read_cdf/TimeSeries crashes if CDF variable doesn't have a 'FILLVAL' attribute

[jgieseler]

Describe the bug

Similar to #5907, I stumbled upon CDF files that aren't fully complying to the CDF ISTP Metadata Guidelines by having a variable that doesn't provide a corresponding FILLVAL attribute. read_cdf() is expecting this attribute to exist, and otherwise crashes (and with it also TimeSeries):

sunpy/sunpy/io/_cdf.py

Lines 83 to 87 in 9f0a37d

	# Set fillval values to NaN
	# It would be nice to properley mask these values to work with
	# non-floating point (ie. int) dtypes, but this is not possible with pandas
	if np.issubdtype(data.dtype, np.floating):
	data[data == attrs['FILLVAL']] = np.nan

But as has been mentioned in #5907 (comment), it's not uncommon that a CDF file is not fully following the ISTP standard (these are by the way recent PSP data files, so I'll anyhow contact the instrument team). If the FILLVAL information isn't provided by the instrument team, one could just skip the replacement call. So I propose to change if np.issubdtype(data.dtype, np.floating): to something like if np.issubdtype(data.dtype, np.floating) and 'FILLVAL' in attrs:

To Reproduce

Because the error message occurs in read_cdf when using TimeSeries, this is what I'm calling here directly:

import datetime as dt
from sunpy.io._cdf import read_cdf
from sunpy.net import Fido
from sunpy.net import attrs as a
from sunpy.timeseries import TimeSeries

dataset = 'PSP_SWP_SPI_SF00_L3_MOM'
trange = a.Time(dt.date(2023,3,14), dt.date(2023,3,15))
result = Fido.search(trange, a.cdaweb.Dataset(dataset))
downloaded_files = Fido.fetch(result)
# ts = TimeSeries(downloaded_files, concatenate=True)
data = read_cdf(downloaded_files.data[0])

Screenshots

No response

System Details

• sunpy version: 5.1.1 and 5.1.2

Installation method

pip

[jgieseler]

I'm just noticing that the corresponding CDF variables have as the VAR_TYPE attribute 'ignore_data'. According to the ISTP Guidelines this is a placeholder, but I'm wondering if the instrument team actually had the intention that these variables should be skipped. 🤔