detect gzipped-FITS files by magic-bytes

[exitflynn]

PR Description

Detection for gzipped-FITS files is now done by checking file signatures instead of checking for file extensions.
First a gzip file-signature is checked and if found, the file's first few bytes are decompressed and checked for the FITS file signature.
The PR closes #6692.

[nabobalis]

Hi @exitflynn, thanks for the PR. If you could add a unit test for this in https://github.com/sunpy/sunpy/blob/main/sunpy/io/tests/test_filetools.py

There should be some gzipped files to use in https://github.com/sunpy/sunpy/tree/main/sunpy/data/test

[exitflynn]

sure. i'll be getting to including PKZIP and BZIP2 detection and the tests tomorrow! (it's quite late here)

[nabobalis]

No problem. I also wouldn't add the other two file formats for now.

[exitflynn]

Hi, I just noticed that a test for detecting gzipped FITS files already exists, since i only re-modified how this was implemented.

[exitflynn]

also i'm not sure why the CI tests have failed, is it something unrelated to my changes that I can ignore? I tried runing the tests before pushing by running test_filetools.py and it didn't raise any issue (this is my first time testing stuff).

[Cadair]

Hi, I just noticed that a test for detecting gzipped FITS files already exists, since i only re-modified how this was implemented.

Could you add a test where the filextension doesn't match but we still detect it properly? This means we test this new behaviour and we make sure that we don't regress in the future?

[Cadair]

also i'm not sure why the CI tests have failed, is it something unrelated to my changes that I can ignore? I tried runing the tests before pushing by running test_filetools.py and it didn't raise any issue (this is my first time testing stuff).

It's not your fault, if you can rebase or merge in main then it should be fixed now #6694

[exitflynn]

right now there are two tests for gzipped-fits files, the test_read_file_fits_gzip() which tests filename.fits.gz, -.fts.gz and -fit.gz files (namely the ones that were included there when detection was done based on the extension) and the test_read_file_fits_gzip_not_gz_extension() which tests gzip_fits_test.file.

I think it'd make more sense to have only one test for gzip-fits files, the latter one since if it passes then there's no need to test for the first one.

Should I remove the first test (renaming the second one to test_read_file_fits_gzip())? If yes, it'd also make the three test -fits.gz files redundant, should i delete those too?

[nabobalis]

If you can combine the two tests, I would prefer that.

[exitflynn]

in that case, is combining them like that alright?

[nabobalis]

Looks fine to me.

[nabobalis]

If you could add a bugfix changelog following https://github.com/sunpy/sunpy/blob/main/changelog/README.rst

and if you could also deal with the pre-commit.ci fail.
You can do that either installing pre-commit, running the tox -e codestyle or checking the failure and fixing it manually.

[exitflynn]

yep, done! (I mean I think that should do it? I can't figure out what's causing the codecov/project check to fail... 🤔 )

[exitflynn]

thanks @ConorMacBride! done 👍

[ConorMacBride]

Thanks @exitflynn!

[exitflynn]

🤠 Thanks Conor, Nabil and Stuart for your time!