Bug: Idempotency key in idempotent_function fails to serialize dates/UUIDs

[jontmy]

Expected Behaviour

The @idempotent_function decorator should be able to serialize any idempotency keys which are Pydantic models containing dates or UUIDs (or more precisely, any type that can be serialized with model_dump(mode="json")).

Current Behaviour

aws_lambda_powertools.utilities.idempotency.exceptions.IdempotencyPersistenceLayerError: Failed to save in progress record to idempotency store - (Object of type UUID is not JSON serializable)

Code snippet

# src/idempotency_serialization/main.py

from uuid import UUID

from aws_lambda_powertools.utilities import idempotency
from pydantic import BaseModel

idempotency_config = idempotency.IdempotencyConfig()
idempotency_persistence_layer = idempotency.DynamoDBPersistenceLayer(
    table_name="idempotency"
)


class IdempotencyKey(BaseModel):
    req_id: UUID


def handle_request(req: IdempotencyKey) -> None:
    return _impl(req=req)


@idempotency.idempotent_function(
    data_keyword_argument="req",
    config=idempotency_config,
    persistence_store=idempotency_persistence_layer,
)
def _impl(req: IdempotencyKey) -> None:
    pass

# tests/test_idempotency.py

from uuid import uuid4

import boto3
from moto import mock_aws


@mock_aws
def test_idempotency() -> None:
    client = boto3.client("dynamodb")
    client.create_table(
        TableName="idempotency",
        KeySchema=[{"AttributeName": "id", "KeyType": "HASH"}],
        AttributeDefinitions=[{"AttributeName": "id", "AttributeType": "S"}],
        BillingMode="PAY_PER_REQUEST",
    )

    from idempotency_serialization.main import IdempotencyKey, handle_request

    handle_request(IdempotencyKey(req_id=uuid4()))

Possible Solution

The _prepare_data function in base.py should probably use .model_dump(mode="json") instead of leaving mode unset, which defaults to "python" and does not serialize UUIDs to strings.

def _prepare_data(data: Any) -> Any:
    """Prepare data for json serialization.

    We will convert Python dataclasses, pydantic models or event source data classes to a dict,
    otherwise return data as-is.
    """

    # Convert from dataclasses
    if hasattr(data, "__dataclass_fields__"):
        import dataclasses

        return dataclasses.asdict(data)

    # Convert from Pydantic model
    if callable(getattr(data, "model_dump", None)):
        return data.model_dump()

    # Convert from event source data class
    if callable(getattr(data, "dict", None)):
        return data.dict()

    # Return raw event
    return getattr(data, "raw_event", data)

Steps to Reproduce

mkdir idempotency-serialization
cd idempotency-serialization
uv init --build-backend uv
uv add aws-lambda-powertools pydantic moto pytest
mkdir -p src/idempotency_serialization
mkdir -p tests
touch src/idempotency_serialization/main.py
touch tests/test_idempotency.py

Then copy the snippets above into src/idempotency_serialization/main.py and tests/test_idempotency.py, and run uv run pytest.

You should get this error:

FAILED tests/test_idempotency.py::test_idempotency - aws_lambda_powertools.utilities.idempotency.exceptions.IdempotencyPersistenceLayerError: Failed to save in progress record to idempotency store - (Object of type UUID is not JSON serializable)

Powertools for AWS Lambda (Python) version

latest

AWS Lambda function runtime

3.14

Packaging format used

PyPi

Debugging logs

[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 Powertools for AWS Lambda Discord: Invite link