Saving and loading (JSON)#

The objects MatLamina, Lamina, Laminate, LaminationParameters and GradABD are saved to and loaded from JSON by the functions of composites.json_io, all available from composites:

`save_json()`, `load_json()`	file, given by its name or as a file object
`to_json()`, `from_json()`	JSON string
`to_dict()`, `from_dict()`	dictionary of JSON types, to embed in another JSON document

from composites import laminated_plate, save_json, load_json

lam = laminated_plate([0, 45, -45, 90], plyt=0.125e-3,
                      laminaprop=(142e9, 8.7e9, 0.28, 5.1e9, 5.1e9, 3.2e9))
save_json(lam, 'laminate.json')
lam2 = load_json('laminate.json')

All the public attributes are stored, including the derived ones, such as the stiffnesses, the mass integrals and the ply stiffnesses, such that the loaded object is identical to the saved one without any recomputation. JSON files are plain text, independent of the Python version and safe to load from untrusted sources, unlike pickle, which remains supported for copy.deepcopy and multiprocessing but is not recommended for files. The output is strict JSON, also readable with JSON.parse in JavaScript, e.g. in the browser with Pyodide, see Installing composites.

Embedding in other documents#

to_dict() returns a dictionary containing only dict, list, str, int, float and None, such that a laminate can be one entry of the input file of a model, read back with from_dict(), see Saving and loading in JSON.

Format version#

Each saved object carries 'type', 'format_version' and 'composites_version', next to the attributes in 'data'. The current format_version is 2, which stores the mass integrals intrhoz3, intrhoz4 and intrhoz6 of a Laminate. from_dict() reads only the format_version of the installed composites and raises a ValueError for an older or a newer one, e.g. for the files of format 1 saved by version 0.9.21, which must be created again. Every key of the format is required: a missing key, at any level, raises a ValueError naming the missing keys and the class, as does an unknown key, such that no attribute is ever given a default value silently.