Harvest throws error when it sees an Ingest LDD file

[mdrum]

🐛 Describe the bug

When running Harvest tool on one of our bundles that contained an Ingest_LDD XML file, the tool reported the following error:
[ERROR] Missing logical identifier: /Users/mdrum/.CMVolumes/tunneled sbn-archive/dsk1/www/archive/pds4/orex/orex.mission/xml_schema/orex_ldd.xml
This is not an error I would expect to see, because the file in question is not a label

📜 To Reproduce

Steps to reproduce the behavior:

1. Download the OSIRIS-REx Mission Bundle https://sbnarchive.psi.edu/pds4/orex/orex.mission.zip
2. Download the attached testcase.cfg.txt, but modify the bundle location to match your download location (and modify the extension)
3. Install registry app, and set up an Elastic Search registry running locally on localhost:9200
4. Run registry-manager create-registry
5. Run harvest -c testcase.cfg

🕵️ Expected behavior

I would expect that the products within the bundle be processed without error, and the XML schema or Ingest LDD files be skipped

📚 Version of Software Used

Registry App 1.0.0

🩺 Test Data / Additional context

Provided in reproduction steps

🏞Screenshots

🖥 System Info

• OS: MacOS 11.6

---

🦄 Related requirements

⚙️ Engineering Details

[jordanpadams]

@mdrum we will look into this, but I was not aware that XML labels could even be archived at this time? We have not built support into any of our tools to handle XML files in an archive. That was part of the whole lblx convo. I imagine validate balks at this as well?

[mdrum]

@jordanpadams Yeah the LDD file isn't being archived, it just happens to be inside the bundle directory alongside the other dictionary artifacts. It would be nice if the tool was smart enough to ignore XML files that weren't labels, but another nice-to-have feature would be if there was some way to filter files within a bundle by path/filename (rather than filtering on product class)

[jordanpadams]

@mdrum I like the latter idea. Are you thinking a regex capability for filtering? Or something like an ignores list?

[mdrum]

@jordanpadams Regex would definitely be more useful, and maybe even begin to allow for some sort of parallelization. A filter for whitelisting or blacklisting by path and name would allow a job to be broken up into pieces.