feat(event_handler): add custom response validation in OpenAPI utility

aws_lambda_powertools/event_handler/api_gateway.py

@@ -19,7 +19,11 @@
 from aws_lambda_powertools.event_handler import content_types
 from aws_lambda_powertools.event_handler.exceptions import NotFoundError, ServiceError
 from aws_lambda_powertools.event_handler.openapi.constants import DEFAULT_API_VERSION, DEFAULT_OPENAPI_VERSION
-from aws_lambda_powertools.event_handler.openapi.exceptions import RequestValidationError, SchemaValidationError
+from aws_lambda_powertools.event_handler.openapi.exceptions import (
+    RequestValidationError,
+    ResponseValidationError,
+    SchemaValidationError,
+)
 from aws_lambda_powertools.event_handler.openapi.types import (
     COMPONENT_REF_PREFIX,
     METHODS_WITH_BODY,
@@ -1496,6 +1500,7 @@ def __init__(
         serializer: Callable[[dict], str] | None = None,
         strip_prefixes: list[str | Pattern] | None = None,
         enable_validation: bool = False,
+        response_validation_error_http_status: HTTPStatus | None = None,
     ):
         """
         Parameters
@@ -1530,6 +1535,7 @@ def __init__(
         self.context: dict = {}  # early init as customers might add context before event resolution
         self.processed_stack_frames = []
         self._response_builder_class = ResponseBuilder[BaseProxyEvent]
+        self._response_validation_error_http_status = response_validation_error_http_status
 
         # Allow for a custom serializer or a concise json serialization
         self._serializer = serializer or partial(json.dumps, separators=(",", ":"), cls=Encoder)
@@ -1539,7 +1545,14 @@ def __init__(
 
             # Note the serializer argument: only use custom serializer if provided by the caller
             # Otherwise, fully rely on the internal Pydantic based mechanism to serialize responses for validation.
-            self.use([OpenAPIValidationMiddleware(validation_serializer=serializer)])
+            self.use(
+                [
+                    OpenAPIValidationMiddleware(
+                        validation_serializer=serializer,
+                        has_response_validation_error=self._response_validation_error_http_status is not None,
+                    ),
+                ],
+            )
 
     def get_openapi_schema(
         self,
@@ -2370,6 +2383,22 @@ def _call_exception_handler(self, exp: Exception, route: Route) -> ResponseBuild
                 route=route,
             )
 
+        # OpenAPIValidationMiddleware will only raise ResponseValidationError when
+        # 'self._response_validation_error_http_status' is not None
+        if isinstance(exp, ResponseValidationError):
+            if self._response_validation_error_http_status is None:
+                raise TypeError
+            errors = [{"loc": e["loc"], "type": e["type"]} for e in exp.errors()]
+            return self._response_builder_class(
+                response=Response(
+                    status_code=self._response_validation_error_http_status,
+                    content_type=content_types.APPLICATION_JSON,
+                    body={"statusCode": self._response_validation_error_http_status, "detail": errors},
+                ),
+                serializer=self._serializer,
+                route=route,
+            )
+
         if isinstance(exp, ServiceError):
             return self._response_builder_class(
                 response=Response(
@@ -2582,6 +2611,7 @@ def __init__(
         serializer: Callable[[dict], str] | None = None,
         strip_prefixes: list[str | Pattern] | None = None,
         enable_validation: bool = False,
+        response_validation_error_http_status: HTTPStatus | None = None,
     ):
         """Amazon API Gateway REST and HTTP API v1 payload resolver"""
         super().__init__(
@@ -2591,6 +2621,7 @@ def __init__(
             serializer,
             strip_prefixes,
             enable_validation,
+            response_validation_error_http_status,
         )
 
     def _get_base_path(self) -> str:
@@ -2664,6 +2695,7 @@ def __init__(
         serializer: Callable[[dict], str] | None = None,
         strip_prefixes: list[str | Pattern] | None = None,
         enable_validation: bool = False,
+        response_validation_error_http_status: HTTPStatus | None = None,
     ):
         """Amazon API Gateway HTTP API v2 payload resolver"""
         super().__init__(
@@ -2673,6 +2705,7 @@ def __init__(
             serializer,
             strip_prefixes,
             enable_validation,
+            response_validation_error_http_status,
         )
 
     def _get_base_path(self) -> str:
@@ -2701,9 +2734,18 @@ def __init__(
         serializer: Callable[[dict], str] | None = None,
         strip_prefixes: list[str | Pattern] | None = None,
         enable_validation: bool = False,
+        response_validation_error_http_status: HTTPStatus | None = None,
     ):
         """Amazon Application Load Balancer (ALB) resolver"""
-        super().__init__(ProxyEventType.ALBEvent, cors, debug, serializer, strip_prefixes, enable_validation)
+        super().__init__(
+            ProxyEventType.ALBEvent,
+            cors,
+            debug,
+            serializer,
+            strip_prefixes,
+            enable_validation,
+            response_validation_error_http_status,
+        )
 
     def _get_base_path(self) -> str:
         # ALB doesn't have a stage variable, so we just return an empty string

aws_lambda_powertools/event_handler/lambda_function_url.py

@@ -8,6 +8,8 @@
 )
 
 if TYPE_CHECKING:
+    from http import HTTPStatus
+
     from aws_lambda_powertools.event_handler import CORSConfig
     from aws_lambda_powertools.utilities.data_classes import LambdaFunctionUrlEvent
 
@@ -57,6 +59,7 @@ def __init__(
         serializer: Callable[[dict], str] | None = None,
         strip_prefixes: list[str | Pattern] | None = None,
         enable_validation: bool = False,
+        response_validation_error_http_status: HTTPStatus | None = None,
     ):
         super().__init__(
             ProxyEventType.LambdaFunctionUrlEvent,
@@ -65,6 +68,7 @@ def __init__(
             serializer,
             strip_prefixes,
             enable_validation,
+            response_validation_error_http_status,
         )
 
     def _get_base_path(self) -> str:

aws_lambda_powertools/event_handler/middlewares/openapi_validation.py

@@ -17,7 +17,7 @@
 )
 from aws_lambda_powertools.event_handler.openapi.dependant import is_scalar_field
 from aws_lambda_powertools.event_handler.openapi.encoders import jsonable_encoder
-from aws_lambda_powertools.event_handler.openapi.exceptions import RequestValidationError
+from aws_lambda_powertools.event_handler.openapi.exceptions import RequestValidationError, ResponseValidationError
 from aws_lambda_powertools.event_handler.openapi.params import Param
 
 if TYPE_CHECKING:
@@ -58,7 +58,11 @@ def get_todos(): list[Todo]:
     ```
     """
 
-    def __init__(self, validation_serializer: Callable[[Any], str] | None = None):
+    def __init__(
+        self,
+        validation_serializer: Callable[[Any], str] | None = None,
+        has_response_validation_error: bool = False,
+    ):
         """
         Initialize the OpenAPIValidationMiddleware.
 
@@ -67,8 +71,14 @@ def __init__(self, validation_serializer: Callable[[Any], str] | None = None):
         validation_serializer : Callable, optional
             Optional serializer to use when serializing the response for validation.
             Use it when you have a custom type that cannot be serialized by the default jsonable_encoder.
+
+        custom_serialize_response_error: ValidationException, optional
+            Optional error type to raise when response to be returned by the endpoint is not
+            serialisable according to field type.
+            Raises RequestValidationError by default.
         """
         self._validation_serializer = validation_serializer
+        self._has_response_validation_error = has_response_validation_error
 
     def handler(self, app: EventHandlerInstance, next_middleware: NextMiddleware) -> Response:
         logger.debug("OpenAPIValidationMiddleware handler")
@@ -165,6 +175,8 @@ def _serialize_response(
             errors: list[dict[str, Any]] = []
             value = _validate_field(field=field, value=response_content, loc=("response",), existing_errors=errors)
             if errors:
+                if self._has_response_validation_error:
+                    raise ResponseValidationError(errors=_normalize_errors(errors), body=response_content)
                 raise RequestValidationError(errors=_normalize_errors(errors), body=response_content)
 
             if hasattr(field, "serialize"):

aws_lambda_powertools/event_handler/openapi/exceptions.py

@@ -23,6 +23,16 @@ def __init__(self, errors: Sequence[Any], *, body: Any = None) -> None:
         self.body = body
 
 
+class ResponseValidationError(ValidationException):
+    """
+    Raised when the response body does not match the OpenAPI schema
+    """
+
+    def __init__(self, errors: Sequence[Any], *, body: Any = None) -> None:
+        super().__init__(errors)
+        self.body = body
+
+
 class SerializationError(Exception):
     """
     Base exception for all encoding errors

aws_lambda_powertools/event_handler/vpc_lattice.py

@@ -8,6 +8,8 @@
 )
 
 if TYPE_CHECKING:
+    from http import HTTPStatus
+
     from aws_lambda_powertools.event_handler import CORSConfig
     from aws_lambda_powertools.utilities.data_classes import VPCLatticeEvent, VPCLatticeEventV2
 
@@ -53,9 +55,18 @@ def __init__(
         serializer: Callable[[dict], str] | None = None,
         strip_prefixes: list[str | Pattern] | None = None,
         enable_validation: bool = False,
+        response_validation_error_http_status: HTTPStatus | None = None,
     ):
         """Amazon VPC Lattice resolver"""
-        super().__init__(ProxyEventType.VPCLatticeEvent, cors, debug, serializer, strip_prefixes, enable_validation)
+        super().__init__(
+            ProxyEventType.VPCLatticeEvent,
+            cors,
+            debug,
+            serializer,
+            strip_prefixes,
+            enable_validation,
+            response_validation_error_http_status,
+        )
 
     def _get_base_path(self) -> str:
         return ""
@@ -102,9 +113,18 @@ def __init__(
         serializer: Callable[[dict], str] | None = None,
         strip_prefixes: list[str | Pattern] | None = None,
         enable_validation: bool = False,
+        response_validation_error_http_status: HTTPStatus | None = None,
     ):
         """Amazon VPC Lattice resolver"""
-        super().__init__(ProxyEventType.VPCLatticeEventV2, cors, debug, serializer, strip_prefixes, enable_validation)
+        super().__init__(
+            ProxyEventType.VPCLatticeEventV2,
+            cors,
+            debug,
+            serializer,
+            strip_prefixes,
+            enable_validation,
+            response_validation_error_http_status,
+        )
 
     def _get_base_path(self) -> str:
         return ""

docs/core/event_handler/api_gateway.md

@@ -309,7 +309,7 @@ Let's rewrite the previous examples to signal our resolver what shape we expect
 
 !!! info "By default, we hide extended error details for security reasons _(e.g., pydantic url, Pydantic code)_."
 
-Any incoming request that fails validation will lead to a `HTTP 422: Unprocessable Entity error` response that will look similar to this:
+Any incoming request or and outgoing response that fails validation will lead to a `HTTP 422: Unprocessable Entity error` response that will look similar to this:
 
 ```json hl_lines="2 3" title="data_validation_error_unsanitized_output.json"
 --8<-- "examples/event_handler_rest/src/data_validation_error_unsanitized_output.json"
@@ -398,6 +398,35 @@ We use the `Annotated` and OpenAPI `Body` type to instruct Event Handler that ou
     --8<-- "examples/event_handler_rest/src/validating_payload_subset_output.json"
     ```
 
+#### Validating responses
+
+The optional `response_validation_error_http_status` argument can be set for all the resolvers to distinguish between failed data validation of payload and response. The desired HTTP status code for failed response validation must be passed to this argument.
+
+Following on from our previous example, we want to distinguish between an invalid payload sent by the user and an invalid response which is being proxying to the user from another endpoint.
+
+=== "customizing_response_validation.py"
+
+    ```python hl_lines="18 30 34 36"
+    --8<-- "examples/event_handler_rest/src/customizing_response_validation.py"
+    ```
+
+    1. This enforces response data validation at runtime. A response with status code set here will be returned if response data is not valid.
+    2. We validate our response body against `Todo`.
+    3. Operation returns a string as oppose to a Todo object. This will lead to a `500` response as set in line 18.
+    4. The distinct `ResponseValidationError` exception can be caught to customise the response—see difference between the sanitized and unsanitized responses.
+
+=== "sanitized_error_response.json"
+
+    ```json hl_lines="2-3"
+    --8<-- "examples/event_handler_rest/src/response_validation_sanitized_error_output.json"
+    ```
+
+=== "unsanitized_error_response.json"
+
+    ```json hl_lines="2-3"
+    --8<-- "examples/event_handler_rest/src/response_validation_error_unsanitized_output.json"
+    ```
+
 #### Validating query strings
 
 !!! info "We will automatically validate and inject incoming query strings via type annotation."

examples/event_handler_rest/src/customizing_response_validation.py

@@ -0,0 +1,49 @@
+from http import HTTPStatus
+from typing import Optional
+
+import requests
+from pydantic import BaseModel, Field
+
+from aws_lambda_powertools import Logger, Tracer
+from aws_lambda_powertools.event_handler import APIGatewayRestResolver, content_types
+from aws_lambda_powertools.event_handler.api_gateway import Response
+from aws_lambda_powertools.event_handler.openapi.exceptions import ResponseValidationError
+from aws_lambda_powertools.logging import correlation_paths
+from aws_lambda_powertools.utilities.typing import LambdaContext
+
+tracer = Tracer()
+logger = Logger()
+app = APIGatewayRestResolver(
+    enable_validation=True,
+    response_validation_error_http_status=HTTPStatus.INTERNAL_SERVER_ERROR, # (1)!
+)
+
+
+class Todo(BaseModel):
+    userId: int
+    id_: Optional[int] = Field(alias="id", default=None)
+    title: str
+    completed: bool
+
+@app.get("/todos_bad_response/<todo_id>")
+@tracer.capture_method
+def get_todo_by_id(todo_id: int) -> Todo: # (2)!
+    todo = requests.get(f"https://jsonplaceholder.typicode.com/todos/{todo_id}")
+    todo.raise_for_status()
+
+    return todo.json()["title"] # (3)!
+
+@app.exception_handler(ResponseValidationError) # (4)!
+def handle_response_validation_error(ex: ResponseValidationError):
+    logger.error("Request failed validation", path=app.current_event.path, errors=ex.errors())
+
+    return Response(
+        status_code=500,
+        content_type=content_types.APPLICATION_JSON,
+        body="Unexpected response.",
+    )
+
+@logger.inject_lambda_context(correlation_id_path=correlation_paths.API_GATEWAY_HTTP)
+@tracer.capture_lambda_handler
+def lambda_handler(event: dict, context: LambdaContext) -> dict:
+    return app.resolve(event, context)

examples/event_handler_rest/src/response_validation_error_unsanitized_output.json

@@ -0,0 +1,8 @@
+{
+  "statusCode": 500,
+  "body": "{\"statusCode\": 500, \"detail\": [{\"type\": \"model_attributes_type\", \"loc\": [\"response\", ]}]}",
+  "isBase64Encoded": false,
+  "headers": {
+    "Content-Type": "application/json"
+  }
+}

examples/event_handler_rest/src/response_validation_sanitized_error_output.json

@@ -0,0 +1,8 @@
+{
+  "statusCode": 500,
+  "body": "Unexpected response.",
+  "isBase64Encoded": false,
+  "headers": {
+    "Content-Type": "application/json"
+  }
+}