Some missing values not being converted to NaN in IGRA2

[danielmwatkins]

Values flagged as missing or invalid in the raw IGRA2 files contain the value -9999 or -8888. On fields that have a decimal value (e.g. 1234 representing 123.4) instead of NaN we sometimes get -999.9.

[avatar101]

@DanielWatkins
Hi, I'm also using IGRA datasets for my work and wrote a code to import it in a pandas data frame. I wish I knew the Siphon module before. Now, I would be happy to test siphon module but, I would like to see the background functions you wrote for reading the data in order to compare and improve my coding style. I wasn't able to find it here, being new to this platform. However, I did find your repo of climstat/igratools.py. Is it the same set of functions that have been implemented?

Cheers,
EDIT: Found the code.

[danielmwatkins]

@avatar101 Glad you found the code! igratools is an old incomplete version of what's in Siphon. The version in Siphon is much simpler than the approach I was taking in igratools. For one thing, I learned about the existence of pandas.read_fwf() which was helpful and faster.

[avatar101]

I'm also using pd.read_fwf() but, your overall approach is much better. If you're interested in the approach I'm using, it is based on this https://stackoverflow.com/questions/48992164/reading-file-between-headers-in-python

It makes the code shorter but, I find it a bit complicated. I find your approach better and certainly, your function is much more versatile and powerful than what I could come up with. However, I still need some time to go through all the code you wrote.

[dopplershift]

@DanielWatkins did your PR #219 fix this?