Bug: APIGatewayProxyEventV2Model not deserializing str to custom model

[nleipi]

Expected Behaviour

It should be possible to use own model classes for body in APIGatewayProxyEventV2Model

Current Behaviour

When I extend APIGatewayProxyEventV2Model model to include my own model for the body following exception is raised 1 validation error for Event\nbody\n value is not a valid dict (type=type_error.dict). I suppose the reason for that is that body of apigw v2 event is a string

Code snippet

from aws_lambda_powertools.utilities.parser import BaseModel
from aws_lambda_powertools.utilities.parser.models import (
    APIGatewayProxyEventV2Model
)


class OrderCancelationRequest(BaseModel):
    reason: str
    orderNo: str


class MyEventModel(APIGatewayProxyEventV2Model):
    body: OrderCancelationRequest

Possible Solution

No response

Steps to Reproduce

parse(model=MyEventModel, event=any valid apigw event)

AWS Lambda Powertools for Python version

latest

AWS Lambda function runtime

3.9

Packaging format used

PyPi

Debugging logs

No response

[boring-cyborg]

Thanks for opening your first issue here! We'll come back to you as soon as we can.
In the meantime, check out the #python channel on our AWS Lambda Powertools Discord: Invite link

[ran-isenberg]

hey @nleipi ,
There's no bug, i got it working. There's a slight gotcha here.
The body is defined as a string, and since you extend the model to be a dataclass and dont use the envelope, you need to transform the string into a dict yourself. That's what the envelope does for you.

Here's a sample code that will get it working for you:

from aws_lambda_powertools.utilities.parser import BaseModel
from aws_lambda_powertools.utilities.parser.models import (
    APIGatewayProxyEventV2Model
)
import json
from pydantic import validator

class OrderCancelationRequest(BaseModel):
    reason: str
    orderNo: str


class MyEventModel(APIGatewayProxyEventV2Model):
    body: OrderCancelationRequest

    @validator("body", pre=True)
    def transform_body_to_dict(cls, value: str):
        return json.loads(value)

@leandrodamascena maybe we should include this example in the docs for extending the models that have a non dict base type like this one.

[nleipi]

I think the correct way to handle this is to use pydantics own Json type as a wrapper for the model.

class MyEventModel(APIGatewayProxyEventV2Model):
    body: Json[OrderCancelationRequest]

That should be definitely highlighted in the documentation.

[ran-isenberg]

I think the correct way to handle this is to use pydantics own Json type as a wrapper for the model.

class MyEventModel(APIGatewayProxyEventV2Model):
    body: Json[OrderCancelationRequest]

That should be definitely highlighted in the documentation.

Nice! even prettier

[rubenfonseca]

Hi @nleipi, thanks for opening this! As @ran-isenberg mentioned, parser envelopes were created exactly with this use case in mind.

In your case, the code would look something like this:

from aws_lambda_powertools.utilities.parser import event_parser, BaseModel, envelopes
from aws_lambda_powertools.utilities.typing import LambdaContext

class OrderCancelationRequest(BaseModel):
    reason: str
    orderNo: str

@event_parser(model=OrderCancelationRequest, envelope=envelopes.ApiGatewayV2Envelope)
def handler(event: OrderCancelationRequest, context: LambdaContext):
    assert event.reason != ""

What do you think of giving this a try?

[nleipi]

I would have used envelope approach, but I need access to pathParamters. Can I access them in other way than through event model?

[ran-isenberg]

In order to get the path params, you must extend the model like you did and not use the envelope.
The APIGatewayProxyEventV2Model defines path params too. You can even change them from dict to a custom class and get even stronger validation.

class MyEventModel(APIGatewayProxyEventV2Model):
    body: OrderCancelationRequest
    pathParameters: MyPathParams

    @validator("body", pre=True)
    def transform_body_to_dict(cls, value: str):
        return json.loads(value)

[nleipi]

Stronger validation of pathParameters is a really good point, thanks!

[rubenfonseca]

Thanks! We already have some documentation about extending built-in modules. Do you think people could benefit from this example too?

[ran-isenberg]

The example describes a simple use case where the business payload param is a dict.
Here, it's a string and requires an extra trick.
It makes to sense to add it.