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                     try {
 141                         $imagick = new \Imagick();
 142                         $imagick->readImageBlob($res->getBody());
 143                         $imagick->setImageFormat('gif');
 144                         $imagick->writeImages($localPath, true);
 145                     } catch (\Exception $e) {
 146                         // if Imagick fail, fallback to the default solution
 147                         imagegif($im, $localPath);
 148                     }
 149                 } else {
 150                     imagegif($im, $localPath);
 151                 }
 152
 153                 $this->logger->debug('DownloadImages: Re-creating gif');
 154                 break;
 155             case 'jpeg':
 156             case 'jpg':
 157                 imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 158                 $this->logger->debug('DownloadImages: Re-creating jpg');
 159                 break;
 160             case 'png':
 161                 imagealphablending($im, false);
 162                 imagesavealpha($im, true);
 163                 imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 164                 $this->logger->debug('DownloadImages: Re-creating png');
 165         }
 166
 167         imagedestroy($im);
 168
 169         return $this->wallabagUrl . '/assets/images/' . $relativePath . '/' . $hashImage . '.' . $ext;
 170     }
 171
 172     /**
 173      * Remove all images for the given entry id.
 174      *
 175      * @param int $entryId ID of the entry
 176      */
 177     public function removeImages($entryId)
 178     {
 179         $relativePath = $this->getRelativePath($entryId);
 180         $folderPath = $this->baseFolder . '/' . $relativePath;
 181
 182         $finder = new Finder();
 183         $finder
 184             ->files()
 185             ->ignoreDotFiles(true)
 186             ->in($folderPath);
 187
 188         foreach ($finder as $file) {
 189             @unlink($file->getRealPath());
 190         }
 191
 192         @rmdir($folderPath);
 193     }
 194
 195     /**
 196      * Get images urls from the srcset image attribute.
 197      *
 198      * @param Crawler $imagesCrawler
 199      *
 200      * @return array An array of urls
 201      */
 202     private function getSrcsetUrls(Crawler $imagesCrawler)
 203     {
 204         $urls = [];
 205         $iterator = $imagesCrawler
 206             ->getIterator();
 207         while ($iterator->valid()) {
 208             $srcsetAttribute = $iterator->current()->getAttribute('srcset');
 209             if ('' !== $srcsetAttribute) {
 210                 // Couldn't start with " OR ' OR a white space
 211                 // Could be one or more white space
 212                 // Must be one or more digits followed by w OR x
 213                 $pattern = "/(?:[^\"'\s]+\s*(?:\d+[wx])+)/";
 214                 preg_match_all($pattern, $srcsetAttribute, $matches);
 215                 $srcset = \call_user_func_array('array_merge', $matches);
 216                 $srcsetUrls = array_map(function ($src) {
 217                     return trim(explode(' ', $src, 2)[0]);
 218                 }, $srcset);
 219                 $urls = array_merge($srcsetUrls, $urls);
 220             }
 221             $iterator->next();
 222         }
 223
 224         return $urls;
 225     }
 226
 227     /**
 228      * Setup base folder where all images are going to be saved.
 229      */
 230     private function setFolder()
 231     {
 232         // if folder doesn't exist, attempt to create one and store the folder name in property $folder
 233         if (!file_exists($this->baseFolder)) {
 234             mkdir($this->baseFolder, 0755, true);
 235         }
 236     }
 237
 238     /**
 239      * Generate the folder where we are going to save images based on the entry url.
 240      *
 241      * @param int $entryId ID of the entry
 242      *
 243      * @return string
 244      */
 245     private function getRelativePath($entryId)
 246     {
 247         $hashId = hash('crc32', $entryId);
 248         $relativePath = $hashId[0] . '/' . $hashId[1] . '/' . $hashId;
 249         $folderPath = $this->baseFolder . '/' . $relativePath;
 250
 251         if (!file_exists($folderPath)) {
 252             mkdir($folderPath, 0777, true);
 253         }
 254
 255         $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
 256
 257         return $relativePath;
 258     }
 259
 260     /**
 261      * Make an $url absolute based on the $base.
 262      *
 263      * @see Graby->makeAbsoluteStr
 264      *
 265      * @param string $base Base url
 266      * @param string $url  Url to make it absolute
 267      *
 268      * @return false|string
 269      */
 270     private function getAbsoluteLink($base, $url)
 271     {
 272         if (preg_match('!^https?://!i', $url)) {
 273             // already absolute
 274             return $url;
 275         }
 276
 277         $base = new \SimplePie_IRI($base);
 278
 279         // remove '//' in URL path (causes URLs not to resolve properly)
 280         if (isset($base->ipath)) {
 281             $base->ipath = preg_replace('!//+!', '/', $base->ipath);
 282         }
 283
 284         if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
 285             return $absolute->get_uri();
 286         }
 287
 288         $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
 289
 290         return false;
 291     }
 292
 293     /**
 294      * Retrieve and validate the extension from the response of the url of the image.
 295      *
 296      * @param Response $res       Guzzle Response
 297      * @param string   $imagePath Path from the src image from the content (used for log only)
 298      *
 299      * @return string|false Extension name or false if validation failed
 300      */
 301     private function getExtensionFromResponse(Response $res, $imagePath)
 302     {
 303         $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
 304         $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
 305
 306         // ok header doesn't have the extension, try a different way
 307         if (empty($ext)) {
 308             $types = [
 309                 'jpeg' => "\xFF\xD8\xFF",
 310                 'gif' => 'GIF',
 311                 'png' => "\x89\x50\x4e\x47\x0d\x0a",
 312             ];
 313             $bytes = substr((string) $res->getBody(), 0, 8);
 314
 315             foreach ($types as $type => $header) {
 316                 if (0 === strpos($bytes, $header)) {
 317                     $ext = $type;
 318                     break;
 319                 }
 320             }
 321
 322             $this->logger->debug('DownloadImages: Checking extension (alternative)', ['ext' => $ext]);
 323         }
 324
 325         if (!\in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
 326             $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: ' . $imagePath);
 327
 328             return false;
 329         }
 330
 331         return $ext;
 332     }
 333 }