starting splendor

overlays/default.nix

@@ -30,7 +30,9 @@
         pytest-xdist
         requests
         ruff
+        scalene
         sqlalchemy
+        textual
         typer
         types-requests
       ]

pyproject.toml

@@ -48,6 +48,10 @@ lint.ignore = [
     "ERA001", # (perm) I don't care about print statements dir
 ]
 
+"python/splendor/**" = [
+    "S311", # (perm) there is no security issue here
+]
+
 [tool.ruff.lint.pydocstyle]
 convention = "google"
 

python/splendor/base.py

@@ -0,0 +1,646 @@
+"""Base logic for the Splendor game."""
+
+from __future__ import annotations
+
+import itertools
+import random
+from dataclasses import dataclass, field
+from typing import TYPE_CHECKING, Literal, Protocol
+
+if TYPE_CHECKING:
+    from collections.abc import Sequence
+
+GemColor = Literal["white", "blue", "green", "red", "black", "gold"]
+
+GEM_COLORS: tuple[GemColor, ...] = (
+    "white",
+    "blue",
+    "green",
+    "red",
+    "black",
+    "gold",
+)
+BASE_COLORS: tuple[GemColor, ...] = (
+    "white",
+    "blue",
+    "green",
+    "red",
+    "black",
+)
+
+GEM_ORDER: list[GemColor] = list(GEM_COLORS)
+GEM_INDEX: dict[GemColor, int] = {c: i for i, c in enumerate(GEM_ORDER)}
+BASE_INDEX: dict[GemColor, int] = {c: i for i, c in enumerate(BASE_COLORS)}
+
+
+@dataclass(frozen=True)
+class Card:
+    """Development card: gives points + a permanent gem discount."""
+
+    tier: int
+    points: int
+    color: GemColor
+    cost: dict[GemColor, int]
+
+
+@dataclass(frozen=True)
+class Noble:
+    """Noble tile: gives points if you have enough bonuses."""
+
+    name: str
+    points: int
+    requirements: dict[GemColor, int]
+
+
+@dataclass
+class PlayerState:
+    """State of a player in the game."""
+
+    strategy: Strategy
+    tokens: dict[GemColor, int] = field(default_factory=lambda: dict.fromkeys(GEM_COLORS, 0))
+    discounts: dict[GemColor, int] = field(default_factory=lambda: dict.fromkeys(GEM_COLORS, 0))
+    cards: list[Card] = field(default_factory=list)
+    reserved: list[Card] = field(default_factory=list)
+    nobles: list[Noble] = field(default_factory=list)
+    card_score: int = 0
+    noble_score: int = 0
+
+    def total_tokens(self) -> int:
+        """Total tokens in player's bank."""
+        return sum(self.tokens.values())
+
+    def update_card_score(self) -> None:
+        """Recalculate card score."""
+        self.card_score = sum(card.points for card in self.cards)
+
+    def update_noble_score(self) -> None:
+        """Recalculate noble score."""
+        self.noble_score = sum(noble.points for noble in self.nobles)
+
+    @property
+    def score(self) -> int:
+        """Total points in player's cards + nobles."""
+        return self.card_score + self.noble_score
+
+    def can_afford(self, card: Card) -> bool:
+        """Check if player can afford card, using discounts + gold."""
+        missing = 0
+        gold = self.tokens["gold"]
+
+        for color, cost in card.cost.items():
+            missing += max(0, cost - self.discounts.get(color, 0) - self.tokens.get(color, 0))
+            if missing > gold:
+                return False
+
+        return True
+
+    def pay_for_card(self, card: Card) -> dict[GemColor, int]:
+        """Pay tokens for card, move card to tableau, return payment for bank."""
+        if not self.can_afford(card):
+            msg = f"{self.name} cannot afford card {card}"
+            raise ValueError(msg)
+
+        payment: dict[GemColor, int] = dict.fromkeys(GEM_COLORS, 0)
+        gold_available = self.tokens["gold"]
+
+        for color in BASE_COLORS:
+            cost = card.cost.get(color, 0)
+            effective_cost = max(0, cost - self.discounts.get(color, 0))
+
+            use = min(self.tokens[color], effective_cost)
+            self.tokens[color] -= use
+            payment[color] += use
+
+            remaining = effective_cost - use
+            if remaining > 0:
+                use_gold = min(gold_available, remaining)
+                gold_available -= use_gold
+                self.tokens["gold"] -= use_gold
+                payment["gold"] += use_gold
+
+        self.cards.append(card)
+        self.discounts[card.color] += 1
+        return payment
+
+
+def get_default_starting_tokens(player_count: int) -> dict[GemColor, int]:
+    """get_default_starting_tokens."""
+    token_count = (player_count * player_count - 3 * player_count + 10) // 2
+    return {
+        "white": token_count,
+        "blue": token_count,
+        "green": token_count,
+        "red": token_count,
+        "black": token_count,
+        "gold": 5,
+    }
+
+
+@dataclass
+class GameConfig:
+    """Game configuration: gems, bank, cards, nobles, etc."""
+
+    win_score: int = 15
+    table_cards_per_tier: int = 4
+    reserve_limit: int = 3
+    token_limit: int = 10
+    turn_limit: int = 1000
+    minimum_tokens_to_buy_2: int = 4
+
+    cards: list[Card] = field(default_factory=list)
+    nobles: list[Noble] = field(default_factory=list)
+
+
+class GameState:
+    """Game state: players, bank, decks, table, available nobles, etc."""
+
+    def __init__(
+        self,
+        config: GameConfig,
+        players: list[PlayerState],
+        bank: dict[GemColor, int],
+        decks_by_tier: dict[int, list[Card]],
+        table_by_tier: dict[int, list[Card]],
+        available_nobles: list[Noble],
+    ) -> None:
+        """Game state."""
+        self.config = config
+        self.players = players
+        self.bank = bank
+        self.decks_by_tier = decks_by_tier
+        self.table_by_tier = table_by_tier
+        self.available_nobles = available_nobles
+        self.noble_min_requirements = 0
+        self.get_noble_min_requirements()
+        self.current_player_index = 0
+        self.finished = False
+
+    def get_noble_min_requirements(self) -> None:
+        """Find the minimum requirement for all available nobles."""
+        test = 0
+
+        for noble in self.available_nobles:
+            test = max(test, min(foo for foo in noble.requirements.values()))
+
+        self.noble_min_requirements = test
+
+    def next_player(self) -> None:
+        """Advance to the next player."""
+        self.current_player_index = (self.current_player_index + 1) % len(self.players)
+
+    @property
+    def current_player(self) -> PlayerState:
+        """Current player."""
+        return self.players[self.current_player_index]
+
+    def refill_table(self) -> None:
+        """Refill face-up cards from decks."""
+        for tier, deck in self.decks_by_tier.items():
+            table = self.table_by_tier[tier]
+            while len(table) < self.config.table_cards_per_tier and deck:
+                table.append(deck.pop())
+
+    def check_winner_simple(self) -> PlayerState | None:
+        """Simplified: end immediately when someone hits win_score."""
+        eligible = [player for player in self.players if player.score >= self.config.win_score]
+        if not eligible:
+            return None
+        eligible.sort(
+            key=lambda p: (p.score, -len(p.cards)),
+            reverse=True,
+        )
+        self.finished = True
+        return eligible[0]
+
+
+class Action(Protocol):
+    """Marker protocol for actions."""
+
+
+@dataclass
+class TakeDifferent(Action):
+    """Take up to 3 different gem colors."""
+
+    colors: list[GemColor]
+
+
+@dataclass
+class TakeDouble(Action):
+    """Take two of the same color."""
+
+    color: GemColor
+
+
+@dataclass
+class BuyCard(Action):
+    """Buy a face-up card."""
+
+    tier: int
+    index: int
+    from_reserved: bool = False
+
+
+@dataclass
+class ReserveCard(Action):
+    """Reserve a face-up card."""
+
+    tier: int
+    index: int | None = None
+    from_deck: bool = False
+
+
+class Strategy(Protocol):
+    """Implement this to make a bot or human controller."""
+
+    def __init__(self, name: str) -> None:
+        """Initialize a strategy."""
+        self.name = name
+
+    def choose_action(self, game: GameState, player: PlayerState) -> Action | None:
+        """Return an Action, or None to concede/end."""
+        raise NotImplementedError
+
+    def choose_discard(
+        self,
+        game: GameState,  # noqa: ARG002
+        player: PlayerState,
+        excess: int,
+    ) -> dict[GemColor, int]:
+        """Called if player has more than token_limit tokens after an action.
+
+        Default: naive auto-discard.
+        """
+        return auto_discard_tokens(player, excess)
+
+    def choose_noble(
+        self,
+        game: GameState,  # noqa: ARG002
+        player: PlayerState,  # noqa: ARG002
+        nobles: list[Noble],
+    ) -> Noble:
+        """Called if player qualifies for multiple nobles. Default: first."""
+        return nobles[0]
+
+
+def auto_discard_tokens(player: PlayerState, excess: int) -> dict[GemColor, int]:
+    """Very dumb discard logic: discard from colors you have the most of."""
+    to_discard: dict[GemColor, int] = dict.fromkeys(GEM_COLORS, 0)
+    remaining = excess
+    while remaining > 0:
+        color = max(player.tokens, key=lambda c: player.tokens[c])
+        if player.tokens[color] == 0:
+            break
+        player.tokens[color] -= 1
+        to_discard[color] += 1
+        remaining -= 1
+    return to_discard
+
+
+def enforce_token_limit(
+    game: GameState,
+    strategy: Strategy,
+    player: PlayerState,
+) -> None:
+    """If player has more than token_limit tokens, force discards."""
+    limit = game.config.token_limit
+    total = player.total_tokens()
+    if total <= limit:
+        return
+    excess = total - limit
+    discards = strategy.choose_discard(game, player, excess)
+    for color, amount in discards.items():
+        available = player.tokens[color]
+        to_remove = min(amount, available)
+        if to_remove <= 0:
+            continue
+        player.tokens[color] -= to_remove
+        game.bank[color] += to_remove
+    remaining = player.total_tokens() - limit
+    if remaining > 0:
+        auto = auto_discard_tokens(player, remaining)
+        for color, amount in auto.items():
+            game.bank[color] += amount
+
+
+def check_nobles_for_player(
+    game: GameState,
+    strategy: Strategy,
+    player: PlayerState,
+) -> None:
+    """Award at most one noble to player if they qualify."""
+    if game.noble_min_requirements > max(player.discounts.values()):
+        return
+
+    candidates = [
+        noble
+        for noble in game.available_nobles
+        if all(player.discounts.get(color, 0) >= requirement for color, requirement in noble.requirements.items())
+    ]
+
+    if not candidates:
+        return
+
+    chosen = candidates[0] if len(candidates) == 1 else strategy.choose_noble(game, player, candidates)
+
+    if chosen not in game.available_nobles:
+        return
+    game.available_nobles.remove(chosen)
+    game.get_noble_min_requirements()
+
+    player.nobles.append(chosen)
+
+
+def apply_take_different(game: GameState, strategy: Strategy, action: TakeDifferent) -> None:
+    """Mutate game state according to action."""
+    player = game.current_player
+    max_token_take = 3
+
+    colors = list(dict.fromkeys(action.colors))
+    colors = [c for c in colors if c in BASE_COLORS and game.bank[c] > 0]
+    if not (1 <= len(colors) <= max_token_take):
+        return
+
+    for color in colors:
+        game.bank[color] -= 1
+        player.tokens[color] += 1
+
+    enforce_token_limit(game, strategy, player)
+
+
+def apply_take_double(game: GameState, strategy: Strategy, action: TakeDouble) -> None:
+    """Mutate game state according to action."""
+    player = game.current_player
+    color = action.color
+    if color not in BASE_COLORS:
+        return
+    if game.bank[color] < game.config.minimum_tokens_to_buy_2:
+        return
+    game.bank[color] -= 2
+    player.tokens[color] += 2
+    enforce_token_limit(game, strategy, player)
+
+
+def apply_buy_card(game: GameState, _strategy: Strategy, action: BuyCard) -> None:
+    """Mutate game state according to action."""
+    player = game.current_player
+
+    if action.from_reserved:
+        if not (0 <= action.index < len(player.reserved)):
+            return
+        card = player.reserved[action.index]
+        if not player.can_afford(card):
+            return
+        player.reserved.pop(action.index)
+        payment = player.pay_for_card(card)
+        for color, amount in payment.items():
+            game.bank[color] += amount
+    else:
+        row = game.table_by_tier.get(action.tier)
+        if row is None or not (0 <= action.index < len(row)):
+            return
+        card = row[action.index]
+        if not player.can_afford(card):
+            return
+        row.pop(action.index)
+        payment = player.pay_for_card(card)
+        for color, amount in payment.items():
+            game.bank[color] += amount
+        game.refill_table()
+
+
+def apply_reserve_card(game: GameState, strategy: Strategy, action: ReserveCard) -> None:
+    """Mutate game state according to action."""
+    player = game.current_player
+
+    if len(player.reserved) >= game.config.reserve_limit:
+        return
+
+    card: Card | None = None
+    if action.from_deck:
+        deck = game.decks_by_tier.get(action.tier)
+        if deck:
+            card = deck.pop()
+    else:
+        row = game.table_by_tier.get(action.tier)
+        if row is None:
+            return
+        if action.index is None or not (0 <= action.index < len(row)):
+            return
+        card = row.pop(action.index)
+        game.refill_table()
+
+    if card is None:
+        return
+    player.reserved.append(card)
+
+    if game.bank["gold"] > 0:
+        game.bank["gold"] -= 1
+        player.tokens["gold"] += 1
+        enforce_token_limit(game, strategy, player)
+
+
+def apply_action(game: GameState, strategy: Strategy, action: Action) -> None:
+    """Mutate game state according to action."""
+    actions = {
+        TakeDifferent: apply_take_different,
+        TakeDouble: apply_take_double,
+        BuyCard: apply_buy_card,
+        ReserveCard: apply_reserve_card,
+    }
+    action_func = actions.get(type(action))
+    if action_func is None:
+        msg = f"Unknown action type: {type(action)}"
+        raise ValueError(msg)
+    action_func(game, strategy, action)
+
+
+def legal_actions(
+    game: GameState,
+    player_index: int | None = None,
+) -> list[Action]:
+    """Enumerate all syntactically legal actions for the given player.
+
+    This enforces:
+    - token-taking rules
+    - reserve limits
+    - affordability for buys
+    """
+    if player_index is None:
+        player_index = game.current_player_index
+    player = game.players[player_index]
+
+    actions: list[Action] = []
+
+    colors_available = [c for c in BASE_COLORS if game.bank[c] > 0]
+    for r in (1, 2, 3):
+        actions.extend(TakeDifferent(colors=list(combo)) for combo in itertools.combinations(colors_available, r))
+
+    actions.extend(
+        TakeDouble(color=color) for color in BASE_COLORS if game.bank[color] >= game.config.minimum_tokens_to_buy_2
+    )
+
+    for tier, row in game.table_by_tier.items():
+        for idx, card in enumerate(row):
+            if player.can_afford(card):
+                actions.append(BuyCard(tier=tier, index=idx))
+
+    for idx, card in enumerate(player.reserved):
+        if player.can_afford(card):
+            actions.append(BuyCard(tier=0, index=idx, from_reserved=True))
+
+    if len(player.reserved) < game.config.reserve_limit:
+        for tier, row in game.table_by_tier.items():
+            for idx, _ in enumerate(row):
+                actions.append(
+                    ReserveCard(tier=tier, index=idx, from_deck=False),
+                )
+        for tier, deck in game.decks_by_tier.items():
+            if deck:
+                actions.append(
+                    ReserveCard(tier=tier, index=None, from_deck=True),
+                )
+
+    return actions
+
+
+def create_random_cards_tier(
+    tier: int,
+    card_count: int,
+    cost_choices: list[int],
+    point_choices: list[int],
+) -> list[Card]:
+    """Create a random set of cards for a given tier."""
+    cards: list[Card] = []
+
+    for color in BASE_COLORS:
+        for _ in range(card_count):
+            cost = dict.fromkeys(GEM_COLORS, 0)
+            for c in BASE_COLORS:
+                if c == color:
+                    continue
+                cost[c] = random.choice(cost_choices)
+            points = random.choice(point_choices)
+            cards.append(Card(tier=tier, points=points, color=color, cost=cost))
+
+    return cards
+
+
+def create_random_cards() -> list[Card]:
+    """Generate a generic but Splendor-ish set of cards.
+
+    This is not the official deck, but structured similarly enough for play.
+    """
+    cards: list[Card] = []
+    cards.extend(
+        create_random_cards_tier(
+            tier=1,
+            card_count=5,
+            cost_choices=[0, 1, 1, 2],
+            point_choices=[0, 0, 1],
+        )
+    )
+    cards.extend(
+        create_random_cards_tier(
+            tier=2,
+            card_count=4,
+            cost_choices=[2, 3, 4],
+            point_choices=[1, 2, 2, 3],
+        )
+    )
+    cards.extend(
+        create_random_cards_tier(
+            tier=3,
+            card_count=3,
+            cost_choices=[4, 5, 6],
+            point_choices=[3, 4, 5],
+        )
+    )
+
+    random.shuffle(cards)
+    return cards
+
+
+def create_random_nobles() -> list[Noble]:
+    """A small set of noble tiles, roughly Splendor-ish."""
+    nobles: list[Noble] = []
+
+    base_requirements: list[dict[GemColor, int]] = [
+        {"white": 3, "blue": 3, "green": 3},
+        {"blue": 3, "green": 3, "red": 3},
+        {"green": 3, "red": 3, "black": 3},
+        {"red": 3, "black": 3, "white": 3},
+        {"black": 3, "white": 3, "blue": 3},
+        {"white": 4, "blue": 4},
+        {"green": 4, "red": 4},
+        {"blue": 4, "black": 4},
+    ]
+
+    for idx, req in enumerate(base_requirements, start=1):
+        nobles.append(
+            Noble(
+                name=f"Noble {idx}",
+                points=3,
+                requirements=dict(req.items()),
+            ),
+        )
+    return nobles
+
+
+def new_game(
+    strategies: Sequence[Strategy],
+    config: GameConfig,
+) -> GameState:
+    """Create a new game state from a config + list of players."""
+    num_players = len(strategies)
+    bank = get_default_starting_tokens(num_players)
+
+    decks_by_tier: dict[int, list[Card]] = {1: [], 2: [], 3: []}
+    for card in config.cards:
+        decks_by_tier.setdefault(card.tier, []).append(card)
+    for deck in decks_by_tier.values():
+        random.shuffle(deck)
+
+    table_by_tier: dict[int, list[Card]] = {1: [], 2: [], 3: []}
+    players = [PlayerState(strategy=strategy) for strategy in strategies]
+
+    nobles = list(config.nobles)
+    random.shuffle(nobles)
+    nobles = nobles[: num_players + 1]
+
+    game = GameState(
+        config=config,
+        players=players,
+        bank=bank,
+        decks_by_tier=decks_by_tier,
+        table_by_tier=table_by_tier,
+        available_nobles=nobles,
+    )
+    game.refill_table()
+    return game
+
+
+def run_game(game: GameState) -> tuple[PlayerState, int]:
+    """Run a full game loop until someone wins or a player returns None."""
+    turn_count = 0
+    while not game.finished:
+        turn_count += 1
+        player = game.current_player
+        strategy = player.strategy
+        action = strategy.choose_action(game, player)
+        if action is None:
+            game.finished = True
+            break
+
+        apply_action(game, strategy, action)
+        check_nobles_for_player(game, strategy, player)
+
+        winner = game.check_winner_simple()
+        if winner is not None:
+            return winner, turn_count
+
+        game.next_player()
+        if turn_count >= game.config.turn_limit:
+            break
+
+    fallback = max(game.players, key=lambda player: player.score)
+    return fallback, turn_count

python/splendor/bot.py

@@ -0,0 +1,58 @@
+from __future__ import annotations
+
+import random
+
+from .base import (
+    BASE_COLORS,
+    Action,
+    BuyCard,
+    GameState,
+    GemColor,
+    PlayerState,
+    ReserveCard,
+    Strategy,
+    TakeDifferent,
+    TakeDouble,
+    auto_discard_tokens,
+)
+
+
+class RandomBot(Strategy):
+    """Dumb bot that follows rules but doesn't think."""
+
+    def __init__(self, name: str = "Bot") -> None:
+        super().__init__(name=name)
+
+    def choose_action(self, game: GameState, player: PlayerState) -> Action | None:
+        affordable: list[tuple[int, int]] = []
+        for tier, row in game.table_by_tier.items():
+            for idx, card in enumerate(row):
+                if player.can_afford(card):
+                    affordable.append((tier, idx))
+        if affordable and random.random() < 0.5:
+            tier, idx = random.choice(affordable)
+            return BuyCard(tier=tier, index=idx)
+
+        if random.random() < 0.2:
+            tier = random.choice([1, 2, 3])
+            row = game.table_by_tier.get(tier, [])
+            if row:
+                idx = random.randrange(len(row))
+                return ReserveCard(tier=tier, index=idx, from_deck=False)
+
+        if random.random() < 0.5:
+            colors_for_double = [c for c in BASE_COLORS if game.bank[c] >= 4]
+            if colors_for_double:
+                return TakeDouble(color=random.choice(colors_for_double))
+
+        colors_for_diff = [c for c in BASE_COLORS if game.bank[c] > 0]
+        random.shuffle(colors_for_diff)
+        return TakeDifferent(colors=colors_for_diff[:3])
+
+    def choose_discard(
+        self,
+        game: GameState,
+        player: PlayerState,
+        excess: int,
+    ) -> dict[GemColor, int]:
+        return auto_discard_tokens(player, excess)

python/splendor/human.py

@@ -0,0 +1,686 @@
+from __future__ import annotations
+
+import sys
+from collections.abc import Mapping
+from typing import Any
+
+from textual.app import App, ComposeResult
+from textual.containers import Horizontal, Vertical
+from textual.widget import Widget
+from textual.widgets import Footer, Header, Input, Static
+
+from .base import (
+    BASE_COLORS,
+    GEM_COLORS,
+    Action,
+    BuyCard,
+    Card,
+    GameState,
+    GemColor,
+    Noble,
+    PlayerState,
+    ReserveCard,
+    Strategy,
+    TakeDifferent,
+    TakeDouble,
+)
+
+# Abbreviations used when rendering costs
+COST_ABBR: dict[GemColor, str] = {
+    "white": "W",
+    "blue": "B",
+    "green": "G",
+    "red": "R",
+    "black": "K",
+    "gold": "O",
+}
+
+# Abbreviations players can type on the command line
+COLOR_ABBR_TO_FULL: dict[str, GemColor] = {
+    "w": "white",
+    "b": "blue",
+    "g": "green",
+    "r": "red",
+    "k": "black",
+    "o": "gold",
+}
+
+
+def parse_color_token(raw: str) -> GemColor:
+    """Convert user input into a GemColor.
+
+    Supports:
+      - full names:  white, blue, green, red, black, gold
+      - abbreviations: w, b, g, r, k, o
+    """
+    key = raw.lower()
+
+    # full color names first
+    if key in BASE_COLORS:
+        return key  # type: ignore[return-value]
+
+    # abbreviations
+    if key in COLOR_ABBR_TO_FULL:
+        return COLOR_ABBR_TO_FULL[key]
+
+    raise ValueError(f"Unknown color: {raw}")
+
+
+def format_cost(cost: Mapping[GemColor, int]) -> str:
+    """Format a cost/requirements dict as colored tokens like 'B:2, R:1'.
+
+    Uses `color_token` internally so colors are guaranteed to match your bank.
+    """
+    parts: list[str] = []
+    for color in GEM_COLORS:
+        n = cost.get(color, 0)
+        if not n:
+            continue
+
+        # color_token gives us e.g. "[blue]blue: 3[/]"
+        token = color_token(color, n)
+
+        # Turn the leading color name into the abbreviation (blue: 3 → B:3)
+        # We only replace the first occurrence.
+        full = f"{color}:"
+        abbr = f"{COST_ABBR[color]}:"
+        token = token.replace(full, abbr, 1)
+
+        parts.append(token)
+
+    return ", ".join(parts) if parts else "-"
+
+
+def format_card(card: Card) -> str:
+    """Readable card line using dataclass fields instead of __str__."""
+    color_abbr = COST_ABBR[card.color]
+    header = f"T{card.tier} {color_abbr} P{card.points}"
+    cost_str = format_cost(card.cost)
+    return f"{header} ({cost_str})"
+
+
+def format_noble(noble: Noble) -> str:
+    """Readable noble line using dataclass fields instead of __str__."""
+    cost_str = format_cost(noble.requirements)
+    return f"{noble.name} +{noble.points} ({cost_str})"
+
+
+def format_tokens(tokens: Mapping[GemColor, int]) -> str:
+    """Colored 'color: n' list for a token dict."""
+    return " ".join(color_token(c, tokens.get(c, 0)) for c in GEM_COLORS)
+
+
+def format_discounts(discounts: Mapping[GemColor, int]) -> str:
+    """Colored discounts, skipping zeros."""
+    parts: list[str] = []
+    for c in GEM_COLORS:
+        n = discounts.get(c, 0)
+        if not n:
+            continue
+        abbr = COST_ABBR[c]
+        fg, bg = COLOR_STYLE[c]
+        parts.append(f"[{fg} on {bg}]{abbr}:{n}[/{fg} on {bg}]")
+    return ", ".join(parts) if parts else "-"
+
+
+COLOR_STYLE: dict[GemColor, tuple[str, str]] = {
+    "white": ("black", "white"),  # fg, bg
+    "blue": ("bright_white", "blue"),
+    "green": ("bright_white", "sea_green4"),
+    "red": ("white", "red3"),
+    "black": ("white", "grey0"),
+    "gold": ("black", "yellow3"),
+}
+
+
+def fmt_gem(color: GemColor) -> str:
+    """Render gem name with fg/bg matching real token color."""
+    fg, bg = COLOR_STYLE[color]
+    return f"[{fg} on {bg}] {color} [/{fg} on {bg}]"
+
+
+def fmt_number(value: int) -> str:
+    return f"[bold cyan]{value}[/]"
+
+
+def color_token(name: GemColor, amount: int) -> str:
+    """Return a Rich-markup colored 'name: n' string."""
+    # Map Splendor colors -> terminal colors
+    color_map: Mapping[GemColor, str] = {
+        "white": "white",
+        "blue": "blue",
+        "green": "green",
+        "red": "red",
+        "black": "grey70",  # 'black' is unreadable on dark backgrounds
+        "gold": "yellow",
+    }
+    style = color_map.get(name, "white")
+    return f"[{style}]{name}: {amount}[/]"
+
+
+class Board(Widget):
+    """Big board widget with the layout you sketched."""
+
+    def __init__(self, game: GameState, me: PlayerState, **kwargs: Any) -> None:
+        super().__init__(**kwargs)
+        self.game = game
+        self.me = me
+
+    def compose(self) -> ComposeResult:
+        # Structure:
+        # ┌ bank row
+        # ├ middle row (tiers | nobles)
+        # └ players row
+        with Vertical(id="board_root"):
+            yield Static(id="bank_box")
+            with Horizontal(id="middle_row"):
+                with Vertical(id="tiers_box"):
+                    yield Static(id="tier1_box")
+                    yield Static(id="tier2_box")
+                    yield Static(id="tier3_box")
+                yield Static(id="nobles_box")
+            yield Static(id="players_box")
+
+    def on_mount(self) -> None:
+        self.refresh_content()
+
+    def refresh_content(self) -> None:
+        self._render_bank()
+        self._render_tiers()
+        self._render_nobles()
+        self._render_players()
+
+    # --- sections ----------------------------------------------------
+
+    def _render_bank(self) -> None:
+        bank = self.game.bank
+        parts: list[str] = ["[b]Bank:[/b]"]
+        # One line, all tokens colored
+        parts.append(format_tokens(bank))
+        self.query_one("#bank_box", Static).update("\n".join(parts))
+
+    def _render_tiers(self) -> None:
+        for tier in (1, 2, 3):
+            box = self.query_one(f"#tier{tier}_box", Static)
+            cards: list[Card] = self.game.table_by_tier.get(tier, [])
+            lines: list[str] = [f"[b]Tier {tier} cards:[/b]"]
+            if not cards:
+                lines.append("  (none)")
+            else:
+                for idx, card in enumerate(cards):
+                    lines.append(f"  [{idx}] {format_card(card)}")
+            box.update("\n".join(lines))
+
+    def _render_nobles(self) -> None:
+        nobles_box = self.query_one("#nobles_box", Static)
+        lines: list[str] = ["[b]Nobles[/b]"]
+        if not self.game.available_nobles:
+            lines.append("  (none)")
+        else:
+            for noble in self.game.available_nobles:
+                lines.append("  - " + format_noble(noble))
+        nobles_box.update("\n".join(lines))
+
+    def _render_players(self) -> None:
+        players_box = self.query_one("#players_box", Static)
+        lines: list[str] = ["[b]Players:[/b]", ""]
+        for player in self.game.players:
+            mark = "*" if player is self.me else " "
+            token_str = format_tokens(player.tokens)
+            discount_str = format_discounts(player.discounts)
+
+            lines.append(
+                f"{mark} {player.name:10}  Score={player.score:2d}  Discounts={discount_str}",
+            )
+            lines.append(f"    Tokens: {token_str}")
+
+            if player.nobles:
+                noble_names = ", ".join(n.name for n in player.nobles)
+                lines.append(f"    Nobles: {noble_names}")
+
+            # Optional: show counts of cards / reserved
+            if player.cards:
+                lines.append(f"    Cards: {len(player.cards)}")
+            if player.reserved:
+                lines.append(f"    Reserved: {len(player.reserved)}")
+
+            lines.append("")
+        players_box.update("\n".join(lines))
+
+
+class ActionApp(App[None]):
+    """Textual app that asks for a single action command and returns an Action."""
+
+    CSS = """
+    Screen {
+        /* 3 rows: command zone, board, footer */
+        layout: grid;
+        grid-size: 1 3;
+        grid-rows: auto 1fr auto;
+    }
+
+    /* Top area with input + instructions */
+    #command_zone {
+        grid-columns: 1;
+        grid-rows: 1;
+        padding: 1 1;
+    }
+
+    /* Board sits in the middle row and can grow */
+    #board {
+        grid-columns: 1;
+        grid-rows: 2;
+        padding: 0 1 1 1;
+    }
+
+    Footer {
+        grid-columns: 1;
+        grid-rows: 3;
+    }
+
+    Input {
+        border: round $accent;
+    }
+
+    /* === Board layout === */
+
+    #board_root {
+        /* outer frame around the whole board area */
+        border: heavy white;
+        padding: 0 1;
+    }
+
+    /* Bank row: full width */
+    #bank_box {
+        border: heavy white;
+        padding: 0 1;
+    }
+
+    /* Middle row: tiers (left) + nobles (right) */
+    #middle_row {
+        layout: horizontal;
+    }
+
+    #tiers_box {
+        border: heavy white;
+        padding: 0 1;
+        width: 70%;
+    }
+
+    #tier1_box,
+    #tier2_box,
+    #tier3_box {
+        border-bottom: heavy white;
+        padding: 0 0 1 0;
+        margin-bottom: 1;
+    }
+
+    #nobles_box {
+        border: heavy white;
+        padding: 0 1;
+        width: 30%;
+    }
+
+    /* Players row: full width at bottom */
+    #players_box {
+        border: heavy white;
+        padding: 0 1;
+    }
+    """
+
+    def __init__(self, game: GameState, player: PlayerState) -> None:
+        super().__init__()
+        self.game = game
+        self.player = player
+        self.result: Action | None = None
+        self.message: str = ""
+
+    def compose(self) -> ComposeResult:  # type: ignore[override]
+        # Row 1: input + Actions text
+        with Vertical(id="command_zone"):
+            yield Input(
+                placeholder="Enter command, e.g. '1 white blue red' or '1 w b r' or 'q'",
+                id="input_line",
+            )
+            yield Static("", id="prompt")
+
+        # Row 2: board
+        yield Board(self.game, self.player, id="board")
+
+        # Row 3: footer
+        yield Footer()
+
+    def on_mount(self) -> None:  # type: ignore[override]
+        self._update_prompt()
+        self.query_one(Input).focus()
+
+    def _update_prompt(self) -> None:
+        lines: list[str] = []
+        lines.append("[bold underline]Actions:[/]")
+        lines.append(
+            " [bold green]1[/] <colors...>  - Take up to 3 different gem colors "
+            "(e.g. [cyan]1 white blue red[/] or [cyan]1 w b r[/])",
+        )
+        lines.append(
+            f" [bold green]2[/] <color>      - Take 2 of the same color (needs {fmt_number(4)} in bank, "
+            "e.g. [cyan]2 blue[/] or [cyan]2 b[/])",
+        )
+        lines.append(
+            " [bold green]3[/] <tier> <idx> - Buy a face-up card (e.g. [cyan]3 1 0[/] for tier 1, index 0)",
+        )
+        lines.append(" [bold green]4[/] <idx>        - Buy a reserved card")
+        lines.append(" [bold green]5[/] <tier> <idx> - Reserve a face-up card")
+        lines.append(" [bold green]6[/] <tier>       - Reserve top card of a deck")
+        lines.append(" [bold red]q[/]                - Quit game")
+        if self.message:
+            lines.append("")
+            lines.append(f"[bold red]Message:[/] {self.message}")
+        self.query_one("#prompt", Static).update("\n".join(lines))
+
+    def on_input_submitted(self, event: Input.Submitted) -> None:  # type: ignore[override]
+        text = (event.value or "").strip()
+        event.input.value = ""
+        if not text:
+            return
+        if text.lower() in {"q", "quit", "0"}:
+            self.result = None
+            self.exit()
+            return
+
+        parts = text.split()
+        cmd = parts[0]
+
+        try:
+            if cmd == "1":
+                # Take up to 3 different gem colors: 1 white blue red  OR  1 w b r
+                color_names = parts[1:]
+                if not color_names:
+                    raise ValueError("Need at least one color (full name or abbreviation).")
+                colors: list[GemColor] = []
+                for name in color_names:
+                    color = parse_color_token(name)
+                    if self.game.bank[color] <= 0:
+                        raise ValueError(f"No tokens left for color: {color}")
+                    colors.append(color)
+                self.result = TakeDifferent(colors=colors[:3])
+                self.exit()
+                return
+
+            if cmd == "2":
+                # TakeDouble: 2 color  (full name or abbreviation)
+                if len(parts) < 2:
+                    raise ValueError("Usage: 2 <color>")
+                raw_color = parts[1]
+                color = parse_color_token(raw_color)
+                if self.game.bank[color] < 4:
+                    raise ValueError("Bank must have at least 4 of that color.")
+                self.result = TakeDouble(color=color)
+                self.exit()
+                return
+
+            if cmd == "3":
+                # Buy face-up card: 3 tier index
+                if len(parts) < 3:
+                    raise ValueError("Usage: 3 <tier> <index>")
+                tier = int(parts[1])
+                idx = int(parts[2])
+                self.result = BuyCard(tier=tier, index=idx)
+                self.exit()
+                return
+
+            if cmd == "4":
+                # Buy reserved card: 4 index
+                if len(parts) < 2:
+                    raise ValueError("Usage: 4 <reserved_index>")
+                idx = int(parts[1])
+                if not (0 <= idx < len(self.player.reserved)):
+                    raise ValueError("Reserved index out of range.")
+                self.result = BuyCard(tier=0, index=idx, from_reserved=True)
+                self.exit()
+                return
+
+            if cmd == "5":
+                # Reserve face-up card: 5 tier index
+                if len(parts) < 3:
+                    raise ValueError("Usage: 5 <tier> <index>")
+                tier = int(parts[1])
+                idx = int(parts[2])
+                self.result = ReserveCard(tier=tier, index=idx, from_deck=False)
+                self.exit()
+                return
+
+            if cmd == "6":
+                # Reserve top of deck: 6 tier
+                if len(parts) < 2:
+                    raise ValueError("Usage: 6 <tier>")
+                tier = int(parts[1])
+                self.result = ReserveCard(tier=tier, index=None, from_deck=True)
+                self.exit()
+                return
+
+            raise ValueError("Unknown command.")
+
+        except ValueError as exc:
+            self.message = str(exc)
+            self._update_prompt()
+            return
+
+
+class DiscardApp(App[None]):
+    """Textual app to choose discards when over token limit."""
+
+    CSS = """
+    Screen {
+        layout: vertical;
+    }
+
+    #command_zone {
+        padding: 1 1;
+    }
+
+    #board {
+        padding: 0 1 1 1;
+    }
+
+    Input {
+        border: round $accent;
+    }
+    """
+
+    def __init__(self, game: GameState, player: PlayerState) -> None:
+        super().__init__()
+        self.game = game
+        self.player = player
+        self.discards: dict[GemColor, int] = dict.fromkeys(GEM_COLORS, 0)
+        self.message: str = ""
+
+    def compose(self) -> ComposeResult:  # type: ignore[override]
+        yield Header(show_clock=False)
+
+        with Vertical(id="command_zone"):
+            yield Input(
+                placeholder="Enter color to discard, e.g. 'blue' or 'b'",
+                id="input_line",
+            )
+            yield Static("", id="prompt")
+
+        # Board directly under the command zone
+        yield Board(self.game, self.player, id="board")
+
+        yield Footer()
+
+    def on_mount(self) -> None:  # type: ignore[override]
+        self._update_prompt()
+        self.query_one(Input).focus()
+
+    def _remaining_to_discard(self) -> int:
+        return self.player.total_tokens() - sum(self.discards.values()) - self.game.config.token_limit
+
+    def _update_prompt(self) -> None:
+        remaining = max(self._remaining_to_discard(), 0)
+        lines: list[str] = []
+        lines.append(
+            "You must discard "
+            f"{fmt_number(remaining)} token(s) "
+            f"to get down to {fmt_number(self.game.config.token_limit)}.",
+        )
+        disc_str = ", ".join(f"{fmt_gem(c)}={fmt_number(self.discards[c])}" for c in GEM_COLORS)
+        lines.append(f"Current planned discards: {{ {disc_str} }}")
+        lines.append(
+            "Type a color name or abbreviation (e.g. 'blue' or 'b') to discard one token.",
+        )
+        if self.message:
+            lines.append("")
+            lines.append(f"[bold red]Message:[/] {self.message}")
+        self.query_one("#prompt", Static).update("\n".join(lines))
+
+    def on_input_submitted(self, event: Input.Submitted) -> None:  # type: ignore[override]
+        raw = (event.value or "").strip()
+        event.input.value = ""
+        if not raw:
+            return
+
+        try:
+            color = parse_color_token(raw)
+        except ValueError:
+            self.message = f"Unknown color: {raw}"
+            self._update_prompt()
+            return
+
+        available = self.player.tokens[color] - self.discards[color]
+        if available <= 0:
+            self.message = f"No more {color} tokens available to discard."
+            self._update_prompt()
+            return
+
+        self.discards[color] += 1
+        if self._remaining_to_discard() <= 0:
+            self.exit()
+            return
+
+        self.message = ""
+        self._update_prompt()
+
+
+# ---------------------------------------------------------------------------
+# Noble choice app
+# ---------------------------------------------------------------------------
+
+
+class NobleChoiceApp(App[None]):
+    """Textual app to choose one noble."""
+
+    CSS = """
+    Screen {
+        layout: vertical;
+    }
+
+    #command_zone {
+        padding: 1 1;
+    }
+
+    #board {
+        padding: 0 1 1 1;
+    }
+
+    Input {
+        border: round $accent;
+    }
+    """
+
+    def __init__(
+        self,
+        game: GameState,
+        player: PlayerState,
+        nobles: list[Noble],
+    ) -> None:
+        super().__init__()
+        self.game = game
+        self.player = player
+        self.nobles = nobles
+        self.result: Noble | None = None
+        self.message: str = ""
+
+    def compose(self) -> ComposeResult:  # type: ignore[override]
+        yield Header(show_clock=False)
+
+        with Vertical(id="command_zone"):
+            yield Input(
+                placeholder="Enter noble index, e.g. '0'",
+                id="input_line",
+            )
+            yield Static("", id="prompt")
+
+        # Board directly under the command zone
+        yield Board(self.game, self.player, id="board")
+
+        yield Footer()
+
+    def on_mount(self) -> None:  # type: ignore[override]
+        self._update_prompt()
+        self.query_one(Input).focus()
+
+    def _update_prompt(self) -> None:
+        lines: list[str] = []
+        lines.append("[bold underline]You qualify for nobles:[/]")
+        for i, noble in enumerate(self.nobles):
+            lines.append(f" [bright_cyan]{i})[/] {format_noble(noble)}")
+        lines.append("Enter the index of the noble you want.")
+        if self.message:
+            lines.append("")
+            lines.append(f"[bold red]Message:[/] {self.message}")
+        self.query_one("#prompt", Static).update("\n".join(lines))
+
+    def on_input_submitted(self, event: Input.Submitted) -> None:  # type: ignore[override]
+        raw = (event.value or "").strip()
+        event.input.value = ""
+        if not raw:
+            return
+        try:
+            idx = int(raw)
+        except ValueError:
+            self.message = "Please enter a valid integer index."
+            self._update_prompt()
+            return
+        if not (0 <= idx < len(self.nobles)):
+            self.message = "Index out of range."
+            self._update_prompt()
+            return
+        self.result = self.nobles[idx]
+        self.exit()
+
+
+class TuiHuman(Strategy):
+    """Textual-based human player Strategy with colorful board."""
+
+    def choose_action(self, game: GameState, player: PlayerState) -> Action | None:
+        if not sys.stdout.isatty():
+            return None
+        app = ActionApp(game, player)
+        app.run()
+        return app.result
+
+    def choose_discard(
+        self,
+        game: GameState,
+        player: PlayerState,
+        excess: int,  # noqa: ARG002
+    ) -> dict[GemColor, int]:
+        if not sys.stdout.isatty():
+            return dict.fromkeys(GEM_COLORS, 0)
+        app = DiscardApp(game, player)
+        app.run()
+        return app.discards
+
+    def choose_noble(
+        self,
+        game: GameState,
+        player: PlayerState,
+        nobles: list[Noble],
+    ) -> Noble:
+        if not sys.stdout.isatty():
+            return nobles[0]
+        app = NobleChoiceApp(game, player, nobles)
+        app.run()
+        assert app.result is not None
+        return app.result

python/splendor/main.py

@@ -0,0 +1,17 @@
+from __future__ import annotations
+
+from .base import new_game, run_game
+from .bot import RandomBot
+from .human import TuiHuman
+
+
+def main() -> None:
+    """Main entry point."""
+    human = TuiHuman()
+    bot = RandomBot()
+    game_state = new_game(["You", "Bot A"])
+    run_game(game_state, [human, bot])
+
+
+if __name__ == "__main__":
+    main()

python/splendor/public_state.py

@@ -0,0 +1,103 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+from .base import (
+    BASE_COLORS,
+    BASE_INDEX,
+    GEM_ORDER,
+    Card,
+    GameState,
+    Noble,
+    PlayerState,
+)
+
+
+@dataclass(frozen=True)
+class ObsCard:
+    """Numeric-ish card view for RL/search."""
+
+    tier: int
+    points: int
+    color_index: int
+    cost: list[int]
+
+
+@dataclass(frozen=True)
+class ObsNoble:
+    points: int
+    requirements: list[int]
+
+
+@dataclass(frozen=True)
+class ObsPlayer:
+    tokens: list[int]
+    discounts: list[int]
+    score: int
+    cards: list[ObsCard]
+    reserved: list[ObsCard]
+    nobles: list[ObsNoble]
+
+
+@dataclass(frozen=True)
+class Observation:
+    current_player: int
+    bank: list[int]
+    players: list[ObsPlayer]
+    table_by_tier: dict[int, list[ObsCard]]
+    decks_remaining: dict[int, int]
+    available_nobles: list[ObsNoble]
+
+
+def _encode_card(card: Card) -> ObsCard:
+    color_index = BASE_INDEX.get(card.color, -1)
+    cost_vec = [card.cost.get(c, 0) for c in BASE_COLORS]
+    return ObsCard(
+        tier=card.tier,
+        points=card.points,
+        color_index=color_index,
+        cost=cost_vec,
+    )
+
+
+def _encode_noble(noble: Noble) -> ObsNoble:
+    req_vec = [noble.requirements.get(c, 0) for c in BASE_COLORS]
+    return ObsNoble(
+        points=noble.points,
+        requirements=req_vec,
+    )
+
+
+def _encode_player(player: PlayerState) -> ObsPlayer:
+    tokens_vec = [player.tokens[c] for c in GEM_ORDER]
+    discounts_vec = [player.discounts[c] for c in GEM_ORDER]
+    cards_enc = [_encode_card(c) for c in player.cards]
+    reserved_enc = [_encode_card(c) for c in player.reserved]
+    nobles_enc = [_encode_noble(n) for n in player.nobles]
+    return ObsPlayer(
+        tokens=tokens_vec,
+        discounts=discounts_vec,
+        score=player.score,
+        cards=cards_enc,
+        reserved=reserved_enc,
+        nobles=nobles_enc,
+    )
+
+
+def to_observation(game: GameState) -> Observation:
+    """Create a structured observation of the full public state."""
+    bank_vec = [game.bank[c] for c in GEM_ORDER]
+    players_enc = [_encode_player(p) for p in game.players]
+    table_enc: dict[int, list[ObsCard]] = {
+        tier: [_encode_card(c) for c in row] for tier, row in game.table_by_tier.items()
+    }
+    decks_remaining = {tier: len(deck) for tier, deck in game.decks_by_tier.items()}
+    nobles_enc = [_encode_noble(n) for n in game.available_nobles]
+    return Observation(
+        current_player=game.current_player_index,
+        bank=bank_vec,
+        players=players_enc,
+        table_by_tier=table_enc,
+        decks_remaining=decks_remaining,
+        available_nobles=nobles_enc,
+    )

python/splendor/sim.py

@@ -0,0 +1,33 @@
+from __future__ import annotations
+
+import copy
+
+from .base import Action, GameState, PlayerState, apply_action, check_nobles_for_player
+from .bot import RandomBot
+
+
+class SimStrategy(RandomBot):
+    """Strategy used in simulate_step.
+
+    We never call choose_action here (caller chooses actions),
+    but we reuse discard/noble-selection logic.
+    """
+
+    def choose_action(self, game: GameState, player: PlayerState) -> Action | None:
+        msg = "SimStrategy.choose_action should not be used in simulate_step"
+        raise RuntimeError(msg)
+
+
+def simulate_step(game: GameState, action: Action) -> GameState:
+    """Return a deep-copied next state after applying action for the current player.
+
+    Useful for tree search / MCTS:
+
+        next_state = simulate_step(state, action)
+    """
+    next_state = copy.deepcopy(game)
+    sim_strategy = SimStrategy()
+    apply_action(next_state, sim_strategy, action)
+    check_nobles_for_player(next_state, sim_strategy, next_state.current_player)
+    next_state.next_player()
+    return next_state

python/splendor/simulat.py

@@ -0,0 +1,27 @@
+from __future__ import annotations
+
+from .base import GameConfig, create_random_nobles, create_random_cards, new_game, run_game
+from .bot import RandomBot
+
+
+def main() -> None:
+    """Main entry point."""
+    turn_limit = 10000
+    for _ in range(1000):
+        bot_a = RandomBot("bot_a")
+        bot_b = RandomBot("bot_b")
+        bot_c = RandomBot("bot_c")
+        bot_d = RandomBot("bot_d")
+        config = GameConfig(
+            cards=create_random_cards(),
+            nobles=create_random_nobles(),
+            turn_limit=turn_limit,
+        )
+        players = (bot_a, bot_b, bot_c, bot_d)
+        game_state = new_game(players, config)
+        winner, turns = run_game(game_state)
+        print(f"Winner is {winner.strategy.name} with {winner.score} points after {turns} turns.")
+
+
+if __name__ == "__main__":
+    main()