Overview

This is a Haskell API running on the Scotty web framework. The API runs on localhost:3000 and provides for basic HTTP requests to retrieve & add users/tweets. This data is retrieved and parsed from the public Twitter API.

The framework is currently hosted and running on Amazon's EC2 server.

For haddock style documentation, open up the index.html file located inside the haddock folder.

Technologies used:

• Aeson for JSON encoding/decoding
• Configurator for configuration
• mysql and mysql-simple for DB access
• OAuth client for Twitter authentication
• resource-pool for DB connections pool
• Scotty for routing
• wai-extra for basic authentication
• Warp as webserver

Configuring the project remotely

To ssh to the EC2 server you must first download the twikell-public.pem key pair for authentication. Once it's downloaded CD into the folder and run in terminal. You only need to do this if you wish to browse the files or change something on the remote server, as it should already be up and running:

1. chmod 600 twikell-public.pem
2. ssh -v -i twikell-public.pem ubuntu@ec2-54-194-118-199.eu-west-1.compute.amazonaws.com
3. All required modules/libraries and DB already setup. Enjoy.

If you ever need to copy new files over to the EC2 server you can do so by scp

scp -i twikell-public.pem -r . ubuntu@ec2-54-194-118-199.eu-west-1.compute.amazonaws.com:twikell-web

Which adds the current directory recursively to the twikell-web folder at the remote side.

Configuring the project locally

Make sure that both the haskell-framework and cabal is set up before proceeding.

1. First you need to install both MySQL-Client and MySQL-Server sudo apt-get install mysql-client mysql-server
2. Make sure that libpcre is installed sudo apt-get install libpcre3 libpcre3-dev
3. Setup all dependent modules and libraries cabal install --only-dependencies
4. Finally run the mysql dump to populate the database.

Any cabal problems; you're on your own.

PS. Something that helped personally was: rm -rf ~/.ghc ~/.cabal and reinstall both. All jokes aside though... dont. Have fun!

Using the framework

CD into the source folder, run GHCI and load the main:

1. cd /path-to-twikell-web/src
2. ghci
3. :l Main.hs
4. main

Or compile and run from the main folder.

For accessing the remote EC2 server, replace all occurences of localhost with 54.194.118.199. If the server is not already running (it should be) then remember to ssh and run it following same steps above.

Authentication

Some routes that perform POST, DELETE and UPDATE methods require authentication.

User: hamza Password: abc123

Tweets

To view all tweets in the DB try:

• http://localhost:3000/tweet
• curl -X GET http://localhost:3000/tweet
  • PS no authentication is required here.

To view tweets belonging to a certain user:

• http://localhost:3000/tweet/X
• curl -X GET http://localhost:3000/tweet/X
  • where x is the screen_user in the db.

To view a certain tweet in based on ID try:

• http://localhost:3000/tweetid/X
• curl -X GET http://localhost:3000/tweetid/X
  • where X is the ID of a tweet in the db.

To populate the DB with tweets from a certain user:

• curl -u hamza:abc123 -X POST http://localhost:3000/admin/tweet/X
  • Where X is the screen_name of a Twitter user. I.e. katyperry.
  • This also adds a user to the DB in case he doesn't exist.

To delete certain tweets from the DB try:

• curl -u hamza:abc123 -X DELETE -d 'id=X' http://localhost:3000/admin/tweet
  • Where X is the id of a Tweet in the db.

Users

To list all users in the DB try:

• http://localhost:3000/user
• curl -X GET http://localhost:3000/user
  • PS no authentication is required here.

To View a user in the DB try:

• http://localhost:3000/user/X
• curl -X GET http://localhost:3000/user/X
  • Where X is the screen_name of a Twitter user. I.e. katyperry.

To Update a user in the DB try:

• curl -u hamza:abc123 -H "Content-Type: application/json" -X PUT -d '{"screen_name":"X","name":"Ben Steer","profile_image_url":"updated.com","location":"London","followers_count":1}' http://localhost:3000/admin/user
  • Where X is the screen_name of a Twitter user. I.e. katyperry.

To populate the DB with users:

• curl -u hamza:abc123 -X POST http://localhost:3000/admin/user/X,Y,Z,...,N)
  • Where X,Y,Z are screen_name's of Twitter users separated by commas. PS. number of N must be less than 100.
  • This also works with a single user, i.e. /X where X can be TheEllenShow

To search for Users in Twitter:

• `http://localhost:3000/search/X
• curl -X GET http://localhost:3000/search/X
  • Where X is the search query to match a Twitter user. I.e. katy.
  • Note that this does not populate the local side DB. No need was found for this.