models/metadata.py in the pikepdf package 1.3.0 through 2.9.2 for Python allows XXE when parsing XMP metadata entries.
{ "nvd_published_at": "2021-04-01T20:15:00Z", "cwe_ids": [ "CWE-611" ], "severity": "HIGH", "github_reviewed": true, "github_reviewed_at": "2021-04-19T21:54:20Z" }