nas/MoviePilot
1
0
Fork 0
mirror of https://github.com/jxxghp/MoviePilot.git synced 2026-03-20 03:57:30 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix(agent)：修复智能体工具调用，优化媒体库查询工具

Browse Source
This commit is contained in:
jxxghp
2026-01-24 09:46:19 +08:00
parent af19f274a7
commit ec40f36114
5 changed files with 97 additions and 117 deletions
Download Patch File Download Diff File Expand all files Collapse all files

90
app/agent/__init__.py
View File

@@ -98,15 +98,14 @@ class MoviePilotAgent:
user_id=self.user_id
)
if messages:
loaded_messages = []
for msg in messages:
if msg.get("role") == "user":
loaded_messages.append(HumanMessage(content=msg.get("content", "")))
chat_history.add_message(HumanMessage(content=msg.get("content", "")))
elif msg.get("role") == "agent":
loaded_messages.append(AIMessage(content=msg.get("content", "")))
chat_history.add_message(AIMessage(content=msg.get("content", "")))
elif msg.get("role") == "tool_call":
metadata = msg.get("metadata", {})
loaded_messages.append(
chat_history.add_message(
AIMessage(
content=msg.get("content", ""),
tool_calls=[
@@ -120,17 +119,12 @@ class MoviePilotAgent:
)
elif msg.get("role") == "tool_result":
metadata = msg.get("metadata", {})
loaded_messages.append(ToolMessage(
chat_history.add_message(ToolMessage(
content=msg.get("content", ""),
tool_call_id=metadata.get("call_id", "unknown")
))
elif msg.get("role") == "system":
loaded_messages.append(SystemMessage(content=msg.get("content", "")))
# 验证并修复工具调用的完整性
validated_messages = self._ensure_tool_call_integrity(loaded_messages)
for msg in validated_messages:
chat_history.add_message(msg)
chat_history.add_message(SystemMessage(content=msg.get("content", "")))
return chat_history
@@ -199,71 +193,6 @@ class MoviePilotAgent:
# 发生错误时返回一个保守的估算值
return len(str(messages)) // 4
def _ensure_tool_call_integrity(self, messages: List[Union[HumanMessage, AIMessage, ToolMessage, SystemMessage]]) \
-> List[Union[HumanMessage, AIMessage, ToolMessage, SystemMessage]]:
"""
确保工具调用的完整性:
1. 带有 tool_calls 的 AIMessage 必须后跟相应的 ToolMessages
2. 移除不完整的 AIMessages（有 tool_calls 但缺少 ToolMessage 响应）及其部分 ToolMessages
3. 过滤掉不对应任何 tool_call 的孤立 ToolMessages
"""
if not messages:
return messages
validated_messages = []
i = 0
while i < len(messages):
msg = messages[i]
# 检查这是否是一个带有 tool_calls 的 AIMessage
if isinstance(msg, AIMessage) and getattr(msg, 'tool_calls', None):
# 提取 tool_call IDs（ToolCall 是 TypedDict，在运行时是 dict）
tool_call_ids = {
tc.get('id') if tc.get('id') is not None else tc.get('name')
for tc in msg.tool_calls
}
# 查找后续的 ToolMessages
j = i + 1
found_tool_messages = []
while j < len(messages) and isinstance(messages[j], ToolMessage):
found_tool_messages.append(messages[j])
j += 1
# 检查所有 tool_calls 是否都有对应的 ToolMessages
found_tool_call_ids = {tm.tool_call_id for tm in found_tool_messages}
# 警告：如果存在不对应任何 tool_call 的额外 ToolMessages
extra_tool_messages = found_tool_call_ids - tool_call_ids
if extra_tool_messages:
logger.warning(f"Found extra ToolMessages that don't correspond to any tool_call: {extra_tool_messages}")
if not tool_call_ids.issubset(found_tool_call_ids):
# 缺少部分 tool_call 响应，跳过此 AIMessage 及其所有 ToolMessages
logger.warning("Removing incomplete tool_call AIMessage and its partial ToolMessages: missing tool_call responses")
i = j # 跳过此 AIMessage 及其所有相关的 ToolMessages
continue
else:
# 添加此 AIMessage 及其对应的 ToolMessages
validated_messages.append(msg)
for tm in found_tool_messages:
if tm.tool_call_id in tool_call_ids:
validated_messages.append(tm)
i = j
continue
# 跳过孤立的 ToolMessages（前面没有带 tool_calls 的 AIMessage）
elif isinstance(msg, ToolMessage):
logger.warning(f"Skipping orphaned ToolMessage with tool_call_id: {msg.tool_call_id}")
i += 1
continue
else:
validated_messages.append(msg)
i += 1
return validated_messages
def _create_agent_executor(self) -> RunnableWithMessageHistory:
"""
创建Agent执行器
@@ -281,8 +210,13 @@ class MoviePilotAgent:
# 包装trimmer，在裁剪后验证工具调用的完整性
def validated_trimmer(messages):
trimmed = base_trimmer(messages)
return self._ensure_tool_call_integrity(trimmed)
# 如果输入是 PromptValue，转换为消息列表
if hasattr(messages, "to_messages"):
messages = messages.to_messages()
trimmed = base_trimmer.invoke(messages)
if len(trimmed) < len(messages):
logger.info(f"LangChain消息上下文已裁剪: {len(messages)} -> {len(trimmed)}")
return trimmed
# 创建Agent执行链
agent = (

2
app/agent/memory/__init__.py
View File

@@ -232,7 +232,7 @@ class ConversationMemoryManager:
return []
# 获取所有消息
return memory.messages
return memory.messages[:-1]
async def get_recent_messages(
self,

18
app/agent/tools/base.py
View File

@@ -1,4 +1,5 @@
import json
import uuid
from abc import ABCMeta, abstractmethod
from typing import Any, Optional
@@ -42,6 +43,9 @@ class MoviePilotTool(BaseTool, metaclass=ABCMeta):
# 获取工具调用前的agent消息
agent_message = await self._callback_handler.get_message()
# 生成唯一的工具调用ID
call_id = f"call_{str(uuid.uuid4())[:16]}"
# 记忆工具调用
await conversation_manager.add_conversation(
session_id=self._session_id,
@@ -49,8 +53,8 @@ class MoviePilotTool(BaseTool, metaclass=ABCMeta):
role="tool_call",
content=agent_message,
metadata={
"call_id": self.__class__.__name__,
"tool_name": self.__class__.__name__,
"call_id": call_id,
"tool_name": self.name,
"parameters": kwargs
}
)
@@ -61,21 +65,21 @@ class MoviePilotTool(BaseTool, metaclass=ABCMeta):
explanation = kwargs.get("explanation")
if explanation:
tool_message = explanation
# 合并agent消息和工具执行消息，一起发送
messages = []
if agent_message:
messages.append(agent_message)
if tool_message:
messages.append(f"⚙️ => {tool_message}")
# 发送合并后的消息
if messages:
merged_message = "\n\n".join(messages)
await self.send_tool_message(merged_message, title="MoviePilot助手")
logger.debug(f'Executing tool {self.name} with args: {kwargs}')
# 执行工具，捕获异常确保结果总是被存储到记忆中
try:
result = await self.run(**kwargs)
@@ -93,13 +97,15 @@ class MoviePilotTool(BaseTool, metaclass=ABCMeta):
formated_result = str(result)
else:
formated_result = json.dumps(result, ensure_ascii=False, indent=2)
await conversation_manager.add_conversation(
session_id=self._session_id,
user_id=self._user_id,
role="tool_result",
content=formated_result,
metadata={
"call_id": self.__class__.__name__
"call_id": call_id,
"tool_name": self.name,
}
)

82
app/agent/tools/impl/query_library_exists.py
View File

@@ -8,6 +8,7 @@ from pydantic import BaseModel, Field
from app.agent.tools.base import MoviePilotTool
from app.chain.mediaserver import MediaServerChain
from app.core.context import MediaInfo
from app.core.meta import MetaBase
from app.log import logger
from app.schemas.types import MediaType
@@ -51,47 +52,88 @@ class QueryLibraryExistsTool(MoviePilotTool):
try:
if not title:
return "请提供媒体标题进行查询"
# 创建 MediaInfo 对象
mediainfo = MediaInfo()
mediainfo.title = title
mediainfo.year = year
# 转换媒体类型
if media_type == "电影":
mediainfo.type = MediaType.MOVIE
elif media_type == "电视剧":
mediainfo.type = MediaType.TV
# media_type == "all" 时不设置类型，让媒体服务器自动判断
# 调用媒体服务器接口实时查询
media_chain = MediaServerChain()
# 1. 识别媒体信息（获取 TMDB ID 和各季的总集数等元数据）
meta = MetaBase(title=title)
if year:
meta.year = str(year)
if media_type == "电影":
meta.type = MediaType.MOVIE
elif media_type == "电视剧":
meta.type = MediaType.TV
# 使用识别方法补充信息
recognize_info = media_chain.recognize_media(meta=meta)
if recognize_info:
mediainfo = recognize_info
else:
# 识别失败，创建基本信息的 MediaInfo
mediainfo = MediaInfo()
mediainfo.title = title
mediainfo.year = year
if media_type == "电影":
mediainfo.type = MediaType.MOVIE
elif media_type == "电视剧":
mediainfo.type = MediaType.TV
# 2. 调用媒体服务器接口实时查询存在信息
existsinfo = media_chain.media_exists(mediainfo=mediainfo)
if not existsinfo:
return "媒体库中未找到相关媒体"
# 如果找到了，获取详细信息
# 3. 如果找到了，获取详细信息并组装结果
result_items = []
if existsinfo.itemid and existsinfo.server:
iteminfo = media_chain.iteminfo(server=existsinfo.server, item_id=existsinfo.itemid)
if iteminfo:
# 使用 model_dump() 转换为字典格式
item_dict = iteminfo.model_dump(exclude_none=True)
# 对于电视剧，补充已存在的季集详情及进度统计
if existsinfo.type == MediaType.TV:
# 注入已存在集信息 (Dict[int, list])
item_dict["seasoninfo"] = existsinfo.seasons
# 统计库中已存在的季集总数
if existsinfo.seasons:
item_dict["existing_episodes_count"] = sum(len(e) for e in existsinfo.seasons.values())
item_dict["seasons_existing_count"] = {str(s): len(e) for s, e in existsinfo.seasons.items()}
# 如果识别到了元数据，补充总计对比和进度概览
if mediainfo.seasons:
item_dict["seasons_total_count"] = {str(s): len(e) for s, e in mediainfo.seasons.items()}
# 进度概览，例如 "Season 1": "3/12"
item_dict["seasons_progress"] = {
f"第{s}季": f"{len(existsinfo.seasons.get(s, []))}/{len(mediainfo.seasons.get(s, []))} 集"
for s in mediainfo.seasons.keys() if (s in existsinfo.seasons or s > 0)
}
result_items.append(item_dict)
if result_items:
return json.dumps(result_items, ensure_ascii=False)
# 如果找到了但没有详细信息，返回基本信息
# 如果找到了但没有获取到 iteminfo，返回基本信息
result_dict = {
"title": mediainfo.title,
"year": mediainfo.year,
"type": existsinfo.type.value if existsinfo.type else None,
"server": existsinfo.server,
"server_type": existsinfo.server_type,
"itemid": existsinfo.itemid,
"seasons": existsinfo.seasons if existsinfo.seasons else {}
}
if existsinfo.type == MediaType.TV and existsinfo.seasons:
result_dict["existing_episodes_count"] = sum(len(e) for e in existsinfo.seasons.values())
result_dict["seasons_existing_count"] = {str(s): len(e) for s, e in existsinfo.seasons.items()}
if mediainfo.seasons:
result_dict["seasons_total_count"] = {str(s): len(e) for s, e in mediainfo.seasons.items()}
return json.dumps([result_dict], ensure_ascii=False)
except Exception as e:
logger.error(f"查询媒体库失败: {e}", exc_info=True)
return f"查询媒体库时发生错误: {str(e)}"