Speed up netCDF4, h5netcdf backends (#9067)

* Speed up netCDF4 backend.

xref #9058

Accessing `.shape` on a netCDF4 variable is ~20-40ms.
This can add up for large numbers of variables, e.g.:
#9058

We already request the shape when creating NetCDF4ArrayWrapper,
so we can reuse that.

* Update h5netcdf backend too

doc/whats-new.rst

@@ -23,6 +23,12 @@ v2024.05.1 (unreleased)
 New Features
 ~~~~~~~~~~~~
 
+Performance
+~~~~~~~~~~~
+
+- Small optimization to the netCDF4 and h5netcdf backends (:issue:`9058`, :pull:`9067`).
+  By `Deepak Cherian <https://github.com/dcherian>`_.
+
 
 Breaking changes
 ~~~~~~~~~~~~~~~~

xarray/backends/h5netcdf_.py

@@ -231,7 +231,7 @@ def open_store_variable(self, name, var):
 
         # save source so __repr__ can detect if it's local or not
         encoding["source"] = self._filename
-        encoding["original_shape"] = var.shape
+        encoding["original_shape"] = data.shape
 
         vlen_dtype = h5py.check_dtype(vlen=var.dtype)
         if vlen_dtype is str:

xarray/backends/netCDF4_.py

@@ -464,7 +464,7 @@ def open_store_variable(self, name: str, var):
         pop_to(attributes, encoding, "least_significant_digit")
         # save source so __repr__ can detect if it's local or not
         encoding["source"] = self._filename
-        encoding["original_shape"] = var.shape
+        encoding["original_shape"] = data.shape
 
         return Variable(dimensions, data, attributes, encoding)