markov-bot

Generates random sentences from your tweet history using the Markov chain, and post them into Twitter.

Usage

1. Obtain tweet.js

Download your Twitter archive from Twitter, which is available on Settings > Account > "Request your archive"

2. Obtain credentials

Register your application on Twitter Application Management with "Read and Write" permissions at least, then set your credentials to the following environment variables.

• MARKOV_BOT_CONSUMER_KEY
• MARKOV_BOT_CONSUMER_SECRET
• MARKOV_BOT_ACCESS_TOKEN
• MARKOV_BOT_ACCESS_TOKEN_SECRET

3. Run

$ docker run \
  -e MARKOV_BOT_CONSUMER_KEY=... \ # or --env-file=...
  -e MARKOV_BOT_CONSUMER_SECRET=... \
  -e MARKOV_BOT_ACCESS_TOKEN=... \
  -e MARKOV_BOT_ACCESS_TOKEN_SECRET=... \
  ryotakameoka/markov-bot \
  --tweet-js=...

Scheduled execution on Heroku

Note that tweet.js must be hosted somewhere (with a public URL). Personally I have a repository for the file and deploy it to Netlify.

Instead, you can also build your own image extending docker.io/ryotakameoka/markov-bot which contains your tweet.js inside and push it to Heroku Container Registry.

Install the Heroku CLI

https://devcenter.heroku.com/articles/heroku-cli#download-and-install

Login to Heroku

$ heroku login

Create an app

Replace <APP_NAME> with an arbitrary name.

$ heroku apps:create <APP_NAME>

Configure environment variables

$ heroku config:set --app=<APP_NAME> MARKOV_BOT_CONSUMER_KEY=...
$ heroku config:set --app=<APP_NAME> MARKOV_BOT_CONSUMER_SECRET=...
$ heroku config:set --app=<APP_NAME> MARKOV_BOT_ACCESS_TOKEN=...
$ heroku config:set --app=<APP_NAME> MARKOV_BOT_ACCESS_TOKEN_SECRET=...

Add Heroku Scheduler to your app

$ heroku addons:create --app <APP_NAME> scheduler:standard

Pull the pre-built Docker image from Docker Hub

$ docker pull docker.io/ryotakameoka/markov-bot

Push the Docker image to Heroku

Note that Heroku Scheduler only supports "web" dyno.

$ heroku container:login
$ docker tag docker.io/ryotakameoka/markov-bot registry.heroku.com/<APP_NAME>/web
$ docker push registry.heroku.com/<APP_NAME>/web
$ heroku container:release --app <APP_NAME> web

Open the Heroku Scheduler in the browser

$ heroku addons:open --app <APP_NAME> scheduler

Press "Add Job" button and fill "Run Command" section as below

$ --tweet-js <URL_OF_TWEET_JS>

Runtime options

Option	Description	Example values
--tweet-js	File path or URL for tweet.js (required)	./tweet.js https://example.com/path/to/tweet.js
--order	Order of Markov chain, in positive integer (optional)	2 3

Related links

• Markov chain - Wikipedia, the free encyclopedia
• morishin/poem-generator

Lisence

This software is distributed under BSD3 lisence.