schema

schema copied to clipboard

There's no need to automatically type check ITERABLE flavours when this can be done by the user. The reason this is needed is because different iterables can often be treated the same way. For example, the default JSON encoder accepts both tuples and lists as array types. Creating a schema with either of them will result in the validation failing if the other is passed instead.

Hmm, I'm not sure what you mean. Doesn't iterable accept both lists and tuples? I'm confused as to the reasoning behind the removal.

Consider the following data:

data = tuple(range(1))

And the following schema:

schema = Schema([int])

Validating raises the following error:

schema.SchemaUnexpectedTypeError: (0,) should be instance of 'list'

As .validate automatically validates the iterable's type. Removing that line of code passes the decision of type validation to the user. By adding a bit of extra logic in script code, the same results can be achieved:

schema = Schema(And(list, Schema([int])))

All while allowing to further abstractify or even bypass such validation:

import collections.abc
schema = Schema(And(collections.abc.Iterable, Schema([int]))) # abstract
schema = Schema([int]) # bypass

Using list to create the data and (int,) in schema would have the the same error failing for tuple.

This request arose when I tried to validate either tuple or list (without knowing what to expect) against an ITERABLE flavoured schema while working with JSON since the default encoder accepts both.