Fodantic

Pydantic-based HTTP forms.

Pydantic is the most widely used data validation library for Python, but it's hard to use it with regular HTTP forms... until now.

Fodantic allow you to quickly wrap your Pydantic models and use them as forms: with support for multiple values, checkboxes, error handling, and integration with your favorite ORM.

A simple example

from fodantic import formable
from pydantic import BaseModel

@formable
class UserModel(BaseModel):
    name: str
    friends: list[int]
    active: bool = True

# This is just an example. Here you would use the
# request POST data of your web framework instead.
# For example, for Flask: `request_data = request.form`
from multidict import MultiDict
request_data = MultiDict([
  ("name", "John Doe"),
  ("friends", "2"),
  ("friends", "3"),
])

# The magic
form = UserModel.as_form(request_data, object=None)

print(form)
#> UserModel.as_form(name='John Doe', friends=[2, 3], active=False)
print(form.fields["name"].value)
#> John Doe
print(form.fields["name"].error)
#> None
print(form.save())  # Can also update the `object` passed as an argument
#> {'name': 'John Doe', 'friends': [2, 3], 'active': False}

Installation

pip install fodantic

Requirements

Python 3.10+
Pydantic 2.*

Documentation

List fields

Fields defined as of type list, tuple, or a derivated type, will be marked as expecting multiple values. A <select multiple> and a group of checkboxes charing the same name (but different values) are the most common examples of how these fields look on a form. Fodantic will use the getall(*) method on the request data to get a list of all the values under the same name.

(*) Also called getlist in many web frameworks.

Booleans fields

Boolean fields are treated special because of how browsers handle checkboxes:

If not checked: the browser doesn't send the field at all, so the missing field will be interpreted as False.
If checked: It sends the "value" attribute, but this is optional, so it could send an empty string instead. So any value other than None will be interpreted as True.