src/Wallabag/CoreBundle/Helper/DownloadImages.php

   1 <?php
   2
   3 namespace Wallabag\CoreBundle\Helper;
   4
   5 use GuzzleHttp\Client;
   6 use GuzzleHttp\Message\Response;
   7 use Psr\Log\LoggerInterface;
   8 use Symfony\Component\DomCrawler\Crawler;
   9 use Symfony\Component\Finder\Finder;
  10 use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
  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      * Process the html and extract image from it, save them to local and return the updated html.
  35      *
  36      * @param int    $entryId ID of the entry
  37      * @param string $html
  38      * @param string $url     Used as a base path for relative image and folder
  39      *
  40      * @return string
  41      */
  42     public function processHtml($entryId, $html, $url)
  43     {
  44         $crawler = new Crawler($html);
  45         $result = $crawler
  46             ->filterXpath('//img')
  47             ->extract(['src']);
  48
  49         $relativePath = $this->getRelativePath($entryId);
  50
  51         // download and save the image to the folder
  52         foreach ($result as $image) {
  53             $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
  54
  55             if (false === $imagePath) {
  56                 continue;
  57             }
  58
  59             // if image contains "&" and we can't find it in the html it might be because it's encoded as &amp;
  60             if (false !== stripos($image, '&') && false === stripos($html, $image)) {
  61                 $image = str_replace('&', '&amp;', $image);
  62             }
  63
  64             $html = str_replace($image, $imagePath, $html);
  65         }
  66
  67         return $html;
  68     }
  69
  70     /**
  71      * Process a single image:
  72      *     - retrieve it
  73      *     - re-saved it (for security reason)
  74      *     - return the new local path.
  75      *
  76      * @param int    $entryId      ID of the entry
  77      * @param string $imagePath    Path to the image to retrieve
  78      * @param string $url          Url from where the image were found
  79      * @param string $relativePath Relative local path to saved the image
  80      *
  81      * @return string Relative url to access the image from the web
  82      */
  83     public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
  84     {
  85         if (null === $relativePath) {
  86             $relativePath = $this->getRelativePath($entryId);
  87         }
  88
  89         $this->logger->debug('DownloadImages: working on image: ' . $imagePath);
  90
  91         $folderPath = $this->baseFolder . '/' . $relativePath;
  92
  93         // build image path
  94         $absolutePath = $this->getAbsoluteLink($url, $imagePath);
  95         if (false === $absolutePath) {
  96             $this->logger->error('DownloadImages: Can not determine the absolute path for that image, skipping.');
  97
  98             return false;
  99         }
 100
 101         try {
 102             $res = $this->client->get($absolutePath);
 103         } catch (\Exception $e) {
 104             $this->logger->error('DownloadImages: Can not retrieve image, skipping.', ['exception' => $e]);
 105
 106             return false;
 107         }
 108
 109         $ext = $this->getExtensionFromResponse($res, $imagePath);
 110         if (false === $res) {
 111             return false;
 112         }
 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->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
 125
 126             return false;
 127         }
 128
 129         switch ($ext) {
 130             case 'gif':
 131                 imagegif($im, $localPath);
 132                 $this->logger->debug('DownloadImages: Re-creating gif');
 133                 break;
 134             case 'jpeg':
 135             case 'jpg':
 136                 imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 137                 $this->logger->debug('DownloadImages: Re-creating jpg');
 138                 break;
 139             case 'png':
 140                 imagealphablending($im, false);
 141                 imagesavealpha($im, true);
 142                 imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 143                 $this->logger->debug('DownloadImages: Re-creating png');
 144         }
 145
 146         imagedestroy($im);
 147
 148         return $this->wallabagUrl . '/assets/images/' . $relativePath . '/' . $hashImage . '.' . $ext;
 149     }
 150
 151     /**
 152      * Remove all images for the given entry id.
 153      *
 154      * @param int $entryId ID of the entry
 155      */
 156     public function removeImages($entryId)
 157     {
 158         $relativePath = $this->getRelativePath($entryId);
 159         $folderPath = $this->baseFolder . '/' . $relativePath;
 160
 161         $finder = new Finder();
 162         $finder
 163             ->files()
 164             ->ignoreDotFiles(true)
 165             ->in($folderPath);
 166
 167         foreach ($finder as $file) {
 168             @unlink($file->getRealPath());
 169         }
 170
 171         @rmdir($folderPath);
 172     }
 173
 174     /**
 175      * Setup base folder where all images are going to be saved.
 176      */
 177     private function setFolder()
 178     {
 179         // if folder doesn't exist, attempt to create one and store the folder name in property $folder
 180         if (!file_exists($this->baseFolder)) {
 181             mkdir($this->baseFolder, 0755, true);
 182         }
 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 }