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 images URLs from it.
  35      *
  36      * @param string $html
  37      *
  38      * @return string[]
  39      */
  40     public static function extractImagesUrlsFromHtml($html)
  41     {
  42         $crawler = new Crawler($html);
  43         $imagesCrawler = $crawler
  44             ->filterXpath('//img');
  45         $imagesUrls = $imagesCrawler
  46             ->extract(['src']);
  47         $imagesSrcsetUrls = self::getSrcsetUrls($imagesCrawler);
  48
  49         return array_unique(array_merge($imagesUrls, $imagesSrcsetUrls));
  50     }
  51
  52     /**
  53      * Process the html and extract image from it, save them to local and return the updated html.
  54      *
  55      * @param int    $entryId ID of the entry
  56      * @param string $html
  57      * @param string $url     Used as a base path for relative image and folder
  58      *
  59      * @return string
  60      */
  61     public function processHtml($entryId, $html, $url)
  62     {
  63         $imagesUrls = self::extractImagesUrlsFromHtml($html);
  64
  65         $relativePath = $this->getRelativePath($entryId);
  66
  67         // download and save the image to the folder
  68         foreach ($imagesUrls as $image) {
  69             $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
  70
  71             if (false === $imagePath) {
  72                 continue;
  73             }
  74
  75             // if image contains "&" and we can't find it in the html it might be because it's encoded as &amp;
  76             if (false !== stripos($image, '&') && false === stripos($html, $image)) {
  77                 $image = str_replace('&', '&amp;', $image);
  78             }
  79
  80             $html = str_replace($image, $imagePath, $html);
  81         }
  82
  83         return $html;
  84     }
  85
  86     /**
  87      * Process a single image:
  88      *     - retrieve it
  89      *     - re-saved it (for security reason)
  90      *     - return the new local path.
  91      *
  92      * @param int    $entryId      ID of the entry
  93      * @param string $imagePath    Path to the image to retrieve
  94      * @param string $url          Url from where the image were found
  95      * @param string $relativePath Relative local path to saved the image
  96      *
  97      * @return string Relative url to access the image from the web
  98      */
  99     public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
 100     {
 101         if (null === $imagePath) {
 102             return false;
 103         }
 104
 105         if (null === $relativePath) {
 106             $relativePath = $this->getRelativePath($entryId);
 107         }
 108
 109         $this->logger->debug('DownloadImages: working on image: ' . $imagePath);
 110
 111         $folderPath = $this->baseFolder . '/' . $relativePath;
 112
 113         // build image path
 114         $absolutePath = $this->getAbsoluteLink($url, $imagePath);
 115         if (false === $absolutePath) {
 116             $this->logger->error('DownloadImages: Can not determine the absolute path for that image, skipping.');
 117
 118             return false;
 119         }
 120
 121         try {
 122             $res = $this->client->get($absolutePath);
 123         } catch (\Exception $e) {
 124             $this->logger->error('DownloadImages: Can not retrieve image, skipping.', ['exception' => $e]);
 125
 126             return false;
 127         }
 128
 129         $ext = $this->getExtensionFromResponse($res, $imagePath);
 130         if (false === $res) {
 131             return false;
 132         }
 133
 134         $hashImage = hash('crc32', $absolutePath);
 135         $localPath = $folderPath . '/' . $hashImage . '.' . $ext;
 136
 137         try {
 138             $im = imagecreatefromstring($res->getBody());
 139         } catch (\Exception $e) {
 140             $im = false;
 141         }
 142
 143         if (false === $im) {
 144             $this->logger->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
 145
 146             return false;
 147         }
 148
 149         switch ($ext) {
 150             case 'gif':
 151                 // use Imagick if available to keep GIF animation
 152                 if (class_exists('\\Imagick')) {
 153                     try {
 154                         $imagick = new \Imagick();
 155                         $imagick->readImageBlob($res->getBody());
 156                         $imagick->setImageFormat('gif');
 157                         $imagick->writeImages($localPath, true);
 158                     } catch (\Exception $e) {
 159                         // if Imagick fail, fallback to the default solution
 160                         imagegif($im, $localPath);
 161                     }
 162                 } else {
 163                     imagegif($im, $localPath);
 164                 }
 165
 166                 $this->logger->debug('DownloadImages: Re-creating gif');
 167                 break;
 168             case 'jpeg':
 169             case 'jpg':
 170                 imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
 171                 $this->logger->debug('DownloadImages: Re-creating jpg');
 172                 break;
 173             case 'png':
 174                 imagealphablending($im, false);
 175                 imagesavealpha($im, true);
 176                 imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
 177                 $this->logger->debug('DownloadImages: Re-creating png');
 178         }
 179
 180         imagedestroy($im);
 181
 182         return $this->wallabagUrl . '/assets/images/' . $relativePath . '/' . $hashImage . '.' . $ext;
 183     }
 184
 185     /**
 186      * Remove all images for the given entry id.
 187      *
 188      * @param int $entryId ID of the entry
 189      */
 190     public function removeImages($entryId)
 191     {
 192         $relativePath = $this->getRelativePath($entryId);
 193         $folderPath = $this->baseFolder . '/' . $relativePath;
 194
 195         $finder = new Finder();
 196         $finder
 197             ->files()
 198             ->ignoreDotFiles(true)
 199             ->in($folderPath);
 200
 201         foreach ($finder as $file) {
 202             @unlink($file->getRealPath());
 203         }
 204
 205         @rmdir($folderPath);
 206     }
 207
 208     /**
 209      * Get images urls from the srcset image attribute.
 210      *
 211      * @param Crawler $imagesCrawler
 212      *
 213      * @return array An array of urls
 214      */
 215     private static function getSrcsetUrls(Crawler $imagesCrawler)
 216     {
 217         $urls = [];
 218         $iterator = $imagesCrawler
 219             ->getIterator();
 220         while ($iterator->valid()) {
 221             $srcsetAttribute = $iterator->current()->getAttribute('srcset');
 222             if ('' !== $srcsetAttribute) {
 223                 // Couldn't start with " OR ' OR a white space
 224                 // Could be one or more white space
 225                 // Must be one or more digits followed by w OR x
 226                 $pattern = "/(?:[^\"'\s]+\s*(?:\d+[wx])+)/";
 227                 preg_match_all($pattern, $srcsetAttribute, $matches);
 228                 $srcset = \call_user_func_array('array_merge', $matches);
 229                 $srcsetUrls = array_map(function ($src) {
 230                     return trim(explode(' ', $src, 2)[0]);
 231                 }, $srcset);
 232                 $urls = array_merge($srcsetUrls, $urls);
 233             }
 234             $iterator->next();
 235         }
 236
 237         return $urls;
 238     }
 239
 240     /**
 241      * Setup base folder where all images are going to be saved.
 242      */
 243     private function setFolder()
 244     {
 245         // if folder doesn't exist, attempt to create one and store the folder name in property $folder
 246         if (!file_exists($this->baseFolder)) {
 247             mkdir($this->baseFolder, 0755, true);
 248         }
 249     }
 250
 251     /**
 252      * Generate the folder where we are going to save images based on the entry url.
 253      *
 254      * @param int $entryId ID of the entry
 255      *
 256      * @return string
 257      */
 258     private function getRelativePath($entryId)
 259     {
 260         $hashId = hash('crc32', $entryId);
 261         $relativePath = $hashId[0] . '/' . $hashId[1] . '/' . $hashId;
 262         $folderPath = $this->baseFolder . '/' . $relativePath;
 263
 264         if (!file_exists($folderPath)) {
 265             mkdir($folderPath, 0777, true);
 266         }
 267
 268         $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
 269
 270         return $relativePath;
 271     }
 272
 273     /**
 274      * Make an $url absolute based on the $base.
 275      *
 276      * @see Graby->makeAbsoluteStr
 277      *
 278      * @param string $base Base url
 279      * @param string $url  Url to make it absolute
 280      *
 281      * @return false|string
 282      */
 283     private function getAbsoluteLink($base, $url)
 284     {
 285         if (preg_match('!^https?://!i', $url)) {
 286             // already absolute
 287             return $url;
 288         }
 289
 290         $base = new \SimplePie_IRI($base);
 291
 292         // remove '//' in URL path (causes URLs not to resolve properly)
 293         if (isset($base->ipath)) {
 294             $base->ipath = preg_replace('!//+!', '/', $base->ipath);
 295         }
 296
 297         if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
 298             return $absolute->get_uri();
 299         }
 300
 301         $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
 302
 303         return false;
 304     }
 305
 306     /**
 307      * Retrieve and validate the extension from the response of the url of the image.
 308      *
 309      * @param Response $res       Guzzle Response
 310      * @param string   $imagePath Path from the src image from the content (used for log only)
 311      *
 312      * @return string|false Extension name or false if validation failed
 313      */
 314     private function getExtensionFromResponse(Response $res, $imagePath)
 315     {
 316         $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
 317         $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
 318
 319         // ok header doesn't have the extension, try a different way
 320         if (empty($ext)) {
 321             $types = [
 322                 'jpeg' => "\xFF\xD8\xFF",
 323                 'gif' => 'GIF',
 324                 'png' => "\x89\x50\x4e\x47\x0d\x0a",
 325             ];
 326             $bytes = substr((string) $res->getBody(), 0, 8);
 327
 328             foreach ($types as $type => $header) {
 329                 if (0 === strpos($bytes, $header)) {
 330                     $ext = $type;
 331                     break;
 332                 }
 333             }
 334
 335             $this->logger->debug('DownloadImages: Checking extension (alternative)', ['ext' => $ext]);
 336         }
 337
 338         if (!\in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
 339             $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: ' . $imagePath);
 340
 341             return false;
 342         }
 343
 344         return $ext;
 345     }
 346 }