Github User/Repository Search [Full Stack Solution] Deployed to Heroku
The solution provides a web interface to user's to search for any user or repository on github, it also exposes the REST Endpoint for same. Some key features are server side caching using redis, well documented API swagger, application metrics dashboard, client side caching using redux-persist, paginated item list with infinite scroll and unit & api testing with cypress.
REDIS_HOST, REDIS_PORT, REDIS_PWD
clone the repo, install dependencies and npm run dev to run locally.
typescript
react, react-router
redux, redux-thunk
redux-persist - to persist store data
Vanilla CSS
Implemented the basic functionality of search across users and repositories, added skeleton loading to avoid loaders and blank page, implemented infinite scroll for pagination (limited to 5 page to check API rate limit), persisted store and cache first strategy before making an API call.
UX for user/repo cards could have been better, also components could have been a bit less coupled.
typescript
express
express-validator - to escape / sanitize API payload
redis - cache
cypress - unit and api testing
swagger - to document api
Endpoints
Health
GET /api/ping
GET /api/health
Search
GET /api/v1/search
1. A client driven endpoint, where client can pass what fields it want (to solve under / over fetching),
2. Though the task mentioned this to be a POST, but I think the usecase is well suited for GET API.
Cache
PUT /clear-cache
1. flushall redis cached data.
Detailed Documentation of API's
well defined typescript interface for providers and controllers, singleton class for redis provider to maintain only one connection, testing with cypress.
Logging - the logger class logs to console for now, which should ideally be using a mature logging library as such winston, morgan or others.
Monitoring - Added a default application metrics dashboard from here.