Pydantic: How to have a single validator for multiple fields?

[jamesob]

This seems pretty basic so I wouldn't be surprised if I'm missing something obvious, but from a close read of the documentation I can't figure out how to have a validator which acts on multiple fields in a BaseModel, i.e. something like

class Model(BaseModel):
  foo: int
  bar: int

  @model_validator(...)
  def compare_foo_and_bar(cls):
    # This is a nonsense example
    if not (foo - bar):
      raise ValueError(...)

[samuelcolvin]

Away from my computer so can't give a full example, but basically, the validator can take values as an argument.

The docs has an example with passwords.

[jamesob]

Cool, thanks. For posterity here's the example mentioned (I think): https://github.com/jamesob/pydantic/blob/fe72ba13f469377057112d2f301f52cd3b206467/docs/examples/validators_simple.py#L14-L20

This looks like values is the raw data coming in - is there a way to do a post-deserialization validation on the resulting BaseModel instance?

[samuelcolvin]

Not on the resulting Model (unless you did it manually yourself), but I don't think values is "raw data", I think it's already been validated.

[jamesob]

Cool, thanks for the great library.

[samuelcolvin]

BUT that means when foo changes, then the value param is whatever foo is, and whenever bar changes, value becomes whatever bar is

No it doesn't. This is just a validator, it's a function that is called when these values are validated, it doesn't mean foo is set to bar or visa-versa.

not telling you which one

No it doesn't. You can use use the field argument to identify which field is changing.

---

More generally: if you want help you would be well advised to change your tone.

If you are going state negative/critical things, take some time to check that what you're saying is correct. Incorrect, negative assertions are a pretty efficient way of pissing people off.

[anikolaienko]

@samuelcolvin
I don't understand, sorry, so if, for instance, I want to verify if either of fields: foo or boo exists but not both at the same time, how can I do that?

Thanks.

[PrettyWood]

Hi @anikolaienko
You could go with something like this

from typing import Dict, Optional

from pydantic import BaseModel, validator


class Model(BaseModel):
    foo: Optional[str]
    boo: Optional[str]

    # Validate the second field 'boo' to have 'foo' in `values` variable
    # We set `always=True` to run the validator even if 'boo' field is not set
    @validator('boo', always=True)
    def ensure_only_foo_or_boo(cls, boo: Optional[str], values: Dict[str, Optional[str]]):
        foo = values.get('foo')
        if foo is not None and boo is not None:
            raise ValueError('only one of "foo" or "boo" can be set')
        if foo is None and boo is None:
            raise ValueError('one of "foo" or "boo" needs to be set')
        return boo

Hope it helps!

[anikolaienko]

Hi @PrettyWood , I appreciate your help! But this doesn't work for me. I still don't understand what is values because in my case values is empty dictionary when boo is not present.

I figured that I can use root_validator(cls, values): very nice and easy for this purpose.