src/Wallabag/CoreBundle/Helper/DownloadImages.php

   1 <?php
   2
   3 namespace Wallabag\CoreBundle\Helper;
   4
   5 use Psr\Log\LoggerInterface;
   6 use Symfony\Component\DomCrawler\Crawler;
   7 use GuzzleHttp\Client;
   8 use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
   9
  10 class DownloadImages
  11 {
  12     const REGENERATE_PICTURES_QUALITY = 80;
  13
  14     private $client;
  15     private $baseFolder;
  16     private $logger;
  17     private $mimeGuesser;
  18     private $wallabagUrl;
  19
  20     public function __construct(Client $client, $baseFolder, LoggerInterface $logger)
  21     {
  22         $this->client = $client;
  23         $this->baseFolder = $baseFolder;
  24         $this->logger = $logger;
  25         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  26
  27         $this->setFolder();
  28     }
  29
  30     /**
  31      * Since we can't inject CraueConfig service because it'll generate a circular reference when injected in the subscriber
  32      * we use a different way to inject the current wallabag url.
  33      *
  34      * @param string $url Usually from `$config->get('wallabag_url')`
  35      */
  36     public function setWallabagUrl($url)
  37     {
  38         $this->wallabagUrl = rtrim($url, '/');
  39     }
  40
  41     /**
  42      * Setup base folder where all images are going to be saved.
  43      */
  44     private function setFolder()
  45     {
  46         // if folder doesn't exist, attempt to create one and store the folder name in property $folder
  47         if (!file_exists($this->baseFolder)) {
  48             mkdir($this->baseFolder, 0777, true);
  49         }
  50     }
  51
  52     /**
  53      * Process the html and extract image from it, save them to local and return the updated html.
  54      *
  55      * @param string $html
  56      * @param string $url  Used as a base path for relative image and folder
  57      *
  58      * @return string
  59      */
  60     public function processHtml($html, $url)
  61     {
  62         $crawler = new Crawler($html);
  63         $result = $crawler
  64             ->filterXpath('//img')
  65             ->extract(array('src'));
  66
  67         $relativePath = $this->getRelativePath($url);
  68
  69         // download and save the image to the folder
  70         foreach ($result as $image) {
  71             $imagePath = $this->processSingleImage($image, $url, $relativePath);
  72
  73             if (false === $imagePath) {
  74                 continue;
  75             }
  76
  77             $html = str_replace($image, $imagePath, $html);
  78         }
  79
  80         return $html;
  81     }
  82
  83     /**
  84      * Process a single image:
  85      *     - retrieve it
  86      *     - re-saved it (for security reason)
  87      *     - return the new local path.
  88      *
  89      * @param string $imagePath    Path to the image to retrieve
  90      * @param string $url          Url from where the image were found
  91      * @param string $relativePath Relative local path to saved the image
  92      *
  93      * @return string Relative url to access the image from the web
  94      */
  95     public function processSingleImage($imagePath, $url, $relativePath = null)
  96     {
  97         if (null == $relativePath) {
  98             $relativePath = $this->getRelativePath($url);
  99         }
 100
 101         $folderPath = $this->baseFolder.'/'.$relativePath;
 102
 103         // build image path
 104         $absolutePath = $this->getAbsoluteLink($url, $imagePath);
 105         if (false === $absolutePath) {
 106             $this->logger->log('error', 'Can not determine the absolute path for that image, skipping.');
 107
 108             return false;
 109         }
 110
 111         try {
 112             $res = $this->client->get($absolutePath);
 113         } catch (\Exception $e) {
 114             $this->logger->log('error', 'Can not retrieve image, skipping.', ['exception' => $e]);
 115
 116             return false;
 117         }
 118
 119         $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
 120         $this->logger->log('debug', 'Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
 121         if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
 122             $this->logger->log('error', 'Processed image with not allowed extension. Skipping '.$imagePath);
 123
 124             return false;
 125         }
 126         $hashImage = hash('crc32', $absolutePath);
 127         $localPath = $folderPath.'/'.$hashImage.'.'.$ext;
 128
 129         try {
 130             $im = imagecreatefromstring($res->getBody());
 131         } catch (\Exception $e) {
 132             $im = false;
 133         }
 134
 135         if (false === $im) {
 136             $this->logger->log('error', 'Error while regenerating image', ['path' => $localPath]);
 137
 138             return false;
 139         }
 140
 141         switch ($ext) {
 142             case 'gif':
 143                 $result = imagegif($im, $localPath);
 144                 $this->logger->log('debug', 'Re-creating gif');
 145                 break;
 146             case 'jpeg':
 147             case 'jpg':
 148                 $result = imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 149                 $this->logger->log('debug', 'Re-creating jpg');
 150                 break;
 151             case 'png':
 152                 $result = imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 153                 $this->logger->log('debug', 'Re-creating png');
 154         }
 155
 156         imagedestroy($im);
 157
 158         return $this->wallabagUrl.'/assets/images/'.$relativePath.'/'.$hashImage.'.'.$ext;
 159     }
 160
 161     /**
 162      * Generate the folder where we are going to save images based on the entry url.
 163      *
 164      * @param string $url
 165      *
 166      * @return string
 167      */
 168     private function getRelativePath($url)
 169     {
 170         $hashUrl = hash('crc32', $url);
 171         $relativePath = $hashUrl[0].'/'.$hashUrl[1].'/'.$hashUrl;
 172         $folderPath = $this->baseFolder.'/'.$relativePath;
 173
 174         if (!file_exists($folderPath)) {
 175             mkdir($folderPath, 0777, true);
 176         }
 177
 178         $this->logger->log('debug', 'Folder used for that url', ['folder' => $folderPath, 'url' => $url]);
 179
 180         return $relativePath;
 181     }
 182
 183     /**
 184      * Make an $url absolute based on the $base.
 185      *
 186      * @see Graby->makeAbsoluteStr
 187      *
 188      * @param string $base Base url
 189      * @param string $url  Url to make it absolute
 190      *
 191      * @return false|string
 192      */
 193     private function getAbsoluteLink($base, $url)
 194     {
 195         if (preg_match('!^https?://!i', $url)) {
 196             // already absolute
 197             return $url;
 198         }
 199
 200         $base = new \SimplePie_IRI($base);
 201
 202         // remove '//' in URL path (causes URLs not to resolve properly)
 203         if (isset($base->ipath)) {
 204             $base->ipath = preg_replace('!//+!', '/', $base->ipath);
 205         }
 206
 207         if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
 208             return $absolute->get_uri();
 209         }
 210
 211         $this->logger->log('error', 'Can not make an absolute link', ['base' => $base, 'url' => $url]);
 212
 213         return false;
 214     }
 215 }