Allow user to pass an aux file to `get_hpc_info`

[hayesla]

It would helpful to allow a use to pass a local aux file to get_hpc_info, rather than needing to do a Fido search everytime its called.

I understand you would want to check for the most up to date files, but its annoying to do it everytime, especially if you're running for many events - of several times etc.

[samaloney]

Notes:

• Use config file and if configure look in local path for AUX files
• Merge local Fido client and use with given path

[samaloney]

So this function could be changed to accept a file but this function is usually called from the coordinate transform stack so no way to pass a file.

A potential way would be to introduce a module level variable OVERRIDE_AUX_FILE then in this code check if this is None and if not assume the variable contains a parsed aux file?

stixpy/stixpy/coordinates/transforms.py

Lines 58 to 72 in e3d8d61

	def get_hpc_info(times, end_time=None):
	r"""
	Get STIX pointing and SO location from L2 aspect files.
	Parameters
	----------
	times : `astropy.time.Time`
	Time/s to get hpc info for at.
	Returns
	-------
	"""
	aux = _get_ephemeris_data(times.min(), end_time or times.max())

and I guess would need to extract this code to allow people to parse the aspect data to the expected format (I can't remember why not using our Product here?)

stixpy/stixpy/coordinates/transforms.py

Lines 194 to 199 in e3d8d61

	hdu = fits.getheader(aux_file, ext=0)
	aux = QTable.read(aux_file, hdu=2)
	date_beg = Time(hdu.get("DATE-BEG"))
	aux["time"] = (
	date_beg + aux["time"] - 32 * u.s
	) # Shift AUX data by half a time bin (starting time vs. bin centre)

This would bring about its own problems, easy to forget and prob not very thread safe. I guess another approach would be to make it a class and with an option to fix / override the returned aux file

[hayesla]

maybe I'm missing something, but could you not have _get_emphemeris_data to take a file passes too? such that you'd have :

get_hpc_info(times, end_time=None, file=None):

and then

_get_ephemeris_data(start_time, end_time=None, file=None):
    ....
    if file is None:
        query = Fido.search(...)
    ....

but not recommend people to do this unless they have a specific reason?

[samaloney]

Yea 100% but when you do six_coord.transform_to(AIAframe) or aia_coord.transform_to(STIXFrame) etc get_hpc_info is called in the coordinate transform stack. But imagine you created the map/coordinate manually using get_hpc_info(times, end_time=None, file=<myfile>) they would be using possibly different aux_files and that could lead to really weird hard to find problems, I think.

So I think you need a way to specify the aux file used in all calculations not just when get_hpc_info is directly called by the user.

Not sure if I'm making any sense here?

I guess if you really only want to specify the an aux file when calling get_hpc_info we can add that? But if you want to do coordinate transforms, reproduction etc with a specific aux file in a consistent way it's more complicated as far as I can see.

[hayesla]

ah yeah sorry I totally get you now - that could definitely end up confusing and very hard to do in a consistent way

My specific use case is to directly use get_hpc_info to make a map from the visibilities, as i'm doing it for 1000's of flares so wanted to bypass doing the query which is slow (i.e. goes and checks etc)

[nicHoch]

@hayesla we discussed a way forward for this issue

this PR #165 introduces a internal ephemeris cache that you could also preload with (many) an anc-ephemeris file

do you think that will solve your problems