pydantic BaseModel cannot be pickled w/ cloudpickle >= 1.3

[marco-neumann-by]

Example

import cloudpickle
import pydantic

class Bar(pydantic.BaseModel):
    pass

if __name__ == '__main__':
    cloudpickle.loads(cloudpickle.dumps(Bar()))

Technical Background

The issue is the interaction between:

https://github.com/samuelcolvin/pydantic/blob/de0657e4a5495de9378db9fe15a17334c2b0fae5/pydantic/main.py#L865-L868

and:

cloudpickle/cloudpickle/cloudpickle.py

Line 460 in cdc704d

is_litteral = getattr(obj, '__values__', None) is not None

Refs

This was originally discussed in #393.

[erccarls]

Is there any workaround for this?

[marco-neumann-by]

I should note that this is only an issue on Python < 3.7. So upgrading is one workaround. Or a PR to cloudpickle that makes the linked inspection code more robust.

[LaurentMazare]

I'm actually running into an issue that seems related with Python 3.8.5 (cloudpickle version 1.6.0, pydantic version 1.7.3).
The example given at the top of this issue works fine however, adding a field to the model makes cloudpickle break whereas the standard pickle module works fine.

Below is some code that reproduces the issue.

import cloudpickle
import pydantic
import pickle

class Bar(pydantic.BaseModel):
    a: int

pickle.loads(pickle.dumps(Bar(a=1))) # This works well
cloudpickle.loads(cloudpickle.dumps(Bar(a=1))) # This fails with the error below

PicklingError: Can't pickle <cyfunction int_validator at 0x7ff32fa23ad0>: attribute lookup lambda12 on pydantic.validators failed

[marco-neumann-by]

@LaurentMazare I've tried to reproduce your issue and first couldn't under Python 3.8. After some digging I've figured out that the difference between your and my environment must be the exact pydantic "flavor" (cythonized vs not cythonized). I think what you reported here is a different issue with a different root cause. So I've opened #408 with all the details and welcome you to continue the discussion there.

[jondequinor]

Will this be included in a release? If so, is there an ETA for that release?