AttributeError: 'int' object has no attribute 'encode' HDF5

user_14707642 0
Viewed 1191

I'm trying to open a HDF5 file in Python using the following code:

with h5py.File('example.hdf5', 'r') as f:
    ls = list(f.keys())
    dat = f.get('data')
    dt = np.array(dat)

However, I get this error when executing the last line:

AttributeError: 'int' object has no attribute 'encode'

dat has the following class:

h5py._hl.group.Group'

Anyone knows where the error could come?

The output from iterating inside the file is the following. How can I access inside each part of the file:

checking hdf5 file
/data is a Group
/data/model_cints is a Dataset
/hdf5_track_times is a Dataset
/meta is a Group
/meta/package is a Group
/meta/package/h5py is an object Dataset
/meta/package/numpy is an object Dataset
/meta/package/pypfilt is an object Dataset
/meta/package/python is an object Dataset
/meta/package/scipy is an object Dataset
/meta/package/toml is an object Dataset
/meta/param is a Group
/meta/param/component is a Group
/meta/param/component/model is an object Dataset
/meta/param/component/obs is a Group
/meta/param/component/obs/LatLon is an object Dataset
/meta/param/component/summary_table is a Group
/meta/param/component/summary_table/model_cints is an object Dataset
/meta/param/component/time is an object Dataset
/meta/param/data_dir is an object Dataset
/meta/param/hist is a Group
/meta/param/hist/extra_cols is a Dataset
/meta/param/hist/px_count is a Dataset
/meta/param/hist/wind_shift is a Dataset
/meta/param/hist/wind_size is a Dataset
/meta/param/last_n_periods is a Dataset
/meta/param/minimal_estimation_run is a Dataset
/meta/param/model is a Group
/meta/param/model/param_max is a Dataset
/meta/param/model/param_min is a Dataset
/meta/param/obs is a Group
/meta/param/obs/LatLon is a Group
/meta/param/obs/LatLon/sdev is a Dataset
/meta/param/out_dir is an object Dataset
/meta/param/prng_seed is a Dataset
/meta/param/random is an object Dataset
/meta/param/resample is a Group
/meta/param/resample/method is an object Dataset
/meta/param/resample/reg_toln is a Dataset
/meta/param/resample/regularisation is a Dataset
/meta/param/resample/regularise_or_fail is a Dataset
/meta/param/resample/threshold is a Dataset
/meta/param/scenario is a Group
/meta/param/scenario/id is an object Dataset
/meta/param/scenario/name is an object Dataset
/meta/param/steps_per_unit is a Dataset
/meta/param/summary is a Group
/meta/param/summary/from_first_day is a Dataset
/meta/param/summary/meta is a Group
/meta/param/summary/meta/packages is an object Dataset
/meta/param/summary/only_forecasts is a Dataset
/meta/param/time is a Group
/meta/param/time/start is a Dataset
/meta/param/time/until is a Dataset
/meta/param/tmp_dir is an object Dataset
/meta/prior is a Group
/meta/prior/lat is an object Dataset
/meta/prior/lon is an object Dataset
/meta/prior/speed_lat is an object Dataset
/meta/prior/speed_lon is an object Dataset
/meta/sim is a Group
/meta/sim/cmdline is an object Dataset
h5py hdf5 numpy arrays python
1 Answers

Here is a code snippet to test your keys for type (Group or Dataset). It uses the visititems() method to recursively walk nodes in your file and report each as a 1) Group, 2) Dataset, 3) Object dataset, or 4) Unknown. Once you find a dataset, you can read and create a NumPy array. However, that is not required. You can work with a h5py dataset object "as-if" it is a NumPy array.

def visitor_func(name, node):
    if isinstance(node, h5py.Group):
        print(node.name, 'is a Group')
    elif isinstance(node, h5py.Dataset):
       if (node.dtype == 'object') :
            print (node.name, 'is an object Dataset')
       else:
            print(node.name, 'is a Dataset')   
    else:
        print(node.name, 'is an unknown type')         

#####    

print ('checking hdf5 file')
with h5py.File('example.hdf5', 'r') as h5f:
    h5f.visititems(visitor_func)
user_104628840
Related