def test_validate_dataset(self):
ds1 = sdf.Dataset('DS1')
ds1.data = 1
errors = sdf._validate_dataset(ds1)
self.assertEqual(["Dataset.data must be a numpy.ndarray"], errors)
ds1.data = np.array([])
errors = sdf._validate_dataset(ds1)
self.assertEqual(["Dataset.data must not be empty"], errors)
ds1.data = np.array(1).astype(np.float32)
errors = sdf._validate_dataset(ds1)
self.assertEqual(["Dataset.data.dtype must be numpy.float64"], errors)
ds1.data = np.array([1.0, 2.0])
errors = sdf._validate_dataset(ds1)
self.assertEqual(
["The number of scales does not match the number of dimensions"],
errors)
ds2 = sdf.Dataset('DS2', data=np.array([0.0, 1.0, 2.0]), is_scale=True)
ds1.scales = [ds2]
errors = sdf._validate_dataset(ds1)
self.assertEqual([], errors)
ds2.data = np.array([[1.0, 2.0], [3.0, 4.0]])
errors = sdf._validate_dataset(ds2)
self.assertEqual(["Scales must be one-dimensional"], errors)
ds2.data = np.array([0, 1.0, 1.0])
errors = sdf._validate_dataset(ds2)
self.assertEqual(["Scales must be strictly monotonic increasing"],
errors)
def _create_dataset(dsobj, datasets):
""" Create a dataset from an h5py dataset """
_, name = os.path.split(dsobj.name)
ds = sdf.Dataset(name, data=dsobj.value)
for attr in dsobj.attrs:
if attr == 'COMMENT':
ds.comment = _to_python_str(dsobj.attrs[attr])
elif attr == 'NAME':
ds.display_name = _to_python_str(dsobj.attrs[attr])
elif attr == 'RELATIVE_QUANTITY' and _to_python_str(dsobj.attrs[attr]) == 'TRUE':
ds.relative_quantity = True
elif attr == 'UNIT':
ds.unit = _to_python_str(dsobj.attrs[attr])
elif attr == 'DISPLAY_UNIT':
ds.display_unit = _to_python_str(dsobj.attrs[attr])
elif attr == 'CLASS' and _to_python_str(dsobj.attrs[attr]) == 'DIMENSION_SCALE':
ds.is_scale = True
elif attr == 'REFERENCE_LIST':
ds.is_scale = True
elif attr in ['REFERENCE_LIST', 'DIMENSION_LIST']:
pass
else:
ds.attributes[attr] = _to_python_str(dsobj.attrs[attr])
ds.scales = [None] * ds.data.ndim
datasets[dsobj] = ds
return ds
def test_data_types(self):
ds_f = sdf.Dataset(name='f',
data=np.asarray([1, 2, 3], dtype=np.float32))
ds_d = sdf.Dataset(name='d',
data=np.asarray([1, 2, 3], dtype=np.float64))
ds_i = sdf.Dataset(name='i',
data=np.asarray([1, 2, 3], dtype=np.int32))
g = sdf.Group(name='/', datasets=[ds_f, ds_d, ds_i])
sdf.save('data_types.sdf', g)
g = sdf.load('data_types.sdf')
self.assertEqual(g['f'].data.dtype, np.float32)
self.assertEqual(g['d'].data.dtype, np.float64)
self.assertEqual(g['i'].data.dtype, np.int32)
def test_hierarchy(self):
# create a scale
ds_time = sdf.Dataset('Time',
comment="A scale",
data=np.linspace(0, 10, 101),
unit='s',
is_scale=True)
ds_sine = sdf.Dataset('sine',
comment="A 1-d dataset /w attached scale",
data=np.sin(ds_time.data),
scales=[ds_time])
# create the root group
g = sdf.Group(name='/',
comment="A test file",
attributes={'A1': "my string"},
datasets=[ds_time, ds_sine])
# create a scalar dataset
ds_alpha = sdf.Dataset(
'alpha',
comment="A scalar /w unit, display unit and display name",
data=np.pi,
display_name='Angle',
unit='rad',
display_unit='deg')
# create a sub group
g1 = sdf.Group(name='g1',
comment="A sub-group",
attributes={'A2': "Attribute in sub group"},
datasets=[ds_alpha])
g.groups.append(g1)
# save the group
sdf.save('roundtrip.sdf', g)
# load the group from the file
g2 = sdf.load('roundtrip.sdf', '/')
def test_roundtrip(self):
# create a scale
ds1 = sdf.Dataset('DS1',
comment='dataset 1',
data=np.array([0.1, 0.2, 0.3]),
display_name='Scale 1',
scale_name='Scale 1',
quantity='Q1',
unit='U1',
display_unit='DU1',
is_scale=True)
# create a 1D dataset
ds2 = sdf.Dataset('DS2',
comment='dataset 2',
data=np.array([1, 2, 3]),
display_name='Dataset 2',
quantity='Q2',
unit='U2',
display_unit='DU2',
scales=[ds1])
# create a group
g = sdf.Group(name='/',
comment='my comment',
attributes={
'A1': 'my string',
'A2': 0.1,
'A3': 1
},
datasets=[ds1, ds2])
# save the group
sdf.save('test.sdf', g)
# load DS2 from the file
ds2r = sdf.load('test.sdf', '/DS2')
# make sure the content is still the same
self.assertEqual(ds2, ds2r)
def load_from_mat(input_filename, schema_data=None):
mat_data = loadmat(input_filename)
try:
del mat_data['__header__']
del mat_data['__globals__']
del mat_data['__version__']
except:
pass
data_list = [sdf.Dataset(key, data=mat_data[key][0].tolist()) for key in mat_data.keys()]
if schema_data == None:
return data_list
else:
return load_from_mat_padding(data_list, schema_data)
def test_roundtrip(self):
# create a scale
ds1 = sdf.Dataset('DS1',
comment="dataset 1",
data=np.array([0.1, 0.2, 0.3]),
display_name='Scale 1',
unit='U1',
display_unit='DU1',
is_scale=True)
# create a 1D dataset
ds2 = sdf.Dataset('DS2',
comment="dataset 2",
data=np.array([1, 2, 3]),
display_name='Dataset 2',
relative_quantity=True,
unit='U2',
display_unit='DU2',
scales=[ds1])
# create a group
g = sdf.Group(name='/',
comment="my comment",
attributes={'A1': 'my string'},
datasets=[ds1, ds2])
g2 = sdf.Group(name='G2')
g.groups.append(g2)
# save the group
sdf.save('test.sdf', g)
# load DS2 from the file
ds2r = sdf.load('test.sdf', '/DS2')
# make sure the content is still the same
self.assertDatasetsEqual(ds2, ds2r)
self.assertDatasetsEqual(ds2.scales[0], ds2r.scales[0])
def test_validate_dataset(self):
ds1 = sdf.Dataset('DS1')
ds1.data = 1
errors = sdf._validate_dataset(ds1)
self.assertEqual(['Dataset.data must not be a numpy.ndarray'], errors)
ds1.data = np.array([])
errors = sdf._validate_dataset(ds1)
self.assertEqual(['Dataset.data must not be empty'], errors)
ds1.data = np.array(1).astype(np.float32)
errors = sdf._validate_dataset(ds1)
self.assertEqual(
['Dataset.data.dtype must be one of numpy.float64 or numpy.int32'],
errors)
ds1.data = np.array([1.0, 2.0])
errors = sdf._validate_dataset(ds1)
self.assertEqual(
['The number of scales does not match the number of dimensions'],
errors)
ds2 = sdf.Dataset('DS2', data=np.array([0, 1, 2]), is_scale=True)
ds1.scales = [ds2]
errors = sdf._validate_dataset(ds1)
self.assertEqual([], errors)
ds2.data = np.array([[1, 2], [3, 4]])
errors = sdf._validate_dataset(ds2)
self.assertEqual(['Scales must be one-dimensional'], errors)
ds2.data = np.array([0, 1, 1])
errors = sdf._validate_dataset(ds2)
self.assertEqual(['Scales must be strictly monotonic increasing'],
errors)
# M is number of mass flux divisions
# N is number of quality divisions
# Q is number of pressure divisions
lenG = len(G)
lenx = len(x)
lenP = len(P)
q = np.zeros((lenG, lenx, lenP))
for i in xrange(lenG):
for j in xrange(lenx):
for k in xrange(lenP):
q[i, j, k] = q_raw[i + k * lenG, j]
# Create the datasets:
ds_G = sdf.Dataset('G',
data=G,
unit='kg/(m2.s)',
is_scale=True,
display_name='Mass Flux')
ds_x = sdf.Dataset('x',
data=x,
unit='1',
is_scale=True,
display_name='Quality')
ds_P = sdf.Dataset('P',
data=P,
unit='Pa',
is_scale=True,
display_name='Pressure')
ds_q = sdf.Dataset('q', data=q, unit='W/m2', scales=[ds_G, ds_x, ds_P])
# Create the root group and write the file:
# get the first sheet
sh = book.sheet_by_index(0)
# get the names, quantities and units
n_t = sh.cell_value(0, 1)
u_t = sh.cell_value(1, 1)
n_u = sh.cell_value(0, 2)
u_u = sh.cell_value(1, 2)
# get the data
col_t = sh.col_values(1, 2, sh.nrows)
col_u = sh.col_values(2, 2, sh.nrows)
# create the data arrays
t = array(col_t)
u = array(col_u)
# create the datasets
ds_t = sdf.Dataset(n_t, data=t, unit=u_t, is_scale=True, display_name='Time')
ds_u = sdf.Dataset(n_u, data=u, unit=u_u, scales=[ds_t])
# create the root group
g = sdf.Group('/', comment='Imported from ' + filename, datasets=[ds_t, ds_u])
# change the file extension
outfile = os.path.splitext(filename)[0] + '.sdf'
# write the SDF file
sdf.save(outfile, g)
"""
Create a simple SDF file
"""
import sdf
import numpy as np
# create the data arrays
t = np.linspace(0, 10, 100)
v = np.sin(t)
# create the datasets
ds_t = sdf.Dataset('t', data=t, unit='s', is_scale=True, display_name='Time')
ds_v = sdf.Dataset('v', data=v, unit='V', scales=[ds_t])
# create the root group
g = sdf.Group('/', comment='A sine voltage', datasets=[ds_t, ds_v])
# write the SDF file
sdf.save('sine.sdf', g)
# read the SDF file
ds_v2 = sdf.load('sine.sdf', '/v', unit='V', scale_units=['s'])
ds_t2 = ds_v2.scales[0]
t2 = ds_t2.data
v2 = ds_v2.data
def test_3D_example(self):
RPM2RADS = 2 * math.pi / 60
kfric = 1 # [Ws/rad] angular damping coefficient [0;100]
kfric3 = 1.5e-6 # [Ws3/rad3] angular damping coefficient (3rd order) [0;10-3]
psi = 0.2 # [Vs] flux linkage [0.001;10]
res = 5e-3 # [Ohm] resistance [0;100]
u_ref = 200 # [V] reference DC voltage [0;1000]
k_u = 5 # linear voltage coefficient [-100;100]
tau = np.arange(0, 230 + 10, 10)
w = np.concatenate((np.arange(
0, 500, 100), np.arange(500, 12e3 + 500, 500))) * RPM2RADS
u = np.asarray([200, 300, 400])
# calculate the power losses
TAU, W, U = np.meshgrid(tau, w, u, indexing='ij')
P_loss = kfric * W + kfric3 * W**3 + (
res * (TAU / psi)**2) + k_u * (U - u_ref)
# create the scales
ds_tau = sdf.Dataset('tau',
comment='Torque',
data=tau,
scale_name='Torque',
quantity='Torque',
unit='N.m',
is_scale=True)
ds_w = sdf.Dataset('w',
comment='Speed',
data=w,
scale_name='Speed',
quantity='AngularVelocity',
unit='rad/s',
display_unit='rpm',
is_scale=True)
ds_u = sdf.Dataset('u',
comment='DC voltage',
data=u,
scale_name='DC voltage',
quantity='Voltage',
unit='V',
is_scale=True)
# create the dataset
ds_P_loss = sdf.Dataset('P_loss',
comment='Power losses',
data=P_loss,
quantity='Voltage',
unit='V',
scales=[ds_tau, ds_w, ds_u])
# create a group
g = sdf.Group(
name='/',
comment=
'Example loss characteristics of an e-machine w.r.t. torque, speed and DC voltage',
attributes={'AUTHOR': 'John Doe'},
datasets=[ds_tau, ds_w, ds_u, ds_P_loss])
errors = sdf.validate(g)
self.assertEqual([], errors)
sdf.save('emachine.sdf', g)
def load_from_xlsx(input_filename):
workbook = xlrd.open_workbook(input_filename)
xlsx_data_list = __get_workbook_dict(workbook)
return [sdf.Dataset(**data) for data in xlsx_data_list]
lenx = len(x)
lendT = len(dT)
lenP = len(P)
alpha = np.zeros((lenx, lendT, lenG, lenP))
for i in xrange(lenx):
for j in xrange(lendT):
for k in xrange(lenG):
for r in xrange(lenP):
alpha[i, j, k, r] = alpha_raw[i + k * lenx + r * lenG * lenx,
j]
# Create the datasets:
ds_x = sdf.Dataset('x',
data=x,
unit='1',
is_scale=True,
display_name='Quality')
ds_dT = sdf.Dataset('dT',
data=dT,
unit='K',
is_scale=True,
display_name='deltaT')
ds_G = sdf.Dataset('G',
data=G,
unit='kg/(m2.s)',
is_scale=True,
display_name='Mass Flux')
ds_P = sdf.Dataset('P',
data=P,
unit='Pa',