assign 'dem_source': string to xr.Dataset in shism.py

[tomsail]

I have a problem at this line when using the sizing from oceanmesh:

Traceback (most recent call last):
  File "/home/tomsail/work/python/pyPoseidon/Tutorial/mesh_model_test.py", line 40, in <module>
    b.create()
  File "/home/tomsail/work/python/pyPoseidon/pyposeidon/schism.py", line 455, in create
    self.mesh = pmesh.set(type="tri2d", **kwargs)
  File "/home/tomsail/work/python/pyPoseidon/pyposeidon/mesh.py", line 36, in set
    return tri2d(**kwargs)
  File "/home/tomsail/work/python/pyPoseidon/pyposeidon/mesh.py", line 226, in __init__
    g, bg = moceanmesh.get(self.boundary.contours, **kwargs)  # create mesh with OCEANMESH
  File "/home/tomsail/work/python/pyPoseidon/pyposeidon/moceanmesh.py", line 136, in get
    gr = make_oceanmesh(contours, **kwargs)
  File "/home/tomsail/work/python/pyPoseidon/pyposeidon/moceanmesh.py", line 411, in make_oceanmesh
    dem = om.DEM(dem_source, crs=crs, bbox=extent)
  File "/home/tomsail/work/python/pyPoseidon/Tutorial/.venv/lib/python3.10/site-packages/oceanmesh/geodata.py", line 777, in __init__
    if dem.exists():
  File "/home/tomsail/work/python/pyPoseidon/Tutorial/.venv/lib/python3.10/site-packages/xarray/core/common.py", line 278, in __getattr__
    raise AttributeError(
AttributeError: 'Dataset' object has no attribute 'exists'

This is cause by the update of this line :

 kwargs.update({"dem_source": self.dem.Dataset})

is that line really needed ? since we already have the variable self.dem attributed?

Problem mentioned in #114 but I am opening a new issue here.

FIY I removed that line in the telemac.py (coming soon on my fork)

[brey]

I think the problem is with the test if dem.exists().

dem should be an instance with dem.Dataset as the xarray object. On the other hand the dem_source is an attribute of the model class indicating the source for the dem.

If you want to check if the Dataset exists in the instance then maybe:

'Dataset' in dem.__dict__

[pmav99]

That might work, but hasattr(dem, "Dataset") is probably a little bit cleaner.