wallabag/src/Wallabag/CoreBundle/Helper/DownloadImages.php
Jeremy Benoist 9743058f7d
Add a command to clean downloaded images
There were a bug in versions prior to 2.4.0 where images weren't properly removed (mostly when coming from the API).
With that command, we'll be able to remove images which aren't associated to any entries.

Like other command you can pass a username to only clean one user.
2020-12-14 22:19:19 +01:00

355 lines
11 KiB
PHP

<?php
namespace Wallabag\CoreBundle\Helper;
use GuzzleHttp\Psr7\Uri;
use GuzzleHttp\Psr7\UriResolver;
use Http\Client\Common\HttpMethodsClient;
use Http\Client\Common\Plugin\ErrorPlugin;
use Http\Client\Common\PluginClient;
use Http\Client\HttpClient;
use Http\Discovery\MessageFactoryDiscovery;
use Http\Message\MessageFactory;
use Psr\Http\Message\ResponseInterface;
use Psr\Log\LoggerInterface;
use Symfony\Component\DomCrawler\Crawler;
use Symfony\Component\Finder\Finder;
use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
class DownloadImages
{
const REGENERATE_PICTURES_QUALITY = 80;
private $client;
private $baseFolder;
private $logger;
private $mimeGuesser;
private $wallabagUrl;
public function __construct(HttpClient $client, $baseFolder, $wallabagUrl, LoggerInterface $logger, MessageFactory $messageFactory = null)
{
$this->client = new HttpMethodsClient(new PluginClient($client, [new ErrorPlugin()]), $messageFactory ?: MessageFactoryDiscovery::find());
$this->baseFolder = $baseFolder;
$this->wallabagUrl = rtrim($wallabagUrl, '/');
$this->logger = $logger;
$this->mimeGuesser = new MimeTypeExtensionGuesser();
$this->setFolder();
}
public function getBaseFolder()
{
return $this->baseFolder;
}
/**
* Process the html and extract images URLs from it.
*
* @param string $html
*
* @return string[]
*/
public static function extractImagesUrlsFromHtml($html)
{
$crawler = new Crawler($html);
$imagesCrawler = $crawler->filterXpath('//img');
$imagesUrls = $imagesCrawler->extract(['src']);
$imagesSrcsetUrls = self::getSrcsetUrls($imagesCrawler);
return array_unique(array_merge($imagesUrls, $imagesSrcsetUrls));
}
/**
* 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)
{
$imagesUrls = self::extractImagesUrlsFromHtml($html);
$relativePath = $this->getRelativePath($entryId);
// download and save the image to the folder
foreach ($imagesUrls as $image) {
$imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
if (false === $imagePath) {
continue;
}
// if image contains "&" and we can't find it in the html it might be because it's encoded as &amp;
if (false !== stripos($image, '&') && false === stripos($html, $image)) {
$image = str_replace('&', '&amp;', $image);
}
$html = str_replace($image, $imagePath, $html);
}
return $html;
}
/**
* 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)
{
if (null === $imagePath) {
return false;
}
if (null === $relativePath) {
$relativePath = $this->getRelativePath($entryId);
}
$this->logger->debug('DownloadImages: working on image: ' . $imagePath);
$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;
}
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;
}
$hashImage = hash('crc32', $absolutePath);
$localPath = $folderPath . '/' . $hashImage . '.' . $ext;
try {
$im = imagecreatefromstring((string) $res->getBody());
} catch (\Exception $e) {
$im = false;
}
if (false === $im) {
$this->logger->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
return false;
}
switch ($ext) {
case 'gif':
// use Imagick if available to keep GIF animation
if (class_exists('\\Imagick')) {
try {
$imagick = new \Imagick();
$imagick->readImageBlob($res->getBody());
$imagick->setImageFormat('gif');
$imagick->writeImages($localPath, true);
} catch (\Exception $e) {
// if Imagick fail, fallback to the default solution
imagegif($im, $localPath);
}
} else {
imagegif($im, $localPath);
}
$this->logger->debug('DownloadImages: Re-creating gif');
break;
case 'jpeg':
case 'jpg':
imagejpeg($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 $this->wallabagUrl . '/assets/images/' . $relativePath . '/' . $hashImage . '.' . $ext;
}
/**
* Remove all images for the given entry id.
*
* @param int $entryId ID of the entry
*/
public function removeImages($entryId)
{
$relativePath = $this->getRelativePath($entryId);
$folderPath = $this->baseFolder . '/' . $relativePath;
$finder = new Finder();
$finder
->files()
->ignoreDotFiles(true)
->in($folderPath);
foreach ($finder as $file) {
@unlink($file->getRealPath());
}
@rmdir($folderPath);
}
/**
* Generate the folder where we are going to save images based on the entry url.
*
* @param int $entryId ID of the entry
* @param bool $createFolder Should we create the folder for the given id?
*
* @return string
*/
public function getRelativePath($entryId, $createFolder = true)
{
$hashId = hash('crc32', $entryId);
$relativePath = $hashId[0] . '/' . $hashId[1] . '/' . $hashId;
$folderPath = $this->baseFolder . '/' . $relativePath;
if (!file_exists($folderPath) && $createFolder) {
mkdir($folderPath, 0777, true);
}
$this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
return $relativePath;
}
/**
* Get images urls from the srcset image attribute.
*
* @return array An array of urls
*/
private static function getSrcsetUrls(Crawler $imagesCrawler)
{
$urls = [];
$iterator = $imagesCrawler->getIterator();
while ($iterator->valid()) {
$srcsetAttribute = $iterator->current()->getAttribute('srcset');
if ('' !== $srcsetAttribute) {
// Couldn't start with " OR ' OR a white space
// Could be one or more white space
// Must be one or more digits followed by w OR x
$pattern = "/(?:[^\"'\s]+\s*(?:\d+[wx])+)/";
preg_match_all($pattern, $srcsetAttribute, $matches);
$srcset = \call_user_func_array('array_merge', $matches);
$srcsetUrls = array_map(function ($src) {
return trim(explode(' ', $src, 2)[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);
}
}
/**
* 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;
}
$base = new Uri($base);
// in case the url has no scheme & host
if ('' === $base->getAuthority() || '' === $base->getScheme()) {
$this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
return false;
}
return (string) UriResolver::resolve($base, new Uri($url));
}
/**
* Retrieve and validate the extension from the response of the url of the image.
*
* @param ResponseInterface $res Http 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(ResponseInterface $res, $imagePath)
{
$ext = $this->mimeGuesser->guess(current($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;
}
}