Tinchat

Chat Application built using NestJS, PostgreSQL, Elasticsearch and ReactJS.

❤️ NestJS

Features

• User creation
  • Registration.
  • Login.
  • Clients are verified via JSON web tokens.
  • A socket will only send messages to the receiver if their jwt token exists in cache.
• User search
  • Clients can look up one another via first name, last name and email
  • Users are indexed into Elasticsearch which makes them available for fast lookups.
• Messaging
  • Messaging is instant, if client is online they will receive the message in real time with the use of Socket.io.
  • Similar to slack you can send message to yourself.

Notable Design Flaws

• If someone gets hold the user's jwt token they can make requests to all routes only authorized to that user despite authenticating themselves.
  • Solution:
    • Using sessions instead of relying purely on JWT validation.
• Everytime the server restarts all the users are logged out.
  • Solution:
    • Using caching that persists outside the instance of the server. For example: Redis.

Endpoints:

• POST 'register'
  • Registers a user and gives you back a jwt token
• POST 'login'
  • Logs in the user and gives you back a jwt token
• POST 'logout'
  • Logs the user out and deletes the jwt token from server cache
• GET 'api/users'
  • Retrieves all user (Protected)
• POST 'api/message'
  • Creates a message (Protected)
• GET 'api/conversation/?with=${with}&page=${page}&limit=${limit}'
  • Retrieves all conversation between two users (Protected)
  • with is the email of the user you are having the conversation with
  • page is a pagination variable to indicate which page of conversation you want to retrieve
  • limit is a pagination variable to indicate how many conversations you want in each page to have
• GET 'api/search/:index'
  • Retrieves all documents of a given index (Protected)
  • index is the name of the Elasticsearch index
• POST 'api/search/:index'
  • Retrives particular documents depending on the query provided (Protected)
  • index is the name of the Elasticsearch index

Dependencies:

• Docker
• Docker Compose

How to run:

You will need the following environment variables configured in order for compose to work. In your root directory create a .env file with the following environment variables filled out.

  NODE_ENV=production
  JWT_EXPIRATION=604800	
  SERVER_PORT=8000
  SOCKET_PORT=8001
  SECRET='<string of your choice>'	
  ES_TIMEOUT=3000	
  # 192.168.99.100 for wsl users
  ES_HOST=localhost
  ES_PORT=9200
  # 192.168.99.100 for wsl users
  DB_HOST=localhost
  DB_PORT=5432
  # 192.168.99.100 for wsl users
  DOCKER_IP=localhost
  REACT_APP_REST_API_ENDPOINT=''
  REACT_APP_SOCKET_ENDPOINT=''
  REACT_APP_COOKIE_MAX_AGE=18000
  REACT_APP_ITEMS_PER_PAGE=15

After you have created the .env open up a terminal and type in: $ make docker-up

Your application should be up on port 8000.