Let torch accessor and dataloader handle either xarray.DataArray or xarray.Dataset inputs

[weiji14]

Description of proposed changes

Convert xarray.Dataset to xarray.DataArray first, so that the .data method works to get the underlying array which can be converted to a torch.Tensor. Also added parametrized unit tests, only for the 2D ('x', 'y') case for both xarray.DataArray and xarray.Dataset.

TODO:

• Preliminary support in .torch accessor to convert xarray.Dataset directly to torch.Tensor
• Add parametrized unit tests for xarray.Dataset and xarray.DataArray inputs
  • In the DataLoader xbatcher/loaders/torch.py
  • In the xarray accessors xbatcher/accessor.py

Fixes #84, continuation of #71

[codecov]

Codecov Report

Merging #85 (106fff9) into main (0ded974) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##              main       #85   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            5         5           
  Lines          179       182    +3     
  Branches        40        37    -3     
=========================================
+ Hits           179       182    +3     

Impacted Files	Coverage Δ
xbatcher/accessors.py	100.00% <100.00%> (ø)
xbatcher/generators.py	100.00% <100.00%> (ø)
xbatcher/loaders/keras.py	100.00% <100.00%> (ø)
xbatcher/loaders/torch.py	100.00% <100.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[maxrjones]

LGTM, thanks @weiji14!

It might be worthwhile to move _as_xarray_dataarray outside of the TorchAccessor class and make the check between the last item in the generator and batch a general testing utility function, but we could do that when they're needed elsewhere in the package.

[weiji14]

It might be worthwhile to move _as_xarray_dataarray outside of the TorchAccessor class and make the check between the last item in the generator and batch a general testing utility function, but we could do that when they're needed elsewhere in the package.

Yep, could probably reuse those utility functions for the keras/tensorflow side too!

Btw, I don't have merge permissions, but happy to be added as a maintainer if extra hands are welcome 😃

[maxrjones]

Btw, I don't have merge permissions, but happy to be added as a maintainer if extra hands are welcome 😃

fantastic! I'll add you as soon as we finish the transfer discussed at #86.