From 548b610a17ae9648903ecdaa68200eabd247a7c3 Mon Sep 17 00:00:00 2001
From: Simounet <contact@simounet.net>
Date: Mon, 29 May 2023 15:12:04 +0200
Subject: [PATCH] Fix images downloading with numeric HTML entity

---
 .../CoreBundle/Helper/DownloadImages.php        | 12 +++++++-----
 .../CoreBundle/Helper/DownloadImagesTest.php    | 17 +++++++++++++++++
 2 files changed, 24 insertions(+), 5 deletions(-)

diff --git a/src/Wallabag/CoreBundle/Helper/DownloadImages.php b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
index 477b509ac..8a4ec9031 100644
--- a/src/Wallabag/CoreBundle/Helper/DownloadImages.php
+++ b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
@@ -86,12 +86,14 @@ class DownloadImages
                 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, $newImage, $html);
+            // if image contains "&" and we can't find it in the html it might be because it's encoded as &amp; or unicode
+            if (false !== stripos($image, '&') && false === stripos($html, $image)) {
+                $imageAmp = str_replace('&', '&amp;', $image);
+                $html = str_replace($imageAmp, $newImage, $html);
+                $imageUnicode = str_replace('&', '&#038;', $image);
+                $html = str_replace($imageUnicode, $newImage, $html);
+            }
         }
 
         return $html;
diff --git a/tests/Wallabag/CoreBundle/Helper/DownloadImagesTest.php b/tests/Wallabag/CoreBundle/Helper/DownloadImagesTest.php
index f0735719d..4b78fb251 100644
--- a/tests/Wallabag/CoreBundle/Helper/DownloadImagesTest.php
+++ b/tests/Wallabag/CoreBundle/Helper/DownloadImagesTest.php
@@ -184,6 +184,23 @@ class DownloadImagesTest extends TestCase
         $this->assertStringNotContainsString('f_auto,q_auto', $res, 'Image srcset attribute were not replaced');
     }
 
+    public function testProcessImageWithNumericHtmlEntitySeparator()
+    {
+        $httpMockClient = new HttpMockClient();
+        $httpMockClient->addResponse(new Response(200, ['content-type' => 'image/jpeg'], file_get_contents(__DIR__ . '/../fixtures/image-no-content-type.jpg')));
+        $httpMockClient->addResponse(new Response(200, ['content-type' => 'image/jpeg'], file_get_contents(__DIR__ . '/../fixtures/image-no-content-type.jpg')));
+        $httpMockClient->addResponse(new Response(200, ['content-type' => 'image/jpeg'], file_get_contents(__DIR__ . '/../fixtures/image-no-content-type.jpg')));
+
+        $logHandler = new TestHandler();
+        $logger = new Logger('test', [$logHandler]);
+
+        $download = new DownloadImages($httpMockClient, sys_get_temp_dir() . '/wallabag_test', 'http://wallabag.io/', $logger);
+        // wordpress.com sites using &#038; as an &amp; alternative
+        $res = $download->processHtml(123, '<img srcset="https://example.com/20191204_133626-scaled.jpg?strip=info&#038;w=600&#038;ssl=1 600w,https://example.com/20191204_133626-scaled.jpg?strip=info&#038;w=900&#038;ssl=1 900w" src="https://example.com/20191204_133626-scaled.jpg?ssl=1"/>', 'https://example.com/about/');
+
+        $this->assertStringNotContainsString('https://example.com', $res, 'Image srcset attribute were not replaced');
+    }
+
     public function testProcessImageWithNullPath()
     {
         $httpMockClient = new HttpMockClient();