assign_mtwcs crashes on background spectra

[stscijgbot-jp]

Issue JP-3787 was created on JIRA by Melanie Clarke:

Testing the master background step in spec3, I came across a crash for an input ASN that contains a background spectrum and is also a moving target.

The assign_mtwcs step attempts to add a moving target WCS to the x1d product, then crashes:

  File "jwst/assign_mtwcs/moving_target_wcs.py", line 42, in assign_moving_target_wcs
mt_ra[i] = model.meta.wcsinfo.mt_ra
^^^^^^^^^^^^^^^^^^
File "stdatamodels/properties.py", line 309, in {}getattr{}
raise AttributeError("No attribute '{0}'".format(attr))
AttributeError: No attribute 'wcsinfo'a

[stscijgbot-jp]

Comment by Melanie Clarke on JIRA:

Ned Molter - I think the issue is that when the ModelLibrary is created from the input ModelContainer, it is not distinguishing between background and science members in the input.

[stscijgbot-jp]

Comment by Melanie Clarke on JIRA:

To reproduce the issue, I have sample data in ██████████████████████████████  Command is:

strun calwebb_spec3 jw01252-o001_20240905t222322_spec3_00003_asn_with_bg.json

[stscijgbot-jp]

Comment by Ned Molter on JIRA:

Melanie Clarke you're right that the cause is a failure to distinguish background and science members.  ModelLibrary attempts to do this, but it doesn't do it properly because it looks for the attribute model.meta.exptype, whereas the JWST schema puts that info into model.meta.asn.exptype.  The model.meta.asn.exptype attributes from this asn are what is expected, which means ModelContainer is setting these correctly when reading the association.  So fixing that in AbstractModelLibrary should fix the problem.

I opened an issue in stpipe about this, but I'm not sure how changing it would affect Roman.  There are also ways to solve this without needing to touch stpipe, and I'll look into those further.

[stscijgbot-jp]

Comment by Ned Molter on JIRA:

It looks like the change that makes the test dataset run correctly will require only stpipe changes.

 

However, we should add a regression test that covers this, and we should not resolve this ticket until that test is added

[stscijgbot-jp]

Comment by Melanie Clarke on JIRA:

I added a regression test in this PR:

#8917

I verified that the test fails with the current stpipe and passes with the expected fix.  I'll leave the PR at draft until there's a fix in stpipe.

[stscijgbot-jp]

Comment by Ned Molter on JIRA:

Fixed by #8918 and spacetelescope/stpipe#201