eCTD XML Compiler

This package appears to implement the advertised eCTD XML skeleton conversion, but there are multiple quality and coherence issues you should address before using it on real or sensitive documents: - Do not run on sensitive production submissions until you audit the code. Review scripts/main.py fully for any unexpected I/O or network calls (the visible portion shows no network activity but the file is truncated in the listing). - Confirm dependency names and install into an isolated virtual environment. requirements.txt lists 'docx' and 'pypdf2' while the code expects python-docx and PyPDF2; install the correct packages (pip install python-docx PyPDF2 lxml) and test imports. - Test with non-sensitive sample documents. The code contains clear bugs (for example, in generate_xml it references module_dir before it is defined) and may raise runtime errors or produce incomplete output. Run python -m py_compile scripts/main.py and then execute with small inputs to see behavior. - Check that the script does not modify files outside the intended output directory and that any CONFIG block it mentions is safe to edit (search the repository for CONFIG). If the script writes or overwrites files, back up originals first. - Confirm .doc support: the tool claims .doc support but python-docx does not parse legacy .doc files; convert .doc to .docx before feeding it in or verify fallback behavior. - Verify MD5 usage: the tool writes MD5 checksums for integrity; be aware MD5 is not cryptographically secure for tamper detection, but is commonly used for simple checksums. If you are not comfortable auditing the code yourself, ask the developer for clarification (source repository, full script, and corrected dependency list) or run the tool inside a sandboxed environment (isolated VM/container) before trusting outputs for regulatory submissions.