wcfsetup/install/files/lib/system/spider/SpiderHandler.class.php

   1 <?php
   2
   3 namespace wcf\system\spider;
   4
   5 use wcf\system\event\EventHandler;
   6 use wcf\system\SingletonFactory;
   7 use wcf\system\spider\event\SpiderCollecting;
   8
   9 /**
  10  * Handles spider related operations.
  11  *
  12  * @author      Olaf Braun
  13  * @copyright   2001-2024 WoltLab GmbH
  14  * @license     GNU Lesser General Public License <http://opensource.org/licenses/lgpl-license.php>
  15  * @since       6.1
  16  */
  17 final class SpiderHandler extends SingletonFactory
  18 {
  19     /**
  20      * @var Spider[]
  21      */
  22     private array $spiders = [];
  23
  24     private string $regex = '';
  25
  26     #[\Override]
  27     protected function init()
  28     {
  29         parent::init();
  30         $event = new SpiderCollecting();
  31         EventHandler::getInstance()->fire($event);
  32
  33         $this->spiders = $event->getSpiders();
  34
  35         $firstCharacter = [];
  36         foreach ($this->spiders as $identifier => $spider) {
  37             if (!isset($firstCharacter[$identifier[0]])) {
  38                 $firstCharacter[$identifier[0]] = [];
  39             }
  40             $firstCharacter[$identifier[0]][] = \substr($identifier, 1);
  41         }
  42
  43         $this->regex = '';
  44         foreach ($firstCharacter as $char => $spiders) {
  45             if ($this->regex !== '') {
  46                 $this->regex .= '|';
  47             }
  48             $this->regex .= \sprintf(
  49                 '(?:%s(?:%s))',
  50                 \preg_quote($char, '/'),
  51                 \implode('|', \array_map(static function ($identifier) {
  52                     return \preg_quote($identifier, '/');
  53                 }, $spiders))
  54             );
  55         }
  56
  57         if ($this->regex === '') {
  58             // This regex will never match anything.
  59             $this->regex = '(?!)';
  60         }
  61         $this->regex = '/' . $this->regex . '/';
  62     }
  63
  64     /**
  65      * Returns the spider with the given identifier.
  66      */
  67     public function getSpider(string $identifier): ?Spider
  68     {
  69         return $this->spiders[$identifier] ?? null;
  70     }
  71
  72     /**
  73      * Finds the spider identifier for the given user agent.
  74      */
  75     public function getIdentifier(string $userAgent): ?string
  76     {
  77         $userAgent = \mb_strtolower($userAgent);
  78         if (\preg_match($this->regex, $userAgent, $matches)) {
  79             return $matches[0];
  80         }
  81
  82         return null;
  83     }
  84 }