cfn-wrapper-python
cfn-wrapper-python copied to clipboard
Python decorator for making Lambda-backed CloudFormation resources
cfn_resource.py
This project is a decorator and validation system that takes the drudgery out of writing custom resources. You still have access to the context and event as normal, but the decorator handles serializing your response and communicating results to CloudFormation.
See cfn-lambda <https://github.com/andrew-templeton/cfn-lambda>
__ from
Andrew Templeton if you're looking to write your custom resources in
Node.js.
Usage
- Copy
cfn_resource.py
into the directory of your lambda function handler.py - Use the
cfn_resource.Resource
event decorators to decorate your handler like inexample.py
- Zip up the contents and upload to Lambda
Once the function is up, copy its ARN and use it as the ServiceToken for
your custom resource <https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-resource-cfn-customresource.html>
.
For more on the requests you may receive, see this document <https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/crpg-ref-requests.html>
.. code-block:: json
{
"AWSTemplateFormatVersion": "2010-09-09",
"Resources": {
"FakeThing": {
"Type": "Custom::MyResource",
"Properties": {
"ServiceToken": "arn:aws:lambda:SOME-REGION:ACCOUNT:function:FunctionName",
"OtherThing": "foobar",
"AnotherThing": 2
}
}
}
}
For more on how custom resources work, see the AWS docs <https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/template-custom-resources.html>
__
Code Sample
For this example, you need to have your handler in Lambda set as
filename.handler
where filename has the below contents.
.. code-block:: python
import cfn_resource
# set `handler` as the entry point for Lambda
handler = cfn_resource.Resource()
@handler.create
def create_thing(event, context):
# do some stuff
return {"PhysicalResourceId": "arn:aws:fake:myID"}
@handler.update
def update_thing(event, context):
# do some stuff
return {"PhysicalResourceId": "arn:aws:fake:myID"}
Running Tests
To run the tests locally, you need Python 2.7 and pip
. Ideally, you
should use a virtualenv.
.. code-block:: sh
$ pip install -r test-requirements.txt
$ py.test
The tests use mock
and py.test
and will give you a terminal
coverage report. Currently the tests cover ~90% of the (very small)
codebase.
License
This code is released under the MIT software license, see LICENSE.txt for details. No warranty of any kind is included, and the copyright notice must be included in redistributions.