xr.open_dataset no groups info

[ThetomekK]

Code Sample, a copy-pastable example if possible

I have to write some data to disk using xr.to_netcdf() method. The data must be structured in groups so i use the group key word. Reading the .nc file from disk leads to empty data variables if group not supplied. Here some samples

data_ds.to_netcdf(path=savepath,mode='w',format='NETCDF4',group='Audio',engine='netcdf4')

datafromdisk = xr.open_dataset(savepath)
datafromdisk

<xarray.Dataset>
Dimensions:  ()
Data variables:
    *empty*

datafromdisk = xr.open_dataset(savepath,group='Audio')

<xarray.Dataset>
Dimensions:  (time: 15360000)
Coordinates:
  * time     (time) datetime64[ns] 2017-05-30T07:40:00 ... 2017-05-30T07:49:59.992280938
Data variables:
    audio    (time) float32 ...
Attributes:
    unit:     Pa

Problem description

Actually this is not a real problem, if you keep in mind what data groups are strored in a .nc file.
At the moment, i work arround with netCDF4 to get infos about possible groups within a .nc file.

from netCDF4 import Dataset
rootgrp = Dataset(savepath)
rootgrp

<class 'netCDF4._netCDF4.Dataset'>
root group (NETCDF4 data model, file format HDF5):
    dimensions(sizes): 
    variables(dimensions): 
    groups: Audio

# or

rootgrp.groups

OrderedDict([('Audio', <class 'netCDF4._netCDF4.Group'>
              group /Audio:
                  unit: Pa
                  dimensions(sizes): time(15360000)
                  variables(dimensions): float32 audio(time), float64 time(time)
                  groups: )])

Expected Output

Well, at least something like this i would appreciate:

datafromdisk = xr.open_dataset(savepath)
datafromdisk

<class 'netCDF4._netCDF4.Dataset'>
root group (NETCDF4 data model, file format HDF5):
    dimensions(sizes): 
    variables(dimensions): 
    groups: Audio

Output of xr.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.7.1 | packaged by conda-forge | (default, Mar 13 2019, 13:32:59) [MSC v.1900 64 bit (AMD64)] python-bits: 64 OS: Windows OS-release: 7 machine: AMD64 processor: Intel64 Family 6 Model 60 Stepping 3, GenuineIntel byteorder: little LC_ALL: None LANG: en LOCALE: None.None libhdf5: 1.10.4 libnetcdf: 4.6.2

xarray: 0.12.1
pandas: 0.24.2
numpy: 1.16.2
scipy: 1.2.1
netCDF4: 1.5.0.1
pydap: None
h5netcdf: None
h5py: None
Nio: None
zarr: None
cftime: 1.0.3.4
nc_time_axis: None
PseudonetCDF: None
rasterio: None
cfgrib: None
iris: None
bottleneck: 1.2.1
dask: 1.2.0
distributed: 1.27.0
matplotlib: 3.0.3
cartopy: None
seaborn: None
setuptools: 41.0.0
pip: 19.0.3
conda: None
pytest: None
IPython: 7.4.0
sphinx: 2.0.1

[dcherian]

I'm in favour of printing out a nice warning message when the netcdf file has groups. Listing them would be even better.

Also it looks like the docs need to be updated to mention netCDF groups: https://xarray.pydata.org/en/stable/io.html#netcdf

[shoyer]

It might make sense to print a warning if a group was not explicitly selected, the root group is empty and there is another non-empty group. Though I'm a little reluctant to do this since empty groups are perfectly valid, and it's a little annoying to get warnings for things that may not be programmer errors.

[ThetomekK]

@shoyer
In my opinion, if xarray offers optional groups in to_netcdf() method, it should consequently provide a. groups attribute, if a.nc file contains groups and is read without group selection.

[shoyer]

see #1092 for discussion about Dataset groups in xarray's data model

[dcherian]

@shoyer What about adding groups to just the repr? That way the user knows there are group names they can pass to open_dataset

<xarray.Dataset>
Dimensions:  ()
Groups: Audio
Data variables:
    *empty*

[zdgriffith]

I work with netCDF groups regularly and am interested in this issue. I agree with @dcherian that having groups added to the Dataset repr when a netCDF file with groups is loaded would be helpful (regardless of if the root group is empty or not) and pretty unobtrusive.

[dcherian]

Actually I don't think that is a good idea any more. A Dataset represents a single group so it's weird to print group info under a dataset repr