volgnie

Get rid of your Twitter followers

shalvah
22
1
Ruby

Volgnie

If, like me, you don’t like having too many Twitter followers, you’ve come to the right place. Volgnie helps you easily purge your Twitter followers. Pick from one of the available criteria, and any followers that don’t meet it will be removed (by blocking/unblocking them).

Note
Retired. This project is retired. I’m tired of struggling with Twitter. Perhaps later, I’ll resurrect it as a self-service option

To use, go to volgnie.com, or read on for technical details.

Tooling

  • Ruby 2.7 and Node.js 14+
  • Serverless framework
  • serverless-offline plugin so we can test locally
  • serverless-rack plugin so we can run a Sinatra web app in a Lambda function
  • OpenTelemetry for tracing
  • Redis for storage

Components

Lambda functions

Tip: Take a look at the serverless.yml file. It’s a concise description of the stack.

There are five key Lambda functions:

  1. web: Handles volgnie.com and the whole web UI. Runs a Sinatra app via serverless-rack. Code is in app/web.rb.
  2. start_purge: When you trigger a purge from the web app, an event (purge_start) is fired, which triggers this function. The start_purge function fetches the user’s followers and following and fires the purge_ready event, which kicks off the purge by triggering the next function…
  3. purge_followers: This takes a batch of followers, check if they match the user’s criteria and purges the ones that fail. Then it sleeps. It does this batch-sleep thing to avoid hitting Twitter’s rate limits.
  4. push_next_batch: This runs at intervals and checks if enough time has passed for us to try the next batch. If so, it fires the purge_ready event again, so purge_followers can process the next batch before sleeping again.
  5. finish_purge: When there are no more batches, the purge_finish event is fired, which triggers this function. The function sends a report of the purge to the user’s email, records metrics and cleans up any unnecessary data.

For more details, see the docs/ folder.

Search

The Twitter API’s search only returns results from a few days, so I had to build a custom search implementation (a better version of what’s currently on oldtweets.today), called Zearch. Unfortunately, it’s not yet open source.

External Services

  • Honeybadger for error reporting, New Relic for OpenTelemetry traces (both optional)
  • SendGrid (production) and Mailtrap (development) for emails
  • AWS SNS (production) for events; in development, events are fired via spawning processes

Setup

  • Install dependencies:
    npm ci --production=false
bundle install
  • Copy .env.example to .env and set the needed environment variables
  • Start Redis (easy way with Docker: docker run --name redis --rm -p 6379:6379 redis)
  • Run the app locally:
    npm run offline

Twitter App Setup

  • Create a Twitter app in the Twitter Developer Portal, get your API key and secret. Then go to your app’s settings and edit the User Authentication Settings:
  • select “Read and Write” under App Permissions
  • Leave “Request email from users” off
  • Select “Web App” under Type of App

Redirect URLs:

Testing

Tests are written with RSpec. To run tests: bundle exec rspec