import copy
import json
import pickle
from hashlib import md5
from pathlib import Path
import numpy as np
from PIL import Image
from scipy.io import savemat
__all__ = [
"array_to_checksum",
"array_to_image",
"array_to_npy",
"array_to_mat",
"array_to_pickle",
"arrays_to_checksum",
"to_json",
"to_mat",
"to_pickle",
]
[docs]
def array_to_checksum(arr):
"""Hash (md5) array values, and return hex-checksum
Parameters
----------
arr : numpy.ndarray
array to be hashed.
Returns
----------
str
hex-string representation of hash (MD5) of given array
"""
return md5(np.ascontiguousarray(arr.round(8))).hexdigest()
[docs]
def array_to_image(arr, filename, format=None, norm=True):
"""Save a 2D numpy array as a grayscale image file.
Parameters
----------
arr : numpy.ndarray
array to be exported. Values will be cropped to [0,255].
filename : Path or str
(full) path to the file to be created.
norm : bool
multiply array by 255, by default True
"""
try:
arr_to_write = np.ascontiguousarray(arr)
except TypeError as e:
raise e from ValueError("arr should be a numpy.ndarray(-like)")
if norm:
arr_to_write = arr_to_write * 255
im = Image.fromarray(arr_to_write.astype("uint8"), mode="L")
im.save(filename)
[docs]
def array_to_npy(arr, filename):
"""Save a numpy array to npy-file.
Parameters
----------
arr : numpy.ndarray
array to be exported.
filename : Path or str
(full) path to the file to be created.
"""
try:
arr_to_write = np.ascontiguousarray(arr)
except TypeError as e:
raise e from ValueError("arr should be a numpy.ndarray(-like)")
filepath = Path(filename).resolve().with_suffix(".npy")
np.save(filepath, arr_to_write)
[docs]
def array_to_mat(arr, filename):
"""Save a numpy array to a mat-file.
Parameters
----------
arr : numpy.ndarray
array to be exported.
filename : Path or str
(full) path to the file to be created.
"""
try:
arr_to_write = np.ascontiguousarray(arr)
except TypeError as e:
raise e from ValueError("arr should be a numpy.ndarray(-like)")
filepath = Path(filename).resolve().with_suffix(".mat")
savemat(filepath, {"arr": arr_to_write})
[docs]
def array_to_pickle(arr, filename):
"""Save a numpy array to a pickle-file.
Parameters
----------
arr : numpy.ndarray
array to be exported.
filename : Path or str
(full) path to the file to be created.
"""
try:
arr_to_write = np.ascontiguousarray(arr)
except TypeError as e:
raise e from ValueError("arr should be a numpy.ndarray(-like)")
filepath = Path(filename).resolve().with_suffix(".pickle")
with filepath.open("wb") as file:
pickle.dump({"arr": arr_to_write}, file, protocol=pickle.HIGHEST_PROTOCOL)
def array_to_json(arr, filename):
"""Save a numpy array to a (pretty) JSON.
Parameters
----------
arr : numpy.ndarray
array to be exported.
filename : Path or str
(full) path to the file to be created.
"""
try:
arr_to_write = np.ascontiguousarray(arr)
except TypeError as e:
raise e from ValueError("arr should be a numpy.ndarray(-like)")
filepath = Path(filename).resolve().with_suffix(".json")
with filepath.open("w", encoding="utf-8") as file:
json.dump(arr_to_write.tolist(), file, ensure_ascii=False, indent=4)
[docs]
def arrays_to_checksum(stim, keys=["img", "mask"]):
"""Hash (md5) values of arrays specified in keys, and save only the hex
Parameters
----------
stim : dict
stimulus dictionary to export.
keys : str of list of str
keys of dict for which the hashing should be performed
Returns
----------
dict[str, Any]
same as input dict but keys now only contain the hex
"""
if isinstance(keys, str):
keys = [
keys,
]
for key in keys:
stim[key] = array_to_checksum(stim[key])
return stim
[docs]
def to_json(stim, filename):
"""Save stimulus-dict(s) as (pretty) JSON
Parameters
----------
stim : dict
stimulus dictionary to export.
filename : Path or str
(full) path to the file to be created.
"""
if not isinstance(stim, dict):
raise ValueError("stim should be a dict")
filepath = Path(filename).resolve().with_suffix(".json")
stim2 = copy.deepcopy(stim)
for key in stim2.keys():
# np.ndarrays are not serializable; change to list
if isinstance(stim2[key], np.ndarray):
stim2[key] = stim2[key].tolist()
with filepath.open("w", encoding="utf-8") as file:
json.dump(stim2, file, ensure_ascii=False, indent=4)
[docs]
def to_mat(stim, filename):
"""Save stimulus-dict(s) as mat-file
Parameters
----------
stim : dict
stimulus dictionary to export.
filename : Path or str
(full) path to the file to be created.
"""
if not isinstance(stim, dict):
raise ValueError("stim should be a dict")
filepath = Path(filename).resolve().with_suffix(".mat")
savemat(filepath, stim)
[docs]
def to_pickle(stim, filename):
"""Save stimulus-dict(s) as pickle-file
Parameters
----------
stim : dict
stimulus dictionary to export.
filename : Path or str
(full) path to the file to be created.
"""
if not isinstance(stim, dict):
raise ValueError("stim should be a dict")
filepath = Path(filename).resolve().with_suffix(".pickle")
stim2 = copy.deepcopy(stim)
for key in stim2.keys():
# certain classes can cause problems for pickles; change to list
if key in ["visual_size", "ppd", "shape"]:
stim2[key] = list(stim2[key])
with filepath.open("wb") as file:
pickle.dump(stim2, file, protocol=pickle.HIGHEST_PROTOCOL)
