hdf5 reading for unyt.quantity.

[aaramburo]

Hi I believe that there is a bug whenever I try to read a unyt.unyt_quantity that was stored in a hdf5 file using write_hdf5. Basically I think it treats the quantity as a non-zero dimensional array, and uses ellipses to get the data. I have tried to use unyt.unyt_quantity.from_hdf5() and unyt.unyt_array.from_hdf5().

----> 1 unyt.unyt_quantity.from_hdf5("...")
File (.../python3.9/site-packages/unyt/array.py:1557), in unyt_array.from_hdf5(cls, filename, dataset_name, group_name)
   1555     g = f
  1555     g = f
   1556 dataset = g[dataset_name]
-> 1557 data = dataset[:]
   1558 units = dataset.attrs.get("units", "")
   1559 unit_lut = default_unit_symbol_lut.copy()

File (objects.pyx:54), in h5py._objects.with_phil.wrapper()

File (objects.pyx:55), in h5py._objects.with_phil.wrapper()

File(..../h5py/_hl/dataset.py:817), in Dataset.__getitem__(self, args, new_dtype)
    815 if self.shape == ():
    816     fspace = self.id.get_space()
--> 817     selection = sel2.select_read(fspace, args)
    818     if selection.mshape is None:
    819         arr = numpy.zeros((), dtype=new_dtype)

File (../h5py/_hl/selections2.py:101), in select_read(fspace, args)
     96 """ Top-level dispatch function for reading.
     97 
     98 At the moment, only supports reading from scalar datasets.
     99 """
    100 if fspace.shape == ():
--> 101     return ScalarReadSelection(fspace, args)
    103 raise NotImplementedError()

File [.../h5py/_hl/selections2.py:86), in ScalarReadSelection.__init__(self, fspace, args)
     84     self.mshape = ()
     85 else:
---> 86     raise ValueError("Illegal slicing argument for scalar dataspace")
     88 self.mspace = h5s.create(h5s.SCALAR)
     89 self.fspace = fspace

ValueError: Illegal slicing argument for scalar dataspace

[neutrinoceros]

Can you provide a self-contained reproducer script ?

[aaramburo]

Hi yes, here is a simple script that for me reproduces the issue

import unyt

variable_quantity = 5*unyt.cm
variable_quantity.write_hdf5("example.hdf5",dataset_name="variable", group_name="example_quantity")
in_quantity = unyt.unyt_quantity.from_hdf5("example.hdf5",dataset_name="variable", group_name="example_quantity")

[neutrinoceros]

Thank you for reporting this. I opened a fix at #560