Python [Pydantic] - default values based other object in the class

Question

I would like to set default email in case email not provided, i.e:

name = a 
last_name = b
email = None

email will become "a_b@email.com"

I tried something like that but obviously didn't work as the name,last_name not define in function.

class User(BaseModel):
     name: Optional[str] = ''
     last_name: Optional[str] = ''
     email: EmailStr
    

    @validator('email')
    def set_email(cls, email):
        if not email:
           return name + last_name + '@email.com' 
        else:
           return email

Update- still not working, i tried:

    @root_validator(pre=True)
    def email_set_config(cls, values):
        email, name ,last_name = values.get('email'), values.get('name') , values.get('last_name')
        if email is None :
            email= name + '_' + name + '@' + last_name 
        return values

Solution:

added to Class :

--Update

Actually it is working without define class Config:

    class Config:
        validate_assignment = True

    @validator('email')
    def set_email(cls, v, values, **kwargs):
        return email or values["name"] + '_' + values["name"] + '@' + ["last_name"]

@NobbyNobbs - thanks for suggestion , not working , post updated . — AviC, Mar 04 '21 at 08:36
actually it works, you don't update values dictionary in your implementation, but set local variable instead. Just fix it :) — NobbyNobbs, Mar 04 '21 at 10:09
@NobbyNobbs - figured out , i update the solution in post , can u explain when root_validator is require? — AviC, Mar 04 '21 at 10:32

obendidi · Answer 1 · 2022-05-12T14:39:19.520

Not enough reputation to comment, so I'll just expand on @NobbyNobbs answer here^^

For your last example using pydantic.validator, you can use the always kwargs to always run a validator even if the value is not provided.

Link in the doc: https://pydantic-docs.helpmanual.io/usage/validators/#validate-always

So your example would be:

from typing import Optional

import pydantic


class User(pydantic.BaseModel):
    first_name: str
    last_name: str
    email: Optional[pydantic.EmailStr]

    @pydantic.validator('email', always=True)
    def set_email(cls, v, values, **kwargs):    
        return v or values["first_name"] + '_' + values["last_name"] + '@email.com'


if __name__ == "__main__":
    print(User(first_name="jon", last_name="doe"))  # first_name='jon' last_name='doe' email='jon_doe@email.com'
    print(User(first_name="jon", last_name="doe", email=None))  # first_name='jon' last_name='doe' email='jon_doe@email.com'

score 3 · Answer 2 · edited Mar 04 '21 at 11:28

I changed your model a bit, now it has two required fields and one optional, which calculated from the others two.

from typing import Optional

import pydantic


class User(pydantic.BaseModel):
    first_name: str
    last_name: str
    email: Optional[pydantic.EmailStr]

First approach to validate your data during instance creation, and have full model context at the same time, is using the @pydantic.root_validator:

    @pydantic.root_validator
    def email_set_config(cls, values):
        email, name, last_name = values.get('email'), values.get('first_name'), values.get('last_name')
        if email is None:
            values["email"] = name + '_' + last_name + '@email.com'
        return values


if __name__ == "__main__":
    u = User(first_name="jon", last_name="doe")
    print(u)  # first_name='jon' last_name='doe' email='jon_doe@email.com'

But if you want to control instantiation process only, I suggest you just override __init__ dunder in your model like this

    def __init__(__pydantic_self__, **data: Any) -> None:
        email, name, last_name = data.get('email'), data.get('first_name'), data.get('last_name')
        if email is None:
            data["email"] = name + '_' + last_name + '@email.com'
        super().__init__(**data)

It's more simple and intuitive approach comparing to validator on my sight, and I would prefer it.

The possible solution with non-root validator, you mention in question, has tricky and a bit unexpected behavior for me.

It works like expected if you pass named email argument to constructor, but doesn't if you not.

    @pydantic.validator('email')
    def set_email(cls, v, values, **kwargs):
        return v or values["first_name"] + '_' + values["last_name"] + '@email.com'


if __name__ == "__main__":
    print(User(first_name="jon", last_name="doe"))  # first_name='jon' last_name='doe' email=None
    print(User(first_name="jon", last_name="doe", email=None))  # first_name='jon' last_name='doe' email='jon_doe@email.com'

NobbyNobbs, thanks much for your great ,detailed explanation! learnt something useful:) — AviC, Mar 04 '21 at 16:51

score 0 · Answer 3 · edited Apr 26 '22 at 22:01

0

In my case the solution was:

@pydantic.root_validator(pre=True)
def method_before_validation(cls, data):

I used It in the root class of a complex object. I have smth like object with orders and separate object with client and I needed to put client's id to each order's object. So we can make data modification with data structure, types e.t.c on fly before the actual validation.

edited Apr 26 '22 at 22:01

Yunnosch

26,130
9
42
54

answered Feb 14 '22 at 10:59

Wotori Movako

147
1
7

Grimmer Kang · Answer 4 · 2021-08-05T07:24:21.920

-4

Pydantic has added default_factory since v1.5, you can use the below code to have the default EmailStr value, as well as the other submodel class object

class User(BaseModel):
     name: Optional[str] = ''
     last_name: Optional[str] = ''
     email: EmailStr = Field(
         default_factory=lambda: EmailStr('foo@example.org'))

edited Aug 05 '21 at 07:24

answered Aug 05 '21 at 06:07

Grimmer Kang

235
3
10

Python [Pydantic] - default values based other object in the class

4 Answers4