gardel/Mofox-Core
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat：合并自我处理器和关系处理器

Browse Source
This commit is contained in:
SengokuCola
2025-06-21 15:46:53 +08:00
parent fd09550af9
commit 0f5fdc2ae5
10 changed files with 327 additions and 598 deletions
Download Patch File Download Diff File Expand all files Collapse all files

187
scripts/log_viewer_optimized.py
View File

@@ -7,6 +7,7 @@ import toml
from datetime import datetime
from collections import defaultdict
import os
import time
class LogIndex:
@@ -334,40 +335,33 @@ class VirtualLogDisplay:
def display_batch(self, start_index, end_index):
"""批量显示日志条目"""
batch_text = []
batch_tags = []
for i in range(start_index, end_index):
log_entry = self.log_index.get_entry_at_filtered_position(i)
if log_entry:
self.append_entry(log_entry, scroll=False)
def append_entry(self, log_entry, scroll=True):
"""将单个日志条目附加到文本小部件"""
# 检查在添加新内容之前视图是否已滚动到底部
should_scroll = scroll and self.text_widget.yview()[1] > 0.99
parts, tags = self.formatter.format_log_entry(log_entry)
# 合并部分为单行文本
line_text = " ".join(parts) + "\n"
batch_text.append(line_text)
# 记录标签信息（简化处理）
if tags and self.formatter.enable_level_colors:
level = log_entry.get("level", "info")
batch_tags.append(
(
"line",
len("".join(batch_text)) - len(line_text),
len("".join(batch_text)) - 1,
f"level_{level}",
)
)
# 获取插入前的末尾位置
start_pos = self.text_widget.index(tk.END + "-1c")
self.text_widget.insert(tk.END, line_text)
# 一次性插入所有文本
if batch_text:
start_pos = self.text_widget.index(tk.END)
all_text = "".join(batch_text)
self.text_widget.insert(tk.END, all_text)
# 为每个部分应用正确的标签
current_len = 0
for part, tag_name in zip(parts, tags):
start_index = f"{start_pos}+{current_len}c"
end_index = f"{start_pos}+{current_len + len(part)}c"
self.text_widget.tag_add(tag_name, start_index, end_index)
current_len += len(part) + 1 # 计入空格
# 应用标签（可选，为了性能可以考虑简化）
for tag_info in batch_tags:
tag_name = tag_info[3]
self.text_widget.tag_add(tag_name, f"{start_pos}+{tag_info[1]}c", f"{start_pos}+{tag_info[2]}c")
if should_scroll:
self.text_widget.see(tk.END)
class AsyncLogLoader:
@@ -459,6 +453,9 @@ class LogViewer:
# 初始化日志文件路径
self.current_log_file = Path("logs/app.log.jsonl")
self.last_file_size = 0
self.watching_thread = None
self.is_watching = tk.BooleanVar(value=True)
# 初始化异步加载器
self.async_loader = AsyncLogLoader(self.on_file_loaded)
@@ -548,6 +545,9 @@ class LogViewer:
ttk.Button(button_frame, text="选择文件", command=self.select_log_file).pack(side=tk.LEFT, padx=2)
ttk.Button(button_frame, text="刷新", command=self.refresh_log_file).pack(side=tk.LEFT, padx=2)
ttk.Checkbutton(button_frame, text="实时更新", variable=self.is_watching, command=self.toggle_watching).pack(
side=tk.LEFT, padx=2
)
# 过滤控制框架
filter_frame = ttk.Frame(self.control_frame)
@@ -583,16 +583,22 @@ class LogViewer:
return
self.log_index = log_index
try:
self.last_file_size = os.path.getsize(self.current_log_file)
except OSError:
self.last_file_size = 0
self.status_var.set(f"已加载 {log_index.total_entries} 条日志")
# 更新模块列表
self.modules = set(log_index.module_index.keys())
module_values = ["全部"] + sorted(list(self.modules))
self.module_combo["values"] = module_values
self.update_module_list()
# 应用过滤并显示
self.filter_logs()
# 如果开启了实时更新，则开始监视
if self.is_watching.get():
self.start_watching()
def on_loading_progress(self, progress, line_count):
"""加载进度回调"""
self.root.after(0, lambda: self.update_progress(progress, line_count))
@@ -604,6 +610,8 @@ class LogViewer:
def load_log_file_async(self):
"""异步加载日志文件"""
self.stop_watching() # 停止任何正在运行的监视器
if not self.current_log_file.exists():
self.status_var.set("文件不存在")
return
@@ -617,6 +625,7 @@ class LogViewer:
self.log_index = LogIndex()
self.modules.clear()
self.selected_modules.clear()
self.module_var.set("全部")
# 开始异步加载
self.async_loader.load_file_async(str(self.current_log_file), self.on_loading_progress)
@@ -672,6 +681,126 @@ class LogViewer:
"""刷新日志文件"""
self.load_log_file_async()
def toggle_watching(self):
"""切换实时更新状态"""
if self.is_watching.get():
self.start_watching()
else:
self.stop_watching()
def start_watching(self):
"""开始监视文件变化"""
if self.watching_thread and self.watching_thread.is_alive():
return # 已经在监视
if not self.current_log_file.exists():
self.is_watching.set(False)
messagebox.showwarning("警告", "日志文件不存在，无法开启实时更新。")
return
self.watching_thread = threading.Thread(target=self.watch_file_loop, daemon=True)
self.watching_thread.start()
def stop_watching(self):
"""停止监视文件变化"""
self.is_watching.set(False)
# 线程通过检查 is_watching 变量来停止，这里不需要强制干预
self.watching_thread = None
def watch_file_loop(self):
"""监视文件循环"""
while self.is_watching.get():
try:
if not self.current_log_file.exists():
self.root.after(
0,
lambda: messagebox.showwarning("警告", "日志文件丢失，已停止实时更新。"),
)
self.root.after(0, self.is_watching.set, False)
break
current_size = os.path.getsize(self.current_log_file)
if current_size > self.last_file_size:
new_entries = self.read_new_logs(self.last_file_size)
self.last_file_size = current_size
if new_entries:
self.root.after(0, self.append_new_logs, new_entries)
elif current_size < self.last_file_size:
# 文件被截断或替换
self.last_file_size = 0
self.root.after(0, self.refresh_log_file)
break # 刷新会重新启动监视（如果需要），所以结束当前循环
except Exception as e:
print(f"监视日志文件时出错: {e}")
self.root.after(0, self.is_watching.set, False)
break
time.sleep(1)
self.watching_thread = None
def read_new_logs(self, from_position):
"""读取新的日志条目并返回它们"""
new_entries = []
new_modules_found = False
with open(self.current_log_file, "r", encoding="utf-8") as f:
f.seek(from_position)
line_count = self.log_index.total_entries
for line in f:
if line.strip():
try:
log_entry = json.loads(line)
self.log_index.add_entry(line_count, log_entry)
new_entries.append(log_entry)
logger_name = log_entry.get("logger_name", "")
if logger_name and logger_name not in self.modules:
self.modules.add(logger_name)
new_modules_found = True
line_count += 1
except json.JSONDecodeError:
continue
if new_modules_found:
self.root.after(0, self.update_module_list)
return new_entries
def append_new_logs(self, new_entries):
"""将新日志附加到显示中"""
# 检查是否应附加或执行完全刷新（例如，如果过滤器处于活动状态）
selected_modules = (
self.selected_modules if (self.selected_modules and "全部" not in self.selected_modules) else None
)
level = self.level_var.get() if self.level_var.get() != "全部" else None
search_text = self.search_var.get().strip() if self.search_var.get().strip() else None
is_filtered = selected_modules or level or search_text
if is_filtered:
# 如果过滤器处于活动状态，我们必须执行完全刷新以应用它们
self.filter_logs()
return
# 如果没有过滤器，只需附加新日志
for entry in new_entries:
self.log_display.append_entry(entry)
# 更新状态
total_count = self.log_index.total_entries
self.status_var.set(f"显示 {total_count} 条日志")
def update_module_list(self):
"""更新模块下拉列表"""
current_selection = self.module_var.get()
self.modules = set(self.log_index.module_index.keys())
module_values = ["全部"] + sorted(list(self.modules))
self.module_combo["values"] = module_values
if current_selection in module_values:
self.module_var.set(current_selection)
else:
self.module_var.set("全部")
def main():
root = tk.Tk()

3
src/chat/focus_chat/heartFC_chat.py
View File

@@ -23,7 +23,6 @@ from src.chat.heart_flow.observation.actions_observation import ActionObservatio
from src.chat.focus_chat.info_processors.tool_processor import ToolProcessor
from src.chat.focus_chat.memory_activator import MemoryActivator
from src.chat.focus_chat.info_processors.base_processor import BaseProcessor
from src.chat.focus_chat.info_processors.self_processor import SelfProcessor
from src.chat.focus_chat.info_processors.expression_selector_processor import ExpressionSelectorProcessor
from src.chat.focus_chat.planners.planner_factory import PlannerFactory
from src.chat.focus_chat.planners.modify_actions import ActionModifier
@@ -45,7 +44,6 @@ PROCESSOR_CLASSES = {
"ChattingInfoProcessor": (ChattingInfoProcessor, None),
"ToolProcessor": (ToolProcessor, "tool_use_processor"),
"WorkingMemoryProcessor": (WorkingMemoryProcessor, "working_memory_processor"),
"SelfProcessor": (SelfProcessor, "self_identify_processor"),
"RelationshipProcessor": (RelationshipProcessor, "relation_processor"),
"ExpressionSelectorProcessor": (ExpressionSelectorProcessor, "expression_selector_processor"),
}
@@ -184,7 +182,6 @@ class HeartFChatting:
if name in [
"ToolProcessor",
"WorkingMemoryProcessor",
"SelfProcessor",
"RelationshipProcessor",
"ExpressionSelectorProcessor",
]:

84
src/chat/focus_chat/info_processors/relationship_processor.py
View File

@@ -13,6 +13,7 @@ from typing import List
from typing import Dict
from src.chat.focus_chat.info.info_base import InfoBase
from src.chat.focus_chat.info.relation_info import RelationInfo
from src.person_info.person_info import PersonInfoManager
from json_repair import repair_json
from src.person_info.person_info import get_person_info_manager
import json
@@ -48,10 +49,11 @@ def init_prompt():
请不要重复调取相同的信息
{name_block}
请你阅读聊天记录，查看是否需要调取某个人的信息，这个人可以是出现在聊天记录中的，也可以是记录中提到的人。
请你阅读聊天记录，查看是否需要调取某个人的信息，这个人可以是出现在聊天记录中的，也可以是记录中提到的人，也可以是你自己({bot_name})。
你不同程度上认识群聊里的人，以及他们谈论到的人，你可以根据聊天记录，回忆起有关他们的信息，帮助你参与聊天
1.你需要提供用户名和你想要提取的信息名称类型来进行调取
2.请注意，提取的信息类型一定要和用户有关，不要提取无关的信息
3.你也可以调取有关自己({bot_name})的信息
请以json格式输出，例如：
@@ -59,7 +61,7 @@ def init_prompt():
"用户A": "ta的昵称",
"用户B": "ta对你的态度",
"用户D": "你对ta的印象",
"person_name": "其他信息",
"{bot_name}": "身份",
"person_name": "其他信息",
}}
@@ -81,6 +83,18 @@ def init_prompt():
"""
Prompt(fetch_info_prompt, "fetch_person_info_prompt")
fetch_bot_info_prompt = """
你是{nickname}，你的昵称有{alias_names}。
以下是你对自己的了解，请你从中提取和"{info_type}"有关的信息，如果无法提取，请输出none：
{person_impression_block}
{points_text_block}
请严格按照以下json输出格式，不要输出多余内容：
{{
"{info_type}": "有关你自己的{info_type}的信息内容"
}}
"""
Prompt(fetch_bot_info_prompt, "fetch_bot_info_prompt")
class RelationshipProcessor(BaseProcessor):
log_prefix = "关系"
@@ -549,6 +563,7 @@ class RelationshipProcessor(BaseProcessor):
prompt = (await global_prompt_manager.get_prompt_async("relationship_prompt")).format(
name_block=name_block,
bot_name=global_config.bot.nickname,
time_now=time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()),
chat_observe_info=chat_observe_info,
info_cache_block=info_cache_block,
@@ -567,17 +582,17 @@ class RelationshipProcessor(BaseProcessor):
person_info_manager = get_person_info_manager()
for person_name, info_type in content_json.items():
is_bot = person_name == global_config.bot.nickname or person_name in global_config.bot.alias_names
if is_bot:
person_id = person_info_manager.get_person_id("system", "bot_id")
logger.info(f"{self.log_prefix} 检测到对bot自身({person_name})的信息查询，使用特殊ID。")
else:
person_id = person_info_manager.get_person_id_by_person_name(person_name)
if not person_id:
logger.warning(f"{self.log_prefix} 未找到用户 {person_name} 的ID，跳过调取信息。")
continue
# 检查是否是bot自己，如果是则跳过
user_id = person_info_manager.get_value_sync(person_id, "user_id")
if user_id == global_config.bot.qq_account:
logger.info(f"{self.log_prefix} 跳过调取bot自己({person_name})的信息。")
continue
self.info_fetching_cache.append(
{
"person_id": person_id,
@@ -747,42 +762,37 @@ class RelationshipProcessor(BaseProcessor):
# 首先检查 info_list 缓存
info_list = await person_info_manager.get_value(person_id, "info_list") or []
cached_info = None
person_name = await person_info_manager.get_value(person_id, "person_name")
print(f"info_list: {info_list}")
# print(f"info_list: {info_list}")
# 查找对应的 info_type
for info_item in info_list:
if info_item.get("info_type") == info_type:
cached_info = info_item.get("info_content")
logger.info(f"{self.log_prefix} [缓存命中] 从 info_list 中找到 {info_type} 信息: {cached_info}")
logger.debug(f"{self.log_prefix} 在info_list中找到 {person_name} 的 {info_type} 信息: {cached_info}")
break
# 如果缓存中有信息，直接使用
if cached_info:
person_name = await person_info_manager.get_value(person_id, "person_name")
if person_id not in self.info_fetched_cache:
self.info_fetched_cache[person_id] = {}
if cached_info == "none":
unknow = True
else:
unknow = False
self.info_fetched_cache[person_id][info_type] = {
"info": cached_info,
"ttl": 8,
"ttl": 4,
"start_time": start_time,
"person_name": person_name,
"unknow": unknow,
"unknow": cached_info == "none",
}
logger.info(f"{self.log_prefix} [缓存使用] 直接使用缓存的 {person_name} 的 {info_type}: {cached_info}")
logger.info(f"{self.log_prefix} 记得 {person_name} 的 {info_type}: {cached_info}")
return
logger.info(f"{self.log_prefix} [缓存命中] 缓存中没有信息")
bot_person_id = PersonInfoManager.get_person_id("system", "bot_id")
is_bot = person_id == bot_person_id
try:
nickname_str = ",".join(global_config.bot.alias_names)
name_block = f"你的名字是{global_config.bot.nickname},你的昵称有{nickname_str}，有人也会用这些昵称称呼你。"
person_name = await person_info_manager.get_value(person_id, "person_name")
person_impression = await person_info_manager.get_value(person_id, "impression")
if person_impression:
@@ -804,13 +814,26 @@ class RelationshipProcessor(BaseProcessor):
self.info_fetched_cache[person_id] = {}
self.info_fetched_cache[person_id][info_type] = {
"info": "none",
"ttl": 8,
"ttl": 4,
"start_time": start_time,
"person_name": person_name,
"unknow": True,
}
logger.info(f"{self.log_prefix} 完全不认识 {person_name}")
await self._save_info_to_cache(person_id, info_type, "none")
return
if is_bot:
prompt = (await global_prompt_manager.get_prompt_async("fetch_bot_info_prompt")).format(
nickname=global_config.bot.nickname,
alias_names=",".join(global_config.bot.alias_names),
info_type=info_type,
person_impression_block=person_impression_block,
points_text_block=points_text_block,
)
else:
nickname_str = ",".join(global_config.bot.alias_names)
name_block = f"你的名字是{global_config.bot.nickname},你的昵称有{nickname_str}，有人也会用这些昵称称呼你。"
prompt = (await global_prompt_manager.get_prompt_async("fetch_person_info_prompt")).format(
name_block=name_block,
info_type=info_type,
@@ -821,14 +844,13 @@ class RelationshipProcessor(BaseProcessor):
)
except Exception:
logger.error(traceback.format_exc())
print(prompt)
return
try:
# 使用小模型进行即时提取
content, _ = await self.instant_llm_model.generate_response_async(prompt=prompt)
logger.info(f"{self.log_prefix} [LLM提取] {person_name} 的 {info_type} 结果: {content}")
if content:
content_json = json.loads(repair_json(content))
@@ -851,17 +873,15 @@ class RelationshipProcessor(BaseProcessor):
await self._save_info_to_cache(person_id, info_type, info_content if not is_unknown else "none")
if not is_unknown:
logger.info(
f"{self.log_prefix} [LLM提取] 成功获取并缓存 {person_name} 的 {info_type}: {info_content}"
)
logger.info(f"{self.log_prefix} 思考得到，{person_name} 的 {info_type}: {content}")
else:
logger.info(f"{self.log_prefix} [LLM提取] {person_name} 的 {info_type} 信息不明确")
logger.info(f"{self.log_prefix} 思考了也不知道{person_name} 的 {info_type} 信息")
else:
logger.warning(
f"{self.log_prefix} [LLM提取] 小模型返回空结果，获取 {person_name} 的 {info_type} 信息失败。"
f"{self.log_prefix} 小模型返回空结果，获取 {person_name} 的 {info_type} 信息失败。"
)
except Exception as e:
logger.error(f"{self.log_prefix} [LLM提取] 执行小模型请求获取用户信息时出错: {e}")
logger.error(f"{self.log_prefix} 执行小模型请求获取用户信息时出错: {e}")
logger.error(traceback.format_exc())
async def _save_info_to_cache(self, person_id: str, info_type: str, info_content: str):

184
src/chat/focus_chat/info_processors/self_processor.py
View File

@@ -1,184 +0,0 @@
from src.chat.heart_flow.observation.chatting_observation import ChattingObservation
from src.chat.heart_flow.observation.observation import Observation
from src.llm_models.utils_model import LLMRequest
from src.config.config import global_config
import time
import traceback
from src.common.logger import get_logger
from src.chat.utils.prompt_builder import Prompt, global_prompt_manager
from src.chat.message_receive.chat_stream import get_chat_manager
from .base_processor import BaseProcessor
from typing import List, Dict
from src.chat.heart_flow.observation.hfcloop_observation import HFCloopObservation
from src.chat.focus_chat.info.info_base import InfoBase
from src.chat.focus_chat.info.self_info import SelfInfo
from src.individuality.individuality import get_individuality
logger = get_logger("processor")
def init_prompt():
indentify_prompt = """
{time_now}，以下是正在进行的聊天内容：
<聊天记录>
{chat_observe_info}
</聊天记录>
{name_block}
请你根据以上聊天记录，思考聊天记录中是否有人提到你自己相关的信息，或者有人询问你的相关信息。
请选择你需要查询的关键词来回答聊天中的问题。如果需要多个关键词，请用逗号隔开。
如果聊天中没有涉及任何关于你的问题，请输出none。
现在请输出你要查询的关键词，注意只输出关键词就好，不要输出其他内容：
"""
Prompt(indentify_prompt, "indentify_prompt")
class SelfProcessor(BaseProcessor):
log_prefix = "自我认同"
def __init__(self, subheartflow_id: str):
super().__init__()
self.subheartflow_id = subheartflow_id
self.info_fetched_cache: Dict[str, Dict[str, any]] = {}
self.llm_model = LLMRequest(
model=global_config.model.utils_small,
request_type="focus.processor.self_identify",
)
name = get_chat_manager().get_stream_name(self.subheartflow_id)
self.log_prefix = f"[{name}] "
async def process_info(self, observations: List[Observation] = None, *infos) -> List[InfoBase]:
"""处理信息对象
Args:
*infos: 可变数量的InfoBase类型的信息对象
Returns:
List[InfoBase]: 处理后的结构化信息列表
"""
self_info_str = await self.self_indentify(observations)
if self_info_str:
self_info = SelfInfo()
self_info.set_self_info(self_info_str)
else:
self_info = None
return None
return [self_info]
async def self_indentify(
self,
observations: List[Observation] = None,
):
"""
在回复前进行思考，生成内心想法并收集工具调用结果
参数:
observations: 观察信息
返回:
如果return_prompt为False:
tuple: (current_mind, past_mind) 当前想法和过去的想法列表
如果return_prompt为True:
tuple: (current_mind, past_mind, prompt) 当前想法、过去的想法列表和使用的prompt
"""
if observations is None:
observations = []
for observation in observations:
if isinstance(observation, ChattingObservation):
# 获取聊天元信息
is_group_chat = observation.is_group_chat
chat_target_info = observation.chat_target_info
chat_target_name = "对方" # 私聊默认名称
if not is_group_chat and chat_target_info:
# 优先使用person_name，其次user_nickname，最后回退到默认值
chat_target_name = (
chat_target_info.get("person_name") or chat_target_info.get("user_nickname") or chat_target_name
)
# 获取聊天内容
chat_observe_info = observation.get_observe_info()
if isinstance(observation, HFCloopObservation):
pass
nickname_str = ""
for nicknames in global_config.bot.alias_names:
nickname_str += f"{nicknames},"
name_block = f"你的名字是{global_config.bot.nickname},你的昵称有{nickname_str}，有人也会用这些昵称称呼你。"
# 获取所有可用的关键词
individuality = get_individuality()
available_keywords = individuality.get_all_keywords()
available_keywords_str = "、".join(available_keywords) if available_keywords else "暂无关键词"
prompt = (await global_prompt_manager.get_prompt_async("indentify_prompt")).format(
name_block=name_block,
time_now=time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()),
chat_observe_info=chat_observe_info[-200:],
available_keywords=available_keywords_str,
bot_name=global_config.bot.nickname,
)
keyword = ""
try:
keyword, _ = await self.llm_model.generate_response_async(prompt=prompt)
# print(f"prompt: {prompt}\nkeyword: {keyword}")
if not keyword:
logger.warning(f"{self.log_prefix} LLM返回空结果，自我识别失败。")
except Exception as e:
# 处理总体异常
logger.error(f"{self.log_prefix} 执行LLM请求或处理响应时出错: {e}")
logger.error(traceback.format_exc())
keyword = "我是谁，我从哪来，要到哪去"
# 解析关键词
keyword = keyword.strip()
if not keyword or keyword == "none":
keyword_set = []
else:
# 只保留非空关键词，去除多余空格
keyword_set = [k.strip() for k in keyword.split(",") if k.strip()]
# 从individuality缓存中查询关键词信息
for keyword in keyword_set:
if keyword not in self.info_fetched_cache:
# 直接从individuality的json缓存中获取关键词信息
fetched_info = individuality.get_keyword_info(keyword)
if fetched_info:
self.info_fetched_cache[keyword] = {
"info": fetched_info,
"ttl": 5,
}
logger.info(f"{self.log_prefix} 从个体特征缓存中获取关键词 '{keyword}' 的信息")
# 管理TTL（生存时间）
expired_keywords = []
for fetched_keyword, info in self.info_fetched_cache.items():
if info["ttl"] > 0:
info["ttl"] -= 1
else:
expired_keywords.append(fetched_keyword)
# 删除过期的关键词
for expired_keyword in expired_keywords:
del self.info_fetched_cache[expired_keyword]
fetched_info_str = ""
for keyword, info in self.info_fetched_cache.items():
fetched_info_str += f"你的：{keyword}信息是: {info['info']}\n"
return fetched_info_str
init_prompt()

28
src/chat/focus_chat/planners/planner_simple.py
View File

@@ -25,34 +25,10 @@ install(extra_lines=3)
def init_prompt():
# Prompt(
# """
# {time_block}
# {indentify_block}你现在正在参与以下的聊天，以下是具体的聊天内容：
# {chat_content_block}
# {self_info_block}
# {relation_info_block}
# {cycle_info_block}
# {moderation_prompt}
# 注意，除了下面动作选项之外，你在群聊里不能做其他任何事情，这是你能力的边界，现在请你选择合适的action:
# {action_options_text}
# 请以动作的输出要求，以严格的 JSON 格式输出，且仅包含 JSON 内容。
# 请输出你提取的JSON，不要有任何其他文字或解释：
# """,
# "simple_planner_prompt",
# )
Prompt(
"""
{time_block}
{indentify_block}你现在正在B站进行直播，你是一个虚拟主播，以下是直播间内容和电脑屏幕内容：
{indentify_block}你现在正在参与以下的聊天，以下是具体的聊天内容：
{chat_content_block}
@@ -62,7 +38,7 @@ def init_prompt():
{cycle_info_block}
{moderation_prompt}
注意，除了下面动作选项之外，你在直播间里不能做其他任何事情，这是你能力的边界，现在请你选择合适的action:
注意，除了下面动作选项之外，你在群聊里不能做其他任何事情，这是你能力的边界，现在请你选择合适的action:
{action_options_text}

6
src/config/official_configs.py
View File

@@ -167,12 +167,6 @@ class FocusChatConfig(ConfigBase):
class FocusChatProcessorConfig(ConfigBase):
"""专注聊天处理器配置类"""
mind_processor: bool = False
"""是否启用思维处理器"""
self_identify_processor: bool = True
"""是否启用自我识别处理器"""
relation_processor: bool = True
"""是否启用关系识别处理器"""

403
src/individuality/individuality.py
View File

@@ -14,37 +14,13 @@ from src.manager.async_task_manager import AsyncTask
from src.llm_models.utils_model import LLMRequest
from src.config.config import global_config
from src.common.logger import get_logger
from src.person_info.person_info import get_person_info_manager
install(extra_lines=3)
logger = get_logger("individuality")
def init_prompt():
"""初始化用于关键词提取的prompts"""
extract_keywords_prompt = """
请分析以下对{bot_name}的描述，提取出其中的独立关键词。每个关键词应该是可以用来从某一角度概括的方面，例如：
性格，身高，喜好，外貌，身份，兴趣，爱好，习惯，等等..........
描述内容：
{personality_sides}
要求：
1. 选择关键词，对{bot_name}的某一方面进行概括
2. 用json格式输出，以下是示例格式：
{{
"性格":"性格开朗",
"兴趣":"喜欢唱歌",
"身份":"大学生",
}}
以上是一个例子，你可以输出多个关键词，现在请你根据描述内容进行总结{bot_name}，输出json格式：
请输出json格式，不要输出任何解释或其他内容
"""
Prompt(extract_keywords_prompt, "extract_keywords_prompt")
class Individuality:
"""个体特征管理类"""
@@ -55,11 +31,8 @@ class Individuality:
self.express_style: PersonalityExpression = PersonalityExpression()
self.name = ""
# 关键词缓存相关
self.keyword_info_cache: dict = {} # {keyword: [info_list]}
self.fetch_info_file_path = "data/personality/fetch_info.json"
self.meta_info_file_path = "data/personality/meta_info.json"
self.bot_person_id = ""
self.meta_info_file_path = "data/personality/meta.json"
async def initialize(
self,
@@ -76,6 +49,13 @@ class Individuality:
personality_sides: 人格侧面描述
identity_detail: 身份细节描述
"""
person_info_manager = get_person_info_manager()
self.bot_person_id = person_info_manager.get_person_id("system", "bot_id")
self.name = bot_nickname
# 检查配置变化，如果变化则清空
await self._check_config_and_clear_if_changed(bot_nickname, personality_core, personality_sides, identity_detail)
# 初始化人格
self.personality = Personality.initialize(
bot_nickname=bot_nickname, personality_core=personality_core, personality_sides=personality_sides
@@ -84,13 +64,31 @@ class Individuality:
# 初始化身份
self.identity = Identity(identity_detail=identity_detail)
# 将所有人设写入impression
impression_parts = []
if personality_core:
impression_parts.append(f"核心人格: {personality_core}")
if personality_sides:
impression_parts.append(f"人格侧面: {'、'.join(personality_sides)}")
if identity_detail:
impression_parts.append(f"身份: {'、'.join(identity_detail)}")
impression_text = "。".join(impression_parts)
if impression_text:
impression_text += "。"
if impression_text:
update_data = {
"platform": "system",
"user_id": "bot_id",
"person_name": self.name,
"nickname": self.name,
}
await person_info_manager.update_one_field(self.bot_person_id, "impression", impression_text, data=update_data)
logger.info(f"已将完整人设更新到bot的impression中")
await self.express_style.extract_and_store_personality_expressions()
self.name = bot_nickname
# 预处理关键词和生成信息缓存
await self._preprocess_personality_keywords(personality_sides, identity_detail)
def to_dict(self) -> dict:
"""将个体特征转换为字典格式"""
return {
@@ -257,227 +255,64 @@ class Individuality:
return self.personality.neuroticism
return None
def _get_config_hash(self, personality_sides: list, identity_detail: list) -> str:
def _get_config_hash(self, bot_nickname: str, personality_core: str, personality_sides: list, identity_detail: list) -> str:
"""获取当前personality和identity配置的哈希值"""
# 将配置转换为字符串并排序，确保一致性
config_str = json.dumps(
{"personality_sides": sorted(personality_sides), "identity_detail": sorted(identity_detail)}, sort_keys=True
)
config_data = {
"nickname": bot_nickname,
"personality_core": personality_core,
"personality_sides": sorted(personality_sides),
"identity_detail": sorted(identity_detail)
}
config_str = json.dumps(config_data, sort_keys=True)
return hashlib.md5(config_str.encode("utf-8")).hexdigest()
async def _check_config_and_clear_if_changed(
self, bot_nickname: str, personality_core: str, personality_sides: list, identity_detail: list
):
"""检查配置是否发生变化，如果变化则清空info_list"""
person_info_manager = get_person_info_manager()
current_hash = self._get_config_hash(bot_nickname, personality_core, personality_sides, identity_detail)
meta_info = self._load_meta_info()
stored_hash = meta_info.get("config_hash")
if current_hash != stored_hash:
logger.info("检测到人格配置发生变化，将清空原有的关键词缓存。")
# 清空数据库中的info_list
update_data = {
"platform": "system",
"user_id": "bot_id",
"person_name": self.name,
"nickname": self.name,
}
await person_info_manager.update_one_field(self.bot_person_id, "info_list", [], data=update_data)
# 更新元信息文件，重置计数器
new_meta_info = {"config_hash": current_hash}
self._save_meta_info(new_meta_info)
def _load_meta_info(self) -> dict:
"""从JSON文件中加载元信息"""
if os.path.exists(self.meta_info_file_path):
try:
with open(self.meta_info_file_path, "r", encoding="utf-8") as f:
return json.load(f)
except Exception as e:
print(f"读取meta_info文件失败: {e}")
except (json.JSONDecodeError, IOError) as e:
logger.error(f"读取meta_info文件失败: {e}, 将创建新文件。")
return {}
return {}
def _save_meta_info(self, meta_info: dict):
"""将元信息保存到JSON文件"""
try:
# 确保目录存在
os.makedirs(os.path.dirname(self.meta_info_file_path), exist_ok=True)
with open(self.meta_info_file_path, "w", encoding="utf-8") as f:
json.dump(meta_info, f, ensure_ascii=False, indent=2)
except Exception as e:
print(f"保存meta_info文件失败: {e}")
except IOError as e:
logger.error(f"保存meta_info文件失败: {e}")
def _check_config_change_and_clear(self, personality_sides: list, identity_detail: list):
"""检查配置是否发生变化，如果变化则清空fetch_info.json"""
current_config_hash = self._get_config_hash(personality_sides, identity_detail)
meta_info = self._load_meta_info()
stored_config_hash = meta_info.get("config_hash", "")
if current_config_hash != stored_config_hash:
logger.info("检测到personality或identity配置发生变化，清空fetch_info数据")
# 清空fetch_info文件
if os.path.exists(self.fetch_info_file_path):
try:
os.remove(self.fetch_info_file_path)
logger.info("已清空fetch_info文件")
except Exception as e:
logger.error(f"清空fetch_info文件失败: {e}")
# 更新元信息
meta_info["config_hash"] = current_config_hash
self._save_meta_info(meta_info)
logger.info("已更新配置哈希值")
def _load_fetch_info_from_file(self) -> dict:
"""从JSON文件中加载已保存的fetch_info数据"""
if os.path.exists(self.fetch_info_file_path):
try:
with open(self.fetch_info_file_path, "r", encoding="utf-8") as f:
return json.load(f)
except Exception as e:
logger.error(f"读取fetch_info文件失败: {e}")
return {}
return {}
def _save_fetch_info_to_file(self, fetch_info_data: dict):
"""将fetch_info数据保存到JSON文件"""
try:
# 确保目录存在
os.makedirs(os.path.dirname(self.fetch_info_file_path), exist_ok=True)
with open(self.fetch_info_file_path, "w", encoding="utf-8") as f:
json.dump(fetch_info_data, f, ensure_ascii=False, indent=2)
except Exception as e:
logger.error(f"保存fetch_info文件失败: {e}")
async def _preprocess_personality_keywords(self, personality_sides: list, identity_detail: list):
"""预处理personality关键词，提取关键词并生成缓存"""
try:
logger.info("开始预处理personality关键词...")
# 检查配置变化
self._check_config_change_and_clear(personality_sides, identity_detail)
# 加载已有的预处理数据（如果存在）
fetch_info_data = self._load_fetch_info_from_file()
logger.info(f"加载已有数据，现有关键词数量: {len(fetch_info_data)}")
# 构建完整描述（personality + identity）
personality_sides_str = ""
for personality_side in personality_sides:
personality_sides_str += f"{personality_side}，"
# 添加identity内容
for detail in identity_detail:
personality_sides_str += f"{detail}，"
if not personality_sides_str:
logger.info("没有personality和identity配置，跳过预处理")
return
# 提取关键词
extract_prompt = (await global_prompt_manager.get_prompt_async("extract_keywords_prompt")).format(
personality_sides=personality_sides_str, bot_name=self.name
)
llm_model = LLMRequest(
model=global_config.model.utils_small,
request_type="individuality.keyword_extract",
)
keywords_result, _ = await llm_model.generate_response_async(prompt=extract_prompt)
logger.info(f"LLM返回的原始关键词结果: '{keywords_result}'")
if not keywords_result or keywords_result.strip() == "none":
logger.info("未提取到有效关键词")
return
# 使用json_repair修复并解析JSON
keyword_dict = json.loads(repair_json(keywords_result))
logger.info(f"成功解析JSON格式的关键词: {keyword_dict}")
# 从字典中提取关键词列表，跳过"keywords"键
keyword_set = []
for key, _value in keyword_dict.items():
if key.lower() != "keywords" and key.strip():
keyword_set.append(key.strip())
logger.info(f"最终提取的关键词列表: {keyword_set}")
logger.info(f"共提取到 {len(keyword_set)} 个关键词")
# 处理每个关键词的信息
updated_count = 0
new_count = 0
for keyword in keyword_set:
try:
logger.info(f"正在处理关键词: '{keyword}' (长度: {len(keyword)})")
# 检查是否已存在该关键词
if keyword in fetch_info_data:
logger.info(f"关键词 '{keyword}' 已存在，将添加新信息...")
action_type = "追加"
else:
logger.info(f"正在为新关键词 '{keyword}' 生成信息...")
action_type = "新增"
fetch_info_data[keyword] = [] # 初始化为空列表
# 从JSON结果中获取关键词的信息
existing_info_from_json = keyword_dict.get(keyword, "")
if (
existing_info_from_json
and existing_info_from_json.strip()
and existing_info_from_json != keyword
):
# 如果JSON中有有效信息且不只是重复关键词本身，直接使用
logger.info(f"从JSON结果中获取到关键词 '{keyword}' 的信息: '{existing_info_from_json}'")
if existing_info_from_json not in fetch_info_data[keyword]:
fetch_info_data[keyword].append(existing_info_from_json)
if action_type == "追加":
updated_count += 1
else:
new_count += 1
logger.info(f"{action_type}关键词 '{keyword}' 的信息成功")
else:
logger.info(f"关键词 '{keyword}' 的信息已存在，跳过重复添加")
else:
logger.info(f"关键词 '{keyword}' 在JSON中没有有效信息，跳过")
except Exception as e:
logger.error(f"为关键词 '{keyword}' 生成信息时出错: {e}")
continue
# 保存合并后的数据到文件和内存缓存
if updated_count > 0 or new_count > 0:
self._save_fetch_info_to_file(fetch_info_data)
logger.info(
f"预处理完成，新增 {new_count} 个关键词，追加 {updated_count} 个关键词信息，总计 {len(fetch_info_data)} 个关键词"
)
else:
logger.info("预处理完成，但没有生成任何新的有效信息")
# 将数据加载到内存缓存
self.keyword_info_cache = fetch_info_data
logger.info(f"关键词缓存已加载，共 {len(self.keyword_info_cache)} 个关键词")
# 注册定时任务（延迟执行，避免阻塞初始化）
import asyncio
asyncio.create_task(self._register_keyword_update_task_delayed())
except Exception as e:
logger.error(f"预处理personality关键词时出错: {e}")
traceback.print_exc()
async def _register_keyword_update_task_delayed(self):
"""延迟注册关键词更新定时任务"""
try:
# 等待一小段时间确保系统完全初始化
import asyncio
await asyncio.sleep(5)
from src.manager.async_task_manager import async_task_manager
logger = get_logger("individuality")
# 创建定时任务
task = KeywordUpdateTask(
personality_sides=list(global_config.personality.personality_sides),
identity_detail=list(global_config.identity.identity_detail),
individuality_instance=self,
)
# 注册任务
await async_task_manager.add_task(task)
logger.info("关键词更新定时任务已注册")
except Exception as e:
logger.error(f"注册关键词更新定时任务失败: {e}")
traceback.print_exc()
def get_keyword_info(self, keyword: str) -> str:
async def get_keyword_info(self, keyword: str) -> str:
"""获取指定关键词的信息
Args:
@@ -486,13 +321,36 @@ class Individuality:
Returns:
str: 随机选择的一条信息，如果没有则返回空字符串
"""
if keyword in self.keyword_info_cache and self.keyword_info_cache[keyword]:
return random.choice(self.keyword_info_cache[keyword])
person_info_manager = get_person_info_manager()
info_list_json = await person_info_manager.get_value(self.bot_person_id, "info_list")
if info_list_json:
try:
# get_value might return a pre-deserialized list if it comes from a cache,
# or a JSON string if it comes from DB.
info_list = json.loads(info_list_json) if isinstance(info_list_json, str) else info_list_json
for item in info_list:
if isinstance(item, dict) and item.get("info_type") == keyword:
return item.get("info_content", "")
except (json.JSONDecodeError, TypeError):
logger.error(f"解析info_list失败: {info_list_json}")
return ""
return ""
def get_all_keywords(self) -> list:
async def get_all_keywords(self) -> list:
"""获取所有已缓存的关键词列表"""
return list(self.keyword_info_cache.keys())
person_info_manager = get_person_info_manager()
info_list_json = await person_info_manager.get_value(self.bot_person_id, "info_list")
keywords = []
if info_list_json:
try:
info_list = json.loads(info_list_json) if isinstance(info_list_json, str) else info_list_json
for item in info_list:
if isinstance(item, dict) and "info_type" in item:
keywords.append(item["info_type"])
except (json.JSONDecodeError, TypeError):
logger.error(f"解析info_list失败: {info_list_json}")
return keywords
individuality = None
@@ -503,66 +361,3 @@ def get_individuality():
if individuality is None:
individuality = Individuality()
return individuality
class KeywordUpdateTask(AsyncTask):
"""关键词更新定时任务"""
def __init__(self, personality_sides: list, identity_detail: list, individuality_instance):
# 调用父类构造函数
super().__init__(
task_name="keyword_update_task",
wait_before_start=3600, # 1小时后开始
run_interval=3600, # 每小时运行一次
)
self.personality_sides = personality_sides
self.identity_detail = identity_detail
self.individuality_instance = individuality_instance
# 任务控制参数
self.max_runs = 20
self.current_runs = 0
self.original_config_hash = individuality_instance._get_config_hash(personality_sides, identity_detail)
async def run(self):
"""执行任务"""
try:
from src.common.logger import get_logger
logger = get_logger("individuality.task")
# 检查是否超过最大运行次数
if self.current_runs >= self.max_runs:
logger.info(f"关键词更新任务已达到最大运行次数({self.max_runs})，停止执行")
# 设置为0间隔来停止循环任务
self.run_interval = 0
return
# 检查配置是否发生变化
current_config_hash = self.individuality_instance._get_config_hash(
self.personality_sides, self.identity_detail
)
if current_config_hash != self.original_config_hash:
logger.info("检测到personality或identity配置发生变化，停止定时任务")
# 设置为0间隔来停止循环任务
self.run_interval = 0
return
self.current_runs += 1
logger.info(f"开始执行关键词更新任务 (第{self.current_runs}/{self.max_runs}次)")
# 执行关键词预处理
await self.individuality_instance._preprocess_personality_keywords(
self.personality_sides, self.identity_detail
)
logger.info(f"关键词更新任务完成 (第{self.current_runs}/{self.max_runs}次)")
except Exception as e:
logger.error(f"关键词更新任务执行失败: {e}")
traceback.print_exc()
# 初始化prompt模板
init_prompt()

1
src/person_info/person_info.py
View File

@@ -47,6 +47,7 @@ person_info_default = {
"info_list": None,
"points": None,
"forgotten_points": None,
"config_hash": None,
}

6
src/plugins/built_in/core_actions/plugin.py
View File

@@ -128,7 +128,7 @@ class NoReplyAction(BaseAction):
_waiting_stages = [10, 60, 600] # 第1、2、3次的等待时间
# 动作参数定义
action_parameters = {}
action_parameters = {"reason": "不回复的原因"}
# 动作使用场景
action_require = ["你发送了消息，目前无人回复"]
@@ -143,6 +143,8 @@ class NoReplyAction(BaseAction):
NoReplyAction._consecutive_count += 1
count = NoReplyAction._consecutive_count
reason = self.action_data.get("reason", "")
# 计算本次等待时间
if count <= len(self._waiting_stages):
# 前3次使用预设时间
@@ -153,7 +155,7 @@ class NoReplyAction(BaseAction):
# 第4次及以后使用WAITING_TIME_THRESHOLD
timeout = self.waiting_timeout
logger.info(f"{self.log_prefix} 选择不回复(第{count}次连续)，等待新消息中... (超时: {timeout}秒)")
logger.info(f"{self.log_prefix} 选择不回复(第{count}次连续)，等待新消息中... (超时: {timeout}秒)，原因: {reason}")
# 等待新消息或达到时间上限
result = await self.wait_for_new_message(timeout)

3
template/bot_config_template.toml
View File

@@ -1,5 +1,5 @@
[inner]
version = "2.24.0"
version = "2.25.0"
#----以下是给开发人员阅读的，如果你只是部署了麦麦，不需要阅读----
#如果你想要修改配置文件，请在修改后将version的值进行变更
@@ -102,7 +102,6 @@ compressed_length = 8 # 不能大于observation_context_size,心流上下文压
compress_length_limit = 4 #最多压缩份数，超过该数值的压缩上下文会被删除
[focus_chat_processor] # 专注聊天处理器，打开可以实现更多功能，但是会增加token消耗
self_identify_processor = true # 是否启用自我识别处理器
relation_processor = true # 是否启用关系识别处理器
tool_use_processor = false # 是否启用工具使用处理器
working_memory_processor = false # 是否启用工作记忆处理器，消耗量大