Factory does not respect randomized collection lengths for optional collection types

[emapple]

Description

When a model has an optional collection attribute, it does not adhere to the randomized collection arguments passed to it, instead using the default length of 1 no matter what.

On a related note -- in the MCVE, removing | None in the type hint will correctly build with a list of random length between the min and max. However, all instances of generated data (via factory.build() or factory.batch()) have the same length. Is there a way to have variable lengths, so each subsequent call has a random, uncorrelated length?

URL to code causing the issue

No response

MCVE

from polyfactory.factories.pydantic_factory import ModelFactory
from pydantic import BaseModel

class Model(BaseModel):
    var: list[str] | None


class Factory(ModelFactory):
    __model__ = Model
    __randomize_collection_length__ = True
    __allow_none_optionals__ = False
    __min_collection_length__ = 2
    __max_collection_length__ = 5

Factory.batch(10)

Output:

# all have length of 1, not respecting constraints
[Model(var=['HbolMJUevblAbkHClEQa']),
 Model(var=['PKriXrefSFPLBYtCRGSE']),
 Model(var=['KclpemfoHstknWHFiLit']),
 Model(var=['ZHCJROlbqnkXTqIuVxnm']),
 Model(var=['DnEUulZAecfGPvdtZBYS']),
 Model(var=['aLJBnHfwxXTnAjVTpHFh']),
 Model(var=['NCOBaqbWTRpdTXezdNNg']),
 Model(var=['yCnTwlCLkqZHwAglrEll']),
 Model(var=['dGlVSdmIlgWDyOPjfwwy']),
 Model(var=['LJZfurSmjxvomWxSCcZR'])]

Steps to reproduce

See MCVE

Screenshots

No response

Logs

No response

Release Version

2.9.0

Platform

• Linux
• Mac
• Windows
• Other (Please specify in the description above)

Note

Check out all issues funded or available for funding here: https://polar.sh/litestar-org

• If you would like to see an issue prioritized, make a pledge towards it!
• We receive the pledge once the issue is completed & verified

[guacs]

Thanks for this :)

This is essentially due to these constraints being encoded into the FieldMeta when they're created for each type. This coupling should be removed, and these length constraints should be handled separately within get_field_value.

[emapple]

Thanks for addressing this!