Add YAML validator; sync allow-lists with METIS_Pipeline classification

[astronomyk]

Summary

• Add YAML/validate_yamls.py — scans every YAML under YAML/, checks (1) PyYAML parseability, (2) static validation via metis_simulations.runRecipes.validate_recipes, (3) acceptance by runSimulationBlock (with simulate stubbed). Writes a per-file markdown report to YAML/yaml_validation_report.md.
• Refactor runRecipes.validateYAML into a reusable module-level validate_recipes(rcps) -> list[str]; fix the early-return bug on line 204 that was silencing the filter/catg/tech/type/nObs/ndit/dit checks.
• Sync simulationDefinitions.py allow-lists with the canonical DPR.CATG/.TECH/.TYPE values from METIS_Pipeline/metisp/workflows/metis/metis_classification.py (treated as authoritative): fix PUP,M → PUP,LM; add IFU; add DARK, FLAT,TWILIGHT, FLAT,LAMP,PINH, RSRF, PERSISTENCE; de-dup lss_m; add all wcu_* modes; add ND-2.8; widen LSS filter lists to include matching-band imaging filters (shared filter wheel).
• Rename DARK,WCU_OFF → DARK,WCUOFF in testYAML.yaml to match the classification form (the three wcuOff*.yaml files were already corrected by the kl/yaml_wcoOff_hotfix on main and merged in here).

After these changes the validator report goes from 14 passing / 65 static failures to 66 passing / 5 static failures. The remaining 5 static failures are genuine YAML bugs for the authors to address (missing filter_name in calib.yaml::IFU_DISTORTION_RAW, empty chophomeN.yaml, etc.).

Test plan

• python YAML/validate_yamls.py runs clean and produces YAML/yaml_validation_report.md.
• Non-zero exit when any file fails; zero exit when everything passes.
• Report contains the three sections (PyYAML / Static / runSimulationBlock acceptance) and a Passing files list; a file only appears in Passing when it clears all three checks.
• from metis_simulations.runRecipes import validate_recipes works without ScopeSim/IRDB installed (the validator pre-stubs scopesimWrapper and raw_script).
• Existing runRecipes.validateYAML(...) callers still work — it's a thin wrapper over validate_recipes.

🤖 Generated with Claude Code

[hugobuddel]

Please do not include code specific to your machine in the repository. This makes the report non-deterministic, meaning that everyone else would get a dirty repository when they run the validation script.

Also, why do we need this file at all in the repoitory if it can be generated?

[hugobuddel]

Now you are only rewriting what is already there, and this got better so this comment should not be a blocker, but still: why not import the canonical source directly?

I know one reason why, because the workflow directory of the pipeline is currently not in the pymetis package. But then please go and fix that, instead of propagating hacks.

[hugobuddel]

Great for making these!

However, the problem is not the availability of these scripts, but that people simply ignore them.

I've created https://github.com/AstarVienna/METIS_Simulations/blob/main/metis_simulations/compare_yaml_with_drld.py that is subsequently only maintained and ran by me.

So a bigger problem is that many people don't seem to care about such consistency at all. Most of us could write such a validator script, especially with help of Claude. But it seems I'm the only one doing so.

Therefore, I think you would be of bigger service to the group by putting on your manager head and make the team care about the importance of such validators. Then we can (have the clanker) write them ourselves.

Edit: we obviously do care about consistency, but we don't care enough to actually ensure consistency.

[astronomyk]

I appreciate the sentiment and duly noted. This script and the report are purely out of necessity for now. When I was playing around with my AIT-hat on, I hit a bunch of yamls that didn't work and I needed a quick way to test them. Fwiw, I've also started a branch that tries to transfer the FITS header setting over to the IRDB/Scopesim. All in the name of consistency. Just a note on the large context here: I think we all agree that METIS_Simulations was originally never meant as production-ready code for a wider audience, and as such we never put the same amount of emphasis on the design and development. But now that we are actually offering it to the AIT team, it warrants a deeper look and some refactoring effort to make it a more sustainable effort. This is a summer project though, as there's too many other things to do atm.

…

------------------------------------- Kieran Leschinski Department of Astrophysics University of Vienna Hugo Buddelmeijer ***@***.***> schrieb am Sa., 18. Apr. 2026, 09:50:

*hugobuddel* left a comment (AstarVienna/METIS_Simulations#181) <#181?email_source=notifications&email_token=ACH66QWEQBP7NADIYFDQK7T4WMXTPA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIMRXGMYTMNZVGE4KM4TFMFZW63VGMF2XI2DPOKSWK5TFNZ2LK4DSL5RW63LNMVXHIX3POBSW4X3DNRUWG2Y#issuecomment-4273167518> Great for making these! However, the problem is not the availability of these scripts, but that people simply ignore them. I've created https://github.com/AstarVienna/METIS_Simulations/blob/main/metis_simulations/compare_yaml_with_drld.py that is subsequently only maintained and ran by me. So a bigger problem is that many people don't seem to care about such consistency at all. Most of us could write such a validator script, especially with help of Claude. But it seems I'm the only one doing so. Therefore, I think you would be of bigger service to the group by putting on your manager head and make the team care about the importance of such validators. Then we can (have the clanker) write them ourselves. — Reply to this email directly, view it on GitHub <#181?email_source=notifications&email_token=ACH66QWEQBP7NADIYFDQK7T4WMXTPA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIMRXGMYTMNZVGE4KM4TFMFZW63VGMF2XI2DPOKSWK5TFNZ2LK4DSL5RW63LNMVXHIX3POBSW4X3DNRUWG2Y#issuecomment-4273167518>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACH66QXKPTQYL6BEHRHAOHT4WMXTPAVCNFSM6AAAAACX45X4O2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DENZTGE3DONJRHA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[hugobuddel]

I think we all agree that METIS_Simulations was originally never meant as production-ready code for a wider audience

Did I create the repository? I think so, since the first commit is mine. It was my goal to have different sets of simulations in here from the beginning. Basically a shared set of 'this is interesting data for METIS', as a consortium-wide collaboration. That's why initially there were different directories for different categories, and I'm happy these directories are back!

I'm fine with how the repository developed; because the people that do the actual work should have the freedom to do it the way they feel comfortable. And the repository is shaping up nicely with the recent developments!

Having yet another validator script does not add too much value I think; in fact, it adds negative value if the script is not ran or maintained. The existing YAML validation script is broken and thus adds only noise. I'll be happy to be proven wrong though.