Open multiple netcdf datasets with xarray, and automatically modify (add a dimension with coordinates based on file path) and then combine them.

xarray-openmfdataset-expand-assign-coords.py

### This works ##### 

FILEPATTERN = '...'
def func(ds):
    var = next(var for var in ds)
    fp = ds[var].encoding['source']
    coordds = ds.assign_coords(path=fp)
    dimds = coordds.expand_dims('path')
    return dimds
fs = fsspec.filesystem('gs')
fps = fs.glob(FILEPATTERN)
ds = xr.open_mfdataset([fs.open(x) for x in fps], concat_dim='path', preprocess=func)


### Not this ### 

FILEPATTERN = '...'
def func(ds):
    var = next(var for var in ds)
    fp = ds[var].encoding['source']
    # Here, I swap the order....
    dimds = coordds.expand_dims('path')
    coordds = ds.assign_coords(path=fp)
    return dimds
fs = fsspec.filesystem('gs')
fps = fs.glob(FILEPATTERN)
ds = xr.open_mfdataset([fs.open(x) for x in fps], concat_dim='path', preprocess=func)


### Although opening ONE of these files and doing the `func` stuff manually does work, in both cases... 
### Some deprecated function usage in `open_mfdataset` with `preprocess` ?