mirror of https://github.com/Mai-with-u/MaiBot.git
长期记忆 第一版
114514
parent
b9085f4b05
commit
ed7e339907
|
|
@ -1,123 +1,171 @@
|
|||
import traceback
|
||||
|
||||
from maim_message import UserInfo
|
||||
import re
|
||||
from typing import Any, Dict
|
||||
from datetime import datetime # 确保导入 datetime
|
||||
from maim_message import UserInfo, MessageRecv # 从 maim_message 导入 MessageRecv
|
||||
from src.config.config import global_config
|
||||
from src.common.logger_manager import get_logger
|
||||
from ..chat.chat_stream import chat_manager
|
||||
from typing import Optional, Dict, Any
|
||||
from ..chat.chat_stream import chat_manager
|
||||
from src.plugins.chat.utils import get_embedding
|
||||
from src.common.database import db
|
||||
from .pfc_manager import PFCManager
|
||||
from src.plugins.chat.message import MessageRecv
|
||||
from src.plugins.storage.storage import MessageStorage
|
||||
from datetime import datetime
|
||||
|
||||
|
||||
logger = get_logger("pfc_processor")
|
||||
|
||||
|
||||
async def _handle_error(error: Exception, context: str, message: Optional[MessageRecv] = None) -> None:
|
||||
async def _handle_error(error: Exception, context: str, message: MessageRecv | None = None) -> None: # 明确 message 类型
|
||||
"""统一的错误处理函数
|
||||
|
||||
Args:
|
||||
error: 捕获到的异常
|
||||
context: 错误发生的上下文描述
|
||||
message: 可选的消息对象，用于记录相关消息内容
|
||||
# ... (方法注释不变) ...
|
||||
"""
|
||||
logger.error(f"{context}: {error}")
|
||||
logger.error(traceback.format_exc())
|
||||
if message and hasattr(message, "raw_message"):
|
||||
# 检查 message 是否 None 以及是否有 raw_message 属性
|
||||
if message and hasattr(message, 'message_info') and hasattr(message.message_info, 'raw_message'): # MessageRecv 结构可能没有直接的 raw_message
|
||||
raw_msg_content = getattr(message.message_info, 'raw_message', None) # 安全获取
|
||||
if raw_msg_content:
|
||||
logger.error(f"相关消息原始内容: {raw_msg_content}")
|
||||
elif message and hasattr(message, 'raw_message'): # 如果 MessageRecv 直接有 raw_message
|
||||
logger.error(f"相关消息原始内容: {message.raw_message}")
|
||||
|
||||
|
||||
class PFCProcessor:
|
||||
"""PFC 处理器，负责处理接收到的信息并计数"""
|
||||
|
||||
def __init__(self):
|
||||
"""初始化 PFC 处理器，创建消息存储实例"""
|
||||
self.storage = MessageStorage()
|
||||
# MessageStorage() 的实例化位置和具体类是什么？
|
||||
# 我们假设它来自 src.plugins.storage.storage
|
||||
# 但由于我们不能修改那个文件，所以这里的 self.storage 将按原样使用
|
||||
from src.plugins.storage.storage import MessageStorage # 明确导入，以便类型提示和理解
|
||||
self.storage: MessageStorage = MessageStorage()
|
||||
self.pfc_manager = PFCManager.get_instance()
|
||||
|
||||
async def process_message(self, message_data: Dict[str, Any]) -> None:
|
||||
async def process_message(self, message_data: dict[str, Any]) -> None: # 使用 dict[str, Any] 替代 Dict
|
||||
"""处理接收到的原始消息数据
|
||||
|
||||
主要流程:
|
||||
1. 消息解析与初始化
|
||||
2. 过滤检查
|
||||
3. 消息存储
|
||||
4. 创建 PFC 流
|
||||
5. 日志记录
|
||||
|
||||
Args:
|
||||
message_data: 原始消息字符串
|
||||
# ... (方法注释不变) ...
|
||||
"""
|
||||
message = None
|
||||
message_obj: MessageRecv | None = None # 初始化为 None，并明确类型
|
||||
try:
|
||||
# 1. 消息解析与初始化
|
||||
message = MessageRecv(message_data)
|
||||
groupinfo = message.message_info.group_info
|
||||
userinfo = message.message_info.user_info
|
||||
messageinfo = message.message_info
|
||||
message_obj = MessageRecv(message_data) # 使用你提供的 message.py 中的 MessageRecv
|
||||
# 确保 message_obj.message_info 存在
|
||||
if not hasattr(message_obj, 'message_info'):
|
||||
logger.error("MessageRecv 对象缺少 message_info 属性。跳过处理。")
|
||||
return
|
||||
|
||||
groupinfo = getattr(message_obj.message_info, 'group_info', None)
|
||||
userinfo = getattr(message_obj.message_info, 'user_info', None)
|
||||
|
||||
if userinfo is None: # 确保 userinfo 存在
|
||||
logger.error("message_obj.message_info 中缺少 user_info。跳过处理。")
|
||||
return
|
||||
if not hasattr(userinfo, 'user_id'): # 确保 user_id 存在
|
||||
logger.error("userinfo 对象中缺少 user_id。跳过处理。")
|
||||
return
|
||||
|
||||
logger.trace(f"准备为{userinfo.user_id}创建/获取聊天流")
|
||||
chat = await chat_manager.get_or_create_stream(
|
||||
platform=messageinfo.platform,
|
||||
platform=message_obj.message_info.platform,
|
||||
user_info=userinfo,
|
||||
group_info=groupinfo,
|
||||
)
|
||||
message.update_chat_stream(chat)
|
||||
message_obj.update_chat_stream(chat) # message.py 中 MessageRecv 有此方法
|
||||
|
||||
# 2. 过滤检查
|
||||
# 处理消息
|
||||
await message.process()
|
||||
# 过滤词/正则表达式过滤
|
||||
if self._check_ban_words(message.processed_plain_text, userinfo) or self._check_ban_regex(
|
||||
message.raw_message, userinfo
|
||||
):
|
||||
await message_obj.process() # 调用 MessageRecv 的异步 process 方法
|
||||
if self._check_ban_words(message_obj.processed_plain_text, userinfo) or \
|
||||
self._check_ban_regex(message_obj.raw_message, userinfo): # MessageRecv 有 raw_message 属性
|
||||
return
|
||||
|
||||
# 3. 消息存储
|
||||
await self.storage.store_message(message, chat)
|
||||
logger.trace(f"存储成功: {message.processed_plain_text}")
|
||||
# 3. 消息存储 (保持原有调用)
|
||||
# 这里的 self.storage.store_message 来自 src/plugins/storage/storage.py
|
||||
# 它内部会将 message_obj 转换为字典并存储
|
||||
await self.storage.store_message(message_obj, chat)
|
||||
logger.trace(f"存储成功 (初步): {message_obj.processed_plain_text}")
|
||||
|
||||
# === 新增：为已存储的消息生成嵌入并更新数据库文档 ===
|
||||
embedding_vector = None
|
||||
text_for_embedding = message_obj.processed_plain_text # 使用处理后的纯文本
|
||||
|
||||
# 在 storage.py 中，会对 processed_plain_text 进行一次过滤
|
||||
# 为了保持一致，我们也在这里应用相同的过滤逻辑
|
||||
# 当然，更优的做法是 store_message 返回过滤后的文本，或在 message_obj 中增加一个 filtered_processed_plain_text 属性
|
||||
# 这里为了简单，我们先重复一次过滤逻辑
|
||||
pattern = r"<MainRule>.*?</MainRule>|<schedule>.*?</schedule>|<UserMessage>.*?</UserMessage>"
|
||||
if text_for_embedding:
|
||||
filtered_text_for_embedding = re.sub(pattern, "", text_for_embedding, flags=re.DOTALL)
|
||||
else:
|
||||
filtered_text_for_embedding = ""
|
||||
|
||||
if filtered_text_for_embedding and filtered_text_for_embedding.strip():
|
||||
try:
|
||||
# request_type 参数根据你的 get_embedding 函数实际需求来定
|
||||
embedding_vector = await get_embedding(filtered_text_for_embedding, request_type="pfc_private_memory")
|
||||
if embedding_vector:
|
||||
logger.debug(f"成功为消息 ID '{message_obj.message_info.message_id}' 生成嵌入向量。")
|
||||
|
||||
# 更新数据库中的对应文档
|
||||
# 确保你有权限访问和操作 db 对象
|
||||
update_result = await db.messages.update_one(
|
||||
{"message_id": message_obj.message_info.message_id, "chat_id": chat.stream_id},
|
||||
{"$set": {"embedding_vector": embedding_vector}}
|
||||
)
|
||||
if update_result.modified_count > 0:
|
||||
logger.info(f"成功为消息 ID '{message_obj.message_info.message_id}' 更新嵌入向量到数据库。")
|
||||
elif update_result.matched_count > 0:
|
||||
logger.warning(f"消息 ID '{message_obj.message_info.message_id}' 已存在嵌入向量或未作修改。")
|
||||
else:
|
||||
logger.error(f"未能找到消息 ID '{message_obj.message_info.message_id}' (chat_id: {chat.stream_id}) 来更新嵌入向量。可能是存储和更新之间存在延迟或问题。")
|
||||
else:
|
||||
logger.warning(f"未能为消息 ID '{message_obj.message_info.message_id}' 的文本 '{filtered_text_for_embedding[:30]}...' 生成嵌入向量。")
|
||||
except Exception as e_embed_update:
|
||||
logger.error(f"为消息 ID '{message_obj.message_info.message_id}' 生成嵌入或更新数据库时发生异常: {e_embed_update}", exc_info=True)
|
||||
else:
|
||||
logger.debug(f"消息 ID '{message_obj.message_info.message_id}' 的过滤后纯文本为空，不生成或更新嵌入。")
|
||||
# === 新增结束 ===
|
||||
|
||||
# 4. 创建 PFC 聊天流
|
||||
await self._create_pfc_chat(message)
|
||||
await self._create_pfc_chat(message_obj)
|
||||
|
||||
# 5. 日志记录
|
||||
# 将时间戳转换为datetime对象
|
||||
current_time = datetime.fromtimestamp(message.message_info.time).strftime("%H:%M:%S")
|
||||
# 确保 message_obj.message_info.time 是 float 类型的时间戳
|
||||
current_time_display = datetime.fromtimestamp(float(message_obj.message_info.time)).strftime("%H:%M:%S")
|
||||
|
||||
# 确保 userinfo.user_nickname 存在
|
||||
user_nickname_display = getattr(userinfo, 'user_nickname', '未知用户')
|
||||
|
||||
logger.info(
|
||||
f"[{current_time}][私聊]{message.message_info.user_info.user_nickname}: {message.processed_plain_text}"
|
||||
f"[{current_time_display}][私聊]{user_nickname_display}: {message_obj.processed_plain_text}"
|
||||
)
|
||||
|
||||
except Exception as e:
|
||||
await _handle_error(e, "消息处理失败", message)
|
||||
await _handle_error(e, "消息处理失败", message_obj) # 传递 message_obj
|
||||
|
||||
async def _create_pfc_chat(self, message: MessageRecv):
|
||||
async def _create_pfc_chat(self, message: MessageRecv): # 明确 message 类型
|
||||
try:
|
||||
chat_id = str(message.chat_stream.stream_id)
|
||||
private_name = str(message.message_info.user_info.user_nickname)
|
||||
private_name = str(message.message_info.user_info.user_nickname) # 假设 UserInfo 有 user_nickname
|
||||
|
||||
if global_config.enable_pfc_chatting:
|
||||
await self.pfc_manager.get_or_create_conversation(chat_id, private_name)
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"创建PFC聊天失败: {e}")
|
||||
logger.error(f"创建PFC聊天失败: {e}", exc_info=True) # 添加 exc_info=True
|
||||
|
||||
@staticmethod
|
||||
def _check_ban_words(text: str, userinfo: UserInfo) -> bool:
|
||||
def _check_ban_words(text: str, userinfo: UserInfo) -> bool: # 明确 userinfo 类型
|
||||
"""检查消息中是否包含过滤词"""
|
||||
for word in global_config.ban_words:
|
||||
if word in text:
|
||||
logger.info(f"[私聊]{userinfo.user_nickname}:{text}")
|
||||
logger.info(f"[私聊]{userinfo.user_nickname}:{text}") # 假设 UserInfo 有 user_nickname
|
||||
logger.info(f"[过滤词识别]消息中含有{word}，filtered")
|
||||
return True
|
||||
return False
|
||||
|
||||
@staticmethod
|
||||
def _check_ban_regex(text: str, userinfo: UserInfo) -> bool:
|
||||
def _check_ban_regex(text: str, userinfo: UserInfo) -> bool: # 明确 userinfo 类型
|
||||
"""检查消息是否匹配过滤正则表达式"""
|
||||
for pattern in global_config.ban_msgs_regex:
|
||||
if pattern.search(text):
|
||||
logger.info(f"[私聊]{userinfo.user_nickname}:{text}")
|
||||
logger.info(f"[正则表达式过滤]消息匹配到{pattern}，filtered")
|
||||
if pattern.search(text): # 假设 ban_msgs_regex 中的元素是已编译的正则对象
|
||||
logger.info(f"[私聊]{userinfo.user_nickname}:{text}") # _nickname
|
||||
logger.info(f"[正则表达式过滤]消息匹配到{pattern.pattern}，filtered") # .pattern 获取原始表达式字符串
|
||||
return True
|
||||
return False
|
||||
return False
|
||||
|
|
@ -1,88 +1,285 @@
|
|||
import traceback
|
||||
import json
|
||||
import re
|
||||
from typing import Dict, Any, Optional, Tuple, List, Union
|
||||
from src.common.logger_manager import get_logger # 确认 logger 的导入路径
|
||||
from src.plugins.memory_system.Hippocampus import HippocampusManager
|
||||
from src.plugins.heartFC_chat.heartflow_prompt_builder import prompt_builder # 确认 prompt_builder 的导入路径
|
||||
from src.plugins.chat.chat_stream import ChatStream
|
||||
from ..person_info.person_info import person_info_manager
|
||||
import math
|
||||
from src.plugins.utils.chat_message_builder import build_readable_messages
|
||||
from .observation_info import ObservationInfo
|
||||
import asyncio # 确保导入 asyncio
|
||||
from typing import Dict, Any, Optional, Tuple, List, Union # 确保导入这些类型
|
||||
|
||||
from src.common.logger_manager import get_logger
|
||||
from src.config.config import global_config
|
||||
from src.common.database import db # << 确认此路径
|
||||
|
||||
# --- 依赖于你项目结构的导入，请务必仔细检查并根据你的实际情况调整 ---
|
||||
from src.plugins.memory_system.Hippocampus import HippocampusManager # << 确认此路径
|
||||
from src.plugins.heartFC_chat.heartflow_prompt_builder import prompt_builder # << 确认此路径
|
||||
from src.plugins.chat.utils import get_embedding # << 确认此路径
|
||||
from src.plugins.utils.chat_message_builder import build_readable_messages # << 确认此路径
|
||||
# --- 依赖导入结束 ---
|
||||
|
||||
from src.plugins.chat.chat_stream import ChatStream # 来自原始 pfc_utils.py
|
||||
from ..person_info.person_info import person_info_manager # 来自原始 pfc_utils.py (相对导入)
|
||||
import math # 来自原始 pfc_utils.py
|
||||
from .observation_info import ObservationInfo # 来自原始 pfc_utils.py (相对导入)
|
||||
|
||||
|
||||
logger = get_logger("pfc_utils")
|
||||
|
||||
|
||||
async def retrieve_contextual_info(text: str, private_name: str) -> Tuple[str, str]:
|
||||
# ==============================================================================
|
||||
# 新增：专门用于检索 PFC 私聊历史对话上下文的函数
|
||||
# ==============================================================================
|
||||
async def find_most_relevant_historical_message(
|
||||
chat_id: str,
|
||||
query_text: str,
|
||||
similarity_threshold: float = 0.3 # 相似度阈值，可以根据效果调整
|
||||
) -> Optional[Dict[str, Any]]:
|
||||
"""
|
||||
根据输入文本检索相关的记忆和知识。
|
||||
|
||||
Args:
|
||||
text: 用于检索的上下文文本 (例如聊天记录)。
|
||||
private_name: 私聊对象的名称，用于日志记录。
|
||||
|
||||
Returns:
|
||||
Tuple[str, str]: (检索到的记忆字符串, 检索到的知识字符串)
|
||||
根据查询文本，在指定 chat_id 的历史消息中查找最相关的消息。
|
||||
"""
|
||||
retrieved_memory_str = "无相关记忆。"
|
||||
if not query_text or not query_text.strip():
|
||||
logger.debug(f"[{chat_id}] (私聊历史)查询文本为空，跳过检索。")
|
||||
return None
|
||||
|
||||
logger.debug(f"[{chat_id}] (私聊历史)开始为查询文本 '{query_text[:50]}...' 检索。")
|
||||
|
||||
# 使用你项目中已有的 get_embedding 函数
|
||||
# request_type 参数需要根据 get_embedding 的实际需求调整
|
||||
query_embedding = await get_embedding(query_text, request_type="pfc_historical_chat_query")
|
||||
if not query_embedding:
|
||||
logger.warning(f"[{chat_id}] (私聊历史)未能为查询文本 '{query_text[:50]}...' 生成嵌入向量。")
|
||||
return None
|
||||
|
||||
pipeline = [
|
||||
{
|
||||
"$match": {
|
||||
"chat_id": chat_id,
|
||||
"embedding_vector": {"$exists": True, "$ne": None, "$not": {"$size": 0}}
|
||||
}
|
||||
},
|
||||
{
|
||||
"$addFields": {
|
||||
"dotProduct": {"$reduce": {"input": {"$range": [0, {"$size": "$embedding_vector"}]}, "initialValue": 0, "in": {"$add": ["$$value", {"$multiply": [{"$arrayElemAt": ["$embedding_vector", "$$this"]}, {"$arrayElemAt": [query_embedding, "$$this"]}]}]}}},
|
||||
"queryVecMagnitude": {"$sqrt": {"$reduce": {"input": query_embedding, "initialValue": 0, "in": {"$add": ["$$value", {"$multiply": ["$$this", "$$this"]}]}}}},
|
||||
"docVecMagnitude": {"$sqrt": {"$reduce": {"input": "$embedding_vector", "initialValue": 0, "in": {"$add": ["$$value", {"$multiply": ["$$this", "$$this"]}]}}}}
|
||||
}
|
||||
},
|
||||
{
|
||||
"$addFields": {
|
||||
"similarity": {
|
||||
"$cond": [
|
||||
{"$and": [{"$gt": ["$queryVecMagnitude", 0]}, {"$gt": ["$docVecMagnitude", 0]}]},
|
||||
{"$divide": ["$dotProduct", {"$multiply": ["$queryVecMagnitude", "$docVecMagnitude"]}]},
|
||||
0
|
||||
]
|
||||
}
|
||||
}
|
||||
},
|
||||
{"$match": {"similarity": {"$gte": similarity_threshold}}},
|
||||
{"$sort": {"similarity": -1}},
|
||||
{"$limit": 1},
|
||||
{"$project": {"_id": 0, "message_id": 1, "time": 1, "chat_id": 1, "user_info": 1, "processed_plain_text": 1, "similarity": 1}} # 可以不返回 embedding_vector 节省带宽
|
||||
]
|
||||
|
||||
try:
|
||||
# 假设 db.messages 是存储PFC私聊消息并带有embedding_vector的集合
|
||||
results = await db.messages.aggregate(pipeline).to_list(length=1)
|
||||
if results and len(results) > 0:
|
||||
most_similar_message = results[0]
|
||||
logger.info(f"[{chat_id}] (私聊历史)找到最相关消息 ID: {most_similar_message.get('message_id')}, 相似度: {most_similar_message.get('similarity'):.4f}")
|
||||
return most_similar_message
|
||||
else:
|
||||
logger.debug(f"[{chat_id}] (私聊历史)未找到相似度超过 {similarity_threshold} 的相关消息。")
|
||||
return None
|
||||
except Exception as e:
|
||||
logger.error(f"[{chat_id}] (私聊历史)在数据库中检索时出错: {e}", exc_info=True)
|
||||
return None
|
||||
|
||||
async def retrieve_chat_context_window(
|
||||
chat_id: str,
|
||||
anchor_message_id: str,
|
||||
anchor_message_time: float,
|
||||
window_size_before: int = 7,
|
||||
window_size_after: int = 7
|
||||
) -> List[Dict[str, Any]]:
|
||||
"""
|
||||
以某条消息为锚点，获取其前后的聊天记录形成一个上下文窗口。
|
||||
"""
|
||||
if not anchor_message_id or anchor_message_time is None:
|
||||
return []
|
||||
|
||||
context_messages: List[Dict[str, Any]] = [] # 明确类型
|
||||
logger.debug(f"[{chat_id}] (私聊历史)准备以消息 ID '{anchor_message_id}' (时间: {anchor_message_time}) 为锚点，获取上下文窗口...")
|
||||
|
||||
try:
|
||||
# 假设 db.messages 是存储PFC私聊消息的集合
|
||||
anchor_message = await db.messages.find_one({"message_id": anchor_message_id, "chat_id": chat_id})
|
||||
|
||||
messages_before_cursor = db.messages.find(
|
||||
{"chat_id": chat_id, "time": {"$lt": anchor_message_time}}
|
||||
).sort("time", -1).limit(window_size_before)
|
||||
messages_before = await messages_before_cursor.to_list(length=window_size_before)
|
||||
messages_before.reverse()
|
||||
|
||||
messages_after_cursor = db.messages.find(
|
||||
{"chat_id": chat_id, "time": {"$gt": anchor_message_time}}
|
||||
).sort("time", 1).limit(window_size_after)
|
||||
messages_after = await messages_after_cursor.to_list(length=window_size_after)
|
||||
|
||||
if messages_before:
|
||||
context_messages.extend(messages_before)
|
||||
if anchor_message:
|
||||
anchor_message.pop("_id", None)
|
||||
context_messages.append(anchor_message)
|
||||
if messages_after:
|
||||
context_messages.extend(messages_after)
|
||||
|
||||
final_window: List[Dict[str, Any]] = [] # 明确类型
|
||||
seen_ids: set[str] = set() # 明确类型
|
||||
for msg in context_messages:
|
||||
msg_id = msg.get("message_id")
|
||||
if msg_id and msg_id not in seen_ids: # 确保 msg_id 存在
|
||||
final_window.append(msg)
|
||||
seen_ids.add(msg_id)
|
||||
|
||||
final_window.sort(key=lambda m: m.get("time", 0))
|
||||
logger.info(f"[{chat_id}] (私聊历史)为锚点 '{anchor_message_id}' 构建了包含 {len(final_window)} 条消息的上下文窗口。")
|
||||
return final_window
|
||||
except Exception as e:
|
||||
logger.error(f"[{chat_id}] (私聊历史)获取消息 ID '{anchor_message_id}' 的上下文窗口时出错: {e}", exc_info=True)
|
||||
return []
|
||||
|
||||
# ==============================================================================
|
||||
# 修改后的 retrieve_contextual_info 函数
|
||||
# ==============================================================================
|
||||
async def retrieve_contextual_info(
|
||||
text: str, # 用于全局记忆和知识检索的主查询文本 (通常是短期聊天记录)
|
||||
private_name: str, # 用于日志
|
||||
chat_id: str, # 用于特定私聊历史的检索
|
||||
historical_chat_query_text: Optional[str] = None # 专门为私聊历史检索准备的查询文本 (例如最新的N条消息合并)
|
||||
) -> Tuple[str, str, str]: # 返回: 全局记忆, 知识, 私聊历史回忆
|
||||
"""
|
||||
检索三种类型的上下文信息：全局压缩记忆、知识库知识、当前私聊的特定历史对话。
|
||||
"""
|
||||
# 初始化返回值
|
||||
retrieved_global_memory_str = "无相关全局记忆。"
|
||||
retrieved_knowledge_str = "无相关知识。"
|
||||
memory_log_msg = "未自动检索到相关记忆。"
|
||||
knowledge_log_msg = "未自动检索到相关知识。"
|
||||
retrieved_historical_chat_str = "无相关私聊历史回忆。"
|
||||
|
||||
if not text or text == "还没有聊天记录。" or text == "[构建聊天记录出错]":
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 无有效上下文，跳过检索。")
|
||||
return retrieved_memory_str, retrieved_knowledge_str
|
||||
|
||||
# 1. 检索记忆 (逻辑来自原 _get_memory_info)
|
||||
try:
|
||||
related_memory = await HippocampusManager.get_instance().get_memory_from_text(
|
||||
text=text,
|
||||
max_memory_num=2,
|
||||
max_memory_length=2,
|
||||
max_depth=3,
|
||||
fast_retrieval=False,
|
||||
)
|
||||
if related_memory:
|
||||
related_memory_info = ""
|
||||
for memory in related_memory:
|
||||
related_memory_info += memory[1] + "\n"
|
||||
if related_memory_info:
|
||||
# 注意：原版提示信息可以根据需要调整
|
||||
retrieved_memory_str = f"你回忆起：\n{related_memory_info.strip()}\n(以上是你的回忆，供参考)\n"
|
||||
memory_log_msg = f"自动检索到记忆: {related_memory_info.strip()[:100]}..."
|
||||
# --- 1. 全局压缩记忆检索 (来自 HippocampusManager) ---
|
||||
# (保持你原始 pfc_utils.py 中这部分的逻辑基本不变)
|
||||
global_memory_log_msg = f"开始全局压缩记忆检索 (基于文本: '{text[:30]}...')"
|
||||
if text and text.strip() and text != "还没有聊天记录。" and text != "[构建聊天记录出错]":
|
||||
try:
|
||||
related_memory = await HippocampusManager.get_instance().get_memory_from_text(
|
||||
text=text,
|
||||
max_memory_num=2,
|
||||
max_memory_length=2, # 你原始代码中这里是2，不是200
|
||||
max_depth=3,
|
||||
fast_retrieval=False, # 你原始代码中这里是False
|
||||
)
|
||||
if related_memory:
|
||||
temp_global_memory_info = ""
|
||||
for memory_item in related_memory:
|
||||
if isinstance(memory_item, (list, tuple)) and len(memory_item) > 1:
|
||||
temp_global_memory_info += str(memory_item[1]) + "\n"
|
||||
elif isinstance(memory_item, str):
|
||||
temp_global_memory_info += memory_item + "\n"
|
||||
|
||||
if temp_global_memory_info.strip():
|
||||
retrieved_global_memory_str = f"你回忆起一些相关的全局记忆：\n{temp_global_memory_info.strip()}\n(以上是你的全局记忆，供参考)\n"
|
||||
global_memory_log_msg = f"自动检索到全局压缩记忆: {temp_global_memory_info.strip()[:100]}..."
|
||||
else:
|
||||
global_memory_log_msg = "全局压缩记忆检索返回为空或格式不符。"
|
||||
else:
|
||||
memory_log_msg = "自动检索记忆返回为空。"
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 记忆检索: {memory_log_msg}")
|
||||
|
||||
except Exception as e:
|
||||
logger.error(
|
||||
f"[私聊][{private_name}] (retrieve_contextual_info) 自动检索记忆时出错: {e}\n{traceback.format_exc()}"
|
||||
)
|
||||
retrieved_memory_str = "检索记忆时出错。\n"
|
||||
|
||||
# 2. 检索知识 (逻辑来自原 action_planner 和 reply_generator)
|
||||
try:
|
||||
# 使用导入的 prompt_builder 实例及其方法
|
||||
knowledge_result = await prompt_builder.get_prompt_info(
|
||||
message=text,
|
||||
threshold=0.38, # threshold 可以根据需要调整
|
||||
)
|
||||
if knowledge_result:
|
||||
retrieved_knowledge_str = knowledge_result # 直接使用返回结果
|
||||
knowledge_log_msg = "自动检索到相关知识。"
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 知识检索: {knowledge_log_msg}")
|
||||
|
||||
except Exception as e:
|
||||
logger.error(
|
||||
f"[私聊][{private_name}] (retrieve_contextual_info) 自动检索知识时出错: {e}\n{traceback.format_exc()}"
|
||||
)
|
||||
retrieved_knowledge_str = "检索知识时出错。\n"
|
||||
|
||||
return retrieved_memory_str, retrieved_knowledge_str
|
||||
global_memory_log_msg = "全局压缩记忆检索返回为空列表。"
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 全局压缩记忆检索: {global_memory_log_msg}")
|
||||
except Exception as e:
|
||||
logger.error(
|
||||
f"[私聊][{private_name}] (retrieve_contextual_info) 检索全局压缩记忆时出错: {e}\n{traceback.format_exc()}"
|
||||
)
|
||||
retrieved_global_memory_str = "[检索全局压缩记忆时出错]\n"
|
||||
else:
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 无有效主查询文本，跳过全局压缩记忆检索。")
|
||||
|
||||
|
||||
# --- 2. 相关知识检索 (来自 prompt_builder) ---
|
||||
# (保持你原始 pfc_utils.py 中这部分的逻辑基本不变)
|
||||
knowledge_log_msg = f"开始知识检索 (基于文本: '{text[:30]}...')"
|
||||
if text and text.strip() and text != "还没有聊天记录。" and text != "[构建聊天记录出错]":
|
||||
try:
|
||||
knowledge_result = await prompt_builder.get_prompt_info(
|
||||
message=text,
|
||||
threshold=0.38,
|
||||
)
|
||||
if knowledge_result and knowledge_result.strip(): # 确保结果不为空
|
||||
retrieved_knowledge_str = knowledge_result # 直接使用返回结果，如果需要也可以包装
|
||||
knowledge_log_msg = f"自动检索到相关知识: {knowledge_result[:100]}..."
|
||||
else:
|
||||
knowledge_log_msg = "知识检索返回为空。"
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 知识检索: {knowledge_log_msg}")
|
||||
except Exception as e:
|
||||
logger.error(
|
||||
f"[私聊][{private_name}] (retrieve_contextual_info) 自动检索知识时出错: {e}\n{traceback.format_exc()}"
|
||||
)
|
||||
retrieved_knowledge_str = "[检索知识时出错]\n"
|
||||
else:
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 无有效主查询文本，跳过知识检索。")
|
||||
|
||||
|
||||
# --- 3. 当前私聊的特定历史对话上下文检索 (新增逻辑) ---
|
||||
query_for_historical_chat = historical_chat_query_text if historical_chat_query_text and historical_chat_query_text.strip() else None
|
||||
historical_chat_log_msg = f"开始私聊历史检索 (查询文本: '{str(query_for_historical_chat)[:30]}...')"
|
||||
|
||||
if query_for_historical_chat:
|
||||
try:
|
||||
most_relevant_message_doc = await find_most_relevant_historical_message(
|
||||
chat_id=chat_id,
|
||||
query_text=query_for_historical_chat,
|
||||
similarity_threshold=0.5 # 你可以根据需要调整这个阈值
|
||||
)
|
||||
if most_relevant_message_doc:
|
||||
anchor_id = most_relevant_message_doc.get("message_id")
|
||||
anchor_time = most_relevant_message_doc.get("time")
|
||||
if anchor_id and anchor_time is not None:
|
||||
context_window_messages = await retrieve_chat_context_window(
|
||||
chat_id=chat_id,
|
||||
anchor_message_id=anchor_id,
|
||||
anchor_message_time=anchor_time,
|
||||
window_size_before=7, # 我们的目标：上7条
|
||||
window_size_after=7 # 我们的目标：下7条 (共15条，包括锚点)
|
||||
)
|
||||
if context_window_messages:
|
||||
formatted_window_str = await build_readable_messages(
|
||||
context_window_messages,
|
||||
replace_bot_name=False, # 在回忆中，保留原始发送者名称
|
||||
merge_messages=False,
|
||||
timestamp_mode="relative", # 可以选择 'absolute' 或 'none'
|
||||
read_mark=0.0
|
||||
)
|
||||
if formatted_window_str and formatted_window_str.strip():
|
||||
retrieved_historical_chat_str = f"你回忆起一段与当前对话相关的历史聊天：\n------\n{formatted_window_str.strip()}\n------\n(以上是针对本次私聊的回忆，供参考)\n"
|
||||
historical_chat_log_msg = f"自动检索到相关私聊历史片段 (锚点ID: {anchor_id}, 相似度: {most_relevant_message_doc.get('similarity'):.3f})"
|
||||
else:
|
||||
historical_chat_log_msg = "检索到的私聊历史对话窗口格式化后为空。"
|
||||
else:
|
||||
historical_chat_log_msg = f"找到了相关锚点消息 (ID: {anchor_id})，但未能构建其上下文窗口。"
|
||||
else:
|
||||
historical_chat_log_msg = "检索到的最相关私聊历史消息文档缺少 message_id 或 time。"
|
||||
else:
|
||||
historical_chat_log_msg = "未找到足够相关的私聊历史对话消息。"
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 私聊历史对话检索: {historical_chat_log_msg}")
|
||||
except Exception as e:
|
||||
logger.error(
|
||||
f"[私聊][{private_name}] (retrieve_contextual_info) 检索私聊历史对话时出错: {e}\n{traceback.format_exc()}"
|
||||
)
|
||||
retrieved_historical_chat_str = "[检索私聊历史对话时出错]\n"
|
||||
else:
|
||||
logger.debug(f"[私聊][{private_name}] (retrieve_contextual_info) 无专门的私聊历史查询文本，跳过私聊历史对话检索。")
|
||||
|
||||
return retrieved_global_memory_str, retrieved_knowledge_str, retrieved_historical_chat_str
|
||||
|
||||
|
||||
# ==============================================================================
|
||||
# 你原始 pfc_utils.py 中的其他函数保持不变
|
||||
# ==============================================================================
|
||||
def get_items_from_json(
|
||||
content: str,
|
||||
private_name: str,
|
||||
|
|
@ -92,121 +289,66 @@ def get_items_from_json(
|
|||
allow_array: bool = True,
|
||||
) -> Tuple[bool, Union[Dict[str, Any], List[Dict[str, Any]]]]:
|
||||
"""从文本中提取JSON内容并获取指定字段
|
||||
|
||||
Args:
|
||||
content: 包含JSON的文本
|
||||
private_name: 私聊名称
|
||||
*items: 要提取的字段名
|
||||
default_values: 字段的默认值，格式为 {字段名: 默认值}
|
||||
required_types: 字段的必需类型，格式为 {字段名: 类型}
|
||||
allow_array: 是否允许解析JSON数组
|
||||
|
||||
Returns:
|
||||
Tuple[bool, Union[Dict[str, Any], List[Dict[str, Any]]]]: (是否成功, 提取的字段字典或字典列表)
|
||||
(保持你原始 pfc_utils.py 中的此函数代码不变)
|
||||
"""
|
||||
cleaned_content = content.strip()
|
||||
result: Union[Dict[str, Any], List[Dict[str, Any]]] = {} # 初始化类型
|
||||
# 匹配 ```json ... ``` 或 ``` ... ```
|
||||
result: Union[Dict[str, Any], List[Dict[str, Any]]] = {}
|
||||
markdown_match = re.search(r"```(?:json)?\s*([\s\S]*?)\s*```", cleaned_content, re.IGNORECASE)
|
||||
if markdown_match:
|
||||
cleaned_content = markdown_match.group(1).strip()
|
||||
logger.debug(f"[私聊][{private_name}] 已去除 Markdown 标记，剩余内容: {cleaned_content[:100]}...")
|
||||
# --- 新增结束 ---
|
||||
|
||||
# 设置默认值
|
||||
default_result: Dict[str, Any] = {} # 用于单对象时的默认值
|
||||
default_result: Dict[str, Any] = {}
|
||||
if default_values:
|
||||
default_result.update(default_values)
|
||||
result = default_result.copy() # 先用默认值初始化
|
||||
|
||||
# 首先尝试解析为JSON数组
|
||||
result = default_result.copy()
|
||||
if allow_array:
|
||||
try:
|
||||
# 尝试直接解析清理后的内容为列表
|
||||
json_array = json.loads(cleaned_content)
|
||||
|
||||
if isinstance(json_array, list):
|
||||
valid_items_list: List[Dict[str, Any]] = []
|
||||
for item in json_array:
|
||||
if not isinstance(item, dict):
|
||||
logger.warning(f"[私聊][{private_name}] JSON数组中的元素不是字典: {item}")
|
||||
for item_json in json_array: # Renamed item to item_json to avoid conflict
|
||||
if not isinstance(item_json, dict):
|
||||
logger.warning(f"[私聊][{private_name}] JSON数组中的元素不是字典: {item_json}")
|
||||
continue
|
||||
|
||||
current_item_result = default_result.copy() # 每个元素都用默认值初始化
|
||||
current_item_result = default_result.copy()
|
||||
valid_item = True
|
||||
|
||||
# 提取并验证字段
|
||||
for field in items:
|
||||
if field in item:
|
||||
current_item_result[field] = item[field]
|
||||
elif field not in default_result: # 如果字段不存在且没有默认值
|
||||
logger.warning(f"[私聊][{private_name}] JSON数组元素缺少必要字段 '{field}': {item}")
|
||||
valid_item = False
|
||||
break # 这个元素无效
|
||||
|
||||
if not valid_item:
|
||||
continue
|
||||
|
||||
# 验证类型
|
||||
for field in items: # items is args from function signature
|
||||
if field in item_json:
|
||||
current_item_result[field] = item_json[field]
|
||||
elif field not in default_result:
|
||||
logger.warning(f"[私聊][{private_name}] JSON数组元素缺少必要字段 '{field}': {item_json}")
|
||||
valid_item = False; break
|
||||
if not valid_item: continue
|
||||
if required_types:
|
||||
for field, expected_type in required_types.items():
|
||||
# 检查 current_item_result 中是否存在该字段 (可能来自 item 或 default_values)
|
||||
if field in current_item_result and not isinstance(
|
||||
current_item_result[field], expected_type
|
||||
):
|
||||
logger.warning(
|
||||
f"[私聊][{private_name}] JSON数组元素字段 '{field}' 类型错误 (应为 {expected_type.__name__}, 实际为 {type(current_item_result[field]).__name__}): {item}"
|
||||
)
|
||||
valid_item = False
|
||||
break
|
||||
|
||||
if not valid_item:
|
||||
continue
|
||||
|
||||
# 验证字符串不为空 (只检查 items 中要求的字段)
|
||||
if field in current_item_result and not isinstance(current_item_result[field], expected_type):
|
||||
logger.warning(f"[私聊][{private_name}] JSON数组元素字段 '{field}' 类型错误 (应为 {expected_type.__name__}, 实际为 {type(current_item_result[field]).__name__}): {item_json}")
|
||||
valid_item = False; break
|
||||
if not valid_item: continue
|
||||
for field in items:
|
||||
if (
|
||||
field in current_item_result
|
||||
and isinstance(current_item_result[field], str)
|
||||
and not current_item_result[field].strip()
|
||||
):
|
||||
logger.warning(f"[私聊][{private_name}] JSON数组元素字段 '{field}' 不能为空字符串: {item}")
|
||||
valid_item = False
|
||||
break
|
||||
|
||||
if valid_item:
|
||||
valid_items_list.append(current_item_result) # 只添加完全有效的项
|
||||
|
||||
if valid_items_list: # 只有当列表不为空时才认为是成功
|
||||
if field in current_item_result and isinstance(current_item_result[field], str) and not current_item_result[field].strip():
|
||||
logger.warning(f"[私聊][{private_name}] JSON数组元素字段 '{field}' 不能为空字符串: {item_json}")
|
||||
valid_item = False; break
|
||||
if valid_item: valid_items_list.append(current_item_result)
|
||||
if valid_items_list:
|
||||
logger.debug(f"[私聊][{private_name}] 成功解析JSON数组，包含 {len(valid_items_list)} 个有效项目。")
|
||||
return True, valid_items_list
|
||||
else:
|
||||
# 如果列表为空（可能所有项都无效），则继续尝试解析为单个对象
|
||||
logger.debug(f"[私聊][{private_name}] 解析为JSON数组，但未找到有效项目，尝试解析单个JSON对象。")
|
||||
# result 重置回单个对象的默认值
|
||||
result = default_result.copy()
|
||||
|
||||
except json.JSONDecodeError:
|
||||
logger.debug(f"[私聊][{private_name}] JSON数组直接解析失败，尝试解析单个JSON对象")
|
||||
# result 重置回单个对象的默认值
|
||||
result = default_result.copy()
|
||||
except Exception as e:
|
||||
logger.error(f"[私聊][{private_name}] 尝试解析JSON数组时发生未知错误: {str(e)}")
|
||||
# result 重置回单个对象的默认值
|
||||
result = default_result.copy()
|
||||
|
||||
# 尝试解析为单个JSON对象
|
||||
try:
|
||||
# 尝试直接解析清理后的内容
|
||||
json_data = json.loads(cleaned_content)
|
||||
if not isinstance(json_data, dict):
|
||||
logger.error(f"[私聊][{private_name}] 解析为单个对象，但结果不是字典类型: {type(json_data)}")
|
||||
return False, default_result # 返回失败和默认值
|
||||
|
||||
return False, default_result
|
||||
except json.JSONDecodeError:
|
||||
# 如果直接解析失败，尝试用正则表达式查找 JSON 对象部分 (作为后备)
|
||||
# 这个正则比较简单，可能无法处理嵌套或复杂的 JSON
|
||||
json_pattern = r"\{[\s\S]*?\}" # 使用非贪婪匹配
|
||||
json_pattern = r"\{[\s\S]*?\}"
|
||||
json_match = re.search(json_pattern, cleaned_content)
|
||||
if json_match:
|
||||
try:
|
||||
|
|
@ -220,133 +362,97 @@ def get_items_from_json(
|
|||
logger.error(f"[私聊][{private_name}] 正则提取的部分 '{potential_json_str[:100]}...' 无法解析为JSON。")
|
||||
return False, default_result
|
||||
else:
|
||||
logger.error(
|
||||
f"[私聊][{private_name}] 无法在返回内容中找到有效的JSON对象部分。原始内容: {cleaned_content[:100]}..."
|
||||
)
|
||||
logger.error(f"[私聊][{private_name}] 无法在返回内容中找到有效的JSON对象部分。原始内容: {cleaned_content[:100]}...")
|
||||
return False, default_result
|
||||
|
||||
# 提取并验证字段 (适用于单个JSON对象)
|
||||
# 确保 result 是字典类型用于更新
|
||||
if not isinstance(result, dict):
|
||||
result = default_result.copy() # 如果之前是列表，重置为字典
|
||||
|
||||
if not isinstance(result, dict): result = default_result.copy()
|
||||
valid_single_object = True
|
||||
for item in items:
|
||||
if item in json_data:
|
||||
result[item] = json_data[item]
|
||||
elif item not in default_result: # 如果字段不存在且没有默认值
|
||||
logger.error(f"[私聊][{private_name}] JSON对象缺少必要字段 '{item}'。JSON内容: {json_data}")
|
||||
valid_single_object = False
|
||||
break # 这个对象无效
|
||||
|
||||
if not valid_single_object:
|
||||
return False, default_result
|
||||
|
||||
# 验证类型
|
||||
for item_field in items: # Renamed item to item_field
|
||||
if item_field in json_data: result[item_field] = json_data[item_field]
|
||||
elif item_field not in default_result:
|
||||
logger.error(f"[私聊][{private_name}] JSON对象缺少必要字段 '{item_field}'。JSON内容: {json_data}")
|
||||
valid_single_object = False; break
|
||||
if not valid_single_object: return False, default_result
|
||||
if required_types:
|
||||
for field, expected_type in required_types.items():
|
||||
if field in result and not isinstance(result[field], expected_type):
|
||||
logger.error(
|
||||
f"[私聊][{private_name}] JSON对象字段 '{field}' 类型错误 (应为 {expected_type.__name__}, 实际为 {type(result[field]).__name__})"
|
||||
)
|
||||
valid_single_object = False
|
||||
break
|
||||
|
||||
if not valid_single_object:
|
||||
return False, default_result
|
||||
|
||||
# 验证字符串不为空 (只检查 items 中要求的字段)
|
||||
logger.error(f"[私聊][{private_name}] JSON对象字段 '{field}' 类型错误 (应为 {expected_type.__name__}, 实际为 {type(result[field]).__name__})")
|
||||
valid_single_object = False; break
|
||||
if not valid_single_object: return False, default_result
|
||||
for field in items:
|
||||
if field in result and isinstance(result[field], str) and not result[field].strip():
|
||||
logger.error(f"[私聊][{private_name}] JSON对象字段 '{field}' 不能为空字符串")
|
||||
valid_single_object = False
|
||||
break
|
||||
|
||||
valid_single_object = False; break
|
||||
if valid_single_object:
|
||||
logger.debug(f"[私聊][{private_name}] 成功解析并验证了单个JSON对象。")
|
||||
return True, result # 返回提取并验证后的字典
|
||||
return True, result
|
||||
else:
|
||||
return False, default_result # 验证失败
|
||||
return False, default_result
|
||||
|
||||
|
||||
async def get_person_id(private_name: str, chat_stream: ChatStream):
|
||||
""" (保持你原始 pfc_utils.py 中的此函数代码不变) """
|
||||
private_user_id_str: Optional[str] = None
|
||||
private_platform_str: Optional[str] = None
|
||||
private_nickname_str = private_name
|
||||
# private_nickname_str = private_name # 这行在你提供的代码中没有被使用，可以考虑移除
|
||||
|
||||
if chat_stream.user_info:
|
||||
private_user_id_str = str(chat_stream.user_info.user_id)
|
||||
private_platform_str = chat_stream.user_info.platform
|
||||
logger.debug(
|
||||
f"[私聊][{private_name}] 从 ChatStream 获取到私聊对象信息: ID={private_user_id_str}, Platform={private_platform_str}, Name={private_nickname_str}"
|
||||
f"[私聊][{private_name}] 从 ChatStream 获取到私聊对象信息: ID={private_user_id_str}, Platform={private_platform_str}, Name={private_name}" # 使用 private_name
|
||||
)
|
||||
elif chat_stream.group_info is None and private_name:
|
||||
pass
|
||||
# elif chat_stream.group_info is None and private_name: # 这个 elif 条件体为空，可以移除
|
||||
# pass
|
||||
|
||||
if private_user_id_str and private_platform_str:
|
||||
try:
|
||||
private_user_id_int = int(private_user_id_str)
|
||||
# person_id = person_info_manager.get_person_id( # get_person_id 可能只查询，不创建
|
||||
# private_platform_str,
|
||||
# private_user_id_int
|
||||
# )
|
||||
# 使用 get_or_create_person 确保用户存在
|
||||
person_id = await person_info_manager.get_or_create_person(
|
||||
platform=private_platform_str,
|
||||
user_id=private_user_id_int,
|
||||
nickname=private_name, # 使用传入的 private_name 作为昵称
|
||||
nickname=private_name,
|
||||
)
|
||||
if person_id is None: # 如果 get_or_create_person 返回 None，说明创建失败
|
||||
if person_id is None:
|
||||
logger.error(f"[私聊][{private_name}] get_or_create_person 未能获取或创建 person_id。")
|
||||
return None # 返回 None 表示失败
|
||||
|
||||
return person_id, private_platform_str, private_user_id_str # 返回获取或创建的 person_id
|
||||
return None
|
||||
return person_id, private_platform_str, private_user_id_str
|
||||
except ValueError:
|
||||
logger.error(f"[私聊][{private_name}] 无法将 private_user_id_str ('{private_user_id_str}') 转换为整数。")
|
||||
return None # 返回 None 表示失败
|
||||
return None
|
||||
except Exception as e_pid:
|
||||
logger.error(f"[私聊][{private_name}] 获取或创建 person_id 时出错: {e_pid}")
|
||||
return None # 返回 None 表示失败
|
||||
return None
|
||||
else:
|
||||
logger.warning(
|
||||
f"[私聊][{private_name}] 未能确定私聊对象的 user_id 或 platform，无法获取 person_id。将在收到消息后尝试。"
|
||||
)
|
||||
return None # 返回 None 表示失败
|
||||
return None
|
||||
|
||||
|
||||
async def adjust_relationship_value_nonlinear(old_value: float, raw_adjustment: float) -> float:
|
||||
# 限制 old_value 范围
|
||||
""" (保持你原始 pfc_utils.py 中的此函数代码不变) """
|
||||
old_value = max(-1000, min(1000, old_value))
|
||||
value = raw_adjustment
|
||||
|
||||
if old_value >= 0:
|
||||
if value >= 0:
|
||||
value = value * math.cos(math.pi * old_value / 2000)
|
||||
if old_value > 500:
|
||||
rdict = await person_info_manager.get_specific_value_list("relationship_value", lambda x: x > 700)
|
||||
# 确保 person_info_manager.get_specific_value_list 是异步的，如果是同步则需要调整
|
||||
rdict = await person_info_manager.get_specific_value_list("relationship_value", lambda x: x > 700 if isinstance(x, (int, float)) else False)
|
||||
high_value_count = len(rdict)
|
||||
if old_value > 700:
|
||||
value *= 3 / (high_value_count + 2)
|
||||
else:
|
||||
value *= 3 / (high_value_count + 3)
|
||||
elif value < 0:
|
||||
value = value * math.exp(old_value / 2000)
|
||||
else:
|
||||
value = 0
|
||||
else:
|
||||
if value >= 0:
|
||||
value = value * math.exp(old_value / 2000)
|
||||
elif value < 0:
|
||||
value = value * math.cos(math.pi * old_value / 2000)
|
||||
else:
|
||||
value = 0
|
||||
|
||||
if old_value > 700: value *= 3 / (high_value_count + 2)
|
||||
else: value *= 3 / (high_value_count + 3)
|
||||
elif value < 0: value = value * math.exp(old_value / 2000)
|
||||
# else: value = 0 # 你原始代码中没有这句，如果value为0，保持为0
|
||||
else: # old_value < 0
|
||||
if value >= 0: value = value * math.exp(old_value / 2000)
|
||||
elif value < 0: value = value * math.cos(math.pi * old_value / 2000)
|
||||
# else: value = 0 # 你原始代码中没有这句
|
||||
return value
|
||||
|
||||
|
||||
async def build_chat_history_text(observation_info: ObservationInfo, private_name: str) -> str:
|
||||
"""构建聊天历史记录文本 (包含未处理消息)"""
|
||||
|
||||
""" (保持你原始 pfc_utils.py 中的此函数代码不变) """
|
||||
chat_history_text = ""
|
||||
try:
|
||||
if hasattr(observation_info, "chat_history_str") and observation_info.chat_history_str:
|
||||
|
|
@ -358,27 +464,32 @@ async def build_chat_history_text(observation_info: ObservationInfo, private_nam
|
|||
)
|
||||
else:
|
||||
chat_history_text = "还没有聊天记录。\n"
|
||||
|
||||
unread_count = getattr(observation_info, "new_messages_count", 0)
|
||||
unread_messages = getattr(observation_info, "unprocessed_messages", [])
|
||||
if unread_count > 0 and unread_messages:
|
||||
bot_qq_str = str(global_config.BOT_QQ)
|
||||
other_unread_messages = [
|
||||
msg for msg in unread_messages if msg.get("user_info", {}).get("user_id") != bot_qq_str
|
||||
]
|
||||
other_unread_count = len(other_unread_messages)
|
||||
if other_unread_count > 0:
|
||||
new_messages_str = await build_readable_messages(
|
||||
other_unread_messages,
|
||||
replace_bot_name=True,
|
||||
merge_messages=False,
|
||||
timestamp_mode="relative",
|
||||
read_mark=0.0,
|
||||
)
|
||||
chat_history_text += f"\n{new_messages_str}\n------\n"
|
||||
bot_qq_str = str(global_config.BOT_QQ) if global_config.BOT_QQ else None # 安全获取
|
||||
if bot_qq_str: # 仅当 bot_qq_str 有效时进行过滤
|
||||
other_unread_messages = [
|
||||
msg for msg in unread_messages if msg.get("user_info", {}).get("user_id") != bot_qq_str
|
||||
]
|
||||
other_unread_count = len(other_unread_messages)
|
||||
if other_unread_count > 0:
|
||||
new_messages_str = await build_readable_messages(
|
||||
other_unread_messages,
|
||||
replace_bot_name=True, # 这里是未处理消息，可能不需要替换机器人名字
|
||||
merge_messages=False,
|
||||
timestamp_mode="relative",
|
||||
read_mark=0.0,
|
||||
)
|
||||
chat_history_text += f"\n{new_messages_str}\n------\n" # 原始代码是加在末尾的
|
||||
else:
|
||||
logger.warning(f"[私聊][{private_name}] BOT_QQ 未配置，无法准确过滤未读消息中的机器人自身消息。")
|
||||
|
||||
except AttributeError as e:
|
||||
logger.warning(f"[私聊][{private_name}] 构建聊天记录文本时属性错误: {e}")
|
||||
chat_history_text = "[获取聊天记录时出错]\n"
|
||||
except Exception as e:
|
||||
logger.error(f"[私聊][{private_name}] 处理聊天记录时发生未知错误: {e}")
|
||||
chat_history_text = "[处理聊天记录时出错]\n"
|
||||
return chat_history_text
|
||||
return chat_history_text
|
||||
|
|
@ -1,5 +1,5 @@
|
|||
import random
|
||||
|
||||
import asyncio
|
||||
from .pfc_utils import retrieve_contextual_info
|
||||
|
||||
from src.common.logger_manager import get_logger
|
||||
|
|
@ -60,6 +60,9 @@ PROMPT_DIRECT_REPLY = """
|
|||
{retrieved_knowledge_str}
|
||||
请你**记住上面的知识**，在回复中有可能会用到。
|
||||
|
||||
你还想到了一些你们之前的聊天记录：
|
||||
{retrieved_historical_chat_str}
|
||||
|
||||
最近的聊天记录：
|
||||
{chat_history_text}
|
||||
|
||||
|
|
@ -68,6 +71,8 @@ PROMPT_DIRECT_REPLY = """
|
|||
{last_rejection_info}
|
||||
|
||||
|
||||
|
||||
|
||||
请根据上述信息，结合聊天记录，回复对方。该回复应该：
|
||||
1. 符合对话目标，以"你"的角度发言（不要自己与自己对话！）
|
||||
2. 符合你的性格特征和身份细节
|
||||
|
|
@ -97,6 +102,9 @@ PROMPT_SEND_NEW_MESSAGE = """
|
|||
{retrieved_knowledge_str}
|
||||
请你**记住上面的知识**，在发消息时有可能会用到。
|
||||
|
||||
你还想到了一些你们之前的聊天记录：
|
||||
{retrieved_historical_chat_str}
|
||||
|
||||
最近的聊天记录：
|
||||
{chat_history_text}
|
||||
|
||||
|
|
@ -223,12 +231,59 @@ class ReplyGenerator:
|
|||
current_emotion_text_str = getattr(conversation_info, "current_emotion_text", "心情平静。")
|
||||
|
||||
persona_text = f"你的名字是{self.name}，{self.personality_info}。"
|
||||
retrieval_context = chat_history_text
|
||||
retrieved_memory_str, retrieved_knowledge_str = await retrieve_contextual_info(
|
||||
retrieval_context, self.private_name
|
||||
)
|
||||
historical_chat_query = ""
|
||||
num_recent_messages_for_query = 3 # 例如，取最近3条作为查询引子
|
||||
if observation_info.chat_history and len(observation_info.chat_history) > 0:
|
||||
# 从 chat_history (已处理并存入 ObservationInfo 的历史) 中取最新N条
|
||||
# 或者，如果 observation_info.unprocessed_messages 更能代表“当前上下文”，也可以考虑用它
|
||||
# 我们先用 chat_history，因为它包含了双方的对话历史，可能更稳定
|
||||
recent_messages_for_query_list = observation_info.chat_history[-num_recent_messages_for_query:]
|
||||
|
||||
# 将这些消息的文本内容合并
|
||||
query_texts_list = []
|
||||
for msg_dict in recent_messages_for_query_list:
|
||||
text_content = msg_dict.get("processed_plain_text", "")
|
||||
if text_content.strip(): # 只添加有内容的文本
|
||||
# 可以选择是否添加发送者信息到查询文本中，例如：
|
||||
# sender_nickname = msg_dict.get("user_info", {}).get("user_nickname", "用户")
|
||||
# query_texts_list.append(f"{sender_nickname}: {text_content}")
|
||||
query_texts_list.append(text_content) # 简单合并文本内容
|
||||
|
||||
if query_texts_list:
|
||||
historical_chat_query = " ".join(query_texts_list).strip()
|
||||
logger.debug(f"[私聊][{self.private_name}] (ReplyGenerator) 生成的私聊历史查询文本 (最近{num_recent_messages_for_query}条): '{historical_chat_query[:100]}...'")
|
||||
else:
|
||||
logger.debug(f"[私聊][{self.private_name}] (ReplyGenerator) 最近{num_recent_messages_for_query}条消息无有效文本内容，不进行私聊历史查询。")
|
||||
else:
|
||||
logger.debug(f"[私聊][{self.private_name}] (ReplyGenerator) 无聊天历史可用于生成私聊历史查询文本。")
|
||||
|
||||
current_chat_id = self.chat_observer.stream_id if self.chat_observer else None
|
||||
if not current_chat_id:
|
||||
logger.error(f"[私聊][{self.private_name}] (ReplyGenerator) 无法获取 current_chat_id，跳过所有上下文检索！")
|
||||
retrieved_global_memory_str = "[获取全局记忆出错：chat_id 未知]"
|
||||
retrieved_knowledge_str = "[获取知识出错：chat_id 未知]"
|
||||
retrieved_historical_chat_str = "[获取私聊历史回忆出错：chat_id 未知]"
|
||||
else:
|
||||
# retrieval_context 之前是用 chat_history_text，现在也用它作为全局记忆和知识的检索上下文
|
||||
retrieval_context_for_global_and_knowledge = chat_history_text
|
||||
|
||||
(
|
||||
retrieved_global_memory_str,
|
||||
retrieved_knowledge_str,
|
||||
retrieved_historical_chat_str # << 新增接收私聊历史回忆
|
||||
) = await retrieve_contextual_info(
|
||||
text=retrieval_context_for_global_and_knowledge, # 用于全局记忆和知识
|
||||
private_name=self.private_name,
|
||||
chat_id=current_chat_id, # << 传递 chat_id
|
||||
historical_chat_query_text=historical_chat_query # << 传递专门的查询文本
|
||||
)
|
||||
# === 调用修改结束 ===
|
||||
|
||||
logger.info(
|
||||
f"[私聊][{self.private_name}] (ReplyGenerator) 统一检索完成。记忆: {'有' if '回忆起' in retrieved_memory_str else '无'} / 知识: {'有' if '出错' not in retrieved_knowledge_str and '无相关知识' not in retrieved_knowledge_str else '无'}"
|
||||
f"[私聊][{self.private_name}] (ReplyGenerator) 上下文检索完成。\n"
|
||||
f" 全局记忆: {'有内容' if '回忆起' in retrieved_global_memory_str else '无或出错'}\n"
|
||||
f" 知识: {'有内容' if '出错' not in retrieved_knowledge_str and '无相关知识' not in retrieved_knowledge_str and retrieved_knowledge_str.strip() else '无或出错'}\n"
|
||||
f" 私聊历史回忆: {'有内容' if '回忆起一段相关的历史聊天' in retrieved_historical_chat_str else '无或出错'}"
|
||||
)
|
||||
|
||||
last_rejection_info_str = ""
|
||||
|
|
@ -292,11 +347,10 @@ class ReplyGenerator:
|
|||
base_format_params = {
|
||||
"persona_text": persona_text,
|
||||
"goals_str": goals_str,
|
||||
"chat_history_text": chat_history_text,
|
||||
"retrieved_memory_str": retrieved_memory_str if retrieved_memory_str else "无相关记忆。", # 确保已定义
|
||||
"retrieved_knowledge_str": retrieved_knowledge_str
|
||||
if retrieved_knowledge_str
|
||||
else "无相关知识。", # 确保已定义
|
||||
"chat_history_text": chat_history_text if chat_history_text.strip() else "还没有聊天记录。", # 当前短期历史
|
||||
"retrieved_global_memory_str": retrieved_global_memory_str if retrieved_global_memory_str.strip() else "无相关全局记忆。",
|
||||
"retrieved_knowledge_str": retrieved_knowledge_str if retrieved_knowledge_str.strip() else "无相关知识。",
|
||||
"retrieved_historical_chat_str": retrieved_historical_chat_str if retrieved_historical_chat_str.strip() else "无相关私聊历史回忆。", # << 新增
|
||||
"last_rejection_info": last_rejection_info_str,
|
||||
"current_time_str": current_time_value,
|
||||
"sender_name": sender_name_str,
|
||||
|
|
|
|||
Loading…
Reference in New Issue