namespace Wallabag\CoreBundle\Helper;
-use Psr\Log\LoggerInterface as Logger;
+use GuzzleHttp\Client;
+use GuzzleHttp\Message\Response;
+use Psr\Log\LoggerInterface;
use Symfony\Component\DomCrawler\Crawler;
-
-define('REGENERATE_PICTURES_QUALITY', 75);
-define('HTTP_PORT', 80);
-define('SSL_PORT', 443);
-define('BASE_URL', '');
+use Symfony\Component\Finder\Finder;
+use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
class DownloadImages
{
- private $folder;
- private $url;
- private $html;
- private $fileName;
+ const REGENERATE_PICTURES_QUALITY = 80;
+
+ private $client;
+ private $baseFolder;
private $logger;
+ private $mimeGuesser;
+ private $wallabagUrl;
- public function __construct($html, $url, Logger $logger)
+ public function __construct(Client $client, $baseFolder, $wallabagUrl, LoggerInterface $logger)
{
- $this->html = $html;
- $this->url = $url;
- $this->setFolder();
+ $this->client = $client;
+ $this->baseFolder = $baseFolder;
+ $this->wallabagUrl = rtrim($wallabagUrl, '/');
$this->logger = $logger;
- }
+ $this->mimeGuesser = new MimeTypeExtensionGuesser();
- public function setFolder($folder = 'assets/images')
- {
- // if folder doesn't exist, attempt to create one and store the folder name in property $folder
- if (!file_exists($folder)) {
- mkdir($folder);
- }
- $this->folder = $folder;
+ $this->setFolder();
}
- public function process()
+ /**
+ * Process the html and extract image from it, save them to local and return the updated html.
+ *
+ * @param int $entryId ID of the entry
+ * @param string $html
+ * @param string $url Used as a base path for relative image and folder
+ *
+ * @return string
+ */
+ public function processHtml($entryId, $html, $url)
{
- //instantiate the symfony DomCrawler Component
- $crawler = new Crawler($this->html);
- // create an array of all scrapped image links
- $this->logger->log('debug', 'Finding images inside document');
- $result = $crawler
- ->filterXpath('//img')
- ->extract(array('src'));
+ $crawler = new Crawler($html);
+ $imagesCrawler = $crawler
+ ->filterXpath('//img');
+ $imagesUrls = $imagesCrawler
+ ->extract(['src']);
+ $imagesSrcsetUrls = $this->getSrcsetUrls($imagesCrawler);
+ $imagesUrls = array_unique(array_merge($imagesUrls, $imagesSrcsetUrls));
+
+ $relativePath = $this->getRelativePath($entryId);
// download and save the image to the folder
- foreach ($result as $image) {
- $file = file_get_contents($image);
+ foreach ($imagesUrls as $image) {
+ $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
+
+ if (false === $imagePath) {
+ continue;
+ }
- // Checks
- $absolute_path = self::getAbsoluteLink($image, $this->url);
- $filename = basename(parse_url($absolute_path, PHP_URL_PATH));
- $fullpath = $this->folder.'/'.$filename;
- self::checks($file, $fullpath, $absolute_path);
- $this->html = str_replace($image, self::getPocheUrl().'/'.$fullpath, $this->html);
+ // if image contains "&" and we can't find it in the html it might be because it's encoded as &
+ if (false !== stripos($image, '&') && false === stripos($html, $image)) {
+ $image = str_replace('&', '&', $image);
+ }
+
+ $html = str_replace($image, $imagePath, $html);
}
- return $this->html;
+ return $html;
}
- private function checks($rawdata, $fullpath, $absolute_path)
+ /**
+ * Process a single image:
+ * - retrieve it
+ * - re-saved it (for security reason)
+ * - return the new local path.
+ *
+ * @param int $entryId ID of the entry
+ * @param string $imagePath Path to the image to retrieve
+ * @param string $url Url from where the image were found
+ * @param string $relativePath Relative local path to saved the image
+ *
+ * @return string Relative url to access the image from the web
+ */
+ public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
{
- $fullpath = urldecode($fullpath);
+ if (null === $imagePath) {
+ return false;
+ }
- if (file_exists($fullpath)) {
- unlink($fullpath);
+ if (null === $relativePath) {
+ $relativePath = $this->getRelativePath($entryId);
}
- // check extension
- $this->logger->log('debug', 'Checking extension');
+ $this->logger->debug('DownloadImages: working on image: ' . $imagePath);
- $file_ext = strrchr($fullpath, '.');
- $whitelist = array('.jpg', '.jpeg', '.gif', '.png');
- if (!(in_array($file_ext, $whitelist))) {
- $this->logger->log('debug', 'processed image with not allowed extension. Skipping '.$fullpath);
+ $folderPath = $this->baseFolder . '/' . $relativePath;
+
+ // build image path
+ $absolutePath = $this->getAbsoluteLink($url, $imagePath);
+ if (false === $absolutePath) {
+ $this->logger->error('DownloadImages: Can not determine the absolute path for that image, skipping.');
return false;
}
- // check headers
- $this->logger->log('debug', 'Checking headers');
- $imageinfo = getimagesize($absolute_path);
- if ($imageinfo['mime'] != 'image/gif' && $imageinfo['mime'] != 'image/jpeg' && $imageinfo['mime'] != 'image/jpg' && $imageinfo['mime'] != 'image/png') {
- $this->logger->log('debug', 'processed image with bad header. Skipping '.$fullpath);
+ try {
+ $res = $this->client->get($absolutePath);
+ } catch (\Exception $e) {
+ $this->logger->error('DownloadImages: Can not retrieve image, skipping.', ['exception' => $e]);
+
+ return false;
+ }
+ $ext = $this->getExtensionFromResponse($res, $imagePath);
+ if (false === $res) {
return false;
}
- // regenerate image
- $this->logger->log('debug', 'regenerating image');
- $im = imagecreatefromstring($rawdata);
- if ($im === false) {
- $this->logger->log('error', 'error while regenerating image '.$fullpath);
+ $hashImage = hash('crc32', $absolutePath);
+ $localPath = $folderPath . '/' . $hashImage . '.' . $ext;
+
+ try {
+ $im = imagecreatefromstring($res->getBody());
+ } catch (\Exception $e) {
+ $im = false;
+ }
+
+ if (false === $im) {
+ $this->logger->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
return false;
}
- switch ($imageinfo['mime']) {
- case 'image/gif':
- $result = imagegif($im, $fullpath);
- $this->logger->log('debug', 'Re-creating gif');
- break;
- case 'image/jpeg':
- case 'image/jpg':
- $result = imagejpeg($im, $fullpath, REGENERATE_PICTURES_QUALITY);
- $this->logger->log('debug', 'Re-creating jpg');
+ switch ($ext) {
+ case 'gif':
+ imagegif($im, $localPath);
+ $this->logger->debug('DownloadImages: Re-creating gif');
break;
- case 'image/png':
- $this->logger->log('debug', 'Re-creating png');
- $result = imagepng($im, $fullpath, ceil(REGENERATE_PICTURES_QUALITY / 100 * 9));
+ case 'jpeg':
+ case 'jpg':
+ imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
+ $this->logger->debug('DownloadImages: Re-creating jpg');
break;
+ case 'png':
+ imagealphablending($im, false);
+ imagesavealpha($im, true);
+ imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
+ $this->logger->debug('DownloadImages: Re-creating png');
}
+
imagedestroy($im);
- return $result;
+ return $this->wallabagUrl . '/assets/images/' . $relativePath . '/' . $hashImage . '.' . $ext;
}
- private static function getAbsoluteLink($relativeLink, $url)
+ /**
+ * Remove all images for the given entry id.
+ *
+ * @param int $entryId ID of the entry
+ */
+ public function removeImages($entryId)
{
- /* return if already absolute URL */
- if (parse_url($relativeLink, PHP_URL_SCHEME) != '') {
- return $relativeLink;
+ $relativePath = $this->getRelativePath($entryId);
+ $folderPath = $this->baseFolder . '/' . $relativePath;
+
+ $finder = new Finder();
+ $finder
+ ->files()
+ ->ignoreDotFiles(true)
+ ->in($folderPath);
+
+ foreach ($finder as $file) {
+ @unlink($file->getRealPath());
}
- /* queries and anchors */
- if ($relativeLink[0] == '#' || $relativeLink[0] == '?') {
- return $url.$relativeLink;
+ @rmdir($folderPath);
+ }
+
+ /**
+ * Get images urls from the srcset image attribute.
+ *
+ * @param Crawler $imagesCrawler
+ *
+ * @return array An array of urls
+ */
+ protected function getSrcsetUrls(Crawler $imagesCrawler)
+ {
+ $urls = [];
+ $iterator = $imagesCrawler
+ ->getIterator();
+ while ($iterator->valid()) {
+ $srcsetAttribute = $iterator->current()->getAttribute('srcset');
+ if ('' !== $srcsetAttribute) {
+ $srcset = array_map('trim', explode(',', $srcsetAttribute));
+ $srcsetUrls = array_map(function ($src) {
+ return explode(' ', $src)[0];
+ }, $srcset);
+ $urls = array_merge($srcsetUrls, $urls);
+ }
+ $iterator->next();
+ }
+
+ return $urls;
+ }
+
+ /**
+ * Setup base folder where all images are going to be saved.
+ */
+ private function setFolder()
+ {
+ // if folder doesn't exist, attempt to create one and store the folder name in property $folder
+ if (!file_exists($this->baseFolder)) {
+ mkdir($this->baseFolder, 0755, true);
+ }
+ }
+
+ /**
+ * Generate the folder where we are going to save images based on the entry url.
+ *
+ * @param int $entryId ID of the entry
+ *
+ * @return string
+ */
+ private function getRelativePath($entryId)
+ {
+ $hashId = hash('crc32', $entryId);
+ $relativePath = $hashId[0] . '/' . $hashId[1] . '/' . $hashId;
+ $folderPath = $this->baseFolder . '/' . $relativePath;
+
+ if (!file_exists($folderPath)) {
+ mkdir($folderPath, 0777, true);
}
- /* parse base URL and convert to local variables:
- $scheme, $host, $path */
- extract(parse_url($url));
+ $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
- /* remove non-directory element from path */
- $path = preg_replace('#/[^/]*$#', '', $path);
+ return $relativePath;
+ }
- /* destroy path if relative url points to root */
- if ($relativeLink[0] == '/') {
- $path = '';
+ /**
+ * Make an $url absolute based on the $base.
+ *
+ * @see Graby->makeAbsoluteStr
+ *
+ * @param string $base Base url
+ * @param string $url Url to make it absolute
+ *
+ * @return false|string
+ */
+ private function getAbsoluteLink($base, $url)
+ {
+ if (preg_match('!^https?://!i', $url)) {
+ // already absolute
+ return $url;
}
- /* dirty absolute URL */
- $abs = $host.$path.'/'.$relativeLink;
+ $base = new \SimplePie_IRI($base);
- /* replace '//' or '/./' or '/foo/../' with '/' */
- $re = array('#(/\.?/)#', '#/(?!\.\.)[^/]+/\.\./#');
- for ($n = 1; $n > 0; $abs = preg_replace($re, '/', $abs, -1, $n)) {
+ // remove '//' in URL path (causes URLs not to resolve properly)
+ if (isset($base->ipath)) {
+ $base->ipath = preg_replace('!//+!', '/', $base->ipath);
}
- /* absolute URL is ready! */
- return $scheme.'://'.$abs;
+ if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
+ return $absolute->get_uri();
+ }
+
+ $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
+
+ return false;
}
- public static function getPocheUrl()
+ /**
+ * Retrieve and validate the extension from the response of the url of the image.
+ *
+ * @param Response $res Guzzle Response
+ * @param string $imagePath Path from the src image from the content (used for log only)
+ *
+ * @return string|false Extension name or false if validation failed
+ */
+ private function getExtensionFromResponse(Response $res, $imagePath)
{
- $baseUrl = '';
- $https = (!empty($_SERVER['HTTPS'])
- && (strtolower($_SERVER['HTTPS']) == 'on'))
- || (isset($_SERVER['SERVER_PORT'])
- && $_SERVER['SERVER_PORT'] == '443') // HTTPS detection.
- || (isset($_SERVER['SERVER_PORT']) //Custom HTTPS port detection
- && $_SERVER['SERVER_PORT'] == SSL_PORT)
- || (isset($_SERVER['HTTP_X_FORWARDED_PROTO'])
- && $_SERVER['HTTP_X_FORWARDED_PROTO'] == 'https');
- $serverport = (!isset($_SERVER['SERVER_PORT'])
- || $_SERVER['SERVER_PORT'] == '80'
- || $_SERVER['SERVER_PORT'] == HTTP_PORT
- || ($https && $_SERVER['SERVER_PORT'] == '443')
- || ($https && $_SERVER['SERVER_PORT'] == SSL_PORT) //Custom HTTPS port detection
- ? '' : ':'.$_SERVER['SERVER_PORT']);
-
- if (isset($_SERVER['HTTP_X_FORWARDED_PORT'])) {
- $serverport = ':'.$_SERVER['HTTP_X_FORWARDED_PORT'];
- }
- // $scriptname = str_replace('/index.php', '/', $_SERVER["SCRIPT_NAME"]);
- // if (!isset($_SERVER["HTTP_HOST"])) {
- // return $scriptname;
- // }
- $host = (isset($_SERVER['HTTP_X_FORWARDED_HOST']) ? $_SERVER['HTTP_X_FORWARDED_HOST'] : (isset($_SERVER['HTTP_HOST']) ? $_SERVER['HTTP_HOST'] : $_SERVER['SERVER_NAME']));
- if (strpos($host, ':') !== false) {
- $serverport = '';
- }
- // check if BASE_URL is configured
- if (BASE_URL) {
- $baseUrl = BASE_URL;
- } else {
- $baseUrl = 'http'.($https ? 's' : '').'://'.$host.$serverport;
- }
-
- return $baseUrl;
+ $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
+ $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
+
+ // ok header doesn't have the extension, try a different way
+ if (empty($ext)) {
+ $types = [
+ 'jpeg' => "\xFF\xD8\xFF",
+ 'gif' => 'GIF',
+ 'png' => "\x89\x50\x4e\x47\x0d\x0a",
+ ];
+ $bytes = substr((string) $res->getBody(), 0, 8);
+
+ foreach ($types as $type => $header) {
+ if (0 === strpos($bytes, $header)) {
+ $ext = $type;
+ break;
+ }
+ }
+
+ $this->logger->debug('DownloadImages: Checking extension (alternative)', ['ext' => $ext]);
+ }
+
+ if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
+ $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: ' . $imagePath);
+
+ return false;
+ }
+
+ return $ext;
}
}
projects / github / wallabag / wallabag.git / blobdiff