DiegoVictor/code-runner

Run code safely in a containerized AWS Lambda with support for multiple languages. Like HackerRank, Codewars, etc.

Code Runner

How would you run user's code in a safe way? This application is my answer, code runner runs user's code in a containerized Lambda, preventing access to credentials or any other type of sensitive information. It simulates a challenge platform (HackerRank, Codewars, etc) where users write their solutions and have a service validating it against various test cases.

Table of Contents

• Requirements
• Install
  • Configuring
    • Postgres
    • Migrations
    • Code Runner Container
    • .env
• Usage
  • Routes
    • Requests
• Deploy
  • ops.yml
  • infrastructure.yml
  • Push Image to ECR Repository
  • template.yml
• Running the tests
  • Coverage report

Requirements

• Node.js ^20.9.0
• Serveless Framework
• AWS Account
  • Elastic Container Registry
  • Lambda
  • API Gateway

Install

npm install

Or simple:

yarn

Configuring

The application uses just one database: Postgres.

Postgres

For the fastest setup it is recommended to use docker-compose, you just need to run:

docker-compose up -d pg

Or if you prefer to create the container manually:

docker run --name pg -e POSTGRES_PASSWORD=docker -p 5432:5432 -d postgres

Migrations

Remember to run the database migrations:

npx prisma migrate dev

See more information on Prisma Migrate.

Code Runner Container

Also, it is necessary to run the code-runner container, this is the one that will execute users' code:

docker-compose up -d code-runner

Don't forget to update you .env file you changed the container settings.

.env

In this file you may configure the database URL, app's port, code-runner container URL and code-runner function name. Rename the .env.example in the root directory to .env and update your settings as needed.

key	description	default
DATABASE_URL	Database connection Url.	postgresql://postgres:docker@localhost:5432/code-runner?schema=public
PORT	Port number where the app will run.	5000
CODERUNNER_CONTAINER_URL	code-runner container URL. All you'll need to update if you change docker-compose.yml is the URL's port.	http://localhost:9000/2015-03-31/functions/function/invocations
CODERUNNER_FUNCTION	code-runner function name, it needs to match with the name on template.yml	CodeRunnerFunction

Usage

First of all start up the server:

npm run dev:server

Or:

yarn dev:server

Make sure to have the Code Runner Container and Postgres running, otherwise you will not be able to execute users' code

Routes

route	HTTP Method	params	description
/challenges	GET	-	Return challenges paginated.
/challenges/:id	GET	id of a challenge.	Return challenge's details.
/challenges	POST	Body with challenge title, description, instructions and inputs.	Create a new challenge.
/challenges/:id/solution	POST	id of a challenge. Body with users' code and language used to solve the problem.	Execute users' code, it uses the results to compare with the expected values provided for the challenge.

Requests

• POST /challenges

Request body:

{
  "title": "Squares",
  "description": "Calculate numbers square",
  "instructions": "For each provided number calculate its square",
  "inputs": [
    {
      "value": 5,
      "expected": 25
    },
    {
      "value": 12,
      "expected": 144
     },
    {
      "value": 25,
      "expected": 625
    }
  ]
}

• POST /challenges/:id/solution

Request body:

{
  "code": "async function run(value){return value * value;}",
  "language": "js",
}

Deploy

Deploy the application in the following order:

1. ops.yml (Optional)
2. infrastructure.yml
3. Push Image to ECR Repository
4. template.yml

ops.yml

Only deployable manually, this is responsible for create permissions needed by github actions, that is, you don't need to deploy this stack if you are not using the pipeline.

sam deploy --stack-name coderunner-ops \
  --template ops.yml \
  --no-fail-on-empty-changeset

infrastructure.yml

(Required) - Responsible for create an ECR Repository and RDS Database Instance and its credentials.

sam deploy --stack-name coderunner-infrastructure \
  --template infrastructure.yml \
  --no-fail-on-empty-changeset

Push Image to ECR Repository

Push container image to ECR Repository created in the previous step:

1. Log in into ECR:

aws ecr get-login-password --region <AWS_REGION> | docker login --username AWS --password-stdin <AWS_ACCOUNT_ID>.dkr.ecr.<AWS_REGION>.amazonaws.com

2. Build the image:

cd container
docker build --platform linux/amd64 -t coderunner:1.0.0 .

3. Tag

docker tag coderunner:1.0.0 <AWS_ACCOUNT_ID>.dkr.ecr.<AWS_REGION>.amazonaws.com/coderunner:1.0.0

4. And push to ECR Repository

docker push <AWS_ACCOUNT_ID>.dkr.ecr.<AWS_REGION>.amazonaws.com/coderunner:1.0.0

template.yml

Deploy application:

sam deploy --stack-name coderunner-dev \
  --no-fail-on-empty-changeset \
  --image-repositories CodeRunnerFunction=<AWS_ACCOUNT_ID>.dkr.ecr.us-east-1.amazonaws.com/coderunner \
  --capabilities CAPABILITY_IAM \
  --role-arn <CLOUDFORMATION_ROLE_ARN> \
  --parameter-overrides \
  ImageUri=<AWS_ACCOUNT_ID>.dkr.ecr.us-east-1.amazonaws.com/coderunner:1.0.0

You can get CLOUDFORMATION_ROLE_ARN from ops.yml if you deployed it, otherwise you will need to create it manually.

Running the tests

Jest was the choice to test the app, to run:

$ yarn test

Or:

$ npm run test

Coverage report

You can see the coverage report inside tests/coverage. They are automatically created after the tests run.