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