cfgrib: Unable to retrieve all variables with cfgrib

Hello,

I’m trying to handle a Grib2 file from which I can extract all the data using directly the command ‘grib_get_data -s stepUnits=0 out.grb’.

So it seems I don’t need all stepUnits.

In the other hand, when using cfgrib it is impossible to read it fully. Here is the command I tried first :

import cfgrib
cfgrib.open_datasets('out.grb')

And here is the result :

<ipython-input-36-ccea25a1a3be> in <module>
      1 import cfgrib
----> 2 cfgrib.open_datasets('out.grb')

~/.local/lib/python3.6/site-packages/cfgrib/xarray_store.py in open_datasets(path, backend_kwargs, **kwargs)
    100     backend_kwargs = backend_kwargs.copy()
    101     backend_kwargs["squeeze"] = False
--> 102     datasets = open_variable_datasets(path, backend_kwargs=backend_kwargs, **kwargs)
    103 
    104     type_of_level_datasets = {}  # type: T.Dict[str, T.List[xr.Dataset]]

~/.local/lib/python3.6/site-packages/cfgrib/xarray_store.py in open_variable_datasets(path, backend_kwargs, **kwargs)
     88         bk["filter_by_keys"] = backend_kwargs.get("filter_by_keys", {}).copy()
     89         bk["filter_by_keys"]["paramId"] = param_id
---> 90         datasets.extend(raw_open_datasets(path, bk, **kwargs))
     91     return datasets
     92 

~/.local/lib/python3.6/site-packages/cfgrib/xarray_store.py in raw_open_datasets(path, backend_kwargs, **kwargs)
     63     datasets = []
     64     try:
---> 65         datasets.append(open_dataset(path, backend_kwargs=backend_kwargs, **kwargs))
     66     except DatasetBuildError as ex:
     67         fbks.extend(ex.args[2])

~/.local/lib/python3.6/site-packages/cfgrib/xarray_store.py in open_dataset(path, **kwargs)
     36         raise ValueError("only engine=='cfgrib' is supported")
     37     kwargs["engine"] = "cfgrib"
---> 38     return xr.open_dataset(path, **kwargs)  # type: ignore
     39 
     40 

~/.local/lib/python3.6/site-packages/xarray/backends/api.py in open_dataset(filename_or_obj, group, decode_cf, mask_and_scale, decode_times, autoclose, concat_characters, decode_coords, engine, chunks, lock, cache, drop_variables, backend_kwargs, use_cftime, decode_timedelta)
    570 
    571         opener = _get_backend_cls(engine)
--> 572         store = opener(filename_or_obj, **extra_kwargs, **backend_kwargs)
    573 
    574     with close_on_error(store):

~/.local/lib/python3.6/site-packages/xarray/backends/cfgrib_.py in __init__(self, filename, lock, **backend_kwargs)
     43             lock = ECCODES_LOCK
     44         self.lock = ensure_lock(lock)
---> 45         self.ds = cfgrib.open_file(filename, **backend_kwargs)
     46 
     47     def open_store_variable(self, name, var):

~/.local/lib/python3.6/site-packages/cfgrib/dataset.py in open_file(path, grib_errors, indexpath, filter_by_keys, read_keys, time_dims, extra_coords, **kwargs)
    718     return Dataset(
    719         *build_dataset_components(
--> 720             index, read_keys=read_keys, time_dims=time_dims, extra_coords=extra_coords, **kwargs
    721         )
    722     )

~/.local/lib/python3.6/site-packages/cfgrib/dataset.py in build_dataset_components(index, errors, encode_cf, squeeze, log, read_keys, time_dims, extra_coords)
    639                 read_keys=read_keys,
    640                 time_dims=time_dims,
--> 641                 extra_coords=extra_coords,
    642             )
    643         except DatasetBuildError as ex:

~/.local/lib/python3.6/site-packages/cfgrib/dataset.py in build_variable_components(index, encode_cf, filter_by_keys, log, errors, squeeze, read_keys, time_dims, extra_coords)
    499         }
    500         attributes.update(COORD_ATTRS.get(coord_name, {}).copy())
--> 501         data = np.array(sorted(values, reverse=attributes.get("stored_direction") == "decreasing"))
    502         dimensions = (coord_name,)  # type: T.Tuple[str, ...]
    503         if squeeze and len(values) == 1:

TypeError: '<' not supported between instances of 'float' and 'str'

After investigating, I tried to filter like that :

cfgrib.open_datasets('out.grb', backend_kwargs={'filter_by_keys': {'stepUnits': 0}})

It works, BUT I do not retrieve all data. Then I tried with :

cfgrib.open_datasets('out.grb', backend_kwargs={'filter_by_keys': {'stepUnits': 1}})

And it fails with the same above error.

So my question is why I would need to get stepUnits=1 (which seems to generate en error) with cfgrib while with eccodes binary grib_get_data there is no problem ?

And how could I do to use cfgrib without making distinction with stepUnits please?

Thanks for your help, Christophe