BookshelfBot is a Discord bot that fetches information for a book and then displays that information within a Discord message.
This bot was inspired by the Reddit GoodReads Bot that I found out about through /r/suggestmeabook. Unfortunately, for this project, I cannot use GoodReads as they're deprecating their API so I have used Google Books instead.
I wrote this in a night. It lacks tests. I'm a terrible Elixir developer. If this breaks then I am sorry. Feel free to post an issue.
/info- Displays information about the bot/book <book_title>- Fetches information about a book using its title
The bot is an OTP application that has two child applications: a GenStage 'consumer' provided by the Nostrum library that will receive data from Discord and act on it and a Cachex process that caches Google Books API responses for better bot performance. You can see these both started as children in lib/bookshelf_bot/application.ex.
The API wrapper for the Google Books service is in lib/bookshelf_bot/google_books. It is very simple and only wraps the https://www.googleapis.com/books/v1/volumes endpoint. This API wrapper is used by the main Discord consumer code to query for book information.
The code that actually handles Discord events is defined in lib/bookshelf_bot_discord. The consumer's behavior is defined in lib/bookshelf_bot_discord/consumer.ex (see the handle_event function matches).
The slash commands are defined individually in the lib/bookshelf_bot_discord/slash_commands and the code that pushes those commands to Discord and handles incoming interactions for them are in lib/bookshelf_bot_discord/slash_commands. The book slash command caches API responses in accordance to the query string used to fetch them. So if someone searched for "Big Foot" as a book title then that string would be normalized to "big+foot" and then any returned book for that title would be stored in an in-memory cache to save subsequent API requests.
- I should probably add a
by <author>feature for when two titles conflict. - The Google Books API actually returns a collection of books and the bot just grabs the first one. This works for now because Google is just really good about approximating which book is the most correct. If this causes problems down the road then I'll have to figure out something more advanced.
- Typespecs are inconsistently applied throughout the project and Dialyzer isn't at all setup.
- There are literally no tests.
- Set
DISCORD_TOKENwithin your shell. - Create a
dev.secret.exsfile based on thedev.secret.exs.examplefile (that config points the bot at a test guild) mix deps.getmix --no-halt
That's the basics on running the bot locally against a dev Discord server.
The production deploy for this bot uses fly.io as a host. It's a Heroku like PaaS that showed up on Hacker News once or twice so I checked them out on a whim. They have great documentation for onboarding an elixir application, have a really good free tier, and are super easy to use. The guide I used to deploy this application is here - it features a Phoenix webapp but I found that it was pretty easy to adapt.
- Botchini: This existing elixir bot was what I started with as a base. I generated an OTP application with
mix new --appand then took chunks out of botchini like a cannibalistic parasite. The code I took I clarified with comments and added some defensive programming. Very grateful it exists. - Nostrum: The Discord library I used.