Losita/mai-bot
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

ref：修改了plan的执行记录展示模式，现在每个动作的执行都会被记录

Browse Source
This commit is contained in:
SengokuCola
2025-10-02 20:11:44 +08:00
parent 6a0a07582e
commit dbdf650b1d
19 changed files with 521 additions and 207 deletions
Download Patch File Download Diff File Expand all files Collapse all files

99
src/chat/planner_actions/planner.py
View File

@@ -44,10 +44,7 @@ def init_prompt():
**聊天内容**
{chat_content_block}
**动作记录**
{actions_before_now_block}
**可用的action**
**可选的action**
reply
动作描述：
1.你可以选择呼叫了你的名字，但是你没有做出回应的消息进行回复
@@ -76,7 +73,10 @@ no_reply_until_call
{action_options_text}
请选择合适的action，并说明触发action的消息id和选择该action的原因。消息id格式:m+数字
**你之前的action执行和思考记录**
{actions_before_now_block}
请选择**可选的**且符合使用条件的action，并说明触发action的消息id(消息id格式:m+数字)
先输出你的选择思考理由，再输出你选择的action，理由是一段平文本，不要分点，精简。
**动作选择要求**
请你根据聊天内容,用户的最新消息和以下标准选择合适的动作:
@@ -99,9 +99,7 @@ no_reply_until_call
"target_message_id":"触发动作的消息id",
//对应参数
}}
```
""",
```""",
"planner_prompt",
)
@@ -133,6 +131,9 @@ class ActionPlanner:
self.last_obs_time_mark = 0.0
self.plan_log:List[Tuple[str,str,ActionPlannerInfo]] = []
def find_message_by_id(
self, message_id: str, message_id_list: List[Tuple[str, "DatabaseMessages"]]
) -> Optional["DatabaseMessages"]:
@@ -157,15 +158,16 @@ class ActionPlanner:
action_json: dict,
message_id_list: List[Tuple[str, "DatabaseMessages"]],
current_available_actions: List[Tuple[str, ActionInfo]],
extracted_reasoning: str = "",
) -> List[ActionPlannerInfo]:
"""解析单个action JSON并返回ActionPlannerInfo列表"""
action_planner_infos = []
try:
action = action_json.get("action", "no_action")
action = action_json.get("action", "no_reply")
reasoning = action_json.get("reason", "未提供原因")
action_data = {key: value for key, value in action_json.items() if key not in ["action", "reason"]}
# 非no_action动作需要target_message_id
# 非no_reply动作需要target_message_id
target_message = None
if target_message_id := action_json.get("target_message_id"):
@@ -202,6 +204,7 @@ class ActionPlanner:
action_data=action_data,
action_message=target_message,
available_actions=available_actions_dict,
action_reasoning=extracted_reasoning if extracted_reasoning else None,
)
)
@@ -216,6 +219,7 @@ class ActionPlanner:
action_data={},
action_message=None,
available_actions=available_actions_dict,
action_reasoning=extracted_reasoning if extracted_reasoning else None,
)
)
@@ -225,12 +229,11 @@ class ActionPlanner:
self,
available_actions: Dict[str, ActionInfo],
loop_start_time: float = 0.0,
) -> Tuple[List[ActionPlannerInfo], Optional["DatabaseMessages"]]:
) -> List[ActionPlannerInfo]:
# sourcery skip: use-named-expression
"""
规划器 (Planner): 使用LLM根据上下文决定做出什么动作。
"""
target_message: Optional["DatabaseMessages"] = None
# 获取聊天上下文
message_list_before_now = get_raw_msg_before_timestamp_with_chat(
@@ -276,7 +279,7 @@ class ActionPlanner:
)
# 调用LLM获取决策
actions = await self._execute_main_planner(
reasoning, actions = await self._execute_main_planner(
prompt=prompt,
message_id_list=message_id_list,
filtered_actions=filtered_actions,
@@ -284,12 +287,22 @@ class ActionPlanner:
loop_start_time=loop_start_time,
)
# 获取target_message（如果有非no_action的动作）
non_no_actions = [a for a in actions if a.action_type != "no_reply"]
if non_no_actions:
target_message = non_no_actions[0].action_message
self.add_plan_log(reasoning, actions)
return actions
def add_plan_log(self, reasoning: str, actions: List[ActionPlannerInfo]):
self.plan_log.append((reasoning, time.time(), actions))
if len(self.plan_log) > 100:
self.plan_log.pop(0)
def get_plan_log_str(self) -> str:
plan_log_str = ""
for reasoning, time, actions in self.plan_log:
time = datetime.fromtimestamp(time).strftime("%H:%M:%S")
plan_log_str += f"{time}:{reasoning}|使用了{','.join([action.action_type for action in actions])}\n"
return plan_log_str
return actions, target_message
async def build_planner_prompt(
self,
@@ -302,18 +315,8 @@ class ActionPlanner:
) -> tuple[str, List[Tuple[str, "DatabaseMessages"]]]:
"""构建 Planner LLM 的提示词 (获取模板并填充数据)"""
try:
# 获取最近执行过的动作
actions_before_now = get_actions_by_timestamp_with_chat(
chat_id=self.chat_id,
timestamp_start=time.time() - 600,
timestamp_end=time.time(),
limit=6,
)
actions_before_now_block = build_readable_actions(actions=actions_before_now)
if actions_before_now_block:
actions_before_now_block = f"你刚刚选择并执行过的action是：\n{actions_before_now_block}"
else:
actions_before_now_block = ""
actions_before_now_block=self.get_plan_log_str()
# 构建聊天上下文描述
chat_context_description = "你现在正在一个群聊中"
@@ -447,7 +450,7 @@ class ActionPlanner:
filtered_actions: Dict[str, ActionInfo],
available_actions: Dict[str, ActionInfo],
loop_start_time: float,
) -> List[ActionPlannerInfo]:
) -> Tuple[str,List[ActionPlannerInfo]]:
"""执行主规划器"""
llm_content = None
actions: List[ActionPlannerInfo] = []
@@ -456,8 +459,8 @@ class ActionPlanner:
# 调用LLM
llm_content, (reasoning_content, _, _) = await self.planner_llm.generate_response_async(prompt=prompt)
# logger.info(f"{self.log_prefix}规划器原始提示词: {prompt}")
# logger.info(f"{self.log_prefix}规划器原始响应: {llm_content}")
logger.info(f"{self.log_prefix}规划器原始提示词: {prompt}")
logger.info(f"{self.log_prefix}规划器原始响应: {llm_content}")
if global_config.debug.show_prompt:
logger.info(f"{self.log_prefix}规划器原始提示词: {prompt}")
@@ -472,7 +475,7 @@ class ActionPlanner:
except Exception as req_e:
logger.error(f"{self.log_prefix}LLM 请求执行失败: {req_e}")
return [
return f"LLM 请求失败，模型出现问题: {req_e}",[
ActionPlannerInfo(
action_type="no_reply",
reasoning=f"LLM 请求失败，模型出现问题: {req_e}",
@@ -485,11 +488,12 @@ class ActionPlanner:
# 解析LLM响应
if llm_content:
try:
if json_objects := self._extract_json_from_markdown(llm_content):
json_objects, extracted_reasoning = self._extract_json_from_markdown(llm_content)
if json_objects:
logger.debug(f"{self.log_prefix}从响应中提取到{len(json_objects)}个JSON对象")
filtered_actions_list = list(filtered_actions.items())
for json_obj in json_objects:
actions.extend(self._parse_single_action(json_obj, message_id_list, filtered_actions_list))
actions.extend(self._parse_single_action(json_obj, message_id_list, filtered_actions_list, extracted_reasoning))
else:
# 尝试解析为直接的JSON
logger.warning(f"{self.log_prefix}LLM没有返回可用动作: {llm_content}")
@@ -502,17 +506,17 @@ class ActionPlanner:
else:
actions = self._create_no_reply("规划器没有获得LLM响应", available_actions)
# 添加循环开始时间到所有非no_action动作
# 添加循环开始时间到所有非no_reply动作
for action in actions:
action.action_data = action.action_data or {}
action.action_data["loop_start_time"] = loop_start_time
logger.debug(f"{self.log_prefix}规划器选择了{len(actions)}个动作: {' '.join([a.action_type for a in actions])}")
return actions
return extracted_reasoning,actions
def _create_no_reply(self, reasoning: str, available_actions: Dict[str, ActionInfo]) -> List[ActionPlannerInfo]:
"""创建no_action"""
"""创建no_reply"""
return [
ActionPlannerInfo(
action_type="no_reply",
@@ -523,15 +527,26 @@ class ActionPlanner:
)
]
def _extract_json_from_markdown(self, content: str) -> List[dict]:
def _extract_json_from_markdown(self, content: str) -> Tuple[List[dict], str]:
# sourcery skip: for-append-to-extend
"""从Markdown格式的内容中提取JSON对象"""
"""从Markdown格式的内容中提取JSON对象和推理内容"""
json_objects = []
reasoning_content = ""
# 使用正则表达式查找```json包裹的JSON内容
json_pattern = r"```json\s*(.*?)\s*```"
matches = re.findall(json_pattern, content, re.DOTALL)
# 提取JSON之前的内容作为推理文本
if matches:
# 找到第一个```json的位置
first_json_pos = content.find("```json")
if first_json_pos > 0:
reasoning_content = content[:first_json_pos].strip()
# 清理推理内容中的注释标记
reasoning_content = re.sub(r"^//\s*", "", reasoning_content, flags=re.MULTILINE)
reasoning_content = reasoning_content.strip()
for match in matches:
try:
# 清理可能的注释和格式问题
@@ -549,7 +564,7 @@ class ActionPlanner:
logger.warning(f"解析JSON块失败: {e}, 块内容: {match[:100]}...")
continue
return json_objects
return json_objects, reasoning_content
init_prompt()