auto-coder 0.1.310__py3-none-any.whl → 0.1.312__py3-none-any.whl

autocoder/memory/active_context_manager.py

@@ -7,6 +7,7 @@ import sys
 import time
 import threading
 import queue
+import json
 from datetime import datetime
 from typing import List, Dict, Optional, Any, Tuple, Set
 from loguru import logger as global_logger
@@ -35,13 +36,14 @@ class ActiveFileContext(BaseModel):
     active_md_path: str = Field(..., description="活动文件路径")
     content: str = Field(..., description="文件完整内容")
     sections: ActiveFileSections = Field(..., description="文件内容各部分")
-    files: List[str] = Field(default_factory=list, description="与该活动文件相关的源文件列表")
+    files: List[str] = Field(default_factory=list,
+                             description="与该活动文件相关的源文件列表")
 
 
 class FileContextsResult(BaseModel):
     """文件上下文查询结果"""
     contexts: Dict[str, ActiveFileContext] = Field(
-        default_factory=dict, 
+        default_factory=dict,
         description="键为目录路径，值为活动文件上下文"
     )
     not_found_files: List[str] = Field(
@@ -53,34 +55,35 @@ class FileContextsResult(BaseModel):
 class ActiveContextManager:
     """
     ActiveContextManager是活动上下文跟踪子系统的主要接口。
-    
+
     该类负责：
     1. 从YAML文件加载任务数据
     2. 映射目录结构
     3. 生成活动上下文文档
     4. 管理异步任务执行
-    
+    5. 持久化任务信息
+
     该类实现了单例模式，确保系统中只有一个实例。
     """
-    
+
     # 类变量，用于存储单例实例
     _instance = None
     _is_initialized = False
-    
+
     # 任务队列和队列处理线程
     _task_queue = None
     _queue_thread = None
     _queue_lock = None
     _is_processing = False
-    
+
     def __new__(cls, llm: byzerllm.ByzerLLM = None, args: AutoCoderArgs = None):
         """
         实现单例模式，确保只创建一个实例
-        
+
         Args:
             llm: ByzerLLM实例，用于生成文档内容
             args: AutoCoderArgs实例，包含配置信息
-        
+
         Returns:
             ActiveContextManager: 单例实例
         """
@@ -88,11 +91,11 @@ class ActiveContextManager:
             cls._instance = super(ActiveContextManager, cls).__new__(cls)
             cls._instance._is_initialized = False
         return cls._instance
-    
-    def __init__(self, llm: byzerllm.ByzerLLM,source_dir:str):
+
+    def __init__(self, llm: byzerllm.ByzerLLM, source_dir: str):
         """
         初始化活动上下文管理器
-        
+
         Args:
             llm: ByzerLLM实例，用于生成文档内容            
         """
@@ -104,39 +107,128 @@ class ActiveContextManager:
         log_dir = os.path.join(source_dir, ".auto-coder", "active-context")
         os.makedirs(log_dir, exist_ok=True)
         log_file = os.path.join(log_dir, "active.log")
-        
+
         # 配置全局日志输出到文件，不输出到控制台
         global_logger.configure(
             handlers=[
                 # 移除控制台输出，只保留文件输出
-                {"sink": log_file, "level": "DEBUG", "rotation": "10 MB", "retention": "1 week", 
+                {"sink": log_file, "level": "DEBUG", "rotation": "10 MB", "retention": "1 week",
                  "format": "{time:YYYY-MM-DD HH:mm:ss} | {level} | {name} | {message}"}
             ]
         )
-        
+
         # 创建专用的logger实例
         self.logger = global_logger.bind(name="ActiveContextManager")
         self.logger.info(f"初始化 ActiveContextManager，日志输出到 {log_file}")
-            
-        self.llm = llm        
-        self.directory_mapper = DirectoryMapper()
-        self.active_package = ActivePackage(llm)
+
+        self.llm = llm
+        self.directory_mapper = DirectoryMapper()        
         self.async_processor = AsyncProcessor()
         self.yml_manager = ActionYmlFileManager(source_dir)
-        self.tasks = {}  # 用于跟踪任务状态
-        self.printer = Printer()
         
+        # 任务持久化文件路径
+        self.tasks_file_path = os.path.join(source_dir, ".auto-coder", "active-context", "tasks.json")
+        
+        # 加载已存在的任务
+        self.tasks = self._load_tasks_from_disk()
+        self.tasks_lock = threading.Lock()  # 添加锁以保护任务字典的操作
+        
+        self.printer = Printer()
+
         # 初始化任务队列和锁
         self.__class__._task_queue = queue.Queue()
         self.__class__._queue_lock = threading.Lock()
-        
+
         # 启动队列处理线程
-        self.__class__._queue_thread = threading.Thread(target=self._process_queue, daemon=True)
+        self.__class__._queue_thread = threading.Thread(
+            target=self._process_queue, daemon=True)
         self.__class__._queue_thread.start()
-        
+
         # 标记为已初始化
-        self._is_initialized = True                    
-    
+        self._is_initialized = True
+        
+    def _load_tasks_from_disk(self) -> Dict[str, Dict[str, Any]]:
+        """
+        从磁盘加载任务信息
+        
+        Returns:
+            Dict: 任务字典
+        """
+        try:
+            if os.path.exists(self.tasks_file_path):
+                with open(self.tasks_file_path, 'r', encoding='utf-8') as f:
+                    tasks_json = json.load(f)
+                
+                # 转换时间字符串为datetime对象
+                for task_id, task in tasks_json.items():
+                    if 'start_time' in task and task['start_time']:
+                        try:
+                            task['start_time'] = datetime.fromisoformat(task['start_time'])
+                        except:
+                            task['start_time'] = None
+                            
+                    if 'completion_time' in task and task['completion_time']:
+                        try:
+                            task['completion_time'] = datetime.fromisoformat(task['completion_time'])
+                        except:
+                            task['completion_time'] = None
+                
+                self.logger.info(f"从 {self.tasks_file_path} 加载了 {len(tasks_json)} 个任务")
+                return tasks_json
+            else:
+                self.logger.info(f"任务文件 {self.tasks_file_path} 不存在，将创建新文件")
+                return {}
+        except Exception as e:
+            self.logger.error(f"加载任务文件失败: {e}")
+            return {}
+            
+    def _save_tasks_to_disk(self):
+        """
+        将任务信息保存到磁盘
+        """
+        try:
+            with self.tasks_lock:  # 使用锁确保线程安全
+                # 创建任务字典的副本并进行序列化处理
+                tasks_copy = {}
+                for task_id, task in self.tasks.items():
+                    # 深拷贝并转换不可序列化的对象
+                    task_copy = {}
+                    for k, v in task.items():
+                        if k in ['start_time', 'completion_time'] and isinstance(v, datetime):
+                            task_copy[k] = v.isoformat()
+                        else:
+                            task_copy[k] = v
+                    tasks_copy[task_id] = task_copy
+                
+                # 确保目录存在
+                os.makedirs(os.path.dirname(self.tasks_file_path), exist_ok=True)
+                
+                # 写入JSON文件
+                with open(self.tasks_file_path, 'w', encoding='utf-8') as f:
+                    json.dump(tasks_copy, f, ensure_ascii=False, indent=2)
+                
+                self.logger.debug(f"成功保存 {len(tasks_copy)} 个任务到 {self.tasks_file_path}")
+        except Exception as e:
+            self.logger.error(f"保存任务到磁盘失败: {e}")
+            
+    def _update_task(self, task_id: str, **kwargs):
+        """
+        更新任务信息并持久化到磁盘
+        
+        Args:
+            task_id: 任务ID
+            **kwargs: 要更新的任务属性
+        """
+        with self.tasks_lock:
+            if task_id in self.tasks:
+                self.tasks[task_id].update(kwargs)
+                self.logger.debug(f"更新任务 {task_id} 信息: {kwargs.keys()}")
+            else:
+                self.logger.warning(f"尝试更新不存在的任务: {task_id}")
+                
+        # 持久化到磁盘
+        self._save_tasks_to_disk()
+
     def _process_queue(self):
         """
         处理任务队列的后台线程
@@ -149,90 +241,102 @@ class ActiveContextManager:
                 if task is None:
                     # None 是退出信号
                     break
-                    
+
                 # 设置处理标志
                 with self._queue_lock:
                     self.__class__._is_processing = True
-                    
+
                 # 解包任务参数
                 task_id, query, changed_urls, current_urls = task
-                
+
                 # 更新任务状态为运行中
-                self.tasks[task_id]['status'] = 'running'                
-                                
-                self._process_changes_async(task_id, query, changed_urls, current_urls)                
-                
+                self._update_task(task_id, status='running')
+
+                self._process_changes_async(
+                    task_id, query, changed_urls, current_urls)
+
                 # 重置处理标志
                 with self._queue_lock:
                     self.__class__._is_processing = False
-                    
+
                 # 标记任务完成
                 self._task_queue.task_done()
-                
+
             except Exception as e:
                 self.logger.error(f"Error in queue processing thread: {e}")
                 # 重置处理标志，确保队列可以继续处理
                 with self._queue_lock:
                     self.__class__._is_processing = False
-    
-    def process_changes(self, args:AutoCoderArgs) -> str:
+
+    def process_changes(self, args: AutoCoderArgs) -> str:
         """
         处理代码变更，创建活动上下文（非阻塞）
-        
+
         Args:
             args: AutoCoderArgs实例，包含配置信息
-        
+
         Returns:
             str: 任务ID，可用于后续查询任务状态
         """
         try:
             # 使用参数中的文件或者指定的文件
             if not args.file:
-                raise ValueError("action file is required")            
-            
+                raise ValueError("action file is required")
+
             file_name = os.path.basename(args.file)
             # 从YAML文件加载数据
             yaml_content = self.yml_manager.load_yaml_content(file_name)
-            
+
             # 提取需要的信息
             query = yaml_content.get('query', '')
             changed_urls = yaml_content.get('add_updated_urls', [])
-            current_urls = yaml_content.get('urls', []) + yaml_content.get('dynamic_urls', [])
-            
+            current_urls = yaml_content.get(
+                'urls', []) + yaml_content.get('dynamic_urls', [])
+
             # 创建任务ID
             task_id = f"active_context_{int(time.time())}_{file_name}"
-            
+
             # 更新任务状态
-            self.tasks[task_id] = {
-                'status': 'queued',
-                'start_time': datetime.now(),
-                'file_name': file_name,
-                'query': query,
-                'changed_urls': changed_urls,
-                'current_urls': current_urls,
-                'queue_position': self._task_queue.qsize() + (1 if self._is_processing else 0)
-            }
-            
+            with self.tasks_lock:
+                self.tasks[task_id] = {
+                    'status': 'queued',
+                    'start_time': datetime.now(),
+                    'file_name': file_name,
+                    'query': query,
+                    'changed_urls': changed_urls,
+                    'current_urls': current_urls,
+                    'queue_position': self._task_queue.qsize() + (1 if self._is_processing else 0),
+                    'total_tokens': 0,  # 初始化token计数
+                    'input_tokens': 0,  # 初始化输入token计数
+                    'output_tokens': 0,  # 初始化输出token计数
+                    'cost': 0.0,  # 初始化费用
+                }
+                # 持久化任务信息
+            self._save_tasks_to_disk()
+
             # 直接启动后台线程处理任务，不通过队列
             thread = threading.Thread(
                 target=self._execute_task_in_background,
-                args=(task_id, query, changed_urls, current_urls),
+                args=(task_id, query, changed_urls, current_urls, args),
                 daemon=True  # 使用守护线程，主程序退出时自动结束
             )
             thread.start()
-            
+
             # 记录任务已启动，并立即返回
             self.logger.info(f"Task {task_id} started in background thread")
             return task_id
-            
+
         except Exception as e:
-            self.logger.error(f"Error in process_changes: {e}")            
+            self.logger.error(f"Error in process_changes: {e}")
             raise
-    
-    def _execute_task_in_background(self, task_id: str, query: str, changed_urls: List[str], current_urls: List[str]):
+
+    def _execute_task_in_background(self, task_id: str, 
+                                    query: str, 
+                                    changed_urls: List[str], current_urls: List[str],
+                                    args: AutoCoderArgs):
         """
         在后台线程中执行任务，处理所有日志重定向
-        
+
         Args:
             task_id: 任务ID
             query: 用户查询
@@ -241,47 +345,48 @@ class ActiveContextManager:
         """
         try:
             # 更新任务状态为运行中
-            self.tasks[task_id]['status'] = 'running'
-            
+            self._update_task(task_id, status='running')
+
             # 重定向输出并执行任务
-            self._process_changes_async(task_id, query, changed_urls, current_urls)
-            
+            self._process_changes_async(
+                task_id, query, changed_urls, current_urls, args)
+
             # 更新任务状态为已完成
-            self.tasks[task_id]['status'] = 'completed'
-            self.tasks[task_id]['completion_time'] = datetime.now()
-            
+            self._update_task(task_id, status='completed', completion_time=datetime.now())
+
         except Exception as e:
             # 记录错误，但不允许异常传播到主线程
             error_msg = f"Background task {task_id} failed: {str(e)}"
             self.logger.error(error_msg)
-            self.tasks[task_id]['status'] = 'failed'
-            self.tasks[task_id]['error'] = error_msg
-    
-    def _process_changes_async(self, task_id: str, query: str, changed_urls: List[str], current_urls: List[str]):
+            self._update_task(task_id, status='failed', error=error_msg)
+
+    def _process_changes_async(self, task_id: str, query: str, changed_urls: List[str], current_urls: List[str], args: AutoCoderArgs):
         """
         实际处理变更的异步方法
-        
+
         Args:
             task_id: 任务ID
             query: 用户查询/需求
             changed_urls: 变更的文件路径列表
             current_urls: 当前相关的文件路径列表
+            args: AutoCoderArgs实例，包含配置信息
         """
         try:
             self.logger.info(f"==== 开始处理任务 {task_id} ====")
             self.logger.info(f"查询内容: {query}")
             self.logger.info(f"变更文件数量: {len(changed_urls)}")
             self.logger.info(f"相关文件数量: {len(current_urls)}")
-            
+
             if changed_urls:
-                self.logger.debug(f"变更文件列表: {', '.join(changed_urls[:5])}{'...' if len(changed_urls) > 5 else ''}")
-            
-            self.tasks[task_id]['status'] = 'running'
-            
+                self.logger.debug(
+                    f"变更文件列表: {', '.join(changed_urls[:5])}{'...' if len(changed_urls) > 5 else ''}")
+
+            self._update_task(task_id, status='running')
+
             # 获取当前任务的文件名
             file_name = self.tasks[task_id].get('file_name')
             self.logger.info(f"任务关联文件: {file_name}")
-            
+
             # 获取文件变更信息
             file_changes = {}
             if file_name:
@@ -294,60 +399,98 @@ class ActiveContextManager:
                     self.logger.info(f"成功获取到 {len(file_changes)} 个文件变更")
                 else:
                     self.logger.warning("未找到提交变更信息")
-            
+
             # 1. 映射目录
             self.logger.info("开始映射目录结构...")
             directory_contexts = self.directory_mapper.map_directories(
                 self.source_dir, changed_urls, current_urls
             )
             self.logger.info(f"目录映射完成，找到 {len(directory_contexts)} 个相关目录")
-            
+
             # 2. 处理每个目录
             processed_dirs = []
+            total_tokens = 0
+            input_tokens = 0
+            output_tokens = 0
+            cost = 0.0
+            
             for i, context in enumerate(directory_contexts):
                 dir_path = context['directory_path']
-                self.logger.info(f"[{i+1}/{len(directory_contexts)}] 开始处理目录: {dir_path}")
+                self.logger.info(
+                    f"[{i+1}/{len(directory_contexts)}] 开始处理目录: {dir_path}")
                 try:
-                    self._process_directory_context(context, query, file_changes)
+                    result = self._process_directory_context(
+                        context, query, file_changes, args)
+                    
+                    # 如果返回了token和费用信息，则累加
+                    if isinstance(result, dict):
+                        dir_tokens = result.get('total_tokens', 0)
+                        dir_input_tokens = result.get('input_tokens', 0)
+                        dir_output_tokens = result.get('output_tokens', 0)
+                        dir_cost = result.get('cost', 0.0)
+                        
+                        total_tokens += dir_tokens
+                        input_tokens += dir_input_tokens
+                        output_tokens += dir_output_tokens
+                        cost += dir_cost
+                        
+                        self.logger.info(f"目录 {dir_path} 处理完成，使用了 {dir_tokens} tokens，费用 {dir_cost:.6f}")
+                    
                     processed_dirs.append(os.path.basename(dir_path))
-                    self.logger.info(f"目录 {dir_path} 处理完成")
+                    
                 except Exception as e:
                     self.logger.error(f"处理目录 {dir_path} 时出错: {str(e)}")
-                
-            # 3. 更新任务状态
-            self.tasks[task_id]['status'] = 'completed'
-            self.tasks[task_id]['completion_time'] = datetime.now()
-            self.tasks[task_id]['processed_dirs'] = processed_dirs
             
-            duration = (datetime.now() - self.tasks[task_id]['start_time']).total_seconds()
+            # 更新任务的token和费用信息
+            self._update_task(
+                task_id, 
+                total_tokens=total_tokens,
+                input_tokens=input_tokens,
+                output_tokens=output_tokens,
+                cost=cost,
+                processed_dirs=processed_dirs
+            )
+
+            # 3. 更新任务状态
+            self._update_task(task_id, 
+                             status='completed', 
+                             completion_time=datetime.now())
+
+            duration = (datetime.now() -
+                        self.tasks[task_id]['start_time']).total_seconds()
             self.logger.info(f"==== 任务 {task_id} 处理完成 ====")
             self.logger.info(f"总耗时: {duration:.2f}秒")
             self.logger.info(f"处理的目录数: {len(processed_dirs)}")
-        
+            self.logger.info(f"使用总tokens: {total_tokens} (输入: {input_tokens}, 输出: {output_tokens})")
+            self.logger.info(f"总费用: {cost:.6f}")
+
         except Exception as e:
             # 记录错误
             self.logger.error(f"任务 {task_id} 失败: {str(e)}", exc_info=True)
-            self.tasks[task_id]['status'] = 'failed'
-            self.tasks[task_id]['error'] = str(e)
-    
-    def _process_directory_context(self, context: Dict[str, Any], query: str, file_changes: Dict[str, Tuple[str, str]] = None):
+            self._update_task(task_id, status='failed', error=str(e))
+
+    def _process_directory_context(self, context: Dict[str, Any], query: str, file_changes: Dict[str, Tuple[str, str]] = None, args: AutoCoderArgs = None):
         """
         处理单个目录上下文
-        
+
         Args:
             context: 目录上下文字典
             query: 用户查询/需求
             file_changes: 文件变更字典，键为文件路径，值为(变更前内容, 变更后内容)的元组
+            args: AutoCoderArgs实例，包含配置信息
+            
+        Returns:
+            Dict: 包含token和费用信息的字典
         """
         try:
             directory_path = context['directory_path']
             self.logger.debug(f"--- 处理目录上下文开始: {directory_path} ---")
-            
+
             # 1. 确保目录存在
             target_dir = self._get_active_context_path(directory_path)
             os.makedirs(target_dir, exist_ok=True)
             self.logger.debug(f"目标目录准备完成: {target_dir}")
-            
+
             # 2. 检查是否有现有的active.md文件
             existing_file_path = os.path.join(target_dir, "active.md")
             if os.path.exists(existing_file_path):
@@ -355,19 +498,20 @@ class ActiveContextManager:
                 try:
                     with open(existing_file_path, 'r', encoding='utf-8') as f:
                         existing_content_preview = f.read(500)
-                    self.logger.debug(f"现有文件内容预览: {existing_content_preview[:100]}...")
+                    self.logger.debug(
+                        f"现有文件内容预览: {existing_content_preview[:100]}...")
                 except Exception as e:
                     self.logger.warning(f"无法读取现有文件内容: {str(e)}")
             else:
                 existing_file_path = None
                 self.logger.info(f"目录 {directory_path} 没有找到现有的 active.md 文件")
-            
+
             # 记录目录中的文件信息
             changed_files = context.get('changed_files', [])
             current_files = context.get('current_files', [])
             self.logger.debug(f"目录中变更文件数: {len(changed_files)}")
             self.logger.debug(f"目录中当前文件数: {len(current_files)}")
-            
+
             # 过滤出当前目录相关的文件变更
             directory_changes = {}
             if file_changes:
@@ -378,69 +522,95 @@ class ActiveContextManager:
                     file_path = file_info['path']
                     dir_files.append(file_path)
                     self.logger.debug(f"添加变更文件: {file_path}")
-                
+
                 for file_info in current_files:
                     file_path = file_info['path']
                     dir_files.append(file_path)
                     self.logger.debug(f"添加当前文件: {file_path}")
-                
+
                 # 从file_changes中获取当前目录文件的变更
                 for file_path, change_info in file_changes.items():
                     if file_path in dir_files:
                         directory_changes[file_path] = change_info
                         old_content, new_content = change_info
-                        old_preview = old_content[:50] if old_content else "(空)"
-                        new_preview = new_content[:50] if new_content else "(空)"
+                        old_preview = old_content[:
+                                                  50] if old_content else "(空)"
+                        new_preview = new_content[:
+                                                  50] if new_content else "(空)"
                         self.logger.debug(f"文件变更: {file_path}")
                         self.logger.debug(f"  旧内容: {old_preview}...")
                         self.logger.debug(f"  新内容: {new_preview}...")
-                
-                self.logger.info(f"找到 {len(directory_changes)} 个与目录 {directory_path} 相关的文件变更")
+
+                self.logger.info(
+                    f"找到 {len(directory_changes)} 个与目录 {directory_path} 相关的文件变更")
             else:
                 self.logger.debug("没有提供文件变更信息")
-            
+
             # 3. 生成活动文件内容
             self.logger.info(f"开始为目录 {directory_path} 生成活动文件内容...")
-            markdown_content = self.active_package.generate_active_file(
-                context, 
+            
+            active_package = ActivePackage(self.llm, args.product_mode)
+            
+            # 调用生成方法，捕获token和费用信息
+            generation_result = active_package.generate_active_file(
+                context,
                 query,
                 existing_file_path=existing_file_path,
-                file_changes=directory_changes
+                file_changes=directory_changes,
+                args=args
             )
             
+            # 检查返回值类型，兼容现有逻辑
+            if isinstance(generation_result, tuple) and len(generation_result) >= 2:
+                markdown_content = generation_result[0]
+                tokens_info = generation_result[1]
+            else:
+                markdown_content = generation_result
+                tokens_info = {
+                    "total_tokens": 0,
+                    "input_tokens": 0,
+                    "output_tokens": 0,
+                    "cost": 0.0
+                }
+
             content_length = len(markdown_content)
             self.logger.debug(f"生成的活动文件内容长度: {content_length} 字符")
             if content_length > 0:
                 self.logger.debug(f"内容预览: {markdown_content[:200]}...")
-            
+
             # 4. 写入文件
             active_md_path = os.path.join(target_dir, "active.md")
             self.logger.info(f"正在写入活动文件: {active_md_path}")
             with open(active_md_path, "w", encoding="utf-8") as f:
                 f.write(markdown_content)
-                
+
             self.logger.info(f"成功创建/更新目录 {directory_path} 的活动文件")
             self.logger.debug(f"--- 处理目录上下文完成: {directory_path} ---")
             
+            # 返回token和费用信息
+            return tokens_info
+
         except Exception as e:
-            self.logger.error(f"处理目录 {context.get('directory_path', 'unknown')} 时出错: {str(e)}", exc_info=True)
+            self.logger.error(
+                f"处理目录 {context.get('directory_path', 'unknown')} 时出错: {str(e)}", exc_info=True)
             raise
-    
+
     def get_task_status(self, task_id: str) -> Dict[str, Any]:
         """
         获取任务状态
-        
+
         Args:
             task_id: 任务ID
-            
+
         Returns:
             Dict: 任务状态信息
         """
         if task_id not in self.tasks:
             return {'status': 'not_found', 'task_id': task_id}
-        
-        task = self.tasks[task_id]
-        
+
+        with self.tasks_lock:
+            task = self.tasks[task_id]
+
         # 计算任务运行时间
         start_time = task.get('start_time')
         if task.get('status') == 'completed':
@@ -456,15 +626,16 @@ class ActiveContextManager:
             else:
                 elapsed = None
             duration = None
-        
+
         # 构建日志文件路径
         log_file_path = None
         if 'file_name' in task:
-            log_dir = os.path.join(self.source_dir, '.auto-coder', 'active-context', 'logs')
+            log_dir = os.path.join(
+                self.source_dir, '.auto-coder', 'active-context', 'logs')
             log_file_path = os.path.join(log_dir, f'{task_id}.log')
             if not os.path.exists(log_file_path):
                 log_file_path = None
-        
+
         # 构建返回结果
         result = {
             'task_id': task_id,
@@ -472,75 +643,81 @@ class ActiveContextManager:
             'file_name': task.get('file_name'),
             'start_time': start_time.strftime("%Y-%m-%d %H:%M:%S") if start_time else None,
         }
-        
+
         # 添加可选信息
         if 'completion_time' in task:
-            result['completion_time'] = task['completion_time'].strftime("%Y-%m-%d %H:%M:%S")
-        
+            result['completion_time'] = task['completion_time'].strftime(
+                "%Y-%m-%d %H:%M:%S")
+
         if elapsed is not None:
             mins, secs = divmod(elapsed, 60)
             hrs, mins = divmod(mins, 60)
             result['elapsed'] = f"{int(hrs):02d}:{int(mins):02d}:{int(secs):02d}"
-        
+
         if duration is not None:
             mins, secs = divmod(duration, 60)
             hrs, mins = divmod(mins, 60)
             result['duration'] = f"{int(hrs):02d}:{int(mins):02d}:{int(secs):02d}"
-        
+
         if 'processed_dirs' in task:
             result['processed_dirs'] = task['processed_dirs']
             result['processed_dirs_count'] = len(task['processed_dirs'])
-        
+
         if 'error' in task:
             result['error'] = task['error']
-        
+            
+        # 添加token和费用信息
+        for key in ['total_tokens', 'input_tokens', 'output_tokens', 'cost']:
+            if key in task:
+                result[key] = task[key]
+
         if log_file_path and os.path.exists(log_file_path):
             result['log_file'] = log_file_path
-            
+
             # 尝试获取日志文件大小
             try:
                 file_size = os.path.getsize(log_file_path)
                 result['log_file_size'] = f"{file_size / 1024:.2f} KB"
             except:
                 pass
-        
+
         return result
-    
+
     def get_all_tasks(self) -> List[Dict[str, Any]]:
         """
         获取所有任务状态
-        
+
         Returns:
             List[Dict]: 所有任务的状态信息
         """
         return [{'task_id': tid, **task} for tid, task in self.tasks.items()]
-    
+
     def get_running_tasks(self) -> List[Dict[str, Any]]:
         """
         获取所有正在运行的任务
-        
+
         Returns:
             List[Dict]: 所有正在运行的任务的状态信息
         """
-        return [{'task_id': tid, **task} for tid, task in self.tasks.items() 
+        return [{'task_id': tid, **task} for tid, task in self.tasks.items()
                 if task['status'] in ['running', 'queued']]
-    
+
     def load_active_contexts_for_files(self, file_paths: List[str]) -> FileContextsResult:
         """
         根据文件路径列表，找到并加载对应的活动上下文文件
-        
+
         Args:
             file_paths: 文件路径列表
-            
+
         Returns:
             FileContextsResult: 包含活动上下文信息的结构化结果
         """
         try:
             result = FileContextsResult()
-            
+
             # 记录未找到对应活动上下文的文件
             found_files: Set[str] = set()
-            
+
             # 1. 获取文件所在的唯一目录列表
             directories = set()
             for file_path in file_paths:
@@ -548,30 +725,31 @@ class ActiveContextManager:
                 dir_path = os.path.dirname(file_path)
                 if os.path.exists(dir_path):
                     directories.add(dir_path)
-            
+
             # 2. 查找每个目录的活动上下文文件
             for dir_path in directories:
                 # 获取活动上下文目录路径
                 active_context_dir = self._get_active_context_path(dir_path)
                 active_md_path = os.path.join(active_context_dir, "active.md")
-                
+
                 # 检查active.md文件是否存在
                 if os.path.exists(active_md_path):
                     try:
                         # 读取文件内容
                         with open(active_md_path, 'r', encoding='utf-8') as f:
                             content = f.read()
-                        
+
                         # 解析文件内容
                         sections_dict = self._parse_active_md_content(content)
                         sections = ActiveFileSections(**sections_dict)
-                        
+
                         # 找到相关的文件
-                        related_files = [f for f in file_paths if dir_path in f]
-                        
+                        related_files = [
+                            f for f in file_paths if dir_path in f]
+
                         # 记录找到了对应活动上下文的文件
                         found_files.update(related_files)
-                        
+
                         # 创建活动文件上下文
                         active_context = ActiveFileContext(
                             directory_path=dir_path,
@@ -580,30 +758,32 @@ class ActiveContextManager:
                             sections=sections,
                             files=related_files
                         )
-                        
+
                         # 添加到结果
                         result.contexts[dir_path] = active_context
-                        
+
                         self.logger.info(f"已加载目录 {dir_path} 的活动上下文文件")
                     except Exception as e:
-                        self.logger.error(f"读取活动上下文文件 {active_md_path} 时出错: {e}")
-            
+                        self.logger.error(
+                            f"读取活动上下文文件 {active_md_path} 时出错: {e}")
+
             # 3. 记录未找到对应活动上下文的文件
-            result.not_found_files = [f for f in file_paths if f not in found_files]
-            
+            result.not_found_files = [
+                f for f in file_paths if f not in found_files]
+
             return result
-            
+
         except Exception as e:
             self.logger.error(f"加载活动上下文失败: {e}")
             return FileContextsResult(not_found_files=file_paths)
-    
+
     def _parse_active_md_content(self, content: str) -> Dict[str, str]:
         """
         解析活动上下文文件内容，提取各个部分
-        
+
         Args:
             content: 活动上下文文件内容
-            
+
         Returns:
             Dict[str, str]: 包含标题、当前变更和文档部分的字典
         """
@@ -613,37 +793,39 @@ class ActiveContextManager:
                 'current_change': '',
                 'document': ''
             }
-            
+
             # 提取标题部分（到第一个二级标题之前）
             header_match = re.search(r'^(.*?)(?=\n## )', content, re.DOTALL)
             if header_match:
                 result['header'] = header_match.group(1).strip()
-            
+
             # 提取当前变更部分
-            current_change_match = re.search(r'## 当前变更\s*\n(.*?)(?=\n## |$)', content, re.DOTALL)
+            current_change_match = re.search(
+                r'## 当前变更\s*\n(.*?)(?=\n## |$)', content, re.DOTALL)
             if current_change_match:
-                result['current_change'] = current_change_match.group(1).strip()
-            
+                result['current_change'] = current_change_match.group(
+                    1).strip()
+
             # 提取文档部分
-            document_match = re.search(r'## 文档\s*\n(.*?)(?=\n## |$)', content, re.DOTALL)
+            document_match = re.search(
+                r'## 文档\s*\n(.*?)(?=\n## |$)', content, re.DOTALL)
             if document_match:
                 result['document'] = document_match.group(1).strip()
-            
+
             return result
         except Exception as e:
             self.logger.error(f"解析活动上下文文件内容时出错: {e}")
             return {'header': '', 'current_change': '', 'document': ''}
-    
+
     def _get_active_context_path(self, directory_path: str) -> str:
         """
         获取活动上下文中对应的目录路径
-        
+
         Args:
             directory_path: 原始目录路径
-            
+
         Returns:
             str: 活动上下文中对应的目录路径
         """
         relative_path = os.path.relpath(directory_path, self.source_dir)
         return os.path.join(self.source_dir, ".auto-coder", "active-context", relative_path)
-