models/metadata.py in the pikepdf package 1.3.0 through 2.9.2 for Python allows XXE when parsing XMP metadata entries.
{ "ubuntu_priority": "medium" }