wallabag/src/Wallabag/CoreBundle/Helper/DownloadImages.php

204 lines
6 KiB
PHP
Raw Normal View History

<?php
namespace Wallabag\CoreBundle\Helper;
2016-10-30 09:48:29 +00:00
use Psr\Log\LoggerInterface;
use Symfony\Component\DomCrawler\Crawler;
2016-10-30 09:48:29 +00:00
use GuzzleHttp\Client;
use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
2016-10-22 07:22:30 +00:00
class DownloadImages
{
2016-10-30 09:48:29 +00:00
const REGENERATE_PICTURES_QUALITY = 80;
private $client;
private $baseFolder;
private $logger;
2016-10-30 09:48:29 +00:00
private $mimeGuesser;
2016-10-30 09:48:29 +00:00
public function __construct(Client $client, $baseFolder, LoggerInterface $logger)
2016-10-22 07:22:30 +00:00
{
2016-10-30 09:48:29 +00:00
$this->client = $client;
$this->baseFolder = $baseFolder;
$this->logger = $logger;
2016-10-30 09:48:29 +00:00
$this->mimeGuesser = new MimeTypeExtensionGuesser();
$this->setFolder();
}
2016-10-30 09:48:29 +00:00
/**
* Setup base folder where all images are going to be saved.
*/
private function setFolder()
2016-10-22 07:22:30 +00:00
{
// if folder doesn't exist, attempt to create one and store the folder name in property $folder
2016-10-30 09:48:29 +00:00
if (!file_exists($this->baseFolder)) {
mkdir($this->baseFolder, 0777, true);
}
}
2016-10-30 09:48:29 +00:00
/**
* Process the html and extract image from it, save them to local and return the updated html.
*
* @param string $html
* @param string $url Used as a base path for relative image and folder
*
* @return string
*/
public function processHtml($html, $url)
2016-10-22 07:22:30 +00:00
{
2016-10-30 09:48:29 +00:00
$crawler = new Crawler($html);
$result = $crawler
->filterXpath('//img')
->extract(array('src'));
2016-10-30 09:48:29 +00:00
$relativePath = $this->getRelativePath($url);
// download and save the image to the folder
foreach ($result as $image) {
2016-10-30 09:48:29 +00:00
$imagePath = $this->processSingleImage($image, $url, $relativePath);
if (false === $imagePath) {
continue;
}
$html = str_replace($image, $imagePath, $html);
}
2016-10-30 09:48:29 +00:00
return $html;
}
2016-10-30 09:48:29 +00:00
/**
* Process a single image:
* - retrieve it
* - re-saved it (for security reason)
* - return the new local path.
*
* @param string $imagePath Path to the image to retrieve
* @param string $url Url from where the image were found
* @param string $relativePath Relative local path to saved the image
*
* @return string Relative url to access the image from the web
*/
public function processSingleImage($imagePath, $url, $relativePath = null)
2016-10-22 07:22:30 +00:00
{
2016-10-30 09:48:29 +00:00
if (null == $relativePath) {
$relativePath = $this->getRelativePath($url);
}
2016-10-30 09:48:29 +00:00
$folderPath = $this->baseFolder.'/'.$relativePath;
2016-10-30 09:48:29 +00:00
// build image path
$absolutePath = $this->getAbsoluteLink($url, $imagePath);
if (false === $absolutePath) {
2016-10-30 10:27:09 +00:00
$this->logger->log('error', 'Can not determine the absolute path for that image, skipping.');
return false;
}
2016-10-30 10:27:09 +00:00
try {
$res = $this->client->get($absolutePath);
} catch (\Exception $e) {
$this->logger->log('error', 'Can not retrieve image, skipping.', ['exception' => $e]);
return false;
}
2016-10-30 09:48:29 +00:00
$ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
$this->logger->log('debug', 'Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
2016-10-30 10:27:09 +00:00
if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
$this->logger->log('error', 'Processed image with not allowed extension. Skipping '.$imagePath);
return false;
}
2016-10-30 09:48:29 +00:00
$hashImage = hash('crc32', $absolutePath);
$localPath = $folderPath.'/'.$hashImage.'.'.$ext;
try {
$im = imagecreatefromstring($res->getBody());
} catch (\Exception $e) {
$im = false;
}
2016-10-30 10:27:09 +00:00
if (false === $im) {
2016-10-30 09:48:29 +00:00
$this->logger->log('error', 'Error while regenerating image', ['path' => $localPath]);
return false;
}
2016-10-30 09:48:29 +00:00
switch ($ext) {
case 'gif':
$result = imagegif($im, $localPath);
2016-10-22 07:22:30 +00:00
$this->logger->log('debug', 'Re-creating gif');
break;
2016-10-30 09:48:29 +00:00
case 'jpeg':
case 'jpg':
$result = imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
2016-10-22 07:22:30 +00:00
$this->logger->log('debug', 'Re-creating jpg');
break;
2016-10-30 09:48:29 +00:00
case 'png':
$result = imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
2016-10-22 07:22:30 +00:00
$this->logger->log('debug', 'Re-creating png');
}
2016-10-30 09:48:29 +00:00
imagedestroy($im);
2016-10-30 09:48:29 +00:00
return '/assets/images/'.$relativePath.'/'.$hashImage.'.'.$ext;
}
2016-10-30 09:48:29 +00:00
/**
* Generate the folder where we are going to save images based on the entry url.
*
* @param string $url
*
* @return string
*/
private function getRelativePath($url)
{
2016-10-30 09:48:29 +00:00
$hashUrl = hash('crc32', $url);
$relativePath = $hashUrl[0].'/'.$hashUrl[1].'/'.$hashUrl;
$folderPath = $this->baseFolder.'/'.$relativePath;
2016-10-30 09:48:29 +00:00
if (!file_exists($folderPath)) {
mkdir($folderPath, 0777, true);
}
2016-10-30 09:48:29 +00:00
$this->logger->log('debug', 'Folder used for that url', ['folder' => $folderPath, 'url' => $url]);
2016-10-30 09:48:29 +00:00
return $relativePath;
}
2016-10-30 09:48:29 +00:00
/**
* 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;
}
2016-10-30 09:48:29 +00:00
$base = new \SimplePie_IRI($base);
2016-10-30 09:48:29 +00:00
// remove '//' in URL path (causes URLs not to resolve properly)
if (isset($base->ipath)) {
$base->ipath = preg_replace('!//+!', '/', $base->ipath);
}
2016-10-30 09:48:29 +00:00
if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
return $absolute->get_uri();
2016-06-28 20:06:00 +00:00
}
2016-10-22 07:22:30 +00:00
2016-10-30 10:27:09 +00:00
$this->logger->log('error', 'Can not make an absolute link', ['base' => $base, 'url' => $url]);
2016-10-30 09:48:29 +00:00
return false;
2016-06-28 20:06:00 +00:00
}
}