aws-cdk-secure-api 0.4.0

A CDK (v2) Construct Library for Secure REST APIs

aws-cdk-secure-api

An unofficial AWS CDK v2 Construct Library for Secure REST APIs.

• Documentation: https://aws-cdk-secure-api.readthedocs.io.

Install

pip install aws-cdk-secure-api

Constructs

• SecureRestApi - A construct to create a (public) REST API secured behind an API key, which needs to be specified in the x-api-key header for all requests.

• IAMSecureRestApi - A construct to create a (public) REST API secured behind AWS IAM authentication, which requires IAM credentials to be signed and included in all requests.

Features

• A CDK Construct which sets up a RestApi secured behind (one of):

  • API key

    • An API key is auto-generated and stored in SSM Parameter Store (which is a free service) as needed.

    • Local cache for the API key, so that API calls are not needed in future CDK deployments.

  • AWS IAM authentication

    • An IAM User (and Policy) is created with minimal permissions to call / invoke the API.

    • The IAM User Credentials (Access Keys) are stored in AWS Secrets Manager.

• Helper methods for all constructs, such as add_resource_and_lambda_methods, to make it easier to integrate a method for an AWS Lambda function for example.

Usage

The SecureRestApi construct represents a Secure REST API in Amazon API Gateway.

Use add_resource, add_lambda_methods, and add_methods to configure the API model, as shown below.

Using a root resource:

from aws_cdk.aws_apigateway import StageOptions
from aws_cdk.aws_lambda import Function, Runtime

from aws_cdk_secure_api import Http, SecureRestApi

# noinspection PyTypeChecker
py_runtime: Runtime = Runtime.PYTHON_3_10

get_handler = Function(self, 'lambda1', runtime=py_runtime, ...)
put_handler = Function(self, 'lambda2', runtime=py_runtime, ...)

api = SecureRestApi(
    self, 'api',
    rest_api_name='My Secure Service',
    # optional: specify a deployment stage
    deploy_options=StageOptions(stage_name='dev')
)

api.add_lambda_methods(get_handler, 'GET')  # GET /
api.add_lambda_methods(put_handler, Http.PUT, Http.POST)  # PUT /, POST /

Using a custom-named resource:

Replace above usage of add_lambda_methods with add_resource_and_lambda_methods, as shown below.

# GET /path1
api.add_resource_and_lambda_methods(get_handler, '/path1', 'GET')
# PUT /path2, POST /path2
api.add_resource_and_lambda_methods(put_handler, '/path2', Http.PUT, Http.POST)

The IAMSecureRestApi construct represents a Secure REST API in Amazon API Gateway, which requires IAM Authorization.

Using a custom-named resource:

from aws_cdk.aws_apigateway import StageOptions
from aws_cdk.aws_lambda import Function, Runtime

from aws_cdk_secure_api import Http, IAMConfig, IAMSecureRestApi

# noinspection PyTypeChecker
py_runtime: Runtime = Runtime.PYTHON_3_10

get_handler = Function(self, 'lambda1', runtime=py_runtime, ...)
put_handler = Function(self, 'lambda2', runtime=py_runtime, ...)

api = IAMSecureRestApi(
    self, 'api',
    rest_api_name='My IAM Secure Service',
    # optional: specify the name of secret to store IAM User Credentials
    config=IAMConfig(secret_name='my-stack/iam-user-access-keys'),
    # optional: specify a deployment stage
    deploy_options=StageOptions(stage_name='dev')
)

# GET /path1
api.add_resource_and_lambda_methods(get_handler, '/path1', 'GET')
# PUT /path2, POST /path2
api.add_resource_and_lambda_methods(put_handler, '/path2', Http.PUT, Http.POST)

AWS Profile

Note that if you normally pass the --profile to the cdk tool, for example such as:

cdk deploy --profile my-aws-profile

The CDK construct won’t be able to detect the AWS profile in this particular case. A few workarounds can be used for this:

1. The environment variable AWS_PROFILE can be set before calling the cdk tool.

2. The profile attribute can be passed in to the config parameter for SecureRestApi.

3. The profile context variable can be passed in to the cdk tool, as shown below:

   cdk deploy --profile my-profile -c profile=my-profile

API Keys

Here is the process that the CDK construct uses for generating or using an API key for a REST API.

1. First, it tries to read the API key from local cache, which is located in your home directory, under ~/.cdk/cache/apigw_api_keys.json.

2. If an API key is found, then it proceeds to use the cached key value, and does not perform the following steps.

3. An API call is made to read the key from AWS SSM Parameter Store. The param name is /{STACK NAME}/api-key, where {STACK NAME} is the name of the CDK stack.

4. If the parameter does not exist, an random API key value is auto-generated, and a new SSM Parameter is created in the same AWS account and region that the CDK stack is deployed to.

5. The API key value is then cached on the local drive, under the ~/.cdk/cache folder.

Stack Outputs

The following stack outputs will additionally be added to the CDK stack:

• APIEndpoint - The base endpoint of the Secure REST API.

  • Note: this output will not show up if override_endpoint_name is disabled in the config parameter.

• APIKey - The API key for the endpoint, which needs to be specified as a value in an HTTP request’s x-api-key header.

• APIIAMUserCredentials - The URL link (to input in a browser) for the Secret stored in AWS Secrets Manager containing the AWS IAM Credentials for invoking the REST API.

Credits

This package was created with Cookiecutter and the rnag/cookiecutter-pypackage project template.

History

0.3.0 (2023-05-17)

Features and Improvements

• Add a helper method add_resource_and_lambda_methods, to set up a new API resource, a lambda integration, and setup HTTP method(s) on the new resource at the same time.

• Update other helper methods – such as add_lambda_methods – to accept an optional resource parameter, which defaults to the “root” API resource (/) by default.

• Add test parameter (boolean) to SecureRestApi – if enabled, then a live API call to AWS SSM (Parameter Store) won’t be performed on an initial run, and instead a dummy API key value is used.

0.2.0 (2023-05-17)

Bugfixes

• Make code compatible with Python 3.11.

Features and Improvements

• Add 3.11 to the list of supported Python versions.

0.1.1 (2022-06-24)

Bugfixes

• Remove typing.Literal usage, so code is compatible with Python 3.7

• Add an import from __future__ import annotations to modules where it was missing.

Features and Improvements

• Update to use the string value of the name attribute for a Http Enum member, instead of the value attribute.

0.1.0 (2022-06-24)

• First release on PyPI.