GitHub Users API

This is an application which provides an API to retrieve detailed information about github users

Requirements

• it should allow retrieval of up to a max of 10 users
• it should return back some basic information about user such as
  • name
  • login
  • company
  • number of followers
  • number of public repos
• the returned users should be sorted alphabetically by name

Use Cases

1. usernames not found in cache or not able to communicate with cache
   • Use github API to fetch information of each user and aggregate response
   • backfill the cache
   • return user information with http status code 200
2. usernames not found in cache and Github API fails for all users
   • return http response with Http Status Code 500 (Internal Server Error)
3. partial usernames found in cache
   • Hit the Github API for fetch information for remaining users
     • if Github API is successful
       • backfill the cache and aggregate response
       • respond with
     • if not Github API fails
       • return partial users and could not able to connect to Github API for remaining Users

Demo

• Hosted on Render: https://users-gh.onrender.com
• API docs

Start Up instructions

• make sure you have docker and docker-compose installed
• make sure you have .env in the current directly

  • HOST=0.0.0.0
    PORT=8080
    REDIS_URL=0.0.0.0:6379
    REDIS_PASSWORD=<some-random-password>
    GITHUB_AUTH_TOKEN=<github-auth-token>
    

  • you can generate github-auth-token (Personal access tokens) form here
  • use make run to start the server
  • Swagger API docs

Run tests

• make sure you have .env setup
• use make test to run tests

Considerations

• Fields which are not existent in the GitHub API response are omitted in the server's API response
• Users in response are sorted by Login field alphabetically in ascending order, case insensitive
• GET request with body is used. Ref
• As Github does not have API for fetching user details for multiple users in a single call and latencies for sequentially fetching 10 users details was over 3 seconds, fetched user details concurrently
• Left out things
  • Use case of rate limits
  • use of contexts for timeouts and cancellation
  • retry mechanism in case of not being able to communicate to github or rate limit reached
  • instrumenting service for observability
  • use of Github GraphQL API instead of REST, which allows fetching user details in bulk in single API call