Flag ".." in <file_name> as error condition

[nutjob4life]

🗒️ Summary

Merge this to have pds-deep-archive raise an exception when encountering <file_name> entries that contain relative paths, such as ../catalog/dataset.cat, which according to this comment is invalid in a PDS label.

⚙️ Test Data and/or Report

Before:

$ .venv/bin/pds-deep-archive --debug --site PDS_ATM --bundle-base-url https://pds-atmospheres.nmsu.edu/PDS/data/ /Users/kelly/Downloads/vo/vo_3002/bundle_voirtm.xml
INFO 👟 PDS Deep Archive, version 1.1.2
…
INFO 🎉 Success! From /Users/kelly/Downloads/vo/vo_3002/bundle_voirtm.xml, generated these output files:
INFO 📄 SIP Manifest: vo_irtm_v1.0_20230404_sip_v1.0.tab
INFO 📄 XML label for the SIP: vo_irtm_v1.0_20230404_sip_v1.0.xml
INFO 👋 That's it for now. Bye.

After:

$ .venv/bin/pds-deep-archive --debug --site PDS_ATM --bundle-base-url https://pds-atmospheres.nmsu.edu/PDS/data/ /Users/kelly/Downloads/vo/vo_3002/bundle_voirtm.xml
INFO 👟 PDS Deep Archive, version 1.1.3
…
CRITICAL 🛑 Cannot proceed as a critical problem has occurred; re-run with --debug for more info.
DEBUG 🖥 Here is the exception: ValueError('Bundle /Users/kelly/Downloads/vo/vo_3002/document/dataset.xml contains a <file_name> ``../catalog/dataset.cat`` which contains a relative path ``..``, which is invalid')
…

♻️ Related Issues

• Canonical path is not processed correctly ('/../ in path) #145

[alexdunnjpl]

v. nice

[nutjob4life]

Thanks @alexdunnjpl @c-suh! 🎉

[c-suh]

@nutjob4life thank you for the quick turnaround! Much appreciated.

[jordanpadams]

@nutjob4life question: does deep-archive provide any support for directory_path_name? according to the spec, this is how someone could specify a path to a file not in the current directory.

[alexdunnjpl]

@jordanpadams that attribute doesn't explicitly allow for relative paths either (as it also requires a root/cwd to be meaningful, which isn't mentioned)... was it definitely intended to support relpaths?

[nutjob4life]

@jordanpadams there is support for directory_path_name; see:

deep-archive/src/pds2/aipgen/utils.py

Lines 200 to 204 in b398f8d

	# any sibling directory_path_name?
	dpnnode = match.getparent().find(f"./{{{PDS_NS_URI}}}directory_path_name")
	fn = match.text.strip()
	dn = None if dpnnode is None else dpnnode.text.strip()
	filepath = os.path.join(dirpath, dn, fn) if dn else os.path.join(dirpath, fn)

[jordanpadams]

@alexdunnjpl I was thinking the same thing. you can start getting really weird with things like ../../../../../path/to/file.dat. this may require an update/clarification to the standard