io.fits creates a corrupt FITS files if a ImageHDU contains zero-dimensional data

[kYwzor]

Description

ImageHDU accepts a ndarray with shape () (zero-dimensional) as a data array. This later causes issues when writing to a file because io.fits assumes that the data has at least 1 dimension, resulting in a corrupt FITS file.

Expected behavior

io.fits should never silently create a corrupt FITS file.

How to Reproduce

Minimal reproducible example:

import numpy as np
from astropy.io import fits

filename = 'corrupted.fits'
hdu = fits.ImageHDU(name='test', data=np.array(1.0))
hdu.writeto(filename, overwrite=True)

Although this causes no errors/warnings, the resulting file is not valid FITS and will fail to properly open with fits.getdata(filename).

Versions

Windows-10-10.0.19044-SP0
Python 3.10.10 (tags/v3.10.10:aad5f6a, Feb 7 2023, 17:20:36) [MSC v.1929 64 bit (AMD64)]
astropy 5.2.1
Numpy 1.24.2
pyerfa 2.0.0.1
Scipy 1.10.0
Matplotlib 3.6.3

[kYwzor]

This seems to have a trivial fix, I will provide a PR soon.