def testEncodeScalar(self):
array_str = array_to_source.array_to_source('array', 4)
expected_str = """array = np.array([
4,
]).reshape(())
"""
self.assertEqual(expected_str, array_str)
def testEncodeArray(self):
array_str = array_to_source.array_to_source('array', [1, 2, 3])
expected_str = """array = np.array([
1,
2,
3,
]).reshape((3,))
"""
self.assertEqual(expected_str, array_str)
def save_ground_truth_part(name, tuple_path, mean, sem, std, sestd):
"""Saves a ground truth part to strings.
This is meant to be called with outputs of
`nest.flatten_with_tuple_paths(ground_truth_mean)`.
Args:
name: Python `str`. Name of the sample transformation.
tuple_path: Tuple path of the part of the ground truth we're saving. See
`nest.flatten_with_tuple_paths`.
mean: Ground truth mean, or `None` if it is absent.
sem: Ground truth stadard error of the mean, or `None` if it is absent.
std: Ground truth standard deviation, or `None` if it is absent.
sestd: Ground truth mean, or `None` if it is absent.
Returns:
array_strs: Python list of strings, representing the encoded arrays (that
were present). Typically these would be joined with a newline and written
out to a module, which can then be passed to `load_ground_truth_part`.
"""
array_strs = []
mean_name, sem_name, std_name, sestd_name = _get_global_variable_names(
name, tuple_path)
if mean is not None:
array_strs.append(array_to_source.array_to_source(mean_name, mean))
if sem is not None:
array_strs.append(array_to_source.array_to_source(sem_name, sem))
if std is not None:
array_strs.append(array_to_source.array_to_source(std_name, std))
if sestd is not None:
array_strs.append(array_to_source.array_to_source(sestd_name, sestd))
return array_strs