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         $imagesCrawler = $crawler
  46             ->filterXpath('//img');
  47         $imagesUrls = $imagesCrawler
  48             ->extract(['src']);
  49         $imagesSrcsetUrls = $this->getSrcsetUrls($imagesCrawler);
  50         $imagesUrls = array_unique(array_merge($imagesUrls, $imagesSrcsetUrls));
  51
  52         $relativePath = $this->getRelativePath($entryId);
  53
  54         // download and save the image to the folder
  55         foreach ($imagesUrls as $image) {
  56             $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
  57
  58             if (false === $imagePath) {
  59                 continue;
  60             }
  61
  62             // if image contains "&" and we can't find it in the html it might be because it's encoded as &amp;
  63             if (false !== stripos($image, '&') && false === stripos($html, $image)) {
  64                 $image = str_replace('&', '&amp;', $image);
  65             }
  66
  67             $html = str_replace($image, $imagePath, $html);
  68         }
  69
  70         return $html;
  71     }
  72
  73     /**
  74      * Process a single image:
  75      *     - retrieve it
  76      *     - re-saved it (for security reason)
  77      *     - return the new local path.
  78      *
  79      * @param int    $entryId      ID of the entry
  80      * @param string $imagePath    Path to the image to retrieve
  81      * @param string $url          Url from where the image were found
  82      * @param string $relativePath Relative local path to saved the image
  83      *
  84      * @return string Relative url to access the image from the web
  85      */
  86     public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
  87     {
  88         if (null === $imagePath) {
  89             return false;
  90         }
  91
  92         if (null === $relativePath) {
  93             $relativePath = $this->getRelativePath($entryId);
  94         }
  95
  96         $this->logger->debug('DownloadImages: working on image: ' . $imagePath);
  97
  98         $folderPath = $this->baseFolder . '/' . $relativePath;
  99
 100         // build image path
 101         $absolutePath = $this->getAbsoluteLink($url, $imagePath);
 102         if (false === $absolutePath) {
 103             $this->logger->error('DownloadImages: Can not determine the absolute path for that image, skipping.');
 104
 105             return false;
 106         }
 107
 108         try {
 109             $res = $this->client->get($absolutePath);
 110         } catch (\Exception $e) {
 111             $this->logger->error('DownloadImages: Can not retrieve image, skipping.', ['exception' => $e]);
 112
 113             return false;
 114         }
 115
 116         $ext = $this->getExtensionFromResponse($res, $imagePath);
 117         if (false === $res) {
 118             return false;
 119         }
 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                 // use Imagick if available to keep GIF animation
 139                 if (class_exists('\\Imagick')) {
 140                     $imagick = new \Imagick();
 141                     $imagick->readImageBlob($res->getBody());
 142                     $imagick->setImageFormat('gif');
 143                     $imagick->writeImages($localPath, true);
 144                 } else {
 145                     imagegif($im, $localPath);
 146                 }
 147
 148                 $this->logger->debug('DownloadImages: Re-creating gif');
 149                 break;
 150             case 'jpeg':
 151             case 'jpg':
 152                 imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 153                 $this->logger->debug('DownloadImages: Re-creating jpg');
 154                 break;
 155             case 'png':
 156                 imagealphablending($im, false);
 157                 imagesavealpha($im, true);
 158                 imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 159                 $this->logger->debug('DownloadImages: Re-creating png');
 160         }
 161
 162         imagedestroy($im);
 163
 164         return $this->wallabagUrl . '/assets/images/' . $relativePath . '/' . $hashImage . '.' . $ext;
 165     }
 166
 167     /**
 168      * Remove all images for the given entry id.
 169      *
 170      * @param int $entryId ID of the entry
 171      */
 172     public function removeImages($entryId)
 173     {
 174         $relativePath = $this->getRelativePath($entryId);
 175         $folderPath = $this->baseFolder . '/' . $relativePath;
 176
 177         $finder = new Finder();
 178         $finder
 179             ->files()
 180             ->ignoreDotFiles(true)
 181             ->in($folderPath);
 182
 183         foreach ($finder as $file) {
 184             @unlink($file->getRealPath());
 185         }
 186
 187         @rmdir($folderPath);
 188     }
 189
 190     /**
 191      * Get images urls from the srcset image attribute.
 192      *
 193      * @param Crawler $imagesCrawler
 194      *
 195      * @return array An array of urls
 196      */
 197     private function getSrcsetUrls(Crawler $imagesCrawler)
 198     {
 199         $urls = [];
 200         $iterator = $imagesCrawler
 201             ->getIterator();
 202         while ($iterator->valid()) {
 203             $srcsetAttribute = $iterator->current()->getAttribute('srcset');
 204             if ('' !== $srcsetAttribute) {
 205                 // Couldn't start with " OR ' OR a white space
 206                 // Could be one or more white space
 207                 // Must be one or more digits followed by w OR x
 208                 $pattern = "/(?:[^\"'\s]+\s*(?:\d+[wx])+)/";
 209                 preg_match_all($pattern, $srcsetAttribute, $matches);
 210                 $srcset = \call_user_func_array('array_merge', $matches);
 211                 $srcsetUrls = array_map(function ($src) {
 212                     return trim(explode(' ', $src, 2)[0]);
 213                 }, $srcset);
 214                 $urls = array_merge($srcsetUrls, $urls);
 215             }
 216             $iterator->next();
 217         }
 218
 219         return $urls;
 220     }
 221
 222     /**
 223      * Setup base folder where all images are going to be saved.
 224      */
 225     private function setFolder()
 226     {
 227         // if folder doesn't exist, attempt to create one and store the folder name in property $folder
 228         if (!file_exists($this->baseFolder)) {
 229             mkdir($this->baseFolder, 0755, true);
 230         }
 231     }
 232
 233     /**
 234      * Generate the folder where we are going to save images based on the entry url.
 235      *
 236      * @param int $entryId ID of the entry
 237      *
 238      * @return string
 239      */
 240     private function getRelativePath($entryId)
 241     {
 242         $hashId = hash('crc32', $entryId);
 243         $relativePath = $hashId[0] . '/' . $hashId[1] . '/' . $hashId;
 244         $folderPath = $this->baseFolder . '/' . $relativePath;
 245
 246         if (!file_exists($folderPath)) {
 247             mkdir($folderPath, 0777, true);
 248         }
 249
 250         $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
 251
 252         return $relativePath;
 253     }
 254
 255     /**
 256      * Make an $url absolute based on the $base.
 257      *
 258      * @see Graby->makeAbsoluteStr
 259      *
 260      * @param string $base Base url
 261      * @param string $url  Url to make it absolute
 262      *
 263      * @return false|string
 264      */
 265     private function getAbsoluteLink($base, $url)
 266     {
 267         if (preg_match('!^https?://!i', $url)) {
 268             // already absolute
 269             return $url;
 270         }
 271
 272         $base = new \SimplePie_IRI($base);
 273
 274         // remove '//' in URL path (causes URLs not to resolve properly)
 275         if (isset($base->ipath)) {
 276             $base->ipath = preg_replace('!//+!', '/', $base->ipath);
 277         }
 278
 279         if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
 280             return $absolute->get_uri();
 281         }
 282
 283         $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
 284
 285         return false;
 286     }
 287
 288     /**
 289      * Retrieve and validate the extension from the response of the url of the image.
 290      *
 291      * @param Response $res       Guzzle Response
 292      * @param string   $imagePath Path from the src image from the content (used for log only)
 293      *
 294      * @return string|false Extension name or false if validation failed
 295      */
 296     private function getExtensionFromResponse(Response $res, $imagePath)
 297     {
 298         $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
 299         $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
 300
 301         // ok header doesn't have the extension, try a different way
 302         if (empty($ext)) {
 303             $types = [
 304                 'jpeg' => "\xFF\xD8\xFF",
 305                 'gif' => 'GIF',
 306                 'png' => "\x89\x50\x4e\x47\x0d\x0a",
 307             ];
 308             $bytes = substr((string) $res->getBody(), 0, 8);
 309
 310             foreach ($types as $type => $header) {
 311                 if (0 === strpos($bytes, $header)) {
 312                     $ext = $type;
 313                     break;
 314                 }
 315             }
 316
 317             $this->logger->debug('DownloadImages: Checking extension (alternative)', ['ext' => $ext]);
 318         }
 319
 320         if (!\in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
 321             $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: ' . $imagePath);
 322
 323             return false;
 324         }
 325
 326         return $ext;
 327     }
 328 }