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 array 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                 return explode('/', current(explode(';', $contenttype)));
  72         }
  73
  74         /**
  75          * Search for chached embeddable data of an url otherwise fetch it
  76          *
  77          * @param string $url         The url of the page which should be scraped
  78          * @param bool   $no_guessing If true the parse doens't search for
  79          *                            preview pictures
  80          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
  81          *                            to avoid endless loops
  82          *
  83          * @return array which contains needed data for embedding
  84          *    string 'url'      => The url of the parsed page
  85          *    string 'type'     => Content type
  86          *    string 'title'    => (optional) The title of the content
  87          *    string 'text'     => (optional) The description for the content
  88          *    string 'image'    => (optional) A preview image of the content (only available if $no_geuessing = false)
  89          *    array  'images'   => (optional) Array of preview pictures
  90          *    string 'keywords' => (optional) The tags which belong to the content
  91          *
  92          * @throws HTTPException\InternalServerErrorException
  93          * @see   ParseUrl::getSiteinfo() for more information about scraping
  94          * embeddable content
  95          */
  96         public static function getSiteinfoCached($url, $no_guessing = false, $do_oembed = true): array
  97         {
  98                 if (empty($url)) {
  99                         return [
 100                                 'url' => '',
 101                                 'type' => 'error',
 102                         ];
 103                 }
 104
 105                 $urlHash = hash('sha256', $url);
 106
 107                 $parsed_url = DBA::selectFirst('parsed_url', ['content'],
 108                         ['url_hash' => $urlHash, 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
 109                 );
 110                 if (!empty($parsed_url['content'])) {
 111                         $data = unserialize($parsed_url['content']);
 112                         return $data;
 113                 }
 114
 115                 $data = self::getSiteinfo($url, $no_guessing, $do_oembed);
 116
 117                 $expires = $data['expires'];
 118
 119                 unset($data['expires']);
 120
 121                 DI::dba()->insert(
 122                         'parsed_url',
 123                         [
 124                                 'url_hash' => $urlHash,
 125                                 'guessing' => !$no_guessing,
 126                                 'oembed'   => $do_oembed,
 127                                 'url'      => $url,
 128                                 'content'  => serialize($data),
 129                                 'created'  => DateTimeFormat::utcNow(),
 130                                 'expires'  => $expires,
 131                         ],
 132                         Database::INSERT_UPDATE
 133                 );
 134
 135                 return $data;
 136         }
 137
 138         /**
 139          * Parse a page for embeddable content information
 140          *
 141          * This method parses to url for meta data which can be used to embed
 142          * the content. If available it prioritizes Open Graph meta tags.
 143          * If this is not available it uses the twitter cards meta tags.
 144          * As fallback it uses standard html elements with meta informations
 145          * like \<title\>Awesome Title\</title\> or
 146          * \<meta name="description" content="An awesome description"\>
 147          *
 148          * @param string $url         The url of the page which should be scraped
 149          * @param bool   $no_guessing If true the parse doens't search for
 150          *                            preview pictures
 151          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
 152          *                            to avoid endless loops
 153          * @param int    $count       Internal counter to avoid endless loops
 154          *
 155          * @return array which contains needed data for embedding
 156          *    string 'url'      => The url of the parsed page
 157          *    string 'type'     => Content type (error, link, photo, image, audio, video)
 158          *    string 'title'    => (optional) The title of the content
 159          *    string 'text'     => (optional) The description for the content
 160          *    string 'image'    => (optional) A preview image of the content (only available if $no_guessing = false)
 161          *    array  'images'   => (optional) Array of preview pictures
 162          *    string 'keywords' => (optional) The tags which belong to the content
 163          *
 164          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 165          * @todo  https://developers.google.com/+/plugins/snippet/
 166          * @verbatim
 167          * <meta itemprop="name" content="Awesome title">
 168          * <meta itemprop="description" content="An awesome description">
 169          * <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
 170          *
 171          * <body itemscope itemtype="http://schema.org/Product">
 172          *   <h1 itemprop="name">Shiny Trinket</h1>
 173          *   <img itemprop="image" src="{image-url}" />
 174          *   <p itemprop="description">Shiny trinkets are shiny.</p>
 175          * </body>
 176          * @endverbatim
 177          */
 178         public static function getSiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1)
 179         {
 180                 if (empty($url)) {
 181                         return [
 182                                 'url' => '',
 183                                 'type' => 'error',
 184                         ];
 185                 }
 186
 187                 // Check if the URL does contain a scheme
 188                 $scheme = parse_url($url, PHP_URL_SCHEME);
 189
 190                 if ($scheme == '') {
 191                         $url = 'http://' . ltrim($url, '/');
 192                 }
 193
 194                 $url = trim($url, "'\"");
 195
 196                 $url = Network::stripTrackingQueryParams($url);
 197
 198                 $siteinfo = [
 199                         'url' => $url,
 200                         'type' => 'link',
 201                         'expires' => DateTimeFormat::utc(self::DEFAULT_EXPIRATION_FAILURE),
 202                 ];
 203
 204                 if ($count > 10) {
 205                         Logger::log('Endless loop detected for ' . $url, Logger::DEBUG);
 206                         return $siteinfo;
 207                 }
 208
 209                 $type = self::getContentType($url);
 210                 Logger::info('Got content-type', ['content-type' => $type, 'url' => $url]);
 211                 if (!empty($type) && in_array($type[0], ['image', 'video', 'audio'])) {
 212                         $siteinfo['type'] = $type[0];
 213                         return $siteinfo;
 214                 }
 215
 216                 if ((count($type) >= 2) && (($type[0] != 'text') || ($type[1] != 'html'))) {
 217                         Logger::info('Unparseable content-type, quitting here, ', ['content-type' => $type, 'url' => $url]);
 218                         return $siteinfo;
 219                 }
 220
 221                 $curlResult = DI::httpRequest()->get($url);
 222                 if (!$curlResult->isSuccess()) {
 223                         return $siteinfo;
 224                 }
 225
 226                 $siteinfo['expires'] = DateTimeFormat::utc(self::DEFAULT_EXPIRATION_SUCCESS);
 227
 228                 // If the file is too large then exit
 229                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 230                         return $siteinfo;
 231                 }
 232
 233                 if ($cacheControlHeader = $curlResult->getHeader('Cache-Control')) {
 234                         if (preg_match('/max-age=([0-9]+)/i', $cacheControlHeader, $matches)) {
 235                                 $maxAge = max(86400, (int)array_pop($matches));
 236                                 $siteinfo['expires'] = DateTimeFormat::utc("now + $maxAge seconds");
 237                         }
 238                 }
 239
 240                 $header = $curlResult->getHeader();
 241                 $body = $curlResult->getBody();
 242
 243                 if ($do_oembed) {
 244                         $oembed_data = OEmbed::fetchURL($url);
 245
 246                         if (!empty($oembed_data->type)) {
 247                                 if (!in_array($oembed_data->type, ['error', 'rich', 'image', 'video', 'audio', ''])) {
 248                                         $siteinfo['type'] = $oembed_data->type;
 249                                 }
 250
 251                                 // See https://github.com/friendica/friendica/pull/5763#discussion_r217913178
 252                                 if ($siteinfo['type'] != 'photo') {
 253                                         if (isset($oembed_data->title)) {
 254                                                 $siteinfo['title'] = trim($oembed_data->title);
 255                                         }
 256                                         if (isset($oembed_data->description)) {
 257                                                 $siteinfo['text'] = trim($oembed_data->description);
 258                                         }
 259                                         if (isset($oembed_data->thumbnail_url)) {
 260                                                 $siteinfo['image'] = $oembed_data->thumbnail_url;
 261                                         }
 262                                 }
 263                         }
 264                 }
 265
 266                 $charset = '';
 267                 // Look for a charset, first in headers
 268                 // Expected form: Content-Type: text/html; charset=ISO-8859-4
 269                 if (preg_match('/charset=([a-z0-9-_.\/]+)/i', $header, $matches)) {
 270                         $charset = trim(trim(trim(array_pop($matches)), ';,'));
 271                 }
 272
 273                 // Then in body that gets precedence
 274                 // Expected forms:
 275                 // - <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
 276                 // - <meta charset="utf-8">
 277                 // - <meta charset=utf-8>
 278                 // - <meta charSet="utf-8">
 279                 // We escape <style> and <script> tags since they can contain irrelevant charset information
 280                 // (see https://github.com/friendica/friendica/issues/9251#issuecomment-698636806)
 281                 Strings::performWithEscapedBlocks($body, '#<(?:style|script).*?</(?:style|script)>#ism', function ($body) use (&$charset) {
 282                         if (preg_match('/charset=["\']?([a-z0-9-_.\/]+)/i', $body, $matches)) {
 283                                 $charset = trim(trim(trim(array_pop($matches)), ';,'));
 284                         }
 285                 });
 286
 287                 $siteinfo['charset'] = $charset;
 288
 289                 if ($charset && strtoupper($charset) != 'UTF-8') {
 290                         // See https://github.com/friendica/friendica/issues/5470#issuecomment-418351211
 291                         $charset = str_ireplace('latin-1', 'latin1', $charset);
 292
 293                         Logger::log('detected charset ' . $charset, Logger::DEBUG);
 294                         $body = iconv($charset, 'UTF-8//TRANSLIT', $body);
 295                 }
 296
 297                 $body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
 298
 299                 $doc = new DOMDocument();
 300                 @$doc->loadHTML($body);
 301
 302                 XML::deleteNode($doc, 'style');
 303                 XML::deleteNode($doc, 'script');
 304                 XML::deleteNode($doc, 'option');
 305                 XML::deleteNode($doc, 'h1');
 306                 XML::deleteNode($doc, 'h2');
 307                 XML::deleteNode($doc, 'h3');
 308                 XML::deleteNode($doc, 'h4');
 309                 XML::deleteNode($doc, 'h5');
 310                 XML::deleteNode($doc, 'h6');
 311                 XML::deleteNode($doc, 'ol');
 312                 XML::deleteNode($doc, 'ul');
 313
 314                 $xpath = new DOMXPath($doc);
 315
 316                 $list = $xpath->query('//meta[@content]');
 317                 foreach ($list as $node) {
 318                         $meta_tag = [];
 319                         if ($node->attributes->length) {
 320                                 foreach ($node->attributes as $attribute) {
 321                                         $meta_tag[$attribute->name] = $attribute->value;
 322                                 }
 323                         }
 324
 325                         if (@$meta_tag['http-equiv'] == 'refresh') {
 326                                 $path = $meta_tag['content'];
 327                                 $pathinfo = explode(';', $path);
 328                                 $content = '';
 329                                 foreach ($pathinfo as $value) {
 330                                         if (substr(strtolower($value), 0, 4) == 'url=') {
 331                                                 $content = substr($value, 4);
 332                                         }
 333                                 }
 334                                 if ($content != '') {
 335                                         $siteinfo = self::getSiteinfo($content, $no_guessing, $do_oembed, ++$count);
 336                                         return $siteinfo;
 337                                 }
 338                         }
 339                 }
 340
 341                 $list = $xpath->query('//title');
 342                 if ($list->length > 0) {
 343                         $siteinfo['title'] = trim($list->item(0)->nodeValue);
 344                 }
 345
 346                 $list = $xpath->query('//meta[@name]');
 347                 foreach ($list as $node) {
 348                         $meta_tag = [];
 349                         if ($node->attributes->length) {
 350                                 foreach ($node->attributes as $attribute) {
 351                                         $meta_tag[$attribute->name] = $attribute->value;
 352                                 }
 353                         }
 354
 355                         if (empty($meta_tag['content'])) {
 356                                 continue;
 357                         }
 358
 359                         $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 360
 361                         switch (strtolower($meta_tag['name'])) {
 362                                 case 'fulltitle':
 363                                         $siteinfo['title'] = trim($meta_tag['content']);
 364                                         break;
 365                                 case 'description':
 366                                         $siteinfo['text'] = trim($meta_tag['content']);
 367                                         break;
 368                                 case 'thumbnail':
 369                                         $siteinfo['image'] = $meta_tag['content'];
 370                                         break;
 371                                 case 'twitter:image':
 372                                         $siteinfo['image'] = $meta_tag['content'];
 373                                         break;
 374                                 case 'twitter:image:src':
 375                                         $siteinfo['image'] = $meta_tag['content'];
 376                                         break;
 377                                 case 'twitter:card':
 378                                         // Detect photo pages
 379                                         if ($meta_tag['content'] == 'summary_large_image') {
 380                                                 $siteinfo['type'] = 'photo';
 381                                         }
 382                                         break;
 383                                 case 'twitter:description':
 384                                         $siteinfo['text'] = trim($meta_tag['content']);
 385                                         break;
 386                                 case 'twitter:title':
 387                                         $siteinfo['title'] = trim($meta_tag['content']);
 388                                         break;
 389                                 case 'dc.title':
 390                                         $siteinfo['title'] = trim($meta_tag['content']);
 391                                         break;
 392                                 case 'dc.description':
 393                                         $siteinfo['text'] = trim($meta_tag['content']);
 394                                         break;
 395                                 case 'keywords':
 396                                         $keywords = explode(',', $meta_tag['content']);
 397                                         break;
 398                                 case 'news_keywords':
 399                                         $keywords = explode(',', $meta_tag['content']);
 400                                         break;
 401                         }
 402                 }
 403
 404                 if (isset($keywords)) {
 405                         $siteinfo['keywords'] = [];
 406                         foreach ($keywords as $keyword) {
 407                                 if (!in_array(trim($keyword), $siteinfo['keywords'])) {
 408                                         $siteinfo['keywords'][] = trim($keyword);
 409                                 }
 410                         }
 411                 }
 412
 413                 $list = $xpath->query('//meta[@property]');
 414                 foreach ($list as $node) {
 415                         $meta_tag = [];
 416                         if ($node->attributes->length) {
 417                                 foreach ($node->attributes as $attribute) {
 418                                         $meta_tag[$attribute->name] = $attribute->value;
 419                                 }
 420                         }
 421
 422                         if (!empty($meta_tag['content'])) {
 423                                 $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 424
 425                                 switch (strtolower($meta_tag['property'])) {
 426                                         case 'og:image':
 427                                                 $siteinfo['image'] = $meta_tag['content'];
 428                                                 break;
 429                                         case 'og:title':
 430                                                 $siteinfo['title'] = trim($meta_tag['content']);
 431                                                 break;
 432                                         case 'og:description':
 433                                                 $siteinfo['text'] = trim($meta_tag['content']);
 434                                                 break;
 435                                 }
 436                         }
 437                 }
 438
 439                 // Prevent to have a photo type without an image
 440                 if ((empty($siteinfo['image']) || !empty($siteinfo['text'])) && ($siteinfo['type'] == 'photo')) {
 441                         $siteinfo['type'] = 'link';
 442                 }
 443
 444                 if (!empty($siteinfo['image'])) {
 445                         $src = self::completeUrl($siteinfo['image'], $url);
 446
 447                         unset($siteinfo['image']);
 448
 449                         $photodata = Images::getInfoFromURLCached($src);
 450
 451                         if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) {
 452                                 $siteinfo['images'][] = ['src' => $src,
 453                                         'width' => $photodata[0],
 454                                         'height' => $photodata[1]];
 455                         }
 456                 }
 457
 458                 if (!empty($siteinfo['text']) && mb_strlen($siteinfo['text']) > self::MAX_DESC_COUNT) {
 459                         $siteinfo['text'] = mb_substr($siteinfo['text'], 0, self::MAX_DESC_COUNT) . '…';
 460                         $pos = mb_strrpos($siteinfo['text'], '.');
 461                         if ($pos > self::MIN_DESC_COUNT) {
 462                                 $siteinfo['text'] = mb_substr($siteinfo['text'], 0, $pos + 1);
 463                         }
 464                 }
 465
 466                 Logger::info('Siteinfo fetched', ['url' => $url, 'siteinfo' => $siteinfo]);
 467
 468                 Hook::callAll('getsiteinfo', $siteinfo);
 469
 470                 return $siteinfo;
 471         }
 472
 473         /**
 474          * Convert tags from CSV to an array
 475          *
 476          * @param string $string Tags
 477          * @return array with formatted Hashtags
 478          */
 479         public static function convertTagsToArray($string)
 480         {
 481                 $arr_tags = str_getcsv($string);
 482                 if (count($arr_tags)) {
 483                         // add the # sign to every tag
 484                         array_walk($arr_tags, ["self", "arrAddHashes"]);
 485
 486                         return $arr_tags;
 487                 }
 488         }
 489
 490         /**
 491          * Add a hasht sign to a string
 492          *
 493          * This method is used as callback function
 494          *
 495          * @param string $tag The pure tag name
 496          * @param int    $k   Counter for internal use
 497          * @return void
 498          */
 499         private static function arrAddHashes(&$tag, $k)
 500         {
 501                 $tag = "#" . $tag;
 502         }
 503
 504         /**
 505          * Add a scheme to an url
 506          *
 507          * The src attribute of some html elements (e.g. images)
 508          * can miss the scheme so we need to add the correct
 509          * scheme
 510          *
 511          * @param string $url    The url which possibly does have
 512          *                       a missing scheme (a link to an image)
 513          * @param string $scheme The url with a correct scheme
 514          *                       (e.g. the url from the webpage which does contain the image)
 515          *
 516          * @return string The url with a scheme
 517          */
 518         private static function completeUrl($url, $scheme)
 519         {
 520                 $urlarr = parse_url($url);
 521
 522                 // If the url does allready have an scheme
 523                 // we can stop the process here
 524                 if (isset($urlarr["scheme"])) {
 525                         return($url);
 526                 }
 527
 528                 $schemearr = parse_url($scheme);
 529
 530                 $complete = $schemearr["scheme"]."://".$schemearr["host"];
 531
 532                 if (!empty($schemearr["port"])) {
 533                         $complete .= ":".$schemearr["port"];
 534                 }
 535
 536                 if (!empty($urlarr["path"])) {
 537                         if (strpos($urlarr["path"], "/") !== 0) {
 538                                 $complete .= "/";
 539                         }
 540
 541                         $complete .= $urlarr["path"];
 542                 }
 543
 544                 if (!empty($urlarr["query"])) {
 545                         $complete .= "?".$urlarr["query"];
 546                 }
 547
 548                 if (!empty($urlarr["fragment"])) {
 549                         $complete .= "#".$urlarr["fragment"];
 550                 }
 551
 552                 return($complete);
 553         }
 554 }