Skip optional fields

Feature Request

We often drop pydantic objects into a NoSQL database where the individual object (document) structure is fairly sparse with quite a few defaults. To save space we often prune out the defaults as they can be regenerated on the fly fairly quickly. To this end, it would be useful to be able to drop results that have been defaulted when serializing automatically.

from pydantic import BaseModel
from typing import Optional

class User(BaseModel):
    id: int
    value: Optional[int]= None # Or some other decorator like Skip[int]?

u = User(id=5)
>>> print(u)
User id=5 value=None
>>> print(u.dict())
{'id': 5, 'value': None}

# Desired feature
>>> print(u.dict(skip_optional=True))
{'id': 5}

>>> u2 = User(id=5, value=5)
{'id': 5, 'value': 5}

Again, happy to add some code here if this is a feature you would consider.

[samuelcolvin]

Yes

On Fri, 8 Feb 2019, 19:59 Daniel Smith <[e-mail] wrote:

Ok, so we should add another slot for fields_set?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/samuelcolvin/pydantic/issues/378#issuecomment-461928019, or mute the thread https://github.com/notifications/unsubscribe-auth/AD2jGW3jfSPDGEZXlHdvDB943Zjt06AEks5vLdcQgaJpZM4aYr1_ .

[PaulRudin]

Incidentally, you can achieve something similar by using a root validator with pre=True to add data for a derived field. So you don’t have a property, but you have the computed value created when the model is created. This works fine for immutable models (which is my use case).