codingdepot/bootbot

A Discord bot that predicts the correct boot choice and visualizes the result.

Boot-Bot

Description

This bot helps you in solve the biggest struggle in your daily life: Which boots should you buy in this particular game of ARAM?

Search no further because now we can employ cutting-edge machine learning to determine your appropriate sense of fashion for every situation. By training a model on dozens thousands of ARAM games in your MMR range you can ensure your choice of footwear is never out of place. Remember: Fashion is not a choice, but a necessity.

Setup

How To Build

Thanks to the power of the crustacean, building is as easy as can be. Just download cargo and run cargo build --release. That's all there is to it. The repository also contains a minimal Dockerfile for all your containerization needs.

If that still sounds like a hassle, the docker images are public. But they are for personal use so I will not take questions regarding their internal well-being.

How To Run

After building the binary, you can run it as you would run any executable file in your OS. However, there are some things the bot needs in order to properly function:

Environment Variables

These variables need to be set in the execution environment:

Name	Function	Required?
RIOT_TOKEN	The token for the Riot API	Yes
DISCORD_TOKEN	The token for the Discord API	Yes
VIP_USER	Discord Snowflake of a privileged user that may use configuration commands	Yes
VIP_GUILD	Discord Snowflakle of a guild that has configuration commands enabled	Yes
GAME_VERSION	The LoL version used for embedded item sprites	Yes
CONFIG_PATH	The base path for startup files	No

If CONFIG_PATH is not set, the program expects these files to be in the same directory as the executable.

Startup Files

Filename	Content	Required?
snowflake_puuid.txt	A Mapping of Discord Snowflakes to Riot PUUIDs, one line per mapping. Syntax: <Snowflake>|<PUUID>	Yes
model.bin	A pre-trained model. If none is provided, the bot won't be able to make predictions until a model has been trained	No

The mapping file is convenient for small, personal use. However, the desired solution is to query Discord connections to get the Riot usernames we need. Also, even when adding a pre-trained model remember to start training a new model after starting if the data is outdated.

Commands

Name	Options	Function	Privileged?
/boots	<[user]>	Suggests the best possible choice of boots for the user if they are in an ARAM match	No
/model	<number of games>	Re-trains the model with the specified number of games. The old model will continue being used until training concludes	Yes
/version	<game version>	Updates the version string for the LoL embed downlaods. Returns an error if the new version leads to an invalid download link	Yes

FAQ

Does this bot work for game modes other than ARAM?

No, the bot only likes ARAM and it refuses to join my Clash team.

How do you handle Cassiopeia?

I do not. Please do not ask the bot for a suggestion when playing Cassiopeia, since it there is a slim chance it leads to the technological singularity.

The Bot recommended <Boots> on <Champion>, why?

Do not question the bot, for it does not question you.

I am a RIOT employee and can't decide whether to grant you a personal API token, what should I do?

Please, I currently can only use the development token for 2 hours before spending 22 hours solving the captcha to generate the next token.

Has anyone ever actually asked these questions?

bootbot's Issues

Change from static PUUIDs to summoner names

Riot PUUIDs are not static!

We need to dynamically fetch the PUUID by name from the summoner API.
On the upside, this will make migrating away from the txt file easier later on.

Investigate using Discord connections to get Riot ID

see how much of a hassle that is

Logging

Use a logging framework to get detailed execution reports.

Create initial documentation

This includes Markdown files as well as code commenting

Create more descriptive errors

Currently most errors are being captured and passed on as "None".

This essentially suppresses the Error information. We should probably define our own error types and wrap any errors with this.
These errors can then be used for more descriptive bot output.

Push a docker image with each new release

Pushing to dockerhub makes it wasier to deploy new versions