Add please connected data support

[DarkwingDuck48]

Feature request

I think it is needed to have related data during schema generation.

Thesis

For example, we have simple schema

schema_definition = lambda: {
    'full_name': field('name', gender=Gender.MALE),
    'surname': field('person.surname', gender=Gender.MALE)
}

>>> {'full_name': 'Irvin Savage', 'surname': 'Delgado'}

Reasoning

It would be better to have related data during schema generation process, when attribure full_name connected to surname and return the same data. In my example with full_name = 'Irvin Savage', surname should return 'Savage', not generated new value.

[lk-geimfari]

I've wanted this feature for a long time, but it's pretty hard to implement for all these languages supported by Mimesis.

[mheguy-flo]

I created a proof of concept for this. In the below example you can get a sense for the API.
The example is a modification of https://mimesis.name/en/master/schema.html#schema-definition where we create the email based on the name.

def custom_email_handler(
    random: Random,
    *,
    domains: list[str],
    _eager_data: dict[str, Any],
    **_kwargs: Any,
) -> str:
    name: str = _eager_data["owner"]["creator"]
    username = name.replace(" ", ".").lower()
    return f"{username}@{random.choice(domains)}"


fieldset = Fieldset()
field = Field()
lazy_field = LazyField()
lazy_field.register_handler("custom_email", custom_email_handler)

schema = EvaluatedSchema(
    {
        "pk": (field, "increment", {}),
        "uid": (field, "uuid", {}),
        "name": (field, "text.word", {}),
        "version": (field, "version", {}),
        "timestamp": (field, "timestamp", {"fmt": TimestampFormat.POSIX}),
        "owner": {
            "email": (lazy_field, "custom_email", {"domains": ["mimesis.name"]}),
            "creator": (field, "full_name", {"gender": Gender.FEMALE}),
        },
        "apiKeys": (fieldset, "token_hex", {"key": lambda s: s[:16], "i": 3}),
    }
)

print(schema.create())

Example output:

{
    "pk": 1,
    "uid": "7c731bdf-656e-4266-aefe-3af7af05dcbf",
    "name": "doc",
    "version": "92.18.33",
    "timestamp": 1726148859,
    "owner": {
        "creator": "Madelaine Price",
        "email": "madelaine.price@mimesis.name"
    },
    "apiKeys": [
        "b404af366115c2fe",
        "d16e1fb9bbd3c354",
        "a63237fed56da108"
    ]
}

@lk-geimfari: If you have an idea of how to implement this with the current Schema class, I would love to hear about it. I couldn't see a way of implementing it with the current API (CallableSchema is essentially a black box).

[lk-geimfari]

I created a proof of concept for this. In the below example you can get a sense for the API.

The example is a modification of https://mimesis.name/en/master/schema.html#schema-definition where we create the email based on the name.

def custom_email_handler(

    random: Random,

    *,

    domains: list[str],

    _eager_data: dict[str, Any],

    **_kwargs: Any,

) -> str:

    name: str = _eager_data["owner"]["creator"]

    username = name.replace(" ", ".").lower()

    return f"{username}@{random.choice(domains)}"





fieldset = Fieldset()

field = Field()

lazy_field = LazyField()

lazy_field.register_handler("custom_email", custom_email_handler)



schema = EvaluatedSchema(

    {

        "pk": (field, "increment", {}),

        "uid": (field, "uuid", {}),

        "name": (field, "text.word", {}),

        "version": (field, "version", {}),

        "timestamp": (field, "timestamp", {"fmt": TimestampFormat.POSIX}),

        "owner": {

            "email": (lazy_field, "custom_email", {"domains": ["mimesis.name"]}),

            "creator": (field, "full_name", {"gender": Gender.FEMALE}),

        },

        "apiKeys": (fieldset, "token_hex", {"key": lambda s: s[:16], "i": 3}),

    }

)



print(schema.create())

Example output:

{

    "pk": 1,

    "uid": "7c731bdf-656e-4266-aefe-3af7af05dcbf",

    "name": "doc",

    "version": "92.18.33",

    "timestamp": 1726148859,

    "owner": {

        "creator": "Madelaine Price",

        "email": "madelaine.price@mimesis.name"

    },

    "apiKeys": [

        "b404af366115c2fe",

        "d16e1fb9bbd3c354",

        "a63237fed56da108"

    ]

}

@lk-geimfari: If you have an idea of how to implement this with the current Schema class, I would love to hear about it. I couldn't see a way of implementing it with the current API (CallableSchema is essentially a black box).

Alas, it's too complicated to implement this feature at this moment.

[mheguy-flo]

I agree it's not viable with the Schema API as-is.
Would you be open to a breaking change of the Schema API and/or a LazySchema class?

The API I demonstrated above isn't theoretical. It's the API from working code. The question is simply if you want it as part of mimesis and if so, how you want to integrate it.

[lk-geimfari]

I'd prefer not to break the existing API, but having a LazyField would be useful.