Issue Summary

When exporting a Confluence page to PDF, macro-embedded content (such as a Markdown file with multiple images) may result in image duplication. Specifically, the first image is duplicated.

This issue occurs even without third-party apps, but it becomes more noticeable when an app retrieves images via URLs with query parameters ?query. It appears that Confluence's caching mechanism ignores query parameters, using only the base URL as the cache key. As a result, when images are determined by these parameters, the wrong image may be displayed.

Steps to Reproduce

1. Prepare a Markdown file in a Bitbucket repository that contains multiple images.
2. Add this file via the file URL to a Confluence page using the macro for the app Include Bitbucket for Confluence but you can also reproduce without our app.
3. Take a PDF export of the page.

Expected Results

PDF export contains all the images.

Actual Results

PDF export contains duplicates of a single image.

Workaround

 
The workaround is the use the actual link of the image not the relative path of the image.
In markdown you can do this, which causes the issue:

![Alt Text](image-path) 

However, when using a complete URL, it doesn't cause any issues.

![Alt Text](https://actual_link_to_image.com)