This project defines a Python class, LambdaHandler
, and associated Pydantic-derived
event classes, for handling API Gateway events from a range of sources, in particular:
- Direct Invocation
- EventBridge
- SQS queues
- SNS topics
When not using the optional FastAPI support (see below), the package's only dependency is pydantic.
from lambda_handler import (
LambdaHandler,
EventBridgeEvent,
SnsEvent,
LambdaResponse,
)
handler = LambdaHandler()
@handler.sns(topic_name="MyTopic")
def handle_mytopic(event: SnsEvent) -> LambdaResponse:
body = frobincate()
return LambdaResponse(status_code=200, body=body)
@handler.event_bridge(resource_name="MyResource")
def handle_myresource(event: EventBridgeEvent) -> LambdaResponse:
body = fizzbuzz()
return LambdaResponse(status_code=200, body=body)
The handler looks after both the event parsing (so your functions should always
accept an event of some *Event
type), and its response as a properly-formatted
dictionary.
A notable omission from the events that are handled by LambdaHandler
directly are
HTTP requests. These can be handled by an instance of FastAPI
, as follows:
from fastapi import FastAPI
from lambda_handler import LambdaHandler
app = FastAPI(title="My HTTP handler")
@app.get("/")
def index():
return "Hello, World!"
handler = LambdaHandler(fastapi_app=app)
The handler will then take care of everything on your behalf. If you'd prefer, you
can set fastapi_app
later instead, and the handler will take care of that, too.
from fastapi import FastAPI
from lambda_handler import LambdaHandler, SnsEvent, LambdaResponse
handler = LambdaHandler()
@handler.sns(topic_name="MyTopic")
def handle_mytopic(event: SnsEvent) -> LambdaResponse:
body = frobincate()
return LambdaResponse(status_code=200, body=body)
app = FastAPI(title="My HTTP handler")
@app.get("/")
def index():
return "Hello, World!"
handler.fastapi_app = app
FastAPI support requires the package to be installed with optional extras:
pip install "lambda-handler[fastapi]"
, and is built on top of the existing
Mangum package.
The *Event
models lambda-handler defines use pydantic
for parsing and validation, and these models are generic. This means that you can
pass a type argument to the class when defining your function, and it will correctly
parse the content of the event (see below) to that type. If this is confusing, it's
easier to see it in action:
from lambda_handler import LambdaHandler, SnsEvent, LambdaResponse
from pydantic import BaseModel
handler = LambdaHandler()
class MyModel(BaseModel):
thing: str
@handler.sns(topic_name=topic_name)
def test_func(event: SnsEvent[MyModel]) -> LambdaResponse:
assert isinstance(event.records[0].sns.message, MyModel)
return LambdaResponse(status_code="200")
Here, we have parametrised SnsEvent
with MyModel
in the signature of test_func
,
meaning that the message
attribute is parsed to a MyModel
instance in the process.
The following attributes are those which are parsed to a Pydantic model for each event type:
Event Type | Parsed Attribute |
---|---|
DirectInvocationEvent |
event.direct_invocation.body |
EventBridgeEvent |
event.detail |
SnsEvent |
event.records[i].sns.message |
SqsEvent |
event.records[i].body |
If you don't want to deal with parsed event objects, you can include the raw=True
parameter to any of the wrapping methods of LambdaHandler
and write a function
that accepts and returns a Dict[str, Any]
instead. Note that, in this case, the
event object will still be parsed by the AwsEvent
subclasses for identification,
but the event object will be passed as-is in dictionary format to the function.
from fastapi import FastAPI
from lambda_handler import LambdaHandler, SnsEvent, LambdaResponse
from typing import Any, Dict
handler = LambdaHandler()
@handler.sns(topic_name="MyTopic")
def handle_mytopic(event: SnsEvent) -> LambdaResponse:
body = frobincate()
return LambdaResponse(status_code=200, body=body)
@handler.sns(topic_name="MyOtherTopic". raw=True)
def handle_mytopic(event: Dict[str, Any]) -> Dict[str, Any]:
body = frobincate()
return {"statusCode": "200"}