/twitter-backend

Twitter backend - Golang

Primary LanguageGo

A Twitter-like Website

Booby Image

This repository contains the backend code for the final project of Fall 2020 Internet Engineering course. In this project, we were instructed to build a twitter-like website. We called our version "Boobier" after a special seabird (more information here), and it's pretty much like Twitter. The frontend of this project is available here. The whole project is deployed on Heroku Cloud Application Platform.

Project Explanation

We chose to write the backend in GoLang since it was the course syllabus, and our database of choice was MongoDB because of its simplicity.

Here are the features that our website is capable of:

  • Tweeting and deleting a tweet
  • Making/Changing a profile (bio, profile picture, and header)
  • Following/Unfollowing other users
  • Notifications and Logs (history of a user's actions)
  • Like, comment, and retweet
  • Searching by username, text, and hashtag
  • Viewing the timeline
  • Seeing who liked and retweeted a tweet
  • Hashtag trends
  • Last but not least, user suggestion

Database and Object Models

We currently have three databases: one for users, another for tweets, and the last one for keeping the hashtags.

Until now, we have considered five objects to be modeled: users, tweets, owners (explained later), hashtags, and events.

1. User Model:

Each user U has the following fields:

  • Name: name of U.
  • Username: unique username.
  • Email and password: email is considered unique and passwords are hashed using bcrypt package.
  • Bio, profile picture, and header: together, they form the user profile.
  • Tweets: an array of ObjectIDs where each id refers to a tweet.
  • Followings/Followers: an array of Owners which represents the profiles U has followed or users that follow U.
  • Notifications: an array of Events and it keeps track of three things:
    • One of U's tweets were liked
    • One of U's tweets were retweeted
    • Someone followed U
  • Logs: Much like the Notifications, but it records U's actions.

2. Tweet Model:

Each tweet T is made of these fields:

  • ID
  • Text
  • Media
  • Date and Time of the tweet
  • Owner: an Owner object which represents the tweet owner.
  • Likes and Retweets: an array of Owners.
  • Parent: generally empty but not when the tweet is a comment for another one.
  • Comments: a list of CommentTweet objects.

The CommentTweet model contains all the Tweet fields except for the Parent and the Comments.

3. Owner Model:

Each owner model is representing a user and only has the Username, Profile picture, Name, Bio, and IsFollowing fields. The IsFollowing field indicates whether the user requesting this object follows the target user or not. For instance, all U's followers has this field equal to true for them.

4. Hashtag Model:

Only records the name of a hashtag, the tweets which it belongs to, and the number of times it was used in general.

5. Event Model:

Three different actions are considered to be an Event: Like, Retweet, and Following as explained before and each event has the following fields:

  • Mode: whether it was a Like, a Retweet, or a Follow action.
  • Source: the user causing the action.
  • Target: the user to whom the action relates.
  • Content: a short, simple description
  • Timestamp
  • Tweet: for Like and Retweet events shows the actual tweet.

The actual requests and corresponding responses can be seen in the code itself and doesn't need much of an explanation.

Search by text

We tried to implement text-based search much like Twitter itself so our search algorithm supports the following examples:

  • "q1 q2 q3": all the tweets with this exact pattern in them.
  • q1 q2 q3: all the tweets having at least one of the queries.
  • "q1" q2 "q3": all the tweets that have q1 and q3. (q2 optional)

You can generate the docs of the backend to get a better sense of requests and responses. To generate these docs (automatically with the help of swag), first install the package and then run swag init, change the listening hostname to localhost and then, the resulting documentation is available at this link.

Contributions

There are many many bugs to be reported, suggestions to be told, ideas to be shared, and other forms of feedback to be told. Any form of these contributions would be a huge help to us improving this project. Thank you in advance.

Website and Team

You can test our website at Boobier.

Our Team: