import h5py
import numpy as np
from h5py import File as FileH5
from ...definitions import PACKAGE_NAME
from ... import __version__
from datetime import datetime
[docs]def save_hdf5(obj, save_path, obj_dict=None):
"""
Save a pyleecan obj in hdf5 format
Parameters
----------
obj: Pyleecan object
object to save
save_path: str
file path
obj_dict : dict
obj.as_dict to save (optionnal to skip call to as_dict)
"""
file5 = None
try:
file5 = FileH5(save_path, "w")
pyleecan_dict_to_hdf5(file5, obj, obj_dict=obj_dict)
file5.close()
except Exception as err:
if file5:
file5.close()
raise (err)
[docs]def pyleecan_dict_to_hdf5(file, obj, obj_dict=None):
"""
Save a dict from a pyleecan object in the hdf5 file
Parameters
----------
file:
hdf5 file
obj: Pyleecan object
object to save
obj_dict : dict
obj.as_dict to save (optionnal to skip call to as_dict)
"""
if obj_dict is not None:
pass
elif isinstance(obj, dict):
obj_dict = dict()
for key, val in obj.items():
obj_dict[key] = val.as_dict(type_handle_ndarray=2)
else:
obj_dict = obj.as_dict(type_handle_ndarray=2)
now = datetime.now()
obj_dict["__save_date__"] = now.strftime("%Y_%m_%d %Hh%Mmin%Ss ")
obj_dict["__version__"] = PACKAGE_NAME + "_" + __version__
for key, val in obj_dict.items():
# Object that need groups
if isinstance(val, dict) or isinstance(val, list):
variable_to_hdf5(file, "", val, key)
# Dataset
elif isinstance(val, np.ndarray):
file[key] = val
elif val == None:
# None is not available in H5 => we use a string
file[key] = np.string_("NoneValue".encode("ISO-8859-2"))
elif isinstance(val, str):
file[key] = np.string_(val.encode("ISO-8859-2"))
else:
file[key] = val
[docs]def list_to_hdf5(file, group_name, name, list_to_save):
"""
Save a list in the hdf5 file : save it as a ndarray if possible
Parameters
----------
file: HDF5 file
file to save the data
group_name: str
name of the group
name: str
name to extend the group or to contain the dataset
list_to_save: list
list to save
"""
np.warnings.filterwarnings("ignore", category=np.VisibleDeprecationWarning)
# Convert into array
array_list = np.array(list_to_save)
# Check the type to split or save as an array
if array_list.dtype.kind in ["O", "U"]:
# Split the list into several datasets
# Create a new group to contain every dataset
group_name = group_name + "/" + name
grp = file.create_group(group_name)
# Add an attribute to load correctly
grp.attrs["length_list"] = len(list_to_save)
# Save every element of the list
for i, element in enumerate(list_to_save):
variable_to_hdf5(file, group_name, element, "list_{}".format(i))
else: # Save as an array
if group_name == "":
file[name] = array_list
file[name].attrs["array_list"] = True
else:
grp = file[group_name]
grp[name] = array_list
# Add an attribute to load correctly
grp[name].attrs["array_list"] = True
[docs]def dict_to_hdf5(file, prefix, dict_to_save):
"""
Save a list in the hdf5 file
"""
for key, value in dict_to_save.items():
if isinstance(key, int):
key = str(key)
variable_to_hdf5(file, prefix, value, key)
[docs]def variable_to_hdf5(file, prefix, variable, name):
# Unable to save matrix of string (U=unicode) => Convert to list
if isinstance(variable, np.ndarray) and "U" in str(variable.dtype):
variable = variable.tolist()
# Pyleecan object dict
if isinstance(variable, dict):
# Create group (Convert key in case of "/")
group_name = prefix + "/" + name.replace("/", "\\x2F")
file.create_group(group_name)
# Call function to create groups and datasets recursively
dict_to_hdf5(file, group_name, variable)
# List
elif isinstance(variable, list):
# Call function to create groups and datasets recursively
list_to_hdf5(file, prefix, name, variable)
# Str
elif isinstance(variable, str):
if len(variable) == 0:
grp = file[prefix]
grp[name] = variable
else:
grp = file[prefix]
# Create a fixed-width ASCII string according
# to http://docs.h5py.org/en/stable/strings.html#exceptions-for-python-3
try:
grp[name] = np.string_(variable.encode("ISO-8859-2"))
except Exception as e:
raise Exception(
"Error while h5 saving variable " + name + ":\n" + str(e)
)
# None
elif variable is None:
# Create dataset
grp = file[prefix]
grp[name] = "NoneValue"
else:
# Create dataset
grp = file[prefix]
grp[name] = variable
