TL;DR ๐ฐ
Learn how to perform integration tests iso prod with aws serverless services. Using lambda-http-interceptor you can easily intercept and mock http calls coming from your deployed lambda functions.
Why should you use it:
- You want to test your lambda functions in a iso-prod environment ๐งช
- You want to save money while running integration tests by not triggering costly third party APIs ๐ธ
- You want to control the behavior of third party APIs to test edge cases ๐ฏ
- You don't want to change your lambda code to make it testable ๐ท
And maybe you can use it for all theses reasons at the same time!
Try lambda-http-interceptor here ๐
Intercept http calls in lambda functions ๐ช
The lib is made of a CDK Construct to instantiate in your stack.
The HttpInterceptor
construct needs to be instantiated in the stack, or inside another Construct. And then the interceptor needs to be applied to the lambda function http calls need to be intercepted from.
The applyHttpInterceptor
uses Aspects
in order to apply it on each NodeLambdaFunction
it finds, thus applyHttpInterceptor
takes any Construct as input.
import { Stack, StackProps } from "aws-cdk-lib";
import { Construct } from "constructs";
import { HttpInterceptor, applyHttpInterceptor } from "lambda-http-interceptor";
import { NodejsFunction } from "aws-cdk-lib/aws-lambda-nodejs";
import { Runtime } from "aws-cdk-lib/aws-lambda";
export class MyStack extends Stack {
constructor(scope: Construct, id: string, props?: StackProps) {
super(scope, id, props);
const interceptor = new HttpInterceptor(this, "HttpInterceptor");
const myLambdaFunctionThatMakesExternalCalls = new NodejsFunction(
this,
"MakeExternalCalls",
{
runtime: Runtime.NODEJS_18_X,
handler: "index.handler",
entry: './handler.ts',
},
);
applyHttpInterceptor(myLambdaFunctionThatDoesExternalCalls, interceptor);
}
}
After deploying, everything is setup on the stack to then perform integration tests.
The second part of the lib is a set of tools to perform integration tests. They are gathered in the HttpLambdaInterceptorClient
class.
import fetch from "node-fetch";
import { expect, describe, it } from "vitest";
process.env.HTTP_INTERCEPTOR_TABLE_NAME = '<table-name-from-construct>'
import { HttpLambdaInterceptorClient } from "lambda-http-interceptor";
import { triggerMyLambdaFunctionThatMakesExternalCalls } from './utils';
describe("my test", () => {
const interceptorClient = new HttpLambdaInterceptorClient(
'<myLambdaFunctionThatMakesExternalCalls-name>',
);
it("tests my lambda function", async () => {
await interceptorClient.createConfigs([
{
url: "https://api-1/*",
response: {
status: 404,
body: JSON.stringify({
errorMessage: "Not found",
}),
},
},
{
url: "https://api-2/path",
response: {
passThrough: true,
},
},
]);
await triggerMyLambdaFunctionThatMakesExternalCalls();
const interceptedCalls = await interceptorClient.pollInterceptedCalls({
numberOfCallsToExpect: 2,
timeout: 5000,
});
expect(resp.interceptedCalls).toBe(2);
});
});
How does it work? ๐
lambda-http-interceptor stores the configurations and the call made in DynamoDB table ๐ช
The HttpInterceptor
instantiates a DynamoDB table in the stack. The table is used to store the configurations of the http calls to intercept. When performing integration tests, filling up the table with configuration is done using the createConfigs
method of the HttpLambdaInterceptorClient
class.
Then assertions can be made on the calls made by the lambda after they are fetched using the pollInterceptedCalls
method of the HttpLambdaInterceptorClient
class.
๐ก Don't forget to give the user you're using to perform the integration tests the right to read in the table. In general, we use AdministratorAccess role for the user performing these tasks.
lambda-http-interceptor uses an internal extension to intercept http calls in lambda functions ๐ก
The internal extension that the interceptor deploys on the lambda functions overrides the http
module of nodejs that is used to make http calls.
For each call made by the lambda, it fetches the http calls configuration stored in DynamoDB and either passes through the call or returns the response value configured at the start.
It keeps track of the http calls listed that are listed in the configuration. If the response of a call doesn't need to be changed but it still needs to be tracked in order to make assertions on it, the configuration of the call doesn't change and the response only contains passthrough: true
.
If you want to deep dive the functioning of the interceptor, you can check out this article that presents extensions really clearly using a simple example.
lambda-http-interceptor has everything built in ๐ช
The setup is fairly easy and it can be used to make assertions on the calls made by your deployed lambda functions. And you don't edit the code of your lambda handler.
The documentation on github is far more exhaustive to get you started.
Try lambda-http-interceptor here ๐
Don't hesitate to star it โญ๏ธ
Top comments (1)
I'm very interested about why you go this way, rather than capturing event payload, and then replaying that against the outer lambda "handler"?