h5py 如何将数据集移动到新组 - 奇怪的错误

Question

I have an hdf5 file that has a dataset 'foo' located at the root. I would like to move it into a new group 'bar'. I have RTFM and attempted the following:

with h5py.File(fname, 'a') as h5file:
    newgroup = h5file.create_group('bar')
    h5file.move('foo', 'bar/foo')

This appears to work fine. Now I try to actually verify that it worked

with h5py.File(fname, 'r') as h5file:
    print(h5file['bar']['foo'].shape)

This actually prints the correct shape of the dataset. But immediately afterwards it throws an exception KeyError: "Unable to open object (object 'foo' doesn't exist)" . Or, more precisely,

/opt/anaconda3/envs/py36qt5/lib/python3.6/site-packages/h5py/_hl/group.py in __getitem__(self, name)
    165                 raise ValueError("Invalid HDF5 object reference")
    166         else:
--> 167             oid = h5o.open(self.id, self._e(name), lapl=self._lapl)
    168 
    169         otype = h5i.get_type(oid)

Any idea what is going wrong? h5py.__version__ = 2.7.1

EDIT :

Here's an even more idiotic observation. The following code works as expected and does not throw the above exception.

with h5py.File(fname, 'r') as h5file:
    for key in h5file['bar'].keys():
        print(key, h5file['bar'][key].shape)

I am at a complete loss...

Answer 1

I am running h5py 2.10 , and this code segment works fine.
I don't know why it doesn't work for you.

import h5py
import numpy as np

fname='SO_64558399.h5'

with h5py.File(fname, 'w') as h5file:
    arr = np.arange(100).reshape(10,10)
    h5file.create_dataset('foo',data=arr)
    
with h5py.File(fname, 'a') as h5file:
    newgroup = h5file.create_group('bar')
    h5file.move('foo', 'bar/foo')
    
with h5py.File(fname, 'r') as h5file:
    print(h5file['bar']['foo'].shape)
    for key in h5file['bar'].keys():
        print(key, h5file['bar'][key].shape)

The printed output is:

(10, 10)
foo (10, 10)