Cato

Online Judge system you can host locally!

Table of contents

• Table of contents
• Description
• Installation
  • Requirements
  • Clone the project
  • Build from source
  • Generate code
  • Linting
• Usage
  • Running in local mode
  • Running in distributed mode
  • Editing problem statements
  • Custom client
• Config
  • CLI Arguments
  • Config files
• TODO
• License

Description

Cato is a simple web-based Online Judge system for competitive programming problems. Designed with local setup in mind, it can be set up with minimal effort, depending only on Docker for containerization of test runs. At the same time, it can also run in a simple distributed host-worker mode, distributing submissions to multiple judge worker machines for more efficient operation.

Installation

Requirements

Docker or a Docker-compatible container runtime must be installed locally to run Cato process in local/distributed worker mode.

To develop the project, OpenAPI Generator is required as it is used to generate the Front End's HTTP API client.

Clone the project

git clone https://github.com/tranHieuDev23/Cato.git

Build from source

# To build for all popular desktop platform
make build-all

# To build for your local platform
make build

Generate code

Cato uses code generation for three purposes:

1. Compile-time dependency injection with github.com/google/wire.
2. Generation of JSON-RPC server and client boilerplate, plus OpenAPI specification using gitlab.com/pjrpc/pjrpc.
3. Generation of TypeScript's HTTP Client using OpenAPI Generator.

To execute all code generation procedure:

make generate

Linting

make lint

Usage

Running in local mode

Just double click on the executable!

The default options is to start the Online Judge server as a standalone process on the local machine, serving both host and worker logic. This is convenient for simple local use cases, with the drawback being that you only have the computational resource of a single machine to evaluate incoming submissions.

Running in distributed mode

To distribute submissions to be evaluated on multiple machines, you can run the system in distributed mode. In this mode, the single HTTP host server is considered to be the source of trust, while one or more worker processes synchronize information with the host over time and periodically fetching the oldest available submission to judge.

Starting the HTTP host server by executing the following command:

cato --distributed

Starting the worker process by executing the following command:

cato --distributed \
    --worker \
    --host-address <address of the host with the prefix http:// or https://> \
    --worker-account worker \
    --worker-password changeme

Make sure that all worker machine have its own Docker (or a Docker-compatible container runtime) running locally and can access the host's HTTP server.

Editing problem statements

QuillJS is the editor for text problem description, while Katex is used for math typesetting, allowing problem setters to write rich problem statements with proper math annotations.

Custom client

Thinking that the web client sucks? You can totally write (or better yet, generate) your own HTTP client for Cato using the generated OpenAPI 3.0 specification.

In fact, this project was created because I discovered that the code generation pipeline using pjrpc and OpenAPI Generator is so powerful that it allows me to finish this project within just 5 days!

Config

CLI Arguments

Config files

Cato use a YAML config file to configure its inner working in more details. By default, if no custom config file is provided, values in the config file configs/local.yaml are used.

TODO

• Problem example list - Small set of small test cases that users can view and copy
• Custom output evaluation method:
  • Evaluation using custom scripts running inside containers
• Contests
  • ACM Scoring Mode - Problems are either completely Accepted or not
  • IOI Scoring Mode - Problems are scored based for passed test cases
  • Ranking dashboard
• Feature flags
  • Enable/disable account registration
  • Enable/disable problem/test case creation
  • Enable/disable submissions
• Account disabling
• Support for more database engines
• Dark mode
• Better documentation

License

MIT