From 9d95edfa116bbfc54e7e894209f7fbe2c49098f1 Mon Sep 17 00:00:00 2001
From: wsy182 <2392948297@qq.com>
Date: Sat, 30 Nov 2024 20:02:04 +0800
Subject: [PATCH] 1

1
---
 scripts/train_chengdu_mahjong_model.py |  2 +-
 src/engine/actions.py                  | 27 +++++++++++++++++---------
 src/engine/game_state.py               |  2 +-
 test.py                                |  2 ++
 4 files changed, 22 insertions(+), 11 deletions(-)
 create mode 100644 test.py

diff --git a/scripts/train_chengdu_mahjong_model.py b/scripts/train_chengdu_mahjong_model.py
index 4ad8138..7afbafb 100644
--- a/scripts/train_chengdu_mahjong_model.py
+++ b/scripts/train_chengdu_mahjong_model.py
@@ -7,7 +7,7 @@ def train_model():
     env = MahjongEnv()
 
     # 使用 PPO 算法训练模型
-    model = PPO("MlpPolicy", env, verbose=1, tensorboard_log="./ppo_mahjong_tensorboard/")
+    model = PPO("MlpPolicy", env, verbose=1, tensorboard_log="../logs/ppo_mahjong_tensorboard/")
 
     # 训练模型，训练总步数为100000
     model.learn(total_timesteps=100000)
diff --git a/src/engine/actions.py b/src/engine/actions.py
index 2659435..f1ccbbe 100644
--- a/src/engine/actions.py
+++ b/src/engine/actions.py
@@ -2,24 +2,26 @@ from loguru import logger
 from src.engine.utils import get_tile_name
 
 
-def draw_tile(self):
+def draw_tile(engine):
     """
     当前玩家摸牌逻辑，记录牌的详细信息和游戏状态。
     """
-    if self.state.remaining_tiles == 0:
+    if engine.state.remaining_tiles == 0:
         logger.warning("牌堆已空，游戏结束！")
-        self.game_over = True
-        return "牌堆已空"
+        engine.game_over = True
+        return 0, True  # 游戏结束时返回 0 和 done = True
 
-    tile = self.state.deck.pop(0)  # 从牌堆中取出一张牌
-    self.state.remaining_tiles -= 1  # 更新剩余牌数
-    self.state.hands[self.state.current_player][tile] += 1  # 加入当前玩家手牌
+    tile = engine.state.deck.pop(0)  # 从牌堆中取出一张牌
+    engine.state.remaining_tiles -= 1  # 更新剩余牌数
+    engine.state.hands[engine.state.current_player][tile] += 1  # 加入当前玩家手牌
 
     tile_name = get_tile_name(tile)  # 获取具体的牌名
     logger.info(
-        f"玩家 {self.state.current_player} 摸到一张牌: {tile_name}（索引 {tile}）。剩余牌堆数量: {self.state.remaining_tiles}"
+        f"玩家 {engine.state.current_player} 摸到一张牌: {tile_name}（索引 {tile}）。剩余牌堆数量: {engine.state.remaining_tiles}"
     )
-    return tile
+
+    # 返回奖励和游戏是否结束的标志
+    return 0, False  # 奖励为 0，done 为 False（游戏继续）
 
 
 def discard_tile(self, tile):
@@ -66,15 +68,22 @@ def gang(self, tile, mode):
         self.state.hands[player][tile] -= 3
         self.state.melds[player].append(("ming_gang", tile))
         logger.info(f"玩家 {player} 明杠: {tile_name}（索引 {tile}）")
+        self.state.scores[player] += 1  # 奖励1分
+        logger.info(f"玩家 {player} 因明杠获得1分")
+
     elif mode == "an" and self.state.hands[player][tile] == 4:
         self.state.hands[player][tile] -= 4
         self.state.melds[player].append(("an_gang", tile))
         logger.info(f"玩家 {player} 暗杠: {tile_name}（索引 {tile}）")
+        self.state.scores[player] += 1  # 奖励1分
+        logger.info(f"玩家 {player} 因暗杠获得1分")
+
     else:
         logger.error(f"玩家 {player} 尝试杠牌失败: {tile_name}（索引 {tile}），条件不满足")
         raise ValueError("杠牌条件不满足")
 
 
+
 def check_blood_battle(self):
     """
     检查游戏是否流局或血战结束，记录状态。
diff --git a/src/engine/game_state.py b/src/engine/game_state.py
index 5ec0303..6aa58b0 100644
--- a/src/engine/game_state.py
+++ b/src/engine/game_state.py
@@ -15,7 +15,7 @@ class ChengduMahjongState:
         # 当前玩家索引
         self.current_player = 0
         # 玩家分数
-        self.scores = 0
+        self.scores = [100, 100, 100, 100]
         # 剩余牌数量
         self.remaining_tiles = 108
         # 胜利玩家列表
diff --git a/test.py b/test.py
new file mode 100644
index 0000000..581416c
--- /dev/null
+++ b/test.py
@@ -0,0 +1,2 @@
+import torch
+print(torch.cuda.is_available())  # 如果返回True，说明可以使用GPU