application/http/MetadataRetriever.php

   1 <?php
   2
   3 declare(strict_types=1);
   4
   5 namespace Shaarli\Http;
   6
   7 use Shaarli\Config\ConfigManager;
   8
   9 /**
  10  * HTTP Tool used to extract metadata from external URL (title, description, etc.).
  11  */
  12 class MetadataRetriever
  13 {
  14     /** @var ConfigManager */
  15     protected $conf;
  16
  17     /** @var HttpAccess */
  18     protected $httpAccess;
  19
  20     public function __construct(ConfigManager $conf, HttpAccess $httpAccess)
  21     {
  22         $this->conf = $conf;
  23         $this->httpAccess = $httpAccess;
  24     }
  25
  26     /**
  27      * Retrieve metadata for given URL.
  28      *
  29      * @return array [
  30      *                  'title' => <remote title>,
  31      *                  'description' => <remote description>,
  32      *                  'tags' => <remote keywords>,
  33      *               ]
  34      */
  35     public function retrieve(string $url): array
  36     {
  37         $charset = null;
  38         $title = null;
  39         $description = null;
  40         $tags = null;
  41         $retrieveDescription = $this->conf->get('general.retrieve_description');
  42
  43         // Short timeout to keep the application responsive
  44         // The callback will fill $charset and $title with data from the downloaded page.
  45         $this->httpAccess->getHttpResponse(
  46             $url,
  47             $this->conf->get('general.download_timeout', 30),
  48             $this->conf->get('general.download_max_size', 4194304),
  49             $this->httpAccess->getCurlDownloadCallback(
  50                 $charset,
  51                 $title,
  52                 $description,
  53                 $tags,
  54                 $retrieveDescription
  55             )
  56         );
  57
  58         if (!empty($title) && strtolower($charset) !== 'utf-8') {
  59             $title = mb_convert_encoding($title, 'utf-8', $charset);
  60         }
  61
  62         return [
  63             'title' => $title,
  64             'description' => $description,
  65             'tags' => $tags,
  66         ];
  67     }
  68 }