1

1

src/engine/dizhu/dizhu_engine.py

@@ -36,23 +36,28 @@ class DiZhuEngine:
 
     def step(self, action):
         """
-        执行玩家的出牌动作。
-        :param action: 当前玩家的动作（出牌或过牌）
+        执行动作并更新状态
+        :param action: 当前玩家的动作（可以是 'pass' 或一个动作列表）
         """
         current_player = self.get_current_player()
 
         if action == "pass":
-            # 玩家选择过牌
-            current_player.history.append("pass")
+            # 过牌作为单独的历史记录
+            current_player.history.append([])
         else:
-            # 玩家出牌，移除对应牌
+            # 确保动作始终为列表
+            if not isinstance(action, list):
+                action = [action]
+            # 检查动作合法性
             if not all(card in current_player.hand_cards for card in action):
-                raise ValueError("玩家手牌不足以完成此次出牌")
+                raise ValueError(f"玩家手牌不足以完成此次出牌: {action}")
+            # 移除出牌
             for card in action:
                 current_player.hand_cards.remove(card)
+            # 记录动作
             current_player.history.append(action)
 
-        # 检查是否游戏结束
+        # 检查游戏是否结束
         if not current_player.hand_cards:
             self.game_over = True
             return f"{current_player.role} 胜利！"
@@ -60,6 +65,7 @@ class DiZhuEngine:
         # 切换到下一个玩家
         self.current_player_index = (self.current_player_index + 1) % 3
 
+
     def get_game_state(self):
         """
         返回当前游戏状态，包括玩家手牌、出牌历史和当前玩家。

src/engine/dizhu/player_state.py

@@ -1,5 +1,7 @@
+from collections import deque
+
 class PlayerState:
     def __init__(self, hand_cards, role):
         self.hand_cards = hand_cards  # 玩家手牌
         self.role = role  # "地主" 或 "农民"
-        self.history = []  # 出牌历史
+        self.history = deque()  # 出牌历史，使用 deque

src/environment/dizhu_env.py

@@ -40,14 +40,22 @@ class DouDiZhuEnv(gym.Env):
     def _get_observation(self):
         """获取当前玩家的状态"""
         current_player = self.engine.get_current_player()
+
+        # 手牌的独热编码
         hand_cards = np.zeros(54, dtype=np.int32)
         for card in current_player.hand_cards:
             hand_cards[card] = 1
 
+        # 出牌历史的独热编码
         history = np.zeros(54, dtype=np.int32)
         for play in current_player.history:
-            for card in play:
-                history[card] = 1
+            if isinstance(play, list):  # 如果是列表
+                for card in play:
+                    history[card] = 1
+            elif isinstance(play, int):  # 如果是单个牌
+                history[play] = 1
+            else:
+                raise ValueError(f"历史记录格式错误: {play}")
 
         return {"hand_cards": hand_cards, "history": history}