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