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 === $imagePath) {
  66                 continue;
  67             }
  68
  69             // if image contains "&"" and we can't find it in the html
  70             // it might be because it's encoded as &amp;
  71             if (false !== stripos($image, '&') && false === stripos($html, $image)) {
  72                 $image = str_replace('&', '&amp;', $image);
  73             }
  74
  75             $html = str_replace($image, $imagePath, $html);
  76         }
  77
  78         return $html;
  79     }
  80
  81     /**
  82      * Process a single image:
  83      *     - retrieve it
  84      *     - re-saved it (for security reason)
  85      *     - return the new local path.
  86      *
  87      * @param int    $entryId      ID of the entry
  88      * @param string $imagePath    Path to the image to retrieve
  89      * @param string $url          Url from where the image were found
  90      * @param string $relativePath Relative local path to saved the image
  91      *
  92      * @return string Relative url to access the image from the web
  93      */
  94     public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
  95     {
  96         if (null === $relativePath) {
  97             $relativePath = $this->getRelativePath($entryId);
  98         }
  99
 100         $this->logger->debug('DownloadImages: working on image: '.$imagePath);
 101
 102         $folderPath = $this->baseFolder.'/'.$relativePath;
 103
 104         // build image path
 105         $absolutePath = $this->getAbsoluteLink($url, $imagePath);
 106         if (false === $absolutePath) {
 107             $this->logger->error('DownloadImages: Can not determine the absolute path for that image, skipping.');
 108
 109             return false;
 110         }
 111
 112         try {
 113             $res = $this->client->get($absolutePath);
 114         } catch (\Exception $e) {
 115             $this->logger->error('DownloadImages: Can not retrieve image, skipping.', ['exception' => $e]);
 116
 117             return false;
 118         }
 119
 120         $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
 121         $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
 122         if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
 123             $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: '.$imagePath);
 124
 125             return false;
 126         }
 127         $hashImage = hash('crc32', $absolutePath);
 128         $localPath = $folderPath.'/'.$hashImage.'.'.$ext;
 129
 130         try {
 131             $im = imagecreatefromstring($res->getBody());
 132         } catch (\Exception $e) {
 133             $im = false;
 134         }
 135
 136         if (false === $im) {
 137             $this->logger->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
 138
 139             return false;
 140         }
 141
 142         switch ($ext) {
 143             case 'gif':
 144                 imagegif($im, $localPath);
 145                 $this->logger->debug('DownloadImages: Re-creating gif');
 146                 break;
 147             case 'jpeg':
 148             case 'jpg':
 149                 imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 150                 $this->logger->debug('DownloadImages: Re-creating jpg');
 151                 break;
 152             case 'png':
 153                 imagealphablending($im, false);
 154                 imagesavealpha($im, true);
 155                 imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 156                 $this->logger->debug('DownloadImages: Re-creating png');
 157         }
 158
 159         imagedestroy($im);
 160
 161         return $this->wallabagUrl.'/assets/images/'.$relativePath.'/'.$hashImage.'.'.$ext;
 162     }
 163
 164     /**
 165      * Remove all images for the given entry id.
 166      *
 167      * @param int $entryId ID of the entry
 168      */
 169     public function removeImages($entryId)
 170     {
 171         $relativePath = $this->getRelativePath($entryId);
 172         $folderPath = $this->baseFolder.'/'.$relativePath;
 173
 174         $finder = new Finder();
 175         $finder
 176             ->files()
 177             ->ignoreDotFiles(true)
 178             ->in($folderPath);
 179
 180         foreach ($finder as $file) {
 181             @unlink($file->getRealPath());
 182         }
 183
 184         @rmdir($folderPath);
 185     }
 186
 187     /**
 188      * Generate the folder where we are going to save images based on the entry url.
 189      *
 190      * @param int $entryId ID of the entry
 191      *
 192      * @return string
 193      */
 194     private function getRelativePath($entryId)
 195     {
 196         $hashId = hash('crc32', $entryId);
 197         $relativePath = $hashId[0].'/'.$hashId[1].'/'.$hashId;
 198         $folderPath = $this->baseFolder.'/'.$relativePath;
 199
 200         if (!file_exists($folderPath)) {
 201             mkdir($folderPath, 0777, true);
 202         }
 203
 204         $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
 205
 206         return $relativePath;
 207     }
 208
 209     /**
 210      * Make an $url absolute based on the $base.
 211      *
 212      * @see Graby->makeAbsoluteStr
 213      *
 214      * @param string $base Base url
 215      * @param string $url  Url to make it absolute
 216      *
 217      * @return false|string
 218      */
 219     private function getAbsoluteLink($base, $url)
 220     {
 221         if (preg_match('!^https?://!i', $url)) {
 222             // already absolute
 223             return $url;
 224         }
 225
 226         $base = new \SimplePie_IRI($base);
 227
 228         // remove '//' in URL path (causes URLs not to resolve properly)
 229         if (isset($base->ipath)) {
 230             $base->ipath = preg_replace('!//+!', '/', $base->ipath);
 231         }
 232
 233         if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
 234             return $absolute->get_uri();
 235         }
 236
 237         $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
 238
 239         return false;
 240     }
 241 }