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 GuzzleHttp\Message\Response;
   9 use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
  10 use Symfony\Component\Finder\Finder;
  11
  12 class DownloadImages
  13 {
  14     const REGENERATE_PICTURES_QUALITY = 80;
  15
  16     private $client;
  17     private $baseFolder;
  18     private $logger;
  19     private $mimeGuesser;
  20     private $wallabagUrl;
  21
  22     public function __construct(Client $client, $baseFolder, $wallabagUrl, LoggerInterface $logger)
  23     {
  24         $this->client = $client;
  25         $this->baseFolder = $baseFolder;
  26         $this->wallabagUrl = rtrim($wallabagUrl, '/');
  27         $this->logger = $logger;
  28         $this->mimeGuesser = new MimeTypeExtensionGuesser();
  29
  30         $this->setFolder();
  31     }
  32
  33     /**
  34      * Setup base folder where all images are going to be saved.
  35      */
  36     private function setFolder()
  37     {
  38         // if folder doesn't exist, attempt to create one and store the folder name in property $folder
  39         if (!file_exists($this->baseFolder)) {
  40             mkdir($this->baseFolder, 0755, true);
  41         }
  42     }
  43
  44     /**
  45      * Process the html and extract image from it, save them to local and return the updated html.
  46      *
  47      * @param int    $entryId ID of the entry
  48      * @param string $html
  49      * @param string $url     Used as a base path for relative image and folder
  50      *
  51      * @return string
  52      */
  53     public function processHtml($entryId, $html, $url)
  54     {
  55         $crawler = new Crawler($html);
  56         $result = $crawler
  57             ->filterXpath('//img')
  58             ->extract(['src']);
  59
  60         $relativePath = $this->getRelativePath($entryId);
  61
  62         // download and save the image to the folder
  63         foreach ($result as $image) {
  64             $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
  65
  66             if (false === $imagePath) {
  67                 continue;
  68             }
  69
  70             // if image contains "&" and we can't find it in the html it might be because it's encoded as &amp;
  71             if (false !== stripos($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 = $this->getExtensionFromResponse($res, $imagePath);
 121         if (false === $res) {
 122             return false;
 123         }
 124
 125         $hashImage = hash('crc32', $absolutePath);
 126         $localPath = $folderPath.'/'.$hashImage.'.'.$ext;
 127
 128         try {
 129             $im = imagecreatefromstring($res->getBody());
 130         } catch (\Exception $e) {
 131             $im = false;
 132         }
 133
 134         if (false === $im) {
 135             $this->logger->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
 136
 137             return false;
 138         }
 139
 140         switch ($ext) {
 141             case 'gif':
 142                 imagegif($im, $localPath);
 143                 $this->logger->debug('DownloadImages: Re-creating gif');
 144                 break;
 145             case 'jpeg':
 146             case 'jpg':
 147                 imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 148                 $this->logger->debug('DownloadImages: Re-creating jpg');
 149                 break;
 150             case 'png':
 151                 imagealphablending($im, false);
 152                 imagesavealpha($im, true);
 153                 imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 154                 $this->logger->debug('DownloadImages: Re-creating png');
 155         }
 156
 157         imagedestroy($im);
 158
 159         return $this->wallabagUrl.'/assets/images/'.$relativePath.'/'.$hashImage.'.'.$ext;
 160     }
 161
 162     /**
 163      * Remove all images for the given entry id.
 164      *
 165      * @param int $entryId ID of the entry
 166      */
 167     public function removeImages($entryId)
 168     {
 169         $relativePath = $this->getRelativePath($entryId);
 170         $folderPath = $this->baseFolder.'/'.$relativePath;
 171
 172         $finder = new Finder();
 173         $finder
 174             ->files()
 175             ->ignoreDotFiles(true)
 176             ->in($folderPath);
 177
 178         foreach ($finder as $file) {
 179             @unlink($file->getRealPath());
 180         }
 181
 182         @rmdir($folderPath);
 183     }
 184
 185     /**
 186      * Generate the folder where we are going to save images based on the entry url.
 187      *
 188      * @param int $entryId ID of the entry
 189      *
 190      * @return string
 191      */
 192     private function getRelativePath($entryId)
 193     {
 194         $hashId = hash('crc32', $entryId);
 195         $relativePath = $hashId[0].'/'.$hashId[1].'/'.$hashId;
 196         $folderPath = $this->baseFolder.'/'.$relativePath;
 197
 198         if (!file_exists($folderPath)) {
 199             mkdir($folderPath, 0777, true);
 200         }
 201
 202         $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
 203
 204         return $relativePath;
 205     }
 206
 207     /**
 208      * Make an $url absolute based on the $base.
 209      *
 210      * @see Graby->makeAbsoluteStr
 211      *
 212      * @param string $base Base url
 213      * @param string $url  Url to make it absolute
 214      *
 215      * @return false|string
 216      */
 217     private function getAbsoluteLink($base, $url)
 218     {
 219         if (preg_match('!^https?://!i', $url)) {
 220             // already absolute
 221             return $url;
 222         }
 223
 224         $base = new \SimplePie_IRI($base);
 225
 226         // remove '//' in URL path (causes URLs not to resolve properly)
 227         if (isset($base->ipath)) {
 228             $base->ipath = preg_replace('!//+!', '/', $base->ipath);
 229         }
 230
 231         if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
 232             return $absolute->get_uri();
 233         }
 234
 235         $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
 236
 237         return false;
 238     }
 239
 240     /**
 241      * Retrieve and validate the extension from the response of the url of the image.
 242      *
 243      * @param Response $res       Guzzle Response
 244      * @param string   $imagePath Path from the src image from the content (used for log only)
 245      *
 246      * @return string|false Extension name or false if validation failed
 247      */
 248     private function getExtensionFromResponse(Response $res, $imagePath)
 249     {
 250         $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
 251         $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
 252
 253         // ok header doesn't have the extension, try a different way
 254         if (empty($ext)) {
 255             $types = [
 256                 'jpeg' => "\xFF\xD8\xFF",
 257                 'gif' => 'GIF',
 258                 'png' => "\x89\x50\x4e\x47\x0d\x0a",
 259             ];
 260             $bytes = substr((string) $res->getBody(), 0, 8);
 261
 262             foreach ($types as $type => $header) {
 263                 if (0 === strpos($bytes, $header)) {
 264                     $ext = $type;
 265                     break;
 266                 }
 267             }
 268
 269             $this->logger->debug('DownloadImages: Checking extension (alternative)', ['ext' => $ext]);
 270         }
 271
 272         if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
 273             $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: '.$imagePath);
 274
 275             return false;
 276         }
 277
 278         return $ext;
 279     }
 280 }