def test_compound(self):
phy_quant = 'Current'
phy_unit = 'nA'
slow_to_fast = True
ret_vals = gen_2pos_2spec(s2f_aux=slow_to_fast, mode='compound')
pos_dims, spec_dims, ndata, data_2d = ret_vals
tran = usid.ArrayTranslator()
with tempfile.TemporaryDirectory() as tmp_dir:
file_path = tmp_dir + 'usid_n_pos_n_spec_compound.h5'
_ = tran.translate(file_path,
'Blah',
data_2d,
phy_quant,
phy_unit,
pos_dims,
spec_dims,
slow_to_fast=slow_to_fast)
objects = hs.load(file_path)
assert isinstance(objects, list)
assert len(objects) == 2
# 1. Validate object type
for new_sig, comp_name in zip(objects, ['amp', 'phas']):
compare_signal_from_usid(file_path,
ndata[comp_name],
new_sig,
sig_type=hs.signals.BaseSignal,
axes_to_spec=['Frequency', 'Bias'],
compound_comp_name=comp_name)
def test_read_all_by_default(self):
slow_to_fast = True
pos_dims, spec_dims, ndata, data_2d = gen_2dim(all_pos=False,
s2f_aux=slow_to_fast)
phy_quant = 'Current'
phy_unit = 'nA'
tran = usid.ArrayTranslator()
with tempfile.TemporaryDirectory() as tmp_dir:
file_path = tmp_dir + 'usid_0_pos_n_spec.h5'
_ = tran.translate(file_path,
'Blah',
data_2d,
phy_quant,
phy_unit,
pos_dims,
spec_dims,
slow_to_fast=slow_to_fast)
ret_vals = gen_2dim(all_pos=True, s2f_aux=slow_to_fast)
pos_dims, spec_dims, ndata_2, data_2d_2 = ret_vals
phy_quant = 'Current'
phy_unit = 'nA'
with h5py.File(file_path, mode='r+') as h5_f:
h5_meas_grp = h5_f.create_group('Measurement_001')
_ = usid.hdf_utils.write_main_dataset(h5_meas_grp,
data_2d_2,
'Spat_Map',
phy_quant,
phy_unit,
pos_dims,
spec_dims,
slow_to_fast=slow_to_fast)
objects = hs.load(file_path)
assert isinstance(objects, list)
assert len(objects) == 2
dset_names = [
'/Measurement_000/Channel_000/Raw_Data', 'Measurement_001/Spat_Map'
]
# 1. Validate object type
for new_sig, ndim_data, dataset_path in zip(objects, [ndata, ndata_2],
dset_names):
compare_signal_from_usid(file_path,
ndim_data,
new_sig,
dataset_path=dataset_path)
def test_pick_specific(self):
slow_to_fast = True
ret_vals = gen_2pos_2spec(s2f_aux=slow_to_fast)
pos_dims, spec_dims, ndata, data_2d = ret_vals
phy_quant = 'Current'
phy_unit = 'nA'
tran = usid.ArrayTranslator()
with tempfile.TemporaryDirectory() as tmp_dir:
file_path = tmp_dir + 'usid_n_pos_n_spec.h5'
_ = tran.translate(file_path,
'Blah',
data_2d,
phy_quant,
phy_unit,
pos_dims,
spec_dims,
slow_to_fast=slow_to_fast)
ret_vals = gen_2dim(all_pos=True, s2f_aux=slow_to_fast)
pos_dims, spec_dims, ndata_2, data_2d_2 = ret_vals
phy_quant = 'Current'
phy_unit = 'nA'
with h5py.File(file_path, mode='r+') as h5_f:
h5_meas_grp = h5_f.create_group('Measurement_001')
h5_chan_grp = h5_meas_grp.create_group('Channel_000')
_ = usid.hdf_utils.write_main_dataset(h5_chan_grp,
data_2d_2,
'Raw_Data',
phy_quant,
phy_unit,
pos_dims,
spec_dims,
slow_to_fast=slow_to_fast)
dataset_path = '/Measurement_001/Channel_000/Raw_Data'
new_sig = hs.load(file_path, dataset_path=dataset_path)
compare_signal_from_usid(file_path,
ndata_2,
new_sig,
dataset_path=dataset_path)
def base_n_pos_m_spec(self, lazy, slow_to_fast=True):
phy_quant = 'Current'
phy_unit = 'nA'
ret_vals = gen_2pos_2spec(s2f_aux=slow_to_fast)
pos_dims, spec_dims, ndata, data_2d = ret_vals
tran = usid.ArrayTranslator()
with tempfile.TemporaryDirectory() as tmp_dir:
file_path = tmp_dir + 'usid_n_pos_n_spec.h5'
_ = tran.translate(file_path,
'Blah',
data_2d,
phy_quant,
phy_unit,
pos_dims,
spec_dims,
slow_to_fast=slow_to_fast)
new_sig = hs.load(file_path, lazy=lazy)
compare_signal_from_usid(file_path,
ndata,
new_sig,
sig_type=hs.signals.BaseSignal,
axes_to_spec=['Frequency', 'Bias'])
def test_non_linear_dimension(self):
pos_dims = [
usid.Dimension('Y', 'um', np.linspace(0, 60, num=5)),
usid.Dimension('X', 'nm', [-250, 750])
]
spec_dims = [
usid.Dimension('Bias', 'V', np.sin(np.linspace(0, 2 * np.pi,
num=7))),
usid.Dimension('Frequency', 'kHz', [300, 350, 400])
]
ndata = np.random.rand(5, 2, 7, 3)
phy_quant = 'Current'
phy_unit = 'nA'
data_2d = ndata.reshape(np.prod(ndata.shape[:2]),
np.prod(ndata.shape[2:]))
tran = usid.ArrayTranslator()
with tempfile.TemporaryDirectory() as tmp_dir:
file_path = tmp_dir + 'usid_n_pos_n_spec_non_lin_dim.h5'
_ = tran.translate(file_path,
'Blah',
data_2d,
phy_quant,
phy_unit,
pos_dims,
spec_dims,
slow_to_fast=True)
with pytest.raises(ValueError):
_ = hs.load(file_path, ignore_non_linear_dims=False)
with pytest.warns(UserWarning) as _:
new_sig = hs.load(file_path)
compare_signal_from_usid(file_path,
ndata,
new_sig,
axes_to_spec=['Frequency', 'Bias'],
invalid_axes=True)
def test_0_pos_n_spec(self):
phy_quant = 'Current'
phy_unit = 'nA'
slow_to_fast = True
pos_dims, spec_dims, ndata, data_2d = gen_2dim(all_pos=False,
s2f_aux=slow_to_fast)
tran = usid.ArrayTranslator()
with tempfile.TemporaryDirectory() as tmp_dir:
file_path = tmp_dir + 'usid_0_pos_n_spec.h5'
_ = tran.translate(file_path,
'Blah',
data_2d,
phy_quant,
phy_unit,
pos_dims,
spec_dims,
slow_to_fast=slow_to_fast)
new_sig = hs.load(file_path)
compare_signal_from_usid(file_path,
ndata,
new_sig,
sig_type=hs.signals.BaseSignal,
axes_to_spec=[])
def file_writer(filename, object2save, **kwds):
"""
Writes a HyperSpy Signal object to a HDF5 file formatted according to USID
Parameters
----------
filename : str
Path to target HDF5 file
object2save : hyperspy.signals.Signal
A HyperSpy signal
"""
append = False
if os.path.exists(filename):
append = True
hs_shape = object2save.data.shape
# Not sure how to safely ignore spurious / additional dimensions
if len(object2save.axes_manager.shape) != len(hs_shape):
raise ValueError('Number of dimensions in data (shape: {}) does not '
'match number of axes: ({})'
'.'.format(hs_shape,
len(object2save.axes_manager.shape)))
parm_dict = _flatten_dict(object2save.metadata.as_dictionary())
temp = object2save.original_metadata.as_dictionary()
parm_dict.update(_flatten_dict(temp, parent_key='Original'))
num_pos_dims = object2save.axes_manager.navigation_dimension
nav_axes = object2save.axes_manager.navigation_axes
sig_axes = object2save.axes_manager.signal_axes
data_2d = object2save.data
# data_2d is assumed to have dimensions arranged from slowest to fastest
# varying dimensions
if num_pos_dims > 0 and object2save.axes_manager.signal_dimension > 0:
# now flatten to 2D:
data_2d = data_2d.reshape(np.prod(hs_shape[:num_pos_dims]),
np.prod(hs_shape[num_pos_dims:]))
pos_dims = _axes_list_to_dimensions(nav_axes,
hs_shape[:num_pos_dims], False)
spec_dims = _axes_list_to_dimensions(sig_axes,
hs_shape[num_pos_dims:], True)
elif num_pos_dims == 0:
# only spectroscopic:
# now flatten to 2D:
data_2d = data_2d.reshape(1, -1)
pos_dims = _axes_list_to_dimensions(nav_axes, [], False)
spec_dims = _axes_list_to_dimensions(sig_axes, hs_shape, True)
else:
# now flatten to 2D:
data_2d = data_2d.reshape(-1, 1)
pos_dims = _axes_list_to_dimensions(nav_axes, hs_shape, False)
spec_dims = _axes_list_to_dimensions(sig_axes, [], True)
# Does HyperSpy store the physical quantity and units somewhere?
phy_quant = 'Unknown Quantity'
phy_units = 'Unknown Units'
dset_name = 'Raw_Data'
if not append:
tran = usid.ArrayTranslator()
_ = tran.translate(filename, dset_name, data_2d, phy_quant, phy_units,
pos_dims, spec_dims, parm_dict=parm_dict,
slow_to_fast=True, **kwds)
else:
with h5py.File(filename, mode='r+') as h5_f:
h5_grp = sidpy.prov_utils.create_indexed_group(h5_f, 'Measurement')
sidpy.hdf_utils.write_simple_attrs(h5_grp, parm_dict)
h5_grp = sidpy.prov_utils.create_indexed_group(h5_grp, 'Channel')
_ = usid.hdf_utils.write_main_dataset(h5_grp, data_2d, dset_name,
phy_quant, phy_units,
pos_dims, spec_dims,
slow_to_fast=True, **kwds)
