micromodel 0.0.0

todo

Micromodel

Static and runtime dictionary validation.

Install

$ pip install micromodel

Why

We had a HUGE Python code base which was using pydantic to provide a validation layer for MongoDB operations. The code was all messy but it worked fine, until we decided to switch the type checker config from "basic" to "strict", then over a thousand of dictionary-related errors popped up, not to mention the annoying conversions from dict to classes that were going on on every validation and that should be checked everytime.

We then decided to make this validation in-loco using a more vanilla approach with only TypedDicts. Now our dictionaries containing MongoDB documents are consistently dicts that match with the static typing.

Usage

import typing
from micromodel import model

Animal = typing.TypedDict('Animal', {
    'name': str,
    'specie': list[typing.Literal[
        'dog',
        'cat',
        'bird'
    ]]
})

# even another TypedDicts can be used!
Person = typing.TypedDict('Person', {
    'name': str,
    'age': int,
    'animal': Animal
})

m = model(Person, {
    'Animal': Animal
})

old_validate = m.validate
def new_validate(target: Person):
    new = target.copy()
    new['name'] = new['name'].capitalize()
    return validate(Person, typing.cast(typing.Any, new))

# hooks can be implemented using monkeypatching
m.validate = new_validate

result = m.validate({
    'name': 'joao',
    'animal': {
        'name': 'thor',
        'specie': [
            'dog',
            # 'turtle' (this would produce both static and runtime errors)
        ]
    }
})

"""
{
  "name": "Joao",
  "animal": {
    "name": "thor",
    "specie": [
      "dog"
    ]
  }
}
"""
print(result)

License

This library is MIT licensed.