Broken decimal parsing when using BeforeValidator about pydantic HOT 7 OPEN

Yes my bad it's only with a plain validator that nothing gets called after the plain validator!

from pydantic.

Oke perfect, thanks :)

from pydantic.

This is the expected behavior of the before (model) validator, the input is untouched. You probably want to use an after model validator: in this case, value["a"] will be a Decimal instance. If you need more flexibility, you can also use a wrap validator

from pydantic.

Please note, that I am not doing any validation. I am using the BeforeValidator as simple hook, to do something else before the validation starts.
From my POV a noop BeforeValidator should not break the validation.
From the docs i would expect that a, b and c are strings at the before validation stage.
Also decimal seems to have another problem with lossy data, which may or may not be related to the data being passed as float instead of a string.

from pydantic.

After thinking some more about it, i guess that pydantic is switching from json validation to python validation, when a BeforeValidator exists. This would be quite strange behaviour.

On the data type in the BeforeValidator side note: probably everything should be a string at this stage. For ints it is no problem, but floats have already lost precision at this point (I can not get it back at this stage), if i want to do my own validation.

from pydantic.

Hmm, I would still expect this to work with a before validator. I'll chat with the team first thing tomorrow and solidify whether or not this is expected behavior.

Thanks for reporting this issue!

from pydantic.

Indeed, this is a bug!

from pydantic.