src/Wallabag/CoreBundle/Service/Extractor.php

   1 <?php
   2
   3 namespace Wallabag\CoreBundle\Service;
   4
   5 use Wallabag\CoreBundle\Helper\Content;
   6 use Wallabag\CoreBundle\Helper\Url;
   7
   8 final class Extractor
   9 {
  10     public static function extract($url)
  11     {
  12         $pageContent = self::getPageContent(new Url(base64_encode($url)));
  13         $title = $pageContent['rss']['channel']['item']['title'] ?: 'Untitled';
  14         $body = $pageContent['rss']['channel']['item']['description'];
  15
  16         $content = new Content();
  17         $content->setTitle($title);
  18         $content->setBody($body);
  19
  20         return $content;
  21     }
  22
  23     /**
  24      * Get the content for a given URL (by a call to FullTextFeed).
  25      *
  26      * @param Url $url
  27      *
  28      * @return mixed
  29      */
  30     public static function getPageContent(Url $url)
  31     {
  32         // Saving and clearing context
  33         $REAL = array();
  34         foreach ($GLOBALS as $key => $value) {
  35             if ($key != 'GLOBALS' && $key != '_SESSION' && $key != 'HTTP_SESSION_VARS') {
  36                 $GLOBALS[$key]  = array();
  37                 $REAL[$key]     = $value;
  38             }
  39         }
  40         // Saving and clearing session
  41         if (isset($_SESSION)) {
  42             $REAL_SESSION = array();
  43             foreach ($_SESSION as $key => $value) {
  44                 $REAL_SESSION[$key] = $value;
  45                 unset($_SESSION[$key]);
  46             }
  47         }
  48
  49         // Running code in different context
  50         $scope = function () {
  51             extract(func_get_arg(1));
  52             $_GET = $_REQUEST = array(
  53                 'url' => $url->getUrl(),
  54                 'max' => 5,
  55                 'links' => 'preserve',
  56                 'exc' => '',
  57                 'format' => 'json',
  58                 'submit' => 'Create Feed',
  59             );
  60             ob_start();
  61             require func_get_arg(0);
  62             $json = ob_get_contents();
  63             ob_end_clean();
  64
  65             return $json;
  66         };
  67
  68         // Silence $scope function to avoid
  69         // issues with FTRSS when error_reporting is to high
  70         // FTRSS generates PHP warnings which break output
  71         $json = @$scope(__DIR__.'/../../../../vendor/wallabag/Fivefilters_Libraries/makefulltextfeed.php', array('url' => $url));
  72
  73         // Clearing and restoring context
  74         foreach ($GLOBALS as $key => $value) {
  75             if ($key != 'GLOBALS' && $key != '_SESSION') {
  76                 unset($GLOBALS[$key]);
  77             }
  78         }
  79         foreach ($REAL as $key => $value) {
  80             $GLOBALS[$key] = $value;
  81         }
  82
  83         // Clearing and restoring session
  84         if (isset($REAL_SESSION)) {
  85             foreach ($_SESSION as $key => $value) {
  86                 unset($_SESSION[$key]);
  87             }
  88
  89             foreach ($REAL_SESSION as $key => $value) {
  90                 $_SESSION[$key] = $value;
  91             }
  92         }
  93
  94         return json_decode($json, true);
  95     }
  96 }