Remote Code Runner is a simple code service for running code on remote server side. Docker is used to execute untrusted code in a sandbox environment.

Environment:

• Ubuntu Linux 18.04
• Docker 19.x
• Python 3.8

MAKE SURE the login user has sudo privileges and does not require entering a password.

Install all required software:

$ sudo apt install git docker.io python3.8

Get source:

$ cd /srv
$ sudo git clone https://github.com/michaelliao/remote-code-runner.git

Generate all from source:

$ cd /srv/remote-code-runner
$ sudo python3.8 generate.py

Download required docker images by warm up script (this may take a long time):

$ cd /srv/remote-code-runner/bin
$ sudo sh warm-up-docker.sh

Start server:

$ cd /srv/remote-code-runner/bin
$ sudo start-runner.sh

Using simple HTTP JSON API:

$ curl http://server-ip:8080/run -H 'Content-Type: application/json' -d '{"language":"python","code":"import math\nprint(math.pi)"}'
{"error": false, "timeout": false, "truncated": false, "output": "3.141592653589793\n"}

API input:

• language: language name, lowercase: java, python, ruby.
• code: language code as JSON string: import math\nprint(math.pi)

API output:

• timeout: boolean, is execution timeout.
• error: boolean, is error output. e.g. compile failed or syntax error.
• truncated: boolean, is output was truncated for too many characters.
• output: string, the output of execution.

How code are executed on the remote server side:

1. Http server runner.py got language name and code from API;
2. Write code into a temperary directory like /tmp/remote-code-runner/1;
3. Execute command like sudo docker run -t --rm -w /app -v /tmp/dir:/app <image-tag> python3 main.py;
4. Write output into API response;
5. Clean up temperary directory.

• Multiple files are not supported.
• There is no way to read input from console. Any user input code will cause timeout.

Remote code runner should only be served in private network. User identity check, rate limit must be done in application server or reverse proxy like Nginx.

How to add a new language:

1. Add configuration in config.json:

{
    ...
    "languages": {
        ...
        "node": {
            "file": "main.js",
            "image": "node:13.12-slim",
            "command": "node main.js"
        }
    }
}

The key node is the language name.

2. Make sure image is downloaded on local:

$ sudo docker run -t --rm node:13.12-slim ls

3. Restart start-runner.sh.