[Feature Request] Support Checker([LIST_OF], LIST_OF=Optional[...])

[mattwang44]

Here is a minimum-reproducible code:

class _Spec:
    data = Checker([LIST_OF], LIST_OF=Optional[str])

validate_data_spec({'data': ['str', None]}, _Spec)
# TypeError: field: _Spec.data, reason: ['str', None] is not a spec of <class '__main__._Spec'>, detail: AttributeError('find')

tried with str | None

class _Spec:
    data = Checker([LIST_OF], LIST_OF=None | str)

validate_data_spec({'data': ['str', None]}, _Spec)
# TypeError: field: _Spec.data, reason: ['str', None] is not a spec of <class '__main__._Spec'>, detail: AttributeError("'types.UnionType' object has no attribute 'find'")

[kilikkuo]

I'm not a fan of returning a list of objects which contains different type of data.
But that is not the scope that DSV should consider about.

IMO, I prefer not mixed-using both DSV defined Checks (i.e. STR, NONE, SPEC,...) and Python typing object (i.e. Optional, Dic, Tuple, Any,...). As it could become a bottomless pit in terms of feature supporting,

For example,
Please support Optional against Checker, make

class _Spec:
    field = Checker([int], optional=True)

equals to

class _Spec:
    field = Optional[Checker([int])]

Besides, the original idea of DSV is to support from python 2.7, it could be a complete different design when Python typing comes into play.

I guess we can start supporting something like

class _Spec:
    data = Checker([LIST_OF], LIST_OF=[None, str, SPEC_A, SPEC_B])

idea ?