src/Util/ParseUrl.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Util;
  23
  24 use DOMDocument;
  25 use DOMXPath;
  26 use Friendica\Content\OEmbed;
  27 use Friendica\Core\Hook;
  28 use Friendica\Core\Logger;
  29 use Friendica\Database\DBA;
  30 use Friendica\DI;
  31
  32 /**
  33  * Get information about a given URL
  34  *
  35  * Class with methods for extracting certain content from an url
  36  */
  37 class ParseUrl
  38 {
  39         /**
  40          * Maximum number of characters for the description
  41          */
  42         const MAX_DESC_COUNT = 250;
  43
  44         /**
  45          * Minimum number of characters for the description
  46          */
  47         const MIN_DESC_COUNT = 100;
  48
  49         /**
  50          * Search for chached embeddable data of an url otherwise fetch it
  51          *
  52          * @param string $url         The url of the page which should be scraped
  53          * @param bool   $no_guessing If true the parse doens't search for
  54          *                            preview pictures
  55          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
  56          *                            to avoid endless loops
  57          *
  58          * @return array which contains needed data for embedding
  59          *    string 'url'      => The url of the parsed page
  60          *    string 'type'     => Content type
  61          *    string 'title'    => (optional) The title of the content
  62          *    string 'text'     => (optional) The description for the content
  63          *    string 'image'    => (optional) A preview image of the content (only available if $no_geuessing = false)
  64          *    array  'images'   => (optional) Array of preview pictures
  65          *    string 'keywords' => (optional) The tags which belong to the content
  66          *
  67          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  68          * @see   ParseUrl::getSiteinfo() for more information about scraping
  69          * embeddable content
  70          */
  71         public static function getSiteinfoCached($url, $no_guessing = false, $do_oembed = true)
  72         {
  73                 if ($url == "") {
  74                         return false;
  75                 }
  76
  77                 $parsed_url = DBA::selectFirst('parsed_url', ['content'],
  78                         ['url' => Strings::normaliseLink($url), 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
  79                 );
  80                 if (!empty($parsed_url['content'])) {
  81                         $data = unserialize($parsed_url['content']);
  82                         return $data;
  83                 }
  84
  85                 $data = self::getSiteinfo($url, $no_guessing, $do_oembed);
  86
  87                 DBA::insert(
  88                         'parsed_url',
  89                         [
  90                                 'url' => substr(Strings::normaliseLink($url), 0, 255), 'guessing' => !$no_guessing,
  91                                 'oembed' => $do_oembed, 'content' => serialize($data),
  92                                 'created' => DateTimeFormat::utcNow()
  93                         ],
  94                         true
  95                 );
  96
  97                 return $data;
  98         }
  99
 100         /**
 101          * Parse a page for embeddable content information
 102          *
 103          * This method parses to url for meta data which can be used to embed
 104          * the content. If available it prioritizes Open Graph meta tags.
 105          * If this is not available it uses the twitter cards meta tags.
 106          * As fallback it uses standard html elements with meta informations
 107          * like \<title\>Awesome Title\</title\> or
 108          * \<meta name="description" content="An awesome description"\>
 109          *
 110          * @param string $url         The url of the page which should be scraped
 111          * @param bool   $no_guessing If true the parse doens't search for
 112          *                            preview pictures
 113          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
 114          *                            to avoid endless loops
 115          * @param int    $count       Internal counter to avoid endless loops
 116          *
 117          * @return array which contains needed data for embedding
 118          *    string 'url'      => The url of the parsed page
 119          *    string 'type'     => Content type
 120          *    string 'title'    => (optional) The title of the content
 121          *    string 'text'     => (optional) The description for the content
 122          *    string 'image'    => (optional) A preview image of the content (only available if $no_guessing = false)
 123          *    array  'images'   => (optional) Array of preview pictures
 124          *    string 'keywords' => (optional) The tags which belong to the content
 125          *
 126          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 127          * @todo  https://developers.google.com/+/plugins/snippet/
 128          * @verbatim
 129          * <meta itemprop="name" content="Awesome title">
 130          * <meta itemprop="description" content="An awesome description">
 131          * <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
 132          *
 133          * <body itemscope itemtype="http://schema.org/Product">
 134          *   <h1 itemprop="name">Shiny Trinket</h1>
 135          *   <img itemprop="image" src="{image-url}" />
 136          *   <p itemprop="description">Shiny trinkets are shiny.</p>
 137          * </body>
 138          * @endverbatim
 139          */
 140         public static function getSiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1)
 141         {
 142                 // Check if the URL does contain a scheme
 143                 $scheme = parse_url($url, PHP_URL_SCHEME);
 144
 145                 if ($scheme == '') {
 146                         $url = 'http://' . ltrim($url, '/');
 147                 }
 148
 149                 $url = trim($url, "'\"");
 150
 151                 $url = Network::stripTrackingQueryParams($url);
 152
 153                 $siteinfo = [
 154                         'url' => $url,
 155                         'type' => 'link',
 156                 ];
 157
 158                 if ($count > 10) {
 159                         Logger::log('Endless loop detected for ' . $url, Logger::DEBUG);
 160                         return $siteinfo;
 161                 }
 162
 163                 $curlResult = DI::httpRequest()->get($url, false, ['content_length' => 1000000]);
 164                 if (!$curlResult->isSuccess()) {
 165                         return $siteinfo;
 166                 }
 167
 168                 // If it isn't a HTML file then exit
 169                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 170                         return $siteinfo;
 171                 }
 172
 173                 $body = $curlResult->getBody();
 174
 175                 if ($do_oembed) {
 176                         $oembed_data = OEmbed::fetchURL($url);
 177
 178                         if (!empty($oembed_data->type)) {
 179                                 if (!in_array($oembed_data->type, ['error', 'rich', ''])) {
 180                                         $siteinfo['type'] = $oembed_data->type;
 181                                 }
 182
 183                                 // See https://github.com/friendica/friendica/pull/5763#discussion_r217913178
 184                                 if ($siteinfo['type'] != 'photo') {
 185                                         if (isset($oembed_data->title)) {
 186                                                 $siteinfo['title'] = trim($oembed_data->title);
 187                                         }
 188                                         if (isset($oembed_data->description)) {
 189                                                 $siteinfo['text'] = trim($oembed_data->description);
 190                                         }
 191                                         if (isset($oembed_data->thumbnail_url)) {
 192                                                 $siteinfo['image'] = $oembed_data->thumbnail_url;
 193                                         }
 194                                 }
 195                         }
 196                 }
 197
 198                 $charset = '';
 199                 // Look for a charset, first in headers
 200                 // Expected form: Content-Type: text/html; charset=ISO-8859-4
 201                 if (preg_match('/charset=([a-z0-9-_.\/]+)/i', $curlResult->getContentType(), $matches)) {
 202                         $charset = trim(trim(trim(array_pop($matches)), ';,'));
 203                 }
 204
 205                 // Then in body that gets precedence
 206                 // Expected forms:
 207                 // - <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
 208                 // - <meta charset="utf-8">
 209                 // - <meta charset=utf-8>
 210                 // - <meta charSet="utf-8">
 211                 // We escape <style> and <script> tags since they can contain irrelevant charset information
 212                 // (see https://github.com/friendica/friendica/issues/9251#issuecomment-698636806)
 213                 Strings::performWithEscapedBlocks($body, '#<(?:style|script).*?</(?:style|script)>#ism', function ($body) use (&$charset) {
 214                         if (preg_match('/charset=["\']?([a-z0-9-_.\/]+)/i', $body, $matches)) {
 215                                 $charset = trim(trim(trim(array_pop($matches)), ';,'));
 216                         }
 217                 });
 218
 219                 $siteinfo['charset'] = $charset;
 220
 221                 if ($charset && strtoupper($charset) != 'UTF-8') {
 222                         // See https://github.com/friendica/friendica/issues/5470#issuecomment-418351211
 223                         $charset = str_ireplace('latin-1', 'latin1', $charset);
 224
 225                         Logger::log('detected charset ' . $charset, Logger::DEBUG);
 226                         $body = iconv($charset, 'UTF-8//TRANSLIT', $body);
 227                 }
 228
 229                 $body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
 230
 231                 $doc = new DOMDocument();
 232                 @$doc->loadHTML($body);
 233
 234                 XML::deleteNode($doc, 'style');
 235                 XML::deleteNode($doc, 'script');
 236                 XML::deleteNode($doc, 'option');
 237                 XML::deleteNode($doc, 'h1');
 238                 XML::deleteNode($doc, 'h2');
 239                 XML::deleteNode($doc, 'h3');
 240                 XML::deleteNode($doc, 'h4');
 241                 XML::deleteNode($doc, 'h5');
 242                 XML::deleteNode($doc, 'h6');
 243                 XML::deleteNode($doc, 'ol');
 244                 XML::deleteNode($doc, 'ul');
 245
 246                 $xpath = new DOMXPath($doc);
 247
 248                 $list = $xpath->query('//meta[@content]');
 249                 foreach ($list as $node) {
 250                         $meta_tag = [];
 251                         if ($node->attributes->length) {
 252                                 foreach ($node->attributes as $attribute) {
 253                                         $meta_tag[$attribute->name] = $attribute->value;
 254                                 }
 255                         }
 256
 257                         if (@$meta_tag['http-equiv'] == 'refresh') {
 258                                 $path = $meta_tag['content'];
 259                                 $pathinfo = explode(';', $path);
 260                                 $content = '';
 261                                 foreach ($pathinfo as $value) {
 262                                         if (substr(strtolower($value), 0, 4) == 'url=') {
 263                                                 $content = substr($value, 4);
 264                                         }
 265                                 }
 266                                 if ($content != '') {
 267                                         $siteinfo = self::getSiteinfo($content, $no_guessing, $do_oembed, ++$count);
 268                                         return $siteinfo;
 269                                 }
 270                         }
 271                 }
 272
 273                 $list = $xpath->query('//title');
 274                 if ($list->length > 0) {
 275                         $siteinfo['title'] = trim($list->item(0)->nodeValue);
 276                 }
 277
 278                 $list = $xpath->query('//meta[@name]');
 279                 foreach ($list as $node) {
 280                         $meta_tag = [];
 281                         if ($node->attributes->length) {
 282                                 foreach ($node->attributes as $attribute) {
 283                                         $meta_tag[$attribute->name] = $attribute->value;
 284                                 }
 285                         }
 286
 287                         if (empty($meta_tag['content'])) {
 288                                 continue;
 289                         }
 290
 291                         $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 292
 293                         switch (strtolower($meta_tag['name'])) {
 294                                 case 'fulltitle':
 295                                         $siteinfo['title'] = trim($meta_tag['content']);
 296                                         break;
 297                                 case 'description':
 298                                         $siteinfo['text'] = trim($meta_tag['content']);
 299                                         break;
 300                                 case 'thumbnail':
 301                                         $siteinfo['image'] = $meta_tag['content'];
 302                                         break;
 303                                 case 'twitter:image':
 304                                         $siteinfo['image'] = $meta_tag['content'];
 305                                         break;
 306                                 case 'twitter:image:src':
 307                                         $siteinfo['image'] = $meta_tag['content'];
 308                                         break;
 309                                 case 'twitter:card':
 310                                         // Detect photo pages
 311                                         if ($meta_tag['content'] == 'summary_large_image') {
 312                                                 $siteinfo['type'] = 'photo';
 313                                         }
 314                                         break;
 315                                 case 'twitter:description':
 316                                         $siteinfo['text'] = trim($meta_tag['content']);
 317                                         break;
 318                                 case 'twitter:title':
 319                                         $siteinfo['title'] = trim($meta_tag['content']);
 320                                         break;
 321                                 case 'dc.title':
 322                                         $siteinfo['title'] = trim($meta_tag['content']);
 323                                         break;
 324                                 case 'dc.description':
 325                                         $siteinfo['text'] = trim($meta_tag['content']);
 326                                         break;
 327                                 case 'keywords':
 328                                         $keywords = explode(',', $meta_tag['content']);
 329                                         break;
 330                                 case 'news_keywords':
 331                                         $keywords = explode(',', $meta_tag['content']);
 332                                         break;
 333                         }
 334                 }
 335
 336                 if (isset($keywords)) {
 337                         $siteinfo['keywords'] = [];
 338                         foreach ($keywords as $keyword) {
 339                                 if (!in_array(trim($keyword), $siteinfo['keywords'])) {
 340                                         $siteinfo['keywords'][] = trim($keyword);
 341                                 }
 342                         }
 343                 }
 344
 345                 $list = $xpath->query('//meta[@property]');
 346                 foreach ($list as $node) {
 347                         $meta_tag = [];
 348                         if ($node->attributes->length) {
 349                                 foreach ($node->attributes as $attribute) {
 350                                         $meta_tag[$attribute->name] = $attribute->value;
 351                                 }
 352                         }
 353
 354                         if (!empty($meta_tag['content'])) {
 355                                 $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 356
 357                                 switch (strtolower($meta_tag['property'])) {
 358                                         case 'og:image':
 359                                                 $siteinfo['image'] = $meta_tag['content'];
 360                                                 break;
 361                                         case 'og:title':
 362                                                 $siteinfo['title'] = trim($meta_tag['content']);
 363                                                 break;
 364                                         case 'og:description':
 365                                                 $siteinfo['text'] = trim($meta_tag['content']);
 366                                                 break;
 367                                 }
 368                         }
 369                 }
 370
 371                 // Prevent to have a photo type without an image
 372                 if ((empty($siteinfo['image']) || !empty($siteinfo['text'])) && ($siteinfo['type'] == 'photo')) {
 373                         $siteinfo['type'] = 'link';
 374                 }
 375
 376                 if (!empty($siteinfo['image'])) {
 377                         $src = self::completeUrl($siteinfo['image'], $url);
 378
 379                         unset($siteinfo['image']);
 380
 381                         $photodata = Images::getInfoFromURLCached($src);
 382
 383                         if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) {
 384                                 $siteinfo['images'][] = ['src' => $src,
 385                                         'width' => $photodata[0],
 386                                         'height' => $photodata[1]];
 387                         }
 388                 }
 389
 390                 if (!empty($siteinfo['text']) && mb_strlen($siteinfo['text']) > self::MAX_DESC_COUNT) {
 391                         $siteinfo['text'] = mb_substr($siteinfo['text'], 0, self::MAX_DESC_COUNT) . '…';
 392                         $pos = mb_strrpos($siteinfo['text'], '.');
 393                         if ($pos > self::MIN_DESC_COUNT) {
 394                                 $siteinfo['text'] = mb_substr($siteinfo['text'], 0, $pos + 1);
 395                         }
 396                 }
 397
 398                 Logger::info('Siteinfo fetched', ['url' => $url, 'siteinfo' => $siteinfo]);
 399
 400                 Hook::callAll('getsiteinfo', $siteinfo);
 401
 402                 return $siteinfo;
 403         }
 404
 405         /**
 406          * Convert tags from CSV to an array
 407          *
 408          * @param string $string Tags
 409          * @return array with formatted Hashtags
 410          */
 411         public static function convertTagsToArray($string)
 412         {
 413                 $arr_tags = str_getcsv($string);
 414                 if (count($arr_tags)) {
 415                         // add the # sign to every tag
 416                         array_walk($arr_tags, ["self", "arrAddHashes"]);
 417
 418                         return $arr_tags;
 419                 }
 420         }
 421
 422         /**
 423          * Add a hasht sign to a string
 424          *
 425          * This method is used as callback function
 426          *
 427          * @param string $tag The pure tag name
 428          * @param int    $k   Counter for internal use
 429          * @return void
 430          */
 431         private static function arrAddHashes(&$tag, $k)
 432         {
 433                 $tag = "#" . $tag;
 434         }
 435
 436         /**
 437          * Add a scheme to an url
 438          *
 439          * The src attribute of some html elements (e.g. images)
 440          * can miss the scheme so we need to add the correct
 441          * scheme
 442          *
 443          * @param string $url    The url which possibly does have
 444          *                       a missing scheme (a link to an image)
 445          * @param string $scheme The url with a correct scheme
 446          *                       (e.g. the url from the webpage which does contain the image)
 447          *
 448          * @return string The url with a scheme
 449          */
 450         private static function completeUrl($url, $scheme)
 451         {
 452                 $urlarr = parse_url($url);
 453
 454                 // If the url does allready have an scheme
 455                 // we can stop the process here
 456                 if (isset($urlarr["scheme"])) {
 457                         return($url);
 458                 }
 459
 460                 $schemearr = parse_url($scheme);
 461
 462                 $complete = $schemearr["scheme"]."://".$schemearr["host"];
 463
 464                 if (!empty($schemearr["port"])) {
 465                         $complete .= ":".$schemearr["port"];
 466                 }
 467
 468                 if (!empty($urlarr["path"])) {
 469                         if (strpos($urlarr["path"], "/") !== 0) {
 470                                 $complete .= "/";
 471                         }
 472
 473                         $complete .= $urlarr["path"];
 474                 }
 475
 476                 if (!empty($urlarr["query"])) {
 477                         $complete .= "?".$urlarr["query"];
 478                 }
 479
 480                 if (!empty($urlarr["fragment"])) {
 481                         $complete .= "#".$urlarr["fragment"];
 482                 }
 483
 484                 return($complete);
 485         }
 486 }