#twitch

ttv-rs

A hopefully-eventually-comprehensive interface to the Twitch API. Somewhat modeled after Twitch4j.

2 releases

0.1.1 Nov 10, 2020
0.1.0 Sep 25, 2020

#22 in #twitch

MIT/Apache

17KB

ttv-rs

Attempt at unifying the Twitch API for the Rust ecosystem.

Why?

This collection of crates was born out of a desire to create a "One library to rule them all!" to interact with the Twitch API using Rust. Initially I thought every sub-crate would likely depend on a different already-existing crate on crates.io, or implement the features itself if none were suitable. However, since beginning this project, I have realized that the twitch-api2 crate, which I originally believed was only intending to cover Helix and TMI, is actually aspiring to cover most of the other endpoints as well eventually (see this issue. Thus, this library is currently only providing a way to combine both twitchchat and twitch-api2 in one dependency, with shorter names 😛

I'm still deciding whether re-exporting twitch_api2 is really the best way to achieve my goals below, but it is looking promising. However, even if that crate ends up covering all of the endpoints on its own, I may end up trying to provide additional convenience features on top of it, such as functions that automatically pick the best endpoint for certain requests, chat features that utilize REST or pubsub endpoints, a Webhooks server, UI, database integration, etc...

Goals

  • Cover the entirety of all API's listed in Twitch's documentation
  • Cover several of the useful unofficial endpoints that are known now or discovered in the future.
  • Match the organization of Twitch's documentation as closely as possible.
    • This makes it easier for users of the crate to find and cross-reference documentation and make use of tutorials for technologies other than Rust, as well as making it easier for contributors to add documentation.
  • Flexibility
    • As often as reasonable, this crate should be implementation-agnostic (e.g. able to support various async runtimes, networking implementations, serialization libraries, etc.)
    • Users should be able to avoid compiling features they don't need.
      • However, a sane set of default features is still critical in my opinion. Currently all active and deprecated API endpoints are enabled by default, but unofficial endpoints are not.

Note that currently I am very impressed with Museun's twitchchat crate, and would love it if all of the other modules mimic its API and interoperate well with it. E.g., you should be able to use the same async executor for all endpoints.

Similarly, I am a fan of the sans-io approach to library design, although it's also nice if a library provides optional integration with other libraries to actually implement the IO, for less boilerplate on the end-user's part.

Also, as I stated, my ideas are heavily informed by the design of Twitch4j; however, I won't be following it precisely. I have some of my own opinions of how things should be organized as well.

Progress

Active Twitch API's

Feature Status Current Implementation
Auth … Testing twitch_oauth2
Chat … Testing twitchchat
Helix … Testing twitch_api2 (with only helix feature)
Extensions ❌ Unimplemented None
PubSub … Testing twitch_api2 (with only pubsub feature)
Webhooks ❌ Unimplemented None

Deprecated Twitch API's

Feature Status Current Implementation
v5 (Kraken) ❌ Unimplemented (libtwitch-rs is currently not license-compatible)
Legacy Drops ❌ Unimplemented None

Unofficial Twitch API's

Feature Status Current Implementation
TMI … Testing twitch_api2 (with only tmi feature)
GraphQL ❌ Unimplemented None

Contributing

Currently I am mostly looking for feedback and input on my vision and ideas, as well as possible existing crates that may fit well into this project. PR's are welcome, but beware that I may be very opinionated about how things are organized and implemented, and may be changing my mind often for the time being.

License

Licensed under MIT and Apache 2.0

Dependencies

~0–17MB
~241K SLoC