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