392                                         break;
 393                                 case 'twitter:title':
 394                                         $siteinfo['title'] = trim($meta_tag['content']);
 395                                         break;
 396                                 case 'dc.title':
 397                                         $siteinfo['title'] = trim($meta_tag['content']);
 398                                         break;
 399                                 case 'dc.description':
 400                                         $siteinfo['text'] = trim($meta_tag['content']);
 401                                         break;
 402                                 case 'keywords':
 403                                         $keywords = explode(',', $meta_tag['content']);
 404                                         break;
 405                                 case 'news_keywords':
 406                                         $keywords = explode(',', $meta_tag['content']);
 407                                         break;
 408                         }
 409                 }
 410
 411                 if (isset($keywords)) {
 412                         $siteinfo['keywords'] = [];
 413                         foreach ($keywords as $keyword) {
 414                                 if (!in_array(trim($keyword), $siteinfo['keywords'])) {
 415                                         $siteinfo['keywords'][] = trim($keyword);
 416                                 }
 417                         }
 418                 }
 419
 420                 $list = $xpath->query('//meta[@property]');
 421                 foreach ($list as $node) {
 422                         $meta_tag = [];
 423                         if ($node->attributes->length) {
 424                                 foreach ($node->attributes as $attribute) {
 425                                         $meta_tag[$attribute->name] = $attribute->value;
 426                                 }
 427                         }
 428
 429                         if (!empty($meta_tag['content'])) {
 430                                 $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 431
 432                                 switch (strtolower($meta_tag['property'])) {
 433                                         case 'og:image':
 434                                                 $siteinfo['image'] = $meta_tag['content'];
 435                                                 break;
 436                                         case 'og:title':
 437                                                 $siteinfo['title'] = trim($meta_tag['content']);
 438                                                 break;
 439                                         case 'og:description':
 440                                                 $siteinfo['text'] = trim($meta_tag['content']);
 441                                                 break;
 442                                 }
 443                         }
 444                 }
 445
 446                 // Prevent to have a photo type without an image
 447                 if ((empty($siteinfo['image']) || !empty($siteinfo['text'])) && ($siteinfo['type'] == 'photo')) {
 448                         $siteinfo['type'] = 'link';
 449                 }
 450
 451                 if (!empty($siteinfo['image'])) {
 452                         $src = self::completeUrl($siteinfo['image'], $url);
 453
 454                         unset($siteinfo['image']);
 455
 456                         $photodata = Images::getInfoFromURLCached($src);
 457
 458                         if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) {
 459                                 $siteinfo['images'][] = ['src' => $src,
 460                                         'width' => $photodata[0],
 461                                         'height' => $photodata[1]];
 462                         }
 463                 }
 464
 465                 if (!empty($siteinfo['text']) && mb_strlen($siteinfo['text']) > self::MAX_DESC_COUNT) {
 466                         $siteinfo['text'] = mb_substr($siteinfo['text'], 0, self::MAX_DESC_COUNT) . '…';
 467                         $pos = mb_strrpos($siteinfo['text'], '.');
 468                         if ($pos > self::MIN_DESC_COUNT) {
 469                                 $siteinfo['text'] = mb_substr($siteinfo['text'], 0, $pos + 1);
 470                         }
 471                 }
 472
 473                 Logger::info('Siteinfo fetched', ['url' => $url, 'siteinfo' => $siteinfo]);
 474
 475                 Hook::callAll('getsiteinfo', $siteinfo);
 476
 477                 return $siteinfo;
 478         }
 479
 480         /**
 481          * Convert tags from CSV to an array
 482          *
 483          * @param string $string Tags
 484          * @return array with formatted Hashtags
 485          */
 486         public static function convertTagsToArray($string)
 487         {
 488                 $arr_tags = str_getcsv($string);
 489                 if (count($arr_tags)) {
 490                         // add the # sign to every tag
 491                         array_walk($arr_tags, ["self", "arrAddHashes"]);
 492
 493                         return $arr_tags;
 494                 }
 495         }
 496
 497         /**
 498          * Add a hasht sign to a string
 499          *
 500          * This method is used as callback function
 501          *
 502          * @param string $tag The pure tag name
 503          * @param int    $k   Counter for internal use
 504          * @return void
 505          */
 506         private static function arrAddHashes(&$tag, $k)
 507         {
 508                 $tag = "#" . $tag;
 509         }
 510
 511         /**
 512          * Add a scheme to an url
 513          *
 514          * The src attribute of some html elements (e.g. images)
 515          * can miss the scheme so we need to add the correct
 516          * scheme
 517          *
 518          * @param string $url    The url which possibly does have
 519          *                       a missing scheme (a link to an image)
 520          * @param string $scheme The url with a correct scheme
 521          *                       (e.g. the url from the webpage which does contain the image)
 522          *
 523          * @return string The url with a scheme
 524          */
 525         private static function completeUrl($url, $scheme)
 526         {
 527                 $urlarr = parse_url($url);
 528
 529                 // If the url does allready have an scheme
 530                 // we can stop the process here
 531                 if (isset($urlarr["scheme"])) {
 532                         return($url);
 533                 }
 534
 535                 $schemearr = parse_url($scheme);
 536
 537                 $complete = $schemearr["scheme"]."://".$schemearr["host"];
 538
 539                 if (!empty($schemearr["port"])) {
 540                         $complete .= ":".$schemearr["port"];
 541                 }
 542
 543                 if (!empty($urlarr["path"])) {
 544                         if (strpos($urlarr["path"], "/") !== 0) {
 545                                 $complete .= "/";
 546                         }
 547
 548                         $complete .= $urlarr["path"];
 549                 }
 550
 551                 if (!empty($urlarr["query"])) {
 552                         $complete .= "?".$urlarr["query"];
 553                 }
 554
 555                 if (!empty($urlarr["fragment"])) {
 556                         $complete .= "#".$urlarr["fragment"];
 557                 }
 558
 559                 return($complete);
 560         }
 561 }