post_dump is passing a list of objects as original object

Hi,

I think post_dump with pass_original=True should pass the original object related to the data serialized and not a list of objects which this object belongs to.

from marshmallow import fields, post_dump, Schema

class DeviceSchema(Schema):
    id = fields.String()

    @post_dump(pass_original=True)
    def __post_dump(self, data, obj):
        print(obj)  # <-- this is a list

devices = [dict(id=1), dict(id=2)]
DeviceSchema().dump(devices, many=True)

In the above example, the parameter obj is a list of devices rather than the device object itself.

What do you think?

[deckar01]

Since schema validators are designed to validate a single object instance unless pass_many is explicitly true, I think the intuitive behavior is for pass_original to return the corresponding original object instance unless pass_many is explicitly true.

• Picking the original preprocessed object out of a list is inconvenient at best and impossible at worst.
• Validation of a nested list is more appropriate on a parent schema instead of an overload of the child schema validator.

Edit: false -> true

[frol]

+1 for @viniciuschiele’s points. It is unexpected to receive a list instead of an object.