Docling Core Remote Code Execution Vulnerability Due to Unsafe PyYAML Usage

A remote code execution vulnerability has been identified in Docling Core versions 2.21.0 prior to 2.48.4. This issue arises when the application processes untrusted YAML data using PyYAML versions prior to 5.4, specifically through the 'full_load' method with the 'FullLoader' option. The vulnerability is present in the 'load_from_yaml' function of the 'docling_core.types.doc.DoclingDocument' class.

Impact

Exploitation of this vulnerability allows for arbitrary remote code execution on the system where Docling Core is used.

Reproduction

To reproduce this vulnerability, load a YAML file containing untrusted data using the 'docling_core.types.doc.DoclingDocument.load_from_yaml()' method. Ensure that PyYAML is version 5.3 or earlier, and that Docling Core is within the vulnerable version range.

Remediation

Upgrade Docling Core to version 2.48.4 or later. If an immediate upgrade is not possible, ensure that PyYAML is updated to version 5.4 or greater.