import uproot
import h5py
import numpy as np
[docs]def convert_h5_to_root(path_h5,
path_root,
nmbr_channels):
"""
Convert a HDF5 file to a ROOT file.
:param path_h5: The path to the hdf5 file that is read.
:type string: string
:param path_root: The path to the root file that is created.
:type path_root: string
:param nmbr_channels: The number of channels of the module.
:type nmbr_channels: int
"""
f_h5 = h5py.File(path_h5, 'r')
f_root = uproot.recreate(path_root)
print('HDF5 has groups: ', f_h5.keys())
for group in f_h5.keys():
print('ADDING TREE: ', group)
print('HDF5 group {} has data sets: {}'.format(group, f_h5[group].keys()))
as_dict = {}
data_types = {}
for dataset in f_h5[group].keys():
# this could create a bug if the dataset randomly has the size of nmbr_channels
if len(f_h5[group][dataset]) == nmbr_channels:
for i in range(nmbr_channels):
name = dataset + '_' + str(i)
data_types[name] = np.float
# handling of the standard event - potentially this could produce bugs in the future for other single
# events that are stores because size is then (channels, features) instead (channels, events, features)
if group == 'stdevent' or group == 'stdevent_tp':
print('-- ADDING BRANCH: ' + name + ', LENGTH: ' + str(1))
as_dict[name] = np.array(f_h5[group][dataset][i]).reshape((1, -1))
else:
print('-- ADDING BRANCH: ' + name + ', LENGTH: ' + str(len(f_h5[group][dataset][i])))
as_dict[name] = np.array(f_h5[group][dataset][i])
else:
name = dataset
print('-- ADDING BRANCH: ' + name + ', LENGTH: ' + str(len(f_h5[group][dataset])))
data_types[name] = np.float
as_dict[name] = np.array(f_h5[group][dataset])
f_root[group] = uproot.newtree(data_types)
f_root[group].extend(as_dict)
print('Converted HDF5 to ROOT file.')