Daily Dose of Quotes: Twitter Bot

Tweets inspiring and motivational quotes every 3 hours! Replies to the tweets of specified users as soon they post a new tweet.

Table of contents

Technologies
Python Libraries
Description
Images
Tweet Limit Calculation
Tracked Users

Technologies:

• Python • MongoDB • Heroku • Twitter API v2 • Git

Python Libraries:

• tweepy • MongoDB Atlas API • PYgithub • requests • loadenv • multithreading

Description

Daily dose of Quotes is a Twitter bot coded in Python which uses the Twitter API endpoint to tweet a motivational quote every 3 hours while tracking a list of users to reply to their latest tweet with a randomly generated quote.
Technology: Python, MongoDB, Heroku, Git.

Detailed Description

• The Twitter bot is powered by a 50 thousand tweet database hosted on MongoDB Atlas.
• It sends a random number as an Index and extracts the corresponding quote using MongoDB Atlas API. Raw data is formatted to show the author's name, tags, and other information while not exceeding the tweet character limit.
• The Tweet data is then posted to Twitter using the tweepy library and Twitter's API v2 endpoint.
• This process running on the first thread sleeps for the next 3 hours.
• Simultaneously, the second thread tracks a list of users hosted on the same database and updates the ID of the latest tweet of every user.
• As soon as the user tweets, within 5 minutes, the bot replies to them with a new random quote and updates the ID of the tweet.

My Learnings

• The project underwent many iterations and setbacks. I started with the 1st version and incrementally upgraded the project to the final version 4.
• I began with the tweepy library in python to GET and POST data via the Twitter V2 endpoint, which used quotes from a third-party quote API. Eventually, the quote API proved unreliable.
• I designed version 2 with a 50-thousand quote database hosted locally, but it had memory problems and frequent crashes due to its sheer size.
• For version 3, I debugged the solution to host the database on MongoDB Atlas and fetched/updated data through its web API.
• I Upgraded the bot to version 4 by implementing user tracking and tweet replying. I had to engineer various logical syntaxes like tracking the optimal number of users without exceeding Twitter's limit, keeping track of tweet IDs to only reply to non-replied or latest tweets, and formatting raw data into a readable Tweet.
• Also assembled the above processes using python multithreading on Heroku worker dyno and streamlined the project by integrating it with Heroku CLI.
• Moreover, I studied eclectic Git commands and practices and maintained failsafe versions of production code in branches.

Images

Tweet Limit Calculation

max tweets allowed: $500,000$
bot scans all day with $5$ tweets per scan.
let the time interval between each scan be: $x$
let the no. of users we want to track be: $y$

$$\therefore y = {x * MaxTweets \over DaysHoursMinutes*TweetsPerScan}$$

$$y = {500000 * x \over 312460*5}$$

$$→ y = 2.24x$$ $\therefore$ Less the time interval, less the no of tracked users & vice versa.

$So$ with 5 minute interval, ie: $x = 5$ $$→ y = 2.24*5 \approx 11$$

$\therefore$ Currently tracking 11 users.

Tracked Users

Bot tracks the most famous personalities on twitter.

1. Adam Grant
2. Elon Musk
3. Narendra Modi
4. NASA
5. Amitabh
6. spaceX
7. Ronaldo
8. isro
9. Joe biden
10. Barack Obama