hundredandten-state

Player observation layer for the card game Hundred and Ten.

GameState is a player-agnostic snapshot of the game at decision time. All seat positions are relative — the observing player is always seat 0, with other players numbered clockwise. No player identifiers appear in the state. This design makes GameState the natural observation space for automation strategies and future ML training.

GameState is usually constructed by an EngineAdapter from a live Game object. But it may be built directly.

from hundredandten.automation.engineadapter import EngineAdapter

state = EngineAdapter.state_from_engine(game, 'player_1')

print(state.status)            # Status.BIDDING, TRICKS, etc.
print(state.hand)              # tuple[Card, ...] — this player's hand
print(state.available_actions) # tuple of AvailableBid / AvailablePlay / etc.
print(state.table.scores)      # tuple[int, ...] — scores indexed by relative seat

GameState Structure

state.available_actions returns the legal actions for the active player given the current phase. Convenience properties available_bids, available_trump_selections, available_discards, and available_plays filter to a specific action type.

Exports

Phase and action types

State structure types

Card knowledge types