Merge pull request #3184 from wallabag/better-way-image-extension

Use an alternative way to detect images

src/Wallabag/CoreBundle/Helper/DownloadImages.php

@@ -5,6 +5,7 @@ namespace Wallabag\CoreBundle\Helper;
 use Psr\Log\LoggerInterface;
 use Symfony\Component\DomCrawler\Crawler;
 use GuzzleHttp\Client;
+use GuzzleHttp\Message\Response;
 use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
 use Symfony\Component\Finder\Finder;
 
@@ -116,13 +117,11 @@ class DownloadImages
             return false;
         }
 
-        $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
+        $ext = $this->getExtensionFromResponse($res, $imagePath);
-        $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
+        if (false === $res) {
-        if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
-            $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: '.$imagePath);
-
             return false;
         }
+
         $hashImage = hash('crc32', $absolutePath);
         $localPath = $folderPath.'/'.$hashImage.'.'.$ext;
 
@@ -237,4 +236,45 @@ class DownloadImages
 
         return false;
     }
+
+    /**
+     * 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)
+    {
+        $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;
+    }
 }

tests/Wallabag/CoreBundle/Helper/DownloadImagesTest.php

@@ -157,4 +157,29 @@ class DownloadImagesTest extends \PHPUnit_Framework_TestCase
 
         $this->assertFalse($res, 'Absolute image can not be determined, so it will not be replaced');
     }
+
+    public function testProcessRealImage()
+    {
+        $client = new Client();
+
+        $mock = new Mock([
+            new Response(200, ['content-type' => null], Stream::factory(file_get_contents(__DIR__.'/../fixtures/image-no-content-type.jpg'))),
+        ]);
+
+        $client->getEmitter()->attach($mock);
+
+        $logHandler = new TestHandler();
+        $logger = new Logger('test', array($logHandler));
+
+        $download = new DownloadImages($client, sys_get_temp_dir().'/wallabag_test', 'http://wallabag.io/', $logger);
+
+        $res = $download->processSingleImage(
+            123,
+            'https://cdn.theconversation.com/files/157200/article/width926/gsj2rjp2-1487348607.jpg',
+            'https://theconversation.com/conversation-avec-gerald-bronner-ce-nest-pas-la-post-verite-qui-nous-menace-mais-lextension-de-notre-credulite-73089'
+        );
+
+        $this->assertContains('http://wallabag.io/assets/images/9/b/9b0ead26/', $res, 'Content-Type was empty but data is ok for an image');
+        $this->assertContains('DownloadImages: Checking extension (alternative)', $logHandler->getRecords()[3]['message']);
+    }
 }