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 use Symfony\Component\Finder\Finder;
  10
  11 class DownloadImages
  12 {
  13     const REGENERATE_PICTURES_QUALITY = 80;
  14
  15     private $client;
  16     private $baseFolder;
  17     private $logger;
  18     private $mimeGuesser;
  19     private $wallabagUrl;
  20
  21     public function __construct(Client $client, $baseFolder, $wallabagUrl, LoggerInterface $logger)
  22     {
  23         $this->client = $client;
  24         $this->baseFolder = $baseFolder;
  25         $this->wallabagUrl = rtrim($wallabagUrl, '/');
  26         $this->logger = $logger;
  27         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  28
  29         $this->setFolder();
  30     }
  31
  32     /**
  33      * Setup base folder where all images are going to be saved.
  34      */
  35     private function setFolder()
  36     {
  37         // if folder doesn't exist, attempt to create one and store the folder name in property $folder
  38         if (!file_exists($this->baseFolder)) {
  39             mkdir($this->baseFolder, 0755, true);
  40         }
  41     }
  42
  43     /**
  44      * Process the html and extract image from it, save them to local and return the updated html.
  45      *
  46      * @param int    $entryId ID of the entry
  47      * @param string $html
  48      * @param string $url     Used as a base path for relative image and folder
  49      *
  50      * @return string
  51      */
  52     public function processHtml($entryId, $html, $url)
  53     {
  54         $crawler = new Crawler($html);
  55         $result = $crawler
  56             ->filterXpath('//img')
  57             ->extract(['src']);
  58
  59         $relativePath = $this->getRelativePath($entryId);
  60
  61         // download and save the image to the folder
  62         foreach ($result as $image) {
  63             $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
  64
  65             if (false === $imagePath) {
  66                 continue;
  67             }
  68
  69             // if image contains "&" and we can't find it in the html it might be because it's encoded as &amp;
  70             if (false !== stripos($image, '&') && false === stripos($html, $image)) {
  71                 $image = str_replace('&', '&amp;', $image);
  72             }
  73
  74             $html = str_replace($image, $imagePath, $html);
  75         }
  76
  77         return $html;
  78     }
  79
  80     /**
  81      * Process a single image:
  82      *     - retrieve it
  83      *     - re-saved it (for security reason)
  84      *     - return the new local path.
  85      *
  86      * @param int    $entryId      ID of the entry
  87      * @param string $imagePath    Path to the image to retrieve
  88      * @param string $url          Url from where the image were found
  89      * @param string $relativePath Relative local path to saved the image
  90      *
  91      * @return string Relative url to access the image from the web
  92      */
  93     public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
  94     {
  95         if (null === $relativePath) {
  96             $relativePath = $this->getRelativePath($entryId);
  97         }
  98
  99         $this->logger->debug('DownloadImages: working on image: '.$imagePath);
 100
 101         $folderPath = $this->baseFolder.'/'.$relativePath;
 102
 103         // build image path
 104         $absolutePath = $this->getAbsoluteLink($url, $imagePath);
 105         if (false === $absolutePath) {
 106             $this->logger->error('DownloadImages: 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->error('DownloadImages: 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->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
 121         if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
 122             $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: '.$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->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
 137
 138             return false;
 139         }
 140
 141         switch ($ext) {
 142             case 'gif':
 143                 imagegif($im, $localPath);
 144                 $this->logger->debug('DownloadImages: Re-creating gif');
 145                 break;
 146             case 'jpeg':
 147             case 'jpg':
 148                 imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 149                 $this->logger->debug('DownloadImages: Re-creating jpg');
 150                 break;
 151             case 'png':
 152                 imagealphablending($im, false);
 153                 imagesavealpha($im, true);
 154                 imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 155                 $this->logger->debug('DownloadImages: Re-creating png');
 156         }
 157
 158         imagedestroy($im);
 159
 160         return $this->wallabagUrl.'/assets/images/'.$relativePath.'/'.$hashImage.'.'.$ext;
 161     }
 162
 163     /**
 164      * Remove all images for the given entry id.
 165      *
 166      * @param int $entryId ID of the entry
 167      */
 168     public function removeImages($entryId)
 169     {
 170         $relativePath = $this->getRelativePath($entryId);
 171         $folderPath = $this->baseFolder.'/'.$relativePath;
 172
 173         $finder = new Finder();
 174         $finder
 175             ->files()
 176             ->ignoreDotFiles(true)
 177             ->in($folderPath);
 178
 179         foreach ($finder as $file) {
 180             @unlink($file->getRealPath());
 181         }
 182
 183         @rmdir($folderPath);
 184     }
 185
 186     /**
 187      * Generate the folder where we are going to save images based on the entry url.
 188      *
 189      * @param int $entryId ID of the entry
 190      *
 191      * @return string
 192      */
 193     private function getRelativePath($entryId)
 194     {
 195         $hashId = hash('crc32', $entryId);
 196         $relativePath = $hashId[0].'/'.$hashId[1].'/'.$hashId;
 197         $folderPath = $this->baseFolder.'/'.$relativePath;
 198
 199         if (!file_exists($folderPath)) {
 200             mkdir($folderPath, 0777, true);
 201         }
 202
 203         $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
 204
 205         return $relativePath;
 206     }
 207
 208     /**
 209      * Make an $url absolute based on the $base.
 210      *
 211      * @see Graby->makeAbsoluteStr
 212      *
 213      * @param string $base Base url
 214      * @param string $url  Url to make it absolute
 215      *
 216      * @return false|string
 217      */
 218     private function getAbsoluteLink($base, $url)
 219     {
 220         if (preg_match('!^https?://!i', $url)) {
 221             // already absolute
 222             return $url;
 223         }
 224
 225         $base = new \SimplePie_IRI($base);
 226
 227         // remove '//' in URL path (causes URLs not to resolve properly)
 228         if (isset($base->ipath)) {
 229             $base->ipath = preg_replace('!//+!', '/', $base->ipath);
 230         }
 231
 232         if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
 233             return $absolute->get_uri();
 234         }
 235
 236         $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
 237
 238         return false;
 239     }
 240 }