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