From 484983094fee90dc5df43ccfc20944224a534bdb Mon Sep 17 00:00:00 2001 From: "github-actions[bot]" Date: Fri, 16 May 2025 04:51:50 +0000 Subject: [PATCH] =?UTF-8?q?=F0=9F=A4=96=20=E8=87=AA=E5=8A=A8=E6=A0=BC?= =?UTF-8?q?=E5=BC=8F=E5=8C=96=E4=BB=A3=E7=A0=81=20[skip=20ci]?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- src/experimental/PFC/PFC_idle/idle_chat.py | 4 +- src/experimental/PFC/action_factory.py | 14 +- src/experimental/PFC/action_handlers.py | 535 +++++++++++++-------- src/experimental/PFC/actions.py | 106 ++-- 4 files changed, 387 insertions(+), 272 deletions(-) diff --git a/src/experimental/PFC/PFC_idle/idle_chat.py b/src/experimental/PFC/PFC_idle/idle_chat.py index 1b551a32..59734b23 100644 --- a/src/experimental/PFC/PFC_idle/idle_chat.py +++ b/src/experimental/PFC/PFC_idle/idle_chat.py @@ -536,7 +536,9 @@ class IdleChat: try: segments = Seg(type="seglist", data=[Seg(type="text", data=content)]) logger.debug(f"[私聊][{self.private_name}]准备发送主动聊天消息: {content}") - await self.message_sender.send_message(chat_stream=chat_stream, segments=segments, reply_to_message=None, content=content) + await self.message_sender.send_message( + chat_stream=chat_stream, segments=segments, reply_to_message=None, content=content + ) logger.info(f"[私聊][{self.private_name}]成功主动发起聊天: {content}") except Exception as e: logger.error(f"[私聊][{self.private_name}]发送主动聊天消息失败: {str(e)}") diff --git a/src/experimental/PFC/action_factory.py b/src/experimental/PFC/action_factory.py index d33590f4..043ad740 100644 --- a/src/experimental/PFC/action_factory.py +++ b/src/experimental/PFC/action_factory.py @@ -2,7 +2,7 @@ from abc import ABC, abstractmethod from typing import Type, TYPE_CHECKING # 从 action_handlers.py 导入具体的处理器类 -from .action_handlers import ( # 调整导入路径 +from .action_handlers import ( # 调整导入路径 ActionHandler, DirectReplyHandler, SendNewMessageHandler, @@ -17,10 +17,12 @@ from .action_handlers import ( # 调整导入路径 ) if TYPE_CHECKING: - from PFC.conversation import Conversation # 调整导入路径 + from PFC.conversation import Conversation # 调整导入路径 + class AbstractActionFactory(ABC): """抽象动作工厂接口。""" + @abstractmethod def create_action_handler(self, action_type: str, conversation: "Conversation") -> ActionHandler: """ @@ -35,8 +37,10 @@ class AbstractActionFactory(ABC): """ pass + class StandardActionFactory(AbstractActionFactory): """标准的动作工厂实现。""" + def create_action_handler(self, action_type: str, conversation: "Conversation") -> ActionHandler: """ 根据动作类型创建并返回具体的动作处理器实例。 @@ -53,10 +57,10 @@ class StandardActionFactory(AbstractActionFactory): "block_and_ignore": BlockAndIgnoreHandler, "wait": WaitHandler, } - handler_class = handler_map.get(action_type) # 获取对应的处理器类 + handler_class = handler_map.get(action_type) # 获取对应的处理器类 # 如果找到对应的处理器类 if handler_class: - return handler_class(conversation) # 创建并返回处理器实例 + return handler_class(conversation) # 创建并返回处理器实例 else: # 如果未找到，返回处理未知动作的默认处理器 - return UnknownActionHandler(conversation) \ No newline at end of file + return UnknownActionHandler(conversation) diff --git a/src/experimental/PFC/action_handlers.py b/src/experimental/PFC/action_handlers.py index 310c2b27..1f183c6e 100644 --- a/src/experimental/PFC/action_handlers.py +++ b/src/experimental/PFC/action_handlers.py @@ -4,7 +4,7 @@ import asyncio import traceback import json import random -from typing import Optional, Set, TYPE_CHECKING, List, Tuple, Dict # 确保导入 Dict +from typing import Optional, Set, TYPE_CHECKING, List, Tuple, Dict # 确保导入 Dict from src.chat.emoji_system.emoji_manager import emoji_manager from src.common.logger_manager import get_logger @@ -27,6 +27,7 @@ class ActionHandler(ABC): 处理动作的抽象基类。 每个具体的动作处理器都应继承此类并实现 execute 方法。 """ + def __init__(self, conversation: "Conversation"): """ 初始化动作处理器。 @@ -44,7 +45,7 @@ class ActionHandler(ABC): observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, - current_action_record: dict + current_action_record: dict, ) -> tuple[bool, str, str]: """ 执行具体的动作逻辑。 @@ -89,15 +90,15 @@ class ActionHandler(ABC): await self.conversation.direct_sender.send_message( chat_stream=self.conversation.chat_stream, segments=final_segments, - reply_to_message=None, # 私聊通常不引用回复 - content=content_for_log # 用于发送器内部的日志记录 + reply_to_message=None, # 私聊通常不引用回复 + content=content_for_log, # 用于发送器内部的日志记录 ) # 注意: my_message_count 的增加现在由具体的发送逻辑（文本或表情）处理后决定 return True except Exception as e: self.logger.error(f"[私聊][{self.conversation.private_name}] 发送消息时失败: {str(e)}") self.logger.error(f"[私聊][{self.conversation.private_name}] {traceback.format_exc()}") - self.conversation.state = ConversationState.ERROR # 发送失败则标记错误状态 + self.conversation.state = ConversationState.ERROR # 发送失败则标记错误状态 return False async def _update_bot_message_in_history( @@ -105,7 +106,7 @@ class ActionHandler(ABC): send_time: float, message_content: str, observation_info: ObservationInfo, - message_id_prefix: str = "bot_sent_" + message_id_prefix: str = "bot_sent_", ): """ 在机器人成功发送消息后，将该消息添加到 ObservationInfo 的聊天历史中。 @@ -122,15 +123,17 @@ class ActionHandler(ABC): # 构造机器人发送的消息字典 bot_message_dict: Dict[str, any] = { - "message_id": f"{message_id_prefix}{send_time:.3f}", # 使用更精确的时间戳 + "message_id": f"{message_id_prefix}{send_time:.3f}", # 使用更精确的时间戳 "time": send_time, "user_info": { "user_id": self.conversation.bot_qq_str, "user_nickname": global_config.BOT_NICKNAME, - "platform": self.conversation.chat_stream.platform if self.conversation.chat_stream else "unknown_platform", + "platform": self.conversation.chat_stream.platform + if self.conversation.chat_stream + else "unknown_platform", }, - "processed_plain_text": message_content, # 历史记录中的纯文本使用传入的 message_content - "detailed_plain_text": message_content, # 详细文本也使用相同内容 + "processed_plain_text": message_content, # 历史记录中的纯文本使用传入的 message_content + "detailed_plain_text": message_content, # 详细文本也使用相同内容 } observation_info.chat_history.append(bot_message_dict) observation_info.chat_history_count = len(observation_info.chat_history) @@ -145,7 +148,7 @@ class ActionHandler(ABC): observation_info.chat_history_count = len(observation_info.chat_history) # 更新用于 Prompt 的历史记录字符串 - history_slice_for_str = observation_info.chat_history[-30:] # 例如最近30条 + history_slice_for_str = observation_info.chat_history[-30:] # 例如最近30条 try: observation_info.chat_history_str = await build_readable_messages( history_slice_for_str, @@ -162,8 +165,8 @@ class ActionHandler(ABC): self, observation_info: ObservationInfo, conversation_info: ConversationInfo, - action_type: str, # 例如 "direct_reply", "send_memes" - event_description_for_emotion: str + action_type: str, # 例如 "direct_reply", "send_memes" + event_description_for_emotion: str, ): """ 在成功发送一条或多条消息（文本/表情）后，处理通用的状态更新。 @@ -175,7 +178,7 @@ class ActionHandler(ABC): action_type (str): 执行的动作类型，用于决定追问逻辑。 event_description_for_emotion (str): 用于情绪更新的事件描述。 """ - current_event_time = time.time() # 获取当前时间作为事件发生时间 + current_event_time = time.time() # 获取当前时间作为事件发生时间 # 更新 IdleChat 的最后消息时间 if self.conversation.idle_chat: @@ -184,7 +187,7 @@ class ActionHandler(ABC): # 清理在本次交互完成（即此函数被调用时）之前的、来自他人的未处理消息 current_unprocessed_messages = getattr(observation_info, "unprocessed_messages", []) message_ids_to_clear: Set[str] = set() - timestamp_before_current_interaction_completion = current_event_time - 0.001 # 确保是严格之前 + timestamp_before_current_interaction_completion = current_event_time - 0.001 # 确保是严格之前 for msg in current_unprocessed_messages: msg_time = msg.get("time") @@ -195,8 +198,8 @@ class ActionHandler(ABC): if ( msg_id and msg_time - and sender_id != self.conversation.bot_qq_str # 是对方的消息 - and msg_time < timestamp_before_current_interaction_completion # 在本次交互完成前 + and sender_id != self.conversation.bot_qq_str # 是对方的消息 + and msg_time < timestamp_before_current_interaction_completion # 在本次交互完成前 ): message_ids_to_clear.add(msg_id) @@ -207,9 +210,7 @@ class ActionHandler(ABC): await observation_info.clear_processed_messages(message_ids_to_clear) # 更新追问状态 (last_successful_reply_action) - other_new_msg_count_during_planning = getattr( - conversation_info, "other_new_messages_during_planning_count", 0 - ) + other_new_msg_count_during_planning = getattr(conversation_info, "other_new_messages_during_planning_count", 0) if action_type in ["direct_reply", "send_new_message", "send_memes"]: if other_new_msg_count_during_planning > 0 and action_type == "direct_reply": # 如果是直接回复，且规划期间有新消息，则下次不应追问 @@ -217,15 +218,12 @@ class ActionHandler(ABC): else: # 否则，记录本次成功的回复/表情动作为下次追问的依据 conversation_info.last_successful_reply_action = action_type - + # 更新关系和情绪状态 await self._update_relationship_and_emotion(observation_info, conversation_info, event_description_for_emotion) async def _update_relationship_and_emotion( - self, - observation_info: ObservationInfo, - conversation_info: ConversationInfo, - event_description: str + self, observation_info: ObservationInfo, conversation_info: ConversationInfo, event_description: str ): """ 辅助方法：调用关系更新器和情绪更新器。 @@ -272,18 +270,18 @@ class ActionHandler(ABC): if emoji_result: emoji_path, full_emoji_description = emoji_result self.logger.info(f"获取到表情包: {emoji_path}, 描述: {full_emoji_description}") - + # 将图片路径转换为纯 Base64 字符串 emoji_b64_content = image_path_to_base64(emoji_path) if not emoji_b64_content: self.logger.error(f"无法将图片 {emoji_path} 转换为Base64。") return None - + # 根据用户提供的片段，Seg type="emoji" data 为纯 Base64 字符串 emoji_segment = Seg(type="emoji", data=emoji_b64_content) # 用于发送器日志的截断描述 log_content_for_emoji = full_emoji_description[-20:] + "..." - + return emoji_segment, full_emoji_description, log_content_for_emoji else: self.logger.warning(f"未能根据查询 '{emoji_query}' 获取到合适的表情包。") @@ -298,12 +296,13 @@ class BaseTextReplyHandler(ActionHandler): 处理基于文本的回复动作的基类，包含生成-检查-重试的循环。 适用于 DirectReplyHandler 和 SendNewMessageHandler。 """ + async def _generate_and_check_text_reply_loop( self, - action_type: str, # "direct_reply" or "send_new_message" + action_type: str, # "direct_reply" or "send_new_message" observation_info: ObservationInfo, conversation_info: ConversationInfo, - max_attempts: int + max_attempts: int, ) -> Tuple[bool, Optional[str], str, bool, bool]: """ 管理生成文本回复并检查其适用性的循环。 @@ -326,10 +325,10 @@ class BaseTextReplyHandler(ActionHandler): 如果 ReplyGenerator 决定发送则为 True，否则为 False。对于 direct_reply，此值恒为 True。 """ reply_attempt_count = 0 - is_suitable = False # 标记内容是否通过检查 - generated_content_to_send: Optional[str] = None # 最终要发送的文本 - final_check_reason = "未开始检查" # 最终检查原因 - need_replan = False # 是否需要重新规划 + is_suitable = False # 标记内容是否通过检查 + generated_content_to_send: Optional[str] = None # 最终要发送的文本 + final_check_reason = "未开始检查" # 最终检查原因 + need_replan = False # 是否需要重新规划 # direct_reply 总是尝试发送；send_new_message 的初始值取决于RG should_send_reply_for_new_message = True if action_type == "direct_reply" else False @@ -338,7 +337,7 @@ class BaseTextReplyHandler(ActionHandler): log_prefix = f"[私聊][{self.conversation.private_name}] 尝试生成/检查 '{action_type}' (第 {reply_attempt_count}/{max_attempts} 次)..." self.logger.info(log_prefix) - self.conversation.state = ConversationState.GENERATING # 设置状态为生成中 + self.conversation.state = ConversationState.GENERATING # 设置状态为生成中 if not self.conversation.reply_generator: raise RuntimeError(f"ReplyGenerator 未为 {self.conversation.private_name} 初始化") @@ -347,56 +346,68 @@ class BaseTextReplyHandler(ActionHandler): observation_info, conversation_info, action_type=action_type ) self.logger.debug(f"{log_prefix} ReplyGenerator.generate 返回: '{raw_llm_output}'") - current_content_for_check = raw_llm_output # 当前待检查的内容 + current_content_for_check = raw_llm_output # 当前待检查的内容 # 如果是 send_new_message 动作，需要解析 JSON 判断是否发送 if action_type == "send_new_message": parsed_json = None try: parsed_json = json.loads(raw_llm_output) - except json.JSONDecodeError: # JSON 解析失败 + except json.JSONDecodeError: # JSON 解析失败 self.logger.error(f"{log_prefix} ReplyGenerator 返回的不是有效的JSON: {raw_llm_output}") conversation_info.last_reply_rejection_reason = "回复生成器未返回有效JSON" conversation_info.last_rejected_reply_content = raw_llm_output - should_send_reply_for_new_message = False # 标记不发送 - is_suitable = True # 决策已做出（不发送），所以认为是 "suitable" 以跳出循环 + should_send_reply_for_new_message = False # 标记不发送 + is_suitable = True # 决策已做出（不发送），所以认为是 "suitable" 以跳出循环 final_check_reason = "回复生成器JSON解析失败，决定不发送" - generated_content_to_send = None # 明确不发送内容 - break # 跳出重试循环 + generated_content_to_send = None # 明确不发送内容 + break # 跳出重试循环 - if parsed_json: # JSON 解析成功 + if parsed_json: # JSON 解析成功 send_decision = parsed_json.get("send", "no").lower() - generated_text_from_json = parsed_json.get("txt", "") # 如果不发送，txt可能是"no" + generated_text_from_json = parsed_json.get("txt", "") # 如果不发送，txt可能是"no" - if send_decision == "yes": # ReplyGenerator 决定发送 + if send_decision == "yes": # ReplyGenerator 决定发送 should_send_reply_for_new_message = True current_content_for_check = generated_text_from_json - self.logger.info(f"{log_prefix} ReplyGenerator 决定发送消息。内容初步为: '{current_content_for_check[:100]}...'") - else: # ReplyGenerator 决定不发送 + self.logger.info( + f"{log_prefix} ReplyGenerator 决定发送消息。内容初步为: '{current_content_for_check[:100]}...'" + ) + else: # ReplyGenerator 决定不发送 should_send_reply_for_new_message = False - is_suitable = True # 决策已做出（不发送） + is_suitable = True # 决策已做出（不发送） final_check_reason = "回复生成器决定不发送" generated_content_to_send = None self.logger.info(f"{log_prefix} ReplyGenerator 决定不发送消息。") - break # 跳出重试循环 - + break # 跳出重试循环 + # 检查生成的内容是否有效（适用于 direct_reply 或 send_new_message 且决定发送的情况） - if not current_content_for_check or \ - current_content_for_check.startswith("抱歉") or \ - current_content_for_check.strip() == "" or \ - (action_type == "send_new_message" and current_content_for_check == "no" and should_send_reply_for_new_message): + if ( + not current_content_for_check + or current_content_for_check.startswith("抱歉") + or current_content_for_check.strip() == "" + or ( + action_type == "send_new_message" + and current_content_for_check == "no" + and should_send_reply_for_new_message + ) + ): warning_msg = f"{log_prefix} 生成内容无效或为错误提示" - if action_type == "send_new_message" and current_content_for_check == "no" and should_send_reply_for_new_message: - warning_msg += " (ReplyGenerator决定发送但文本为'no')" + if ( + action_type == "send_new_message" + and current_content_for_check == "no" + and should_send_reply_for_new_message + ): + warning_msg += " (ReplyGenerator决定发送但文本为'no')" self.logger.warning(warning_msg + "，将进行下一次尝试 (如果适用)。") - final_check_reason = "生成内容无效" # 更新检查原因 + final_check_reason = "生成内容无效" # 更新检查原因 conversation_info.last_reply_rejection_reason = final_check_reason conversation_info.last_rejected_reply_content = current_content_for_check - await asyncio.sleep(0.5) # 暂停后重试 - continue # 进入下一次循环 + await asyncio.sleep(0.5) # 暂停后重试 + continue # 进入下一次循环 # --- 内容检查 --- - self.conversation.state = ConversationState.CHECKING # 设置状态为检查中 + self.conversation.state = ConversationState.CHECKING # 设置状态为检查中 if not self.conversation.reply_checker: raise RuntimeError(f"ReplyChecker 未为 {self.conversation.private_name} 初始化") @@ -414,51 +425,60 @@ class BaseTextReplyHandler(ActionHandler): if global_config.enable_pfc_reply_checker: self.logger.debug(f"{log_prefix} 调用 ReplyChecker 检查 (配置已启用)...") is_suitable_check, reason_check, need_replan_check = await self.conversation.reply_checker.check( - reply=current_content_for_check, goal=current_goal_str, - chat_history=chat_history_for_check, chat_history_text=chat_history_text_for_check, - current_time_str=current_time_value_for_check, retry_count=(reply_attempt_count - 1) + reply=current_content_for_check, + goal=current_goal_str, + chat_history=chat_history_for_check, + chat_history_text=chat_history_text_for_check, + current_time_str=current_time_value_for_check, + retry_count=(reply_attempt_count - 1), ) self.logger.info( f"{log_prefix} ReplyChecker 结果: 合适={is_suitable_check}, 原因='{reason_check}', 需重规划={need_replan_check}" ) - else: # ReplyChecker 未启用 + else: # ReplyChecker 未启用 is_suitable_check, reason_check, need_replan_check = True, "ReplyChecker 已通过配置关闭", False self.logger.debug(f"{log_prefix} [配置关闭] ReplyChecker 已跳过，默认回复为合适。") - is_suitable = is_suitable_check # 更新内容是否合适 - final_check_reason = reason_check # 更新检查原因 - need_replan = need_replan_check # 更新是否需要重规划 + is_suitable = is_suitable_check # 更新内容是否合适 + final_check_reason = reason_check # 更新检查原因 + need_replan = need_replan_check # 更新是否需要重规划 - if not is_suitable: # 如果内容不合适 + if not is_suitable: # 如果内容不合适 conversation_info.last_reply_rejection_reason = final_check_reason conversation_info.last_rejected_reply_content = current_content_for_check if final_check_reason == "机器人尝试发送重复消息" and not need_replan: self.logger.warning(f"{log_prefix} 回复因自身重复被拒绝。将重试。") - elif not need_replan and reply_attempt_count < max_attempts: # 如果不需要重规划且还有尝试次数 + elif not need_replan and reply_attempt_count < max_attempts: # 如果不需要重规划且还有尝试次数 self.logger.warning(f"{log_prefix} 回复不合适: {final_check_reason}。将重试。") - else: # 需要重规划或已达到最大尝试次数 + else: # 需要重规划或已达到最大尝试次数 self.logger.warning(f"{log_prefix} 回复不合适且(需要重规划或已达最大次数): {final_check_reason}") - break # 结束循环 - await asyncio.sleep(0.5) # 重试前暂停 - else: # 内容合适 - generated_content_to_send = current_content_for_check # 设置最终要发送的内容 - conversation_info.last_reply_rejection_reason = None # 清除上次拒绝原因 - conversation_info.last_rejected_reply_content = None # 清除上次拒绝内容 - break # 成功，跳出循环 - + break # 结束循环 + await asyncio.sleep(0.5) # 重试前暂停 + else: # 内容合适 + generated_content_to_send = current_content_for_check # 设置最终要发送的内容 + conversation_info.last_reply_rejection_reason = None # 清除上次拒绝原因 + conversation_info.last_rejected_reply_content = None # 清除上次拒绝内容 + break # 成功，跳出循环 + # 确保 send_new_message 在 RG 决定不发送时，is_suitable 为 True，generated_content_to_send 为 None if action_type == "send_new_message" and not should_send_reply_for_new_message: - is_suitable = True # 决策已完成（不发送） - generated_content_to_send = None # 确认不发送任何内容 + is_suitable = True # 决策已完成（不发送） + generated_content_to_send = None # 确认不发送任何内容 - return is_suitable, generated_content_to_send, final_check_reason, need_replan, should_send_reply_for_new_message + return ( + is_suitable, + generated_content_to_send, + final_check_reason, + need_replan, + should_send_reply_for_new_message, + ) async def _process_and_send_reply_with_optional_emoji( self, - action_type: str, # "direct_reply" or "send_new_message" + action_type: str, # "direct_reply" or "send_new_message" observation_info: ObservationInfo, conversation_info: ConversationInfo, - max_reply_attempts: int + max_reply_attempts: int, ) -> Tuple[bool, bool, List[str], Optional[str], bool, str, bool]: """ 核心共享方法：处理文本生成/检查，获取表情，并按顺序发送。 @@ -483,15 +503,20 @@ class BaseTextReplyHandler(ActionHandler): sent_emoji_successfully = False final_reason_parts: List[str] = [] full_emoji_description_if_sent: Optional[str] = None - + # 1. 处理文本部分 - is_suitable_text, generated_text_content, text_check_reason, need_replan_text, rg_decided_to_send_text = \ - await self._generate_and_check_text_reply_loop( - action_type=action_type, - observation_info=observation_info, - conversation_info=conversation_info, - max_attempts=max_reply_attempts - ) + ( + is_suitable_text, + generated_text_content, + text_check_reason, + need_replan_text, + rg_decided_to_send_text, + ) = await self._generate_and_check_text_reply_loop( + action_type=action_type, + observation_info=observation_info, + conversation_info=conversation_info, + max_attempts=max_reply_attempts, + ) text_to_send: Optional[str] = None # 对于 send_new_message，只有当 RG 决定发送且内容合适时才有文本 @@ -502,90 +527,95 @@ class BaseTextReplyHandler(ActionHandler): elif action_type == "direct_reply": if is_suitable_text and generated_text_content: text_to_send = generated_text_content - - rg_decided_not_to_send_text = (action_type == "send_new_message" and not rg_decided_to_send_text) + + rg_decided_not_to_send_text = action_type == "send_new_message" and not rg_decided_to_send_text # 2. 处理表情部分 - emoji_prepared_info: Optional[Tuple[Seg, str, str]] = None # (segment, full_description, log_description) + emoji_prepared_info: Optional[Tuple[Seg, str, str]] = None # (segment, full_description, log_description) emoji_query = conversation_info.current_emoji_query if emoji_query: emoji_prepared_info = await self._fetch_and_prepare_emoji_segment(emoji_query) # 清理查询，无论是否成功获取，避免重复使用 - conversation_info.current_emoji_query = None # 重要：在这里清理 - + conversation_info.current_emoji_query = None # 重要：在这里清理 + # 3. 决定发送顺序并发送 send_order: List[str] = [] - if text_to_send and emoji_prepared_info: # 文本和表情都有 + if text_to_send and emoji_prepared_info: # 文本和表情都有 send_order = ["text", "emoji"] if random.random() < 0.5 else ["emoji", "text"] - elif text_to_send: # 只有文本 + elif text_to_send: # 只有文本 send_order = ["text"] - elif emoji_prepared_info: # 只有表情 (可能是 direct_reply 带表情，或 send_new_message 时 RG 不发文本但有表情) + elif emoji_prepared_info: # 只有表情 (可能是 direct_reply 带表情，或 send_new_message 时 RG 不发文本但有表情) send_order = ["emoji"] for item_type in send_order: - current_send_time = time.time() # 每次发送前获取精确时间 + current_send_time = time.time() # 每次发送前获取精确时间 if item_type == "text" and text_to_send: - self.conversation.generated_reply = text_to_send # 用于日志和历史记录 + self.conversation.generated_reply = text_to_send # 用于日志和历史记录 text_segment = Seg(type="text", data=text_to_send) if await self._send_reply_or_segments([text_segment], text_to_send): sent_text_successfully = True await self._update_bot_message_in_history(current_send_time, text_to_send, observation_info) if self.conversation.conversation_info: - self.conversation.conversation_info.current_instance_message_count +=1 - self.conversation.conversation_info.my_message_count += 1 # 文本发送成功，增加计数 + self.conversation.conversation_info.current_instance_message_count += 1 + self.conversation.conversation_info.my_message_count += 1 # 文本发送成功，增加计数 final_reason_parts.append(f"成功发送文本 ('{text_to_send[:20]}...')") else: final_reason_parts.append("发送文本失败") # 如果文本发送失败，通常不应继续发送表情，除非有特殊需求 - break + break elif item_type == "emoji" and emoji_prepared_info: emoji_segment, full_emoji_desc, log_emoji_desc = emoji_prepared_info if await self._send_reply_or_segments([emoji_segment], log_emoji_desc): sent_emoji_successfully = True full_emoji_description_if_sent = full_emoji_desc - await self._update_bot_message_in_history(current_send_time, full_emoji_desc, observation_info, "bot_emoji_") + await self._update_bot_message_in_history( + current_send_time, full_emoji_desc, observation_info, "bot_emoji_" + ) if self.conversation.conversation_info: - self.conversation.conversation_info.current_instance_message_count +=1 - self.conversation.conversation_info.my_message_count += 1 # 表情发送成功，增加计数 + self.conversation.conversation_info.current_instance_message_count += 1 + self.conversation.conversation_info.my_message_count += 1 # 表情发送成功，增加计数 final_reason_parts.append(f"成功发送表情 ({full_emoji_desc})") else: final_reason_parts.append("发送表情失败") # 如果表情发送失败，但文本已成功，也应记录 - if not text_to_send : # 如果只有表情且表情失败 + if not text_to_send: # 如果只有表情且表情失败 break - + return ( sent_text_successfully, sent_emoji_successfully, final_reason_parts, full_emoji_description_if_sent, need_replan_text, - text_check_reason if not is_suitable_text else "文本检查通过或未执行", # 返回文本检查失败的原因 - rg_decided_not_to_send_text + text_check_reason if not is_suitable_text else "文本检查通过或未执行", # 返回文本检查失败的原因 + rg_decided_not_to_send_text, ) class DirectReplyHandler(BaseTextReplyHandler): """处理直接回复动作（direct_reply）的处理器。""" + async def execute( self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, - current_action_record: dict + current_action_record: dict, ) -> tuple[bool, str, str]: """ 执行直接回复动作。 会尝试生成文本回复，并根据 current_emoji_query 发送附带表情。 """ if not observation_info or not conversation_info: - self.logger.error(f"[私聊][{self.conversation.private_name}] DirectReplyHandler: ObservationInfo 或 ConversationInfo 为空。") + self.logger.error( + f"[私聊][{self.conversation.private_name}] DirectReplyHandler: ObservationInfo 或 ConversationInfo 为空。" + ) return False, "error", "内部信息缺失，无法执行直接回复" - action_successful = False # 整体动作是否成功 - final_status = "recall" # 默认最终状态 - final_reason = "直接回复动作未成功执行" # 默认最终原因 + action_successful = False # 整体动作是否成功 + final_status = "recall" # 默认最终状态 + final_reason = "直接回复动作未成功执行" # 默认最终原因 max_reply_attempts: int = getattr(global_config, "pfc_max_reply_attempts", 3) ( @@ -595,12 +625,12 @@ class DirectReplyHandler(BaseTextReplyHandler): full_emoji_desc, need_replan_from_text_check, text_check_failure_reason, - _ # rg_decided_not_to_send_text, direct_reply 不关心这个 + _, # rg_decided_not_to_send_text, direct_reply 不关心这个 ) = await self._process_and_send_reply_with_optional_emoji( action_type="direct_reply", observation_info=observation_info, conversation_info=conversation_info, - max_reply_attempts=max_reply_attempts + max_reply_attempts=max_reply_attempts, ) # 根据发送结果决定最终状态 @@ -608,7 +638,7 @@ class DirectReplyHandler(BaseTextReplyHandler): action_successful = True final_status = "done" final_reason = "; ".join(reason_parts) if reason_parts else "成功完成操作" - + # 统一调用发送后状态更新 event_desc_parts = [] if sent_text_successfully and self.conversation.generated_reply: @@ -618,16 +648,18 @@ class DirectReplyHandler(BaseTextReplyHandler): event_desc = " ".join(event_desc_parts) if event_desc_parts else "机器人发送了消息" await self._update_post_send_states(observation_info, conversation_info, "direct_reply", event_desc) - elif need_replan_from_text_check: # 文本检查要求重规划 + elif need_replan_from_text_check: # 文本检查要求重规划 final_status = "recall" final_reason = f"文本回复检查要求重新规划: {text_check_failure_reason}" - conversation_info.last_successful_reply_action = None # 重置追问状态 - else: # 文本和表情都未能发送，或者文本检查失败且不需重规划（已达最大尝试） + conversation_info.last_successful_reply_action = None # 重置追问状态 + else: # 文本和表情都未能发送，或者文本检查失败且不需重规划（已达最大尝试） final_status = "max_checker_attempts_failed" if not need_replan_from_text_check else "recall" - final_reason = f"直接回复失败。文本检查: {text_check_failure_reason}. " + ("; ".join(reason_parts) if reason_parts else "") + final_reason = f"直接回复失败。文本检查: {text_check_failure_reason}. " + ( + "; ".join(reason_parts) if reason_parts else "" + ) action_successful = False - conversation_info.last_successful_reply_action = None # 重置追问状态 - + conversation_info.last_successful_reply_action = None # 重置追问状态 + # 清理 my_message_count (如果动作整体不成功，但部分发送了，需要调整) if not action_successful and conversation_info: # _process_and_send_reply_with_optional_emoji 内部会增加 my_message_count @@ -642,13 +674,14 @@ class DirectReplyHandler(BaseTextReplyHandler): class SendNewMessageHandler(BaseTextReplyHandler): """处理发送新消息动作（send_new_message）的处理器。""" + async def execute( self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, - current_action_record: dict + current_action_record: dict, ) -> tuple[bool, str, str]: """ 执行发送新消息动作。 @@ -656,12 +689,14 @@ class SendNewMessageHandler(BaseTextReplyHandler): 同时，也可能根据 current_emoji_query 发送附带表情。 """ if not observation_info or not conversation_info: - self.logger.error(f"[私聊][{self.conversation.private_name}] SendNewMessageHandler: ObservationInfo 或 ConversationInfo 为空。") + self.logger.error( + f"[私聊][{self.conversation.private_name}] SendNewMessageHandler: ObservationInfo 或 ConversationInfo 为空。" + ) return False, "error", "内部信息缺失，无法执行发送新消息" - action_successful = False # 整体动作是否成功 - final_status = "recall" # 默认最终状态 - final_reason = "发送新消息动作未成功执行" # 默认最终原因 + action_successful = False # 整体动作是否成功 + final_status = "recall" # 默认最终状态 + final_reason = "发送新消息动作未成功执行" # 默认最终原因 max_reply_attempts: int = getattr(global_config, "pfc_max_reply_attempts", 3) ( @@ -671,35 +706,39 @@ class SendNewMessageHandler(BaseTextReplyHandler): full_emoji_desc, need_replan_from_text_check, text_check_failure_reason, - rg_decided_not_to_send_text # 重要：获取RG是否决定不发文本 + rg_decided_not_to_send_text, # 重要：获取RG是否决定不发文本 ) = await self._process_and_send_reply_with_optional_emoji( action_type="send_new_message", observation_info=observation_info, conversation_info=conversation_info, - max_reply_attempts=max_reply_attempts + max_reply_attempts=max_reply_attempts, ) # 根据发送结果和RG的决策决定最终状态 - if rg_decided_not_to_send_text: # ReplyGenerator 明确决定不发送文本 - if sent_emoji_successfully: # 但表情成功发送了 + if rg_decided_not_to_send_text: # ReplyGenerator 明确决定不发送文本 + if sent_emoji_successfully: # 但表情成功发送了 action_successful = True - final_status = "done" # 整体算完成，因为有内容发出 + final_status = "done" # 整体算完成，因为有内容发出 final_reason = f"回复生成器决定不发送文本，但成功发送了附带表情 ({full_emoji_desc or '未知表情'})" # 即使只发了表情，也算一次交互，可以更新post_send_states event_desc = f"你发送了表情: '{full_emoji_desc or '未知表情'}' (文本未发送)" await self._update_post_send_states(observation_info, conversation_info, "send_new_message", event_desc) - else: # RG不发文本，表情也没发出去或失败 - action_successful = True # 决策本身是成功的（决定不发） - final_status = "done_no_reply" # 标记为完成但无回复 - final_reason = text_check_failure_reason if text_check_failure_reason and text_check_failure_reason != "文本检查通过或未执行" else "回复生成器决定不发送消息，且无表情或表情发送失败" - conversation_info.last_successful_reply_action = None # 因为没有文本发出 - if self.conversation.conversation_info: # 确保 my_message_count 被重置 - self.conversation.conversation_info.my_message_count = 0 - elif sent_text_successfully or sent_emoji_successfully: # RG决定发文本（或未明确反对），且至少有一个发出去了 + else: # RG不发文本，表情也没发出去或失败 + action_successful = True # 决策本身是成功的（决定不发） + final_status = "done_no_reply" # 标记为完成但无回复 + final_reason = ( + text_check_failure_reason + if text_check_failure_reason and text_check_failure_reason != "文本检查通过或未执行" + else "回复生成器决定不发送消息，且无表情或表情发送失败" + ) + conversation_info.last_successful_reply_action = None # 因为没有文本发出 + if self.conversation.conversation_info: # 确保 my_message_count 被重置 + self.conversation.conversation_info.my_message_count = 0 + elif sent_text_successfully or sent_emoji_successfully: # RG决定发文本（或未明确反对），且至少有一个发出去了 action_successful = True final_status = "done" final_reason = "; ".join(reason_parts) if reason_parts else "成功完成操作" - + event_desc_parts = [] if sent_text_successfully and self.conversation.generated_reply: event_desc_parts.append(f"你发送了新消息: '{self.conversation.generated_reply[:30]}...'") @@ -708,46 +747,51 @@ class SendNewMessageHandler(BaseTextReplyHandler): event_desc = " ".join(event_desc_parts) if event_desc_parts else "机器人发送了消息" await self._update_post_send_states(observation_info, conversation_info, "send_new_message", event_desc) - elif need_replan_from_text_check: # 文本检查要求重规划 + elif need_replan_from_text_check: # 文本检查要求重规划 final_status = "recall" final_reason = f"文本回复检查要求重新规划: {text_check_failure_reason}" conversation_info.last_successful_reply_action = None - else: # 文本和表情都未能发送（且RG没有明确说不发文本），或者文本检查失败且不需重规划 + else: # 文本和表情都未能发送（且RG没有明确说不发文本），或者文本检查失败且不需重规划 final_status = "max_checker_attempts_failed" if not need_replan_from_text_check else "recall" - final_reason = f"发送新消息失败。文本检查: {text_check_failure_reason}. " + ("; ".join(reason_parts) if reason_parts else "") + final_reason = f"发送新消息失败。文本检查: {text_check_failure_reason}. " + ( + "; ".join(reason_parts) if reason_parts else "" + ) action_successful = False conversation_info.last_successful_reply_action = None - + if not action_successful and conversation_info: # 同 DirectReplyHandler，my_message_count 的精确回滚依赖 last_successful_reply_action 的清除 pass - + return action_successful, final_status, final_reason.strip() class SayGoodbyeHandler(ActionHandler): """处理发送告别语动作（say_goodbye）的处理器。""" + async def execute( self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, - current_action_record: dict + current_action_record: dict, ) -> tuple[bool, str, str]: """ 执行发送告别语的动作。 会生成告别文本并发送，然后标记对话结束。 """ if not observation_info or not conversation_info: - self.logger.error(f"[私聊][{self.conversation.private_name}] SayGoodbyeHandler: ObservationInfo 或 ConversationInfo 为空。") + self.logger.error( + f"[私聊][{self.conversation.private_name}] SayGoodbyeHandler: ObservationInfo 或 ConversationInfo 为空。" + ) return False, "error", "内部信息缺失，无法执行告别" action_successful = False - final_status = "recall" # 默认状态 - final_reason = "告别语动作未成功执行" # 默认原因 + final_status = "recall" # 默认状态 + final_reason = "告别语动作未成功执行" # 默认原因 - self.conversation.state = ConversationState.GENERATING # 设置状态为生成中 + self.conversation.state = ConversationState.GENERATING # 设置状态为生成中 if not self.conversation.reply_generator: raise RuntimeError(f"ReplyGenerator 未为 {self.conversation.private_name} 初始化") @@ -759,96 +803,103 @@ class SayGoodbyeHandler(ActionHandler): f"[私聊][{self.conversation.private_name}] 动作 'say_goodbye': 生成内容: '{generated_content[:100]}...'" ) - if not generated_content or generated_content.startswith("抱歉"): # 如果生成内容无效 + if not generated_content or generated_content.startswith("抱歉"): # 如果生成内容无效 self.logger.warning( f"[私聊][{self.conversation.private_name}] 动作 'say_goodbye': 生成内容为空或为错误提示，取消发送。" ) final_reason = "生成告别内容无效" - final_status = "done" # 即使不发送，结束对话的决策也算完成 - self.conversation.should_continue = False # 标记对话结束 - action_successful = True # 动作（决策结束）本身算成功 - else: # 如果生成内容有效 + final_status = "done" # 即使不发送，结束对话的决策也算完成 + self.conversation.should_continue = False # 标记对话结束 + action_successful = True # 动作（决策结束）本身算成功 + else: # 如果生成内容有效 self.conversation.generated_reply = generated_content - self.conversation.state = ConversationState.SENDING # 设置状态为发送中 + self.conversation.state = ConversationState.SENDING # 设置状态为发送中 text_segment = Seg(type="text", data=self.conversation.generated_reply) send_success = await self._send_reply_or_segments([text_segment], self.conversation.generated_reply) send_end_time = time.time() - if send_success: # 如果发送成功 + if send_success: # 如果发送成功 action_successful = True final_status = "done" final_reason = "成功发送告别语" - self.conversation.should_continue = False # 标记对话结束 + self.conversation.should_continue = False # 标记对话结束 if self.conversation.conversation_info: - self.conversation.conversation_info.current_instance_message_count +=1 - self.conversation.conversation_info.my_message_count += 1 # 告别语也算一次发言 - await self._update_bot_message_in_history(send_end_time, self.conversation.generated_reply, observation_info) + self.conversation.conversation_info.current_instance_message_count += 1 + self.conversation.conversation_info.my_message_count += 1 # 告别语也算一次发言 + await self._update_bot_message_in_history( + send_end_time, self.conversation.generated_reply, observation_info + ) event_desc = f"你发送了告别消息: '{self.conversation.generated_reply[:50]}...'" # 注意：由于 should_continue 已设为 False，后续的 idle chat 更新可能意义不大，但情绪更新仍可进行 await self._update_post_send_states(observation_info, conversation_info, "say_goodbye", event_desc) - else: # 如果发送失败 + else: # 如果发送失败 final_status = "recall" final_reason = "发送告别语失败" action_successful = False - self.conversation.should_continue = True # 发送失败则不立即结束对话，让其自然流转 + self.conversation.should_continue = True # 发送失败则不立即结束对话，让其自然流转 return action_successful, final_status, final_reason class SendMemesHandler(ActionHandler): """处理发送表情包动作（send_memes）的处理器。""" + async def execute( self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, - current_action_record: dict + current_action_record: dict, ) -> tuple[bool, str, str]: """ 执行发送表情包的动作。 会根据 current_emoji_query 获取并发送表情。 """ if not observation_info or not conversation_info: - self.logger.error(f"[私聊][{self.conversation.private_name}] SendMemesHandler: ObservationInfo 或 ConversationInfo 为空。") + self.logger.error( + f"[私聊][{self.conversation.private_name}] SendMemesHandler: ObservationInfo 或 ConversationInfo 为空。" + ) return False, "error", "内部信息缺失，无法发送表情包" action_successful = False - final_status = "recall" # 默认状态 + final_status = "recall" # 默认状态 final_reason_prefix = "发送表情包" - final_reason = f"{final_reason_prefix}失败：未知原因" # 默认原因 - self.conversation.state = ConversationState.GENERATING # 或 SENDING_MEME + final_reason = f"{final_reason_prefix}失败：未知原因" # 默认原因 + self.conversation.state = ConversationState.GENERATING # 或 SENDING_MEME emoji_query = conversation_info.current_emoji_query - if not emoji_query: # 如果没有表情查询 + if not emoji_query: # 如果没有表情查询 final_reason = f"{final_reason_prefix}失败：缺少表情包查询语句" # 此动作不依赖文本回复的追问状态，所以不修改 last_successful_reply_action return False, "recall", final_reason - + # 清理表情查询，因为我们要处理它了 conversation_info.current_emoji_query = None emoji_prepared_info = await self._fetch_and_prepare_emoji_segment(emoji_query) - if emoji_prepared_info: # 如果成功获取并准备了表情 + if emoji_prepared_info: # 如果成功获取并准备了表情 emoji_segment, full_emoji_description, log_emoji_description = emoji_prepared_info send_success = await self._send_reply_or_segments([emoji_segment], log_emoji_description) send_end_time = time.time() - if send_success: # 如果发送成功 + if send_success: # 如果发送成功 action_successful = True final_status = "done" final_reason = f"{final_reason_prefix}成功发送 ({full_emoji_description})" if self.conversation.conversation_info: - self.conversation.conversation_info.current_instance_message_count +=1 - self.conversation.conversation_info.my_message_count += 1 # 表情也算一次发言 - await self._update_bot_message_in_history(send_end_time, full_emoji_description, observation_info, "bot_meme_") + self.conversation.conversation_info.current_instance_message_count += 1 + self.conversation.conversation_info.my_message_count += 1 # 表情也算一次发言 + await self._update_bot_message_in_history( + send_end_time, full_emoji_description, observation_info, "bot_meme_" + ) event_desc = f"你发送了一个表情包 ({full_emoji_description})" await self._update_post_send_states(observation_info, conversation_info, "send_memes", event_desc) - else: # 如果发送失败 + else: # 如果发送失败 final_status = "recall" final_reason = f"{final_reason_prefix}失败：发送时出错" - else: # 如果未能获取或准备表情 + else: # 如果未能获取或准备表情 final_reason = f"{final_reason_prefix}失败：未找到或准备表情失败 ({emoji_query})" # last_successful_reply_action 保持不变 @@ -857,81 +908,143 @@ class SendMemesHandler(ActionHandler): class RethinkGoalHandler(ActionHandler): """处理重新思考目标动作（rethink_goal）的处理器。""" - async def execute(self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, current_action_record: dict) -> tuple[bool, str, str]: + + async def execute( + self, + reason: str, + observation_info: Optional[ObservationInfo], + conversation_info: Optional[ConversationInfo], + action_start_time: float, + current_action_record: dict, + ) -> tuple[bool, str, str]: """执行重新思考对话目标的动作。""" if not conversation_info or not observation_info: - self.logger.error(f"[私聊][{self.conversation.private_name}] RethinkGoalHandler: ObservationInfo 或 ConversationInfo 为空。") + self.logger.error( + f"[私聊][{self.conversation.private_name}] RethinkGoalHandler: ObservationInfo 或 ConversationInfo 为空。" + ) return False, "error", "内部信息缺失，无法重新思考目标" - self.conversation.state = ConversationState.RETHINKING # 设置状态为重新思考中 + self.conversation.state = ConversationState.RETHINKING # 设置状态为重新思考中 if not self.conversation.goal_analyzer: raise RuntimeError(f"GoalAnalyzer 未为 {self.conversation.private_name} 初始化") - await self.conversation.goal_analyzer.analyze_goal(conversation_info, observation_info) # 调用目标分析器 + await self.conversation.goal_analyzer.analyze_goal(conversation_info, observation_info) # 调用目标分析器 event_desc = "你重新思考了对话目标和方向" - await self._update_relationship_and_emotion(observation_info, conversation_info, event_desc) # 更新关系和情绪 + await self._update_relationship_and_emotion(observation_info, conversation_info, event_desc) # 更新关系和情绪 return True, "done", "成功重新思考目标" + class ListeningHandler(ActionHandler): """处理倾听动作（listening）的处理器。""" - async def execute(self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, current_action_record: dict) -> tuple[bool, str, str]: + + async def execute( + self, + reason: str, + observation_info: Optional[ObservationInfo], + conversation_info: Optional[ConversationInfo], + action_start_time: float, + current_action_record: dict, + ) -> tuple[bool, str, str]: """执行倾听对方发言的动作。""" if not conversation_info or not observation_info: - self.logger.error(f"[私聊][{self.conversation.private_name}] ListeningHandler: ObservationInfo 或 ConversationInfo 为空。") + self.logger.error( + f"[私聊][{self.conversation.private_name}] ListeningHandler: ObservationInfo 或 ConversationInfo 为空。" + ) return False, "error", "内部信息缺失，无法执行倾听" - self.conversation.state = ConversationState.LISTENING # 设置状态为倾听中 + self.conversation.state = ConversationState.LISTENING # 设置状态为倾听中 if not self.conversation.waiter: raise RuntimeError(f"Waiter 未为 {self.conversation.private_name} 初始化") - await self.conversation.waiter.wait_listening(conversation_info) # 调用等待器的倾听方法 + await self.conversation.waiter.wait_listening(conversation_info) # 调用等待器的倾听方法 event_desc = "你决定耐心倾听对方的发言" - await self._update_relationship_and_emotion(observation_info, conversation_info, event_desc) # 更新关系和情绪 + await self._update_relationship_and_emotion(observation_info, conversation_info, event_desc) # 更新关系和情绪 # listening 动作完成后，状态会由新消息或超时驱动，最终回到 ANALYZING return True, "done", "进入倾听状态" + class EndConversationHandler(ActionHandler): """处理结束对话动作（end_conversation）的处理器。""" - async def execute(self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, current_action_record: dict) -> tuple[bool, str, str]: + + async def execute( + self, + reason: str, + observation_info: Optional[ObservationInfo], + conversation_info: Optional[ConversationInfo], + action_start_time: float, + current_action_record: dict, + ) -> tuple[bool, str, str]: """执行结束当前对话的动作。""" - self.logger.info(f"[私聊][{self.conversation.private_name}] 动作 'end_conversation': 收到最终结束指令，停止对话...") - self.conversation.should_continue = False # 标记对话不应继续，主循环会因此退出 + self.logger.info( + f"[私聊][{self.conversation.private_name}] 动作 'end_conversation': 收到最终结束指令，停止对话..." + ) + self.conversation.should_continue = False # 标记对话不应继续，主循环会因此退出 # 注意：最终的关系评估通常在 Conversation.stop() 方法中进行 return True, "done", "对话结束指令已执行" + class BlockAndIgnoreHandler(ActionHandler): """处理屏蔽并忽略对话动作（block_and_ignore）的处理器。""" - async def execute(self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, current_action_record: dict) -> tuple[bool, str, str]: + + async def execute( + self, + reason: str, + observation_info: Optional[ObservationInfo], + conversation_info: Optional[ConversationInfo], + action_start_time: float, + current_action_record: dict, + ) -> tuple[bool, str, str]: """执行屏蔽并忽略当前对话一段时间的动作。""" - if not conversation_info or not observation_info: # 防御性检查 - self.logger.error(f"[私聊][{self.conversation.private_name}] BlockAndIgnoreHandler: ObservationInfo 或 ConversationInfo 为空。") + if not conversation_info or not observation_info: # 防御性检查 + self.logger.error( + f"[私聊][{self.conversation.private_name}] BlockAndIgnoreHandler: ObservationInfo 或 ConversationInfo 为空。" + ) return False, "error", "内部信息缺失，无法执行屏蔽" self.logger.info(f"[私聊][{self.conversation.private_name}] 动作 'block_and_ignore': 不想再理你了...") - ignore_duration_seconds = 10 * 60 # 例如忽略10分钟，可以配置 - self.conversation.ignore_until_timestamp = time.time() + ignore_duration_seconds # 设置忽略截止时间 - self.conversation.state = ConversationState.IGNORED # 设置状态为已忽略 + ignore_duration_seconds = 10 * 60 # 例如忽略10分钟，可以配置 + self.conversation.ignore_until_timestamp = time.time() + ignore_duration_seconds # 设置忽略截止时间 + self.conversation.state = ConversationState.IGNORED # 设置状态为已忽略 event_desc = "当前对话让你感到不适，你决定暂时不再理会对方" - await self._update_relationship_and_emotion(observation_info, conversation_info, event_desc) # 更新关系和情绪 + await self._update_relationship_and_emotion(observation_info, conversation_info, event_desc) # 更新关系和情绪 # should_continue 仍为 True，但主循环会检查 ignore_until_timestamp return True, "done", f"已屏蔽并忽略对话 {ignore_duration_seconds // 60} 分钟" + class WaitHandler(ActionHandler): """处理等待动作（wait）的处理器。""" - async def execute(self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, current_action_record: dict) -> tuple[bool, str, str]: + + async def execute( + self, + reason: str, + observation_info: Optional[ObservationInfo], + conversation_info: Optional[ConversationInfo], + action_start_time: float, + current_action_record: dict, + ) -> tuple[bool, str, str]: """执行等待对方回复的动作。""" - if not conversation_info or not observation_info: # 防御性检查 - self.logger.error(f"[私聊][{self.conversation.private_name}] WaitHandler: ObservationInfo 或 ConversationInfo 为空。") + if not conversation_info or not observation_info: # 防御性检查 + self.logger.error( + f"[私聊][{self.conversation.private_name}] WaitHandler: ObservationInfo 或 ConversationInfo 为空。" + ) return False, "error", "内部信息缺失，无法执行等待" - self.conversation.state = ConversationState.WAITING # 设置状态为等待中 + self.conversation.state = ConversationState.WAITING # 设置状态为等待中 if not self.conversation.waiter: raise RuntimeError(f"Waiter 未为 {self.conversation.private_name} 初始化") - timeout_occurred = await self.conversation.waiter.wait(conversation_info) # 调用等待器的常规等待方法 + timeout_occurred = await self.conversation.waiter.wait(conversation_info) # 调用等待器的常规等待方法 event_desc = "你等待对方回复，但对方长时间没有回应" if timeout_occurred else "你选择等待对方的回复" - await self._update_relationship_and_emotion(observation_info, conversation_info, event_desc) # 更新关系和情绪 + await self._update_relationship_and_emotion(observation_info, conversation_info, event_desc) # 更新关系和情绪 # wait 动作完成后，状态会由新消息或超时驱动，最终回到 ANALYZING return True, "done", "等待动作完成" + class UnknownActionHandler(ActionHandler): """处理未知或无效动作的处理器。""" - async def execute(self, reason: str, observation_info: Optional[ObservationInfo], conversation_info: Optional[ConversationInfo], action_start_time: float, current_action_record: dict) -> tuple[bool, str, str]: - """处理无法识别的动作类型。""" - action_name = current_action_record.get("action", "未知动作类型") # 从记录中获取动作名 - self.logger.warning(f"[私聊][{self.conversation.private_name}] 接收到未知的动作类型: {action_name}") - return False, "recall", f"未知的动作类型: {action_name}" # 标记为需要重新规划 + async def execute( + self, + reason: str, + observation_info: Optional[ObservationInfo], + conversation_info: Optional[ConversationInfo], + action_start_time: float, + current_action_record: dict, + ) -> tuple[bool, str, str]: + """处理无法识别的动作类型。""" + action_name = current_action_record.get("action", "未知动作类型") # 从记录中获取动作名 + self.logger.warning(f"[私聊][{self.conversation.private_name}] 接收到未知的动作类型: {action_name}") + return False, "recall", f"未知的动作类型: {action_name}" # 标记为需要重新规划 diff --git a/src/experimental/PFC/actions.py b/src/experimental/PFC/actions.py index e984e363..5c8233ba 100644 --- a/src/experimental/PFC/actions.py +++ b/src/experimental/PFC/actions.py @@ -5,17 +5,17 @@ import traceback from typing import Optional, TYPE_CHECKING from src.common.logger_manager import get_logger -from .pfc_types import ConversationState # 调整导入路径 -from .observation_info import ObservationInfo # 调整导入路径 -from .conversation_info import ConversationInfo # 调整导入路径 +from .pfc_types import ConversationState # 调整导入路径 +from .observation_info import ObservationInfo # 调整导入路径 +from .conversation_info import ConversationInfo # 调整导入路径 # 导入工厂类 -from .action_factory import StandardActionFactory # 调整导入路径 +from .action_factory import StandardActionFactory # 调整导入路径 if TYPE_CHECKING: - from .conversation import Conversation # 调整导入路径 + from .conversation import Conversation # 调整导入路径 -logger = get_logger("pfc_actions") # 模块级别日志记录器 +logger = get_logger("pfc_actions") # 模块级别日志记录器 async def handle_action( @@ -40,41 +40,39 @@ async def handle_action( # 如果 conversation_info 和 done_action 存在且不为空 if conversation_info and hasattr(conversation_info, "done_action") and conversation_info.done_action: # 更新最后一个动作记录的状态和原因 - if conversation_info.done_action: # 再次检查列表是否不为空 - conversation_info.done_action[-1].update( - {"status": "error", "final_reason": "ObservationInfo is None"} - ) - conversation_instance.state = ConversationState.ERROR # 设置对话状态为错误 + if conversation_info.done_action: # 再次检查列表是否不为空 + conversation_info.done_action[-1].update({"status": "error", "final_reason": "ObservationInfo is None"}) + conversation_instance.state = ConversationState.ERROR # 设置对话状态为错误 return # 检查 conversation_info 是否为空 if not conversation_info: logger.error(f"[私聊][{conversation_instance.private_name}] ConversationInfo 为空，无法处理动作 '{action}'。") - conversation_instance.state = ConversationState.ERROR # 设置对话状态为错误 + conversation_instance.state = ConversationState.ERROR # 设置对话状态为错误 return logger.info(f"[私聊][{conversation_instance.private_name}] 开始处理动作: {action}, 原因: {reason}") - action_start_time = time.time() # 记录动作开始时间 + action_start_time = time.time() # 记录动作开始时间 # 当前动作记录 current_action_record = { - "action": action, # 动作类型 - "plan_reason": reason, # 规划原因 - "status": "start", # 初始状态为 "start" - "time": datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"), # 当前时间 - "final_reason": None, # 最终原因，默认为 None + "action": action, # 动作类型 + "plan_reason": reason, # 规划原因 + "status": "start", # 初始状态为 "start" + "time": datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"), # 当前时间 + "final_reason": None, # 最终原因，默认为 None } # 如果 done_action 不存在或为空，则初始化 if not hasattr(conversation_info, "done_action") or conversation_info.done_action is None: conversation_info.done_action = [] - conversation_info.done_action.append(current_action_record) # 添加当前动作记录 - action_index = len(conversation_info.done_action) - 1 # 获取当前动作记录的索引 + conversation_info.done_action.append(current_action_record) # 添加当前动作记录 + action_index = len(conversation_info.done_action) - 1 # 获取当前动作记录的索引 - action_successful: bool = False # 动作是否成功，默认为 False - final_status: str = "recall" # 最终状态，默认为 "recall" - final_reason: str = "动作未成功执行" # 最终原因，默认为 "动作未成功执行" + action_successful: bool = False # 动作是否成功，默认为 False + final_status: str = "recall" # 最终状态，默认为 "recall" + final_reason: str = "动作未成功执行" # 最终原因，默认为 "动作未成功执行" - factory = StandardActionFactory() # 创建标准动作工厂实例 - action_handler = factory.create_action_handler(action, conversation_instance) # 创建动作处理器 + factory = StandardActionFactory() # 创建标准动作工厂实例 + action_handler = factory.create_action_handler(action, conversation_instance) # 创建动作处理器 try: # 执行动作处理器 @@ -86,33 +84,33 @@ async def handle_action( # 此部分之前位于每个 if/elif 块内部 # 如果动作不是回复类型的动作 if action not in ["direct_reply", "send_new_message", "say_goodbye", "send_memes"]: - conversation_info.last_successful_reply_action = None # 清除上次成功回复动作 - conversation_info.last_reply_rejection_reason = None # 清除上次回复拒绝原因 - conversation_info.last_rejected_reply_content = None # 清除上次拒绝的回复内容 + conversation_info.last_successful_reply_action = None # 清除上次成功回复动作 + conversation_info.last_reply_rejection_reason = None # 清除上次回复拒绝原因 + conversation_info.last_rejected_reply_content = None # 清除上次拒绝的回复内容 # 如果动作不是发送表情包或发送表情包失败，则清除表情查询 if action != "send_memes" or not action_successful: if hasattr(conversation_info, "current_emoji_query"): conversation_info.current_emoji_query = None - except asyncio.CancelledError: # 捕获任务取消错误 + except asyncio.CancelledError: # 捕获任务取消错误 logger.warning(f"[私聊][{conversation_instance.private_name}] 处理动作 '{action}' 时被取消。") - final_status = "cancelled" # 设置最终状态为 "cancelled" + final_status = "cancelled" # 设置最终状态为 "cancelled" final_reason = "动作处理被取消" # 如果 conversation_info 存在 if conversation_info: - conversation_info.last_successful_reply_action = None # 清除上次成功回复动作 - raise # 重新抛出异常，由循环处理 - except Exception as handle_err: # 捕获其他异常 + conversation_info.last_successful_reply_action = None # 清除上次成功回复动作 + raise # 重新抛出异常，由循环处理 + except Exception as handle_err: # 捕获其他异常 logger.error(f"[私聊][{conversation_instance.private_name}] 处理动作 '{action}' 时出错: {handle_err}") logger.error(f"[私聊][{conversation_instance.private_name}] {traceback.format_exc()}") - final_status = "error" # 设置最终状态为 "error" + final_status = "error" # 设置最终状态为 "error" final_reason = f"处理动作时出错: {handle_err}" - conversation_instance.state = ConversationState.ERROR # 设置对话状态为错误 + conversation_instance.state = ConversationState.ERROR # 设置对话状态为错误 # 如果 conversation_info 存在 if conversation_info: - conversation_info.last_successful_reply_action = None # 清除上次成功回复动作 - action_successful = False # 确保动作为不成功 + conversation_info.last_successful_reply_action = None # 清除上次成功回复动作 + action_successful = False # 确保动作为不成功 finally: # 更新动作历史记录 @@ -130,45 +128,43 @@ async def handle_action( final_reason = f"动作 {action} 成功完成" # 如果是发送表情包且 current_emoji_query 存在（理想情况下从处理器获取描述） if action == "send_memes" and conversation_info.current_emoji_query: - pass # 占位符 - 表情描述最好从处理器的执行结果中获取并用于原因 + pass # 占位符 - 表情描述最好从处理器的执行结果中获取并用于原因 # 更新动作记录 conversation_info.done_action[action_index].update( { - "status": final_status, # 最终状态 - "time_completed": datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"), # 完成时间 - "final_reason": final_reason, # 最终原因 - "duration_ms": int((time.time() - action_start_time) * 1000), # 持续时间（毫秒） + "status": final_status, # 最终状态 + "time_completed": datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"), # 完成时间 + "final_reason": final_reason, # 最终原因 + "duration_ms": int((time.time() - action_start_time) * 1000), # 持续时间（毫秒） } ) - else: # 如果无法更新动作历史记录 + else: # 如果无法更新动作历史记录 logger.error( f"[私聊][{conversation_instance.private_name}] 无法更新动作历史记录，done_action 无效或索引 {action_index} 超出范围。" ) - # 根据最终状态设置对话状态 if final_status in ["done", "done_no_reply", "recall"]: - conversation_instance.state = ConversationState.ANALYZING # 设置为分析中 + conversation_instance.state = ConversationState.ANALYZING # 设置为分析中 elif final_status in ["error", "max_checker_attempts_failed"]: - conversation_instance.state = ConversationState.ERROR # 设置为错误 + conversation_instance.state = ConversationState.ERROR # 设置为错误 # 其他状态如 LISTENING, WAITING, IGNORED, ENDED 在各自的处理器内部或由循环设置。 # 此处移至 try 块以确保即使在发生异常之前也运行 # 如果动作不是回复类型的动作 if action not in ["direct_reply", "send_new_message", "say_goodbye", "send_memes"]: - if conversation_info: # 再次检查 conversation_info 是否不为 None - conversation_info.last_successful_reply_action = None # 清除上次成功回复动作 - conversation_info.last_reply_rejection_reason = None # 清除上次回复拒绝原因 - conversation_info.last_rejected_reply_content = None # 清除上次拒绝的回复内容 + if conversation_info: # 再次检查 conversation_info 是否不为 None + conversation_info.last_successful_reply_action = None # 清除上次成功回复动作 + conversation_info.last_reply_rejection_reason = None # 清除上次回复拒绝原因 + conversation_info.last_rejected_reply_content = None # 清除上次拒绝的回复内容 # 如果动作不是发送表情包或发送表情包失败 if action != "send_memes" or not action_successful: # 如果 conversation_info 存在且有 current_emoji_query 属性 if conversation_info and hasattr(conversation_info, "current_emoji_query"): - conversation_info.current_emoji_query = None # 清除当前表情查询 + conversation_info.current_emoji_query = None # 清除当前表情查询 - - log_final_reason_msg = final_reason if final_reason else "无明确原因" # 记录的最终原因消息 + log_final_reason_msg = final_reason if final_reason else "无明确原因" # 记录的最终原因消息 # 如果最终状态为 "done"，动作成功，且是直接回复或发送新消息，并且有生成的回复 if ( final_status == "done" @@ -179,8 +175,8 @@ async def handle_action( ): log_final_reason_msg += f" (发送内容: '{conversation_instance.generated_reply[:30]}...')" # elif final_status == "done" and action_successful and action == "send_memes": - # 表情包的日志记录在其处理器内部或通过下面的通用日志处理 + # 表情包的日志记录在其处理器内部或通过下面的通用日志处理 logger.info( f"[私聊][{conversation_instance.private_name}] 动作 '{action}' 处理完成。最终状态: {final_status}, 原因: {log_final_reason_msg}" - ) \ No newline at end of file + )