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