Casino

Casino is a library that provides the backend for playing various card games including poker. The library comprises a deck of cards as well as the implementations (hand ranking, betting, etc.) for the included games. The goal of this library is to provide well-tested code that can be safely relied upon for building card games on top of it.

Crates

• casino_cards: A library that provides a deck of playing cards that can be used for various card games.
• casino_poker: A library that provides hand ranking & the backend for poker games.

Using the library

Add casino_poker (which re-exports casino_cards) to your project and drive the TexasHoldEm engine with a thin loop, supplying a PokerAgent per player. The engine owns all money and rules; your code decides how each player acts. See the casino_poker README for a complete, runnable example covering hand evaluation, playing a hand, and observing the event stream.

$ cargo test --workspace      # run the engine + card-library test suites
$ cargo doc --no-deps --open  # browse the API docs

Architecture notes

• The poker engine (casino_poker) owns all money and rules: per-player contributions, betting, all-ins, and side pots (computed at showdown by layering total contributions, so any number of unequal all-ins resolve correctly). Hand strength is evaluated with a kicker-correct ComparableHand, cross-checked in tests against an independent brute-force oracle.
• Players act through a PokerAgent trait that receives an owned PlayerView snapshot — including the opponent roster (seats) and button — and can observe the hero-redacted public_events() stream, which keys players by a stable PlayerRef (id + name) for per-opponent modeling. Humans and any AI — including a future model-backed opponent — implement the same trait, so a smarter agent can be swapped in without engine changes, and the owned/serializable view can be handed to an external model.
• The engine is I/O-free: instead of printing, it emits serializable GameEvents to a perspective-aware GameObserver. Network callers use public_events() or client_view() for enforced private-card filtering.

Rules simplifications

The button advances by tracking the player on it; when that player busts, the button falls to the first remaining seat rather than implementing full dead-button rules.