"array" type should be (list, tuple)

[igrishaev]

Currently, it's impossible to use tuples as arrays, only lists.
Array type should be a tuple like (list, tuple).

default_types = {
            "array" : (list, tuple), "boolean" : bool, "integer" : int_types, ...

or

default_types = {
            "array" : array_types, "boolean" : bool, "integer" : int_types, ...

, where

array_types = (list, tuple)

[Julian]

Hey!

simplejson will never give you a tuple when loading a JSON array.

That said, if you want to actually allow tuples, you can provide a mapping of types that you want to consider to be arrays by providing a types dictionary to a validator's __init__.

[igrishaev]

Julian, thank you for you advice, I've figured out with types={'array': (list, tuple)} parameter in .validate()

I needed to validate some python dict with nested tuples.

[eirnym]

JSON parsers are not only source for your library. For example I check all requests and all responses.

[Julian]

Yup -- I'm glad, I think the current APIs should be usable for whatever use case you have in mind, please feel free to reach out if you have any issues.

[eirnym]

Could you please extend array type if you think so?

[Julian]

What is it you mean by extend the array type? As the post two comments above you mentioned, Draft4Validatot(schema, types=dict(array=(list, tuple))) is how you say that you want JSON arrays to allow tuples

Once again be concrete on where you are having issues.

[CMCDragonkai]

Is there a way to do Draft4Validator(...) without specifying the draft version? Just Validator(..., types=...)? In my case the version is specified within the schema. Or is that possible by just using jsonschema.validate(instance=..., schema=..., types=dict(array=(list, tuple)))?

[pekkaklarck]

Something has apparently changed and using Draft4Validator(schema, types=dict(array=(list, tuple))) proposed by @Julian above doesn't work anymore. Testing with v4.17.3:

>>> Draft4Validator(schema, types=dict(array=(list, tuple)))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: create.<locals>.Validator.__init__() got an unexpected keyword argument 'types'

I honestly don't see why tuples couldn't be accepted by default. It's true that json.load won't return tuples, but then again json.dump accepts them just fine. I have data that uses tuples that I'd like to validate before converting it to JSON and this limitation is pretty annoying.

[stojan-jovic]

Previous method is deprecated from the jsonschema version 4.x (or something like that), but now the same can be achieved as follows:

from jsonschema import validators, Draft201909Validator  # Provide draft version as you need...


def list_and_tuple(checker, instance):
    return isinstance(instance, (list, tuple))


type_checker = Draft201909Validator.TYPE_CHECKER.redefine(
    'array', list_and_tuple,
)

CustomArrayValidator = validators.extend(
    Draft201909Validator,
    type_checker=type_checker,
)

# A sample schema, like what we'd get from json.load()
schema = {
    'type': 'object',
    'properties': {
        'price': {'type': 'number'},
        'name': {'type': 'string'},
        'features': {'type': 'array'}
    }
}
CustomArrayValidator.check_schema(schema)

# If no exception is raised by validate(), the instance is valid.
validator = CustomArrayValidator(schema=schema)
validator.validate(
    instance={
        'name': 'Eggs',
        'price': 34.99,
        # 'features': [1, 2, 3],  # Valid!
        'features': (1, 2, 3),  # Valid!
        # 'features': {1, 2, 3}  # Not valid!
    }
)

It requires much more code than earlier, but it's possible to make tuples valid.

Btw, I agree that it should be supported by default.

[mmulich]

I'm in agreement with @pekkaklarck, @stojan-jovic and others, tuple should be a valid default.

Building off @stojan-jovic's workaround, I've updated a portion of their code to grab the latest validator.

import jsonschema

LatestValidator = jsonschema.validators._LATEST_VERSION

def list_and_tuple(checker, instance):
    return isinstance(instance, (list, tuple))

type_checker = LatestValidator.TYPE_CHECKER.redefine(
    'array', list_and_tuple,
)

CustomValidator = jsonschema.validators.extend(
    LatestValidator,
    type_checker=type_checker,
)

Thanks @stojan-jovic!