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'] ?: '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      * @return mixed
  28      */
  29     public static function getPageContent(Url $url)
  30     {
  31         // Saving and clearing context
  32         $REAL = array();
  33         foreach ($GLOBALS as $key => $value) {
  34             if ($key != 'GLOBALS' && $key != '_SESSION' && $key != 'HTTP_SESSION_VARS') {
  35                 $GLOBALS[$key]  = array();
  36                 $REAL[$key]     = $value;
  37             }
  38         }
  39         // Saving and clearing session
  40         if (isset($_SESSION)) {
  41             $REAL_SESSION = array();
  42             foreach ($_SESSION as $key => $value) {
  43                 $REAL_SESSION[$key] = $value;
  44                 unset($_SESSION[$key]);
  45             }
  46         }
  47
  48         // Running code in different context
  49         $scope = function () {
  50             extract(func_get_arg(1));
  51             $_GET = $_REQUEST = array(
  52                 "url" => $url->getUrl(),
  53                 "max" => 5,
  54                 "links" => "preserve",
  55                 "exc" => "",
  56                 "format" => "json",
  57                 "submit" => "Create Feed",
  58             );
  59             ob_start();
  60             require func_get_arg(0);
  61             $json = ob_get_contents();
  62             ob_end_clean();
  63
  64             return $json;
  65         };
  66
  67         // Silence $scope function to avoid
  68         // issues with FTRSS when error_reporting is to high
  69         // FTRSS generates PHP warnings which break output
  70         $json = @$scope(__DIR__."/../../../../vendor/wallabag/Fivefilters_Libraries/makefulltextfeed.php", array("url" => $url));
  71
  72         // Clearing and restoring context
  73         foreach ($GLOBALS as $key => $value) {
  74             if ($key != "GLOBALS" && $key != "_SESSION") {
  75                 unset($GLOBALS[$key]);
  76             }
  77         }
  78         foreach ($REAL as $key => $value) {
  79             $GLOBALS[$key] = $value;
  80         }
  81
  82         // Clearing and restoring session
  83         if (isset($REAL_SESSION)) {
  84             foreach ($_SESSION as $key => $value) {
  85                 unset($_SESSION[$key]);
  86             }
  87
  88             foreach ($REAL_SESSION as $key => $value) {
  89                 $_SESSION[$key] = $value;
  90             }
  91         }
  92
  93         return json_decode($json, true);
  94     }
  95 }