Convert binary punch file to netCDF

[iriadigos]

I am trying to convert a punch file to netCDF using the "AC_tools" but it shows me the following error when I use the "convert_to_netCDF" function:

_File "/home/i/irs11/miniconda3/lib/python3.7/site-packages/AC_tools/bpch2netCDF.py", line 209, in bpch_to_netCDF bpch_data = datasets.load(bpch_files)

NameError: name 'datasets' is not defined_

I'm using the following packages: python 3.7, iris 2.2.0, numpy 1.17.2, netcdf4 1.5.3 ..... I don't have the pygchem package because I am not able to install it in python 3.7. Consequently, I've tried to use an environment with python 2.7, but in such case AC_tools needs python=>3.5 to be installed. So I got stuck.

Any suggestion?

[tsherwen]

Hello @iriadigos,

Thanks for writing.

Yes. To use the iris backend via PyGChem in AC_tools, you need to run in a python 2 environment. This is due to PyGChem. AC_tools should install fine in Python 2 environment. Is your install of AC_tools not successful in a python2 environment?

When using the convert_to_netCDF function, you need to be in a Python2 environment. For doing anything else using AC_tools, I would recommend using a python3 environment.

[iriadigos]

No, I am not able to install AC_tools in a python2 environment. This is the error:

ERROR: Package 'AC-tools' requires a different Python: 2.7.17 not in '>=3.5'

[tsherwen]

OK. You are installing via the pip method listed on the README, correct?

I've not re-installed a AC_tools in python2 since switching to the pip method. I think the second approach listed on the wiki should work. Would be able to give that a go?

I am assuming you have successfully installed PyGChem and iris in your Python2 environment already?

[iriadigos]

I have successfully installed AC_tools with the second approach. PyGChem 0.3.0 and iris 2.2.1 are also installed in the environment.

But again, I have the same error when executing "convert_to_netCDF" function:

Creating a netCDF from 1 file(s). This can take some time... Traceback (most recent call last):

File "", line 1, in _ds=AC.convert_to_netCDF(folder=folder,bpch_file_type='tracavg.*')

_File "/home/i/irs11/miniconda3/envs/py27/lib/python2.7/site-packages/AC_tools-0.1.1-py2.7.egg/AC_tools/bpch2netCDF.py", line 62, in convert_to_netCDF file_type=bpch_filetype, verbose=verbose)

_File "/home/i/irs11/miniconda3/envs/py27/lib/python2.7/site-packages/AC_tools-0.1.1-py2.7.egg/AC_tools/bpch2netCDF.py", line 209, in bpch_to_netCDF bpch_data = datasets.load(bpchfiles)

NameError: global name 'datasets' is not defined

[tsherwen]

Please could I check what (if any) error messages you get when you paste in the following lines?

import pygchem
from pygchem import datasets
import pygchem.datafields as datasets

I just checked and PyChem version I have from running the same process is '0.3.0dev' and the iris version is '1.13.0'. You may need to force the iris version to be the same, as I think some large happened at v2.

[iriadigos]

Importing pygchem does not show any error. However, the other two lines show these errors:

`from pygchem import datasets Traceback (most recent call last):

File "", line 1, in from pygchem import datasets

File "/home/i/irs11/miniconda3/envs/py27/lib/python2.7/site-packages/PyGChem-0.3.0.dev0-py2.7.egg/pygchem/datasets.py", line 97, in _load_backend(DEFAULT_BACKEND)

File "/home/i/irs11/miniconda3/envs/py27/lib/python2.7/site-packages/PyGChem-0.3.0.dev0-py2.7.egg/pygchem/datasets.py", line 41, in _load_backend .format(backend_name))

File "/home/i/irs11/miniconda3/envs/py27/lib/python2.7/importlib/init.py", line 37, in import_module import(name)

File "/home/i/irs11/miniconda3/envs/py27/lib/python2.7/site-packages/PyGChem-0.3.0.dev0-py2.7.egg/pygchem/dataset_backends/backend_iris.py", line 27, in import iris.unit

ImportError: No module named unit

import pygchem.datafields as datasets Traceback (most recent call last):

File "", line 1, in import pygchem.datafields as datasets

ImportError: No module named datafields`

[iriadigos]

Finally, it works with the iris version '1.13.0'. Thank you very much.

[tsherwen]

Good to hear!

I will add some information on this to the README to help others who encounter this in the future.