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(array('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             $html = str_replace($image, $imagePath, $html);
  70         }
  71
  72         return $html;
  73     }
  74
  75     /**
  76      * Process a single image:
  77      *     - retrieve it
  78      *     - re-saved it (for security reason)
  79      *     - return the new local path.
  80      *
  81      * @param int    $entryId      ID of the entry
  82      * @param string $imagePath    Path to the image to retrieve
  83      * @param string $url          Url from where the image were found
  84      * @param string $relativePath Relative local path to saved the image
  85      *
  86      * @return string Relative url to access the image from the web
  87      */
  88     public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
  89     {
  90         if (null === $relativePath) {
  91             $relativePath = $this->getRelativePath($entryId);
  92         }
  93
  94         $this->logger->debug('DownloadImages: working on image: '.$imagePath);
  95
  96         $folderPath = $this->baseFolder.'/'.$relativePath;
  97
  98         // build image path
  99         $absolutePath = $this->getAbsoluteLink($url, $imagePath);
 100         if (false === $absolutePath) {
 101             $this->logger->error('DownloadImages: Can not determine the absolute path for that image, skipping.');
 102
 103             return false;
 104         }
 105
 106         try {
 107             $res = $this->client->get($absolutePath);
 108         } catch (\Exception $e) {
 109             $this->logger->error('DownloadImages: Can not retrieve image, skipping.', ['exception' => $e]);
 110
 111             return false;
 112         }
 113
 114         $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
 115         $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
 116         if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
 117             $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping '.$imagePath);
 118
 119             return false;
 120         }
 121         $hashImage = hash('crc32', $absolutePath);
 122         $localPath = $folderPath.'/'.$hashImage.'.'.$ext;
 123
 124         try {
 125             $im = imagecreatefromstring($res->getBody());
 126         } catch (\Exception $e) {
 127             $im = false;
 128         }
 129
 130         if (false === $im) {
 131             $this->logger->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
 132
 133             return false;
 134         }
 135
 136         switch ($ext) {
 137             case 'gif':
 138                 imagegif($im, $localPath);
 139                 $this->logger->debug('DownloadImages: Re-creating gif');
 140                 break;
 141             case 'jpeg':
 142             case 'jpg':
 143                 imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 144                 $this->logger->debug('DownloadImages: Re-creating jpg');
 145                 break;
 146             case 'png':
 147                 imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 148                 $this->logger->debug('DownloadImages: Re-creating png');
 149         }
 150
 151         imagedestroy($im);
 152
 153         return $this->wallabagUrl.'/assets/images/'.$relativePath.'/'.$hashImage.'.'.$ext;
 154     }
 155
 156     /**
 157      * Remove all images for the given entry id.
 158      *
 159      * @param int $entryId ID of the entry
 160      */
 161     public function removeImages($entryId)
 162     {
 163         $relativePath = $this->getRelativePath($entryId);
 164         $folderPath = $this->baseFolder.'/'.$relativePath;
 165
 166         $finder = new Finder();
 167         $finder
 168             ->files()
 169             ->ignoreDotFiles(true)
 170             ->in($folderPath);
 171
 172         foreach ($finder as $file) {
 173             @unlink($file->getRealPath());
 174         }
 175
 176         @rmdir($folderPath);
 177     }
 178
 179     /**
 180      * Generate the folder where we are going to save images based on the entry url.
 181      *
 182      * @param int $entryId ID of the entry
 183      *
 184      * @return string
 185      */
 186     private function getRelativePath($entryId)
 187     {
 188         $hashId = hash('crc32', $entryId);
 189         $relativePath = $hashId[0].'/'.$hashId[1].'/'.$hashId;
 190         $folderPath = $this->baseFolder.'/'.$relativePath;
 191
 192         if (!file_exists($folderPath)) {
 193             mkdir($folderPath, 0777, true);
 194         }
 195
 196         $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
 197
 198         return $relativePath;
 199     }
 200
 201     /**
 202      * Make an $url absolute based on the $base.
 203      *
 204      * @see Graby->makeAbsoluteStr
 205      *
 206      * @param string $base Base url
 207      * @param string $url  Url to make it absolute
 208      *
 209      * @return false|string
 210      */
 211     private function getAbsoluteLink($base, $url)
 212     {
 213         if (preg_match('!^https?://!i', $url)) {
 214             // already absolute
 215             return $url;
 216         }
 217
 218         $base = new \SimplePie_IRI($base);
 219
 220         // remove '//' in URL path (causes URLs not to resolve properly)
 221         if (isset($base->ipath)) {
 222             $base->ipath = preg_replace('!//+!', '/', $base->ipath);
 223         }
 224
 225         if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
 226             return $absolute->get_uri();
 227         }
 228
 229         $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
 230
 231         return false;
 232     }
 233 }