JOJO
99f4d3d1dd
事故：socketio 用户任务（不在 task_manager 注册）与完成通知轮询器 派发的通知任务并发写入共享 conversation_history，产生 assistant 乱序段， 导致 Kimi API 400（tool_call_id is not found）与通知永久丢失。 - 新增 server/main_task_gate.py：对话级主任务门闸（单写者不变量， 一个 WebTerminal 同一时刻只允许一个主聊天任务） - process_message_task 作为唯一入口获取/认领门闸，拿不到即拒绝； 通知链 token 经 session_data 移交，失败释放并回滚通知标记； _run_chat_task finally 兜底释放防泄漏 - execute_tool_calls 改守护包装（try/finally 复位 _tool_loop_active）， 杜绝并发交错导致标志卡死 - build_messages 增加孤儿 tool 消息剥离防御层（止血，不替代门闸） - AGENTS.md 新增 §12 记录门闸架构与改代码硬约束
451 lines
23 KiB
Python
451 lines
23 KiB
Python
import asyncio
|
||
import json
|
||
from pathlib import Path
|
||
from typing import Any, Dict, List, Optional, Set
|
||
|
||
try:
|
||
from config import (
|
||
OUTPUT_FORMATS, DATA_DIR, PROMPTS_DIR, NEED_CONFIRMATION,
|
||
MAX_TERMINALS, TERMINAL_BUFFER_SIZE, TERMINAL_DISPLAY_SIZE,
|
||
MAX_READ_FILE_CHARS, READ_TOOL_DEFAULT_MAX_CHARS,
|
||
READ_TOOL_DEFAULT_CONTEXT_BEFORE, READ_TOOL_DEFAULT_CONTEXT_AFTER,
|
||
READ_TOOL_MAX_CONTEXT_BEFORE, READ_TOOL_MAX_CONTEXT_AFTER,
|
||
READ_TOOL_DEFAULT_MAX_MATCHES, READ_TOOL_MAX_MATCHES,
|
||
READ_TOOL_MAX_FILE_SIZE,
|
||
TERMINAL_SANDBOX_MOUNT_PATH,
|
||
TERMINAL_SANDBOX_MODE,
|
||
TERMINAL_SANDBOX_CPUS,
|
||
TERMINAL_SANDBOX_MEMORY,
|
||
PROJECT_MAX_STORAGE_MB,
|
||
CUSTOM_TOOLS_ENABLED,
|
||
)
|
||
except ImportError:
|
||
import sys
|
||
project_root = Path(__file__).resolve().parents[2]
|
||
if str(project_root) not in sys.path:
|
||
sys.path.insert(0, str(project_root))
|
||
from config import (
|
||
OUTPUT_FORMATS, DATA_DIR, PROMPTS_DIR, NEED_CONFIRMATION,
|
||
MAX_TERMINALS, TERMINAL_BUFFER_SIZE, TERMINAL_DISPLAY_SIZE,
|
||
MAX_READ_FILE_CHARS, READ_TOOL_DEFAULT_MAX_CHARS,
|
||
READ_TOOL_DEFAULT_CONTEXT_BEFORE, READ_TOOL_DEFAULT_CONTEXT_AFTER,
|
||
READ_TOOL_MAX_CONTEXT_BEFORE, READ_TOOL_MAX_CONTEXT_AFTER,
|
||
READ_TOOL_DEFAULT_MAX_MATCHES, READ_TOOL_MAX_MATCHES,
|
||
READ_TOOL_MAX_FILE_SIZE,
|
||
TERMINAL_SANDBOX_MOUNT_PATH,
|
||
TERMINAL_SANDBOX_MODE,
|
||
TERMINAL_SANDBOX_CPUS,
|
||
TERMINAL_SANDBOX_MEMORY,
|
||
PROJECT_MAX_STORAGE_MB,
|
||
CUSTOM_TOOLS_ENABLED,
|
||
)
|
||
|
||
from modules.file_manager import FileManager
|
||
from modules.search_engine import SearchEngine
|
||
from modules.terminal_ops import TerminalOperator
|
||
from modules.memory_manager import MemoryManager
|
||
from modules.terminal_manager import TerminalManager
|
||
from modules.todo_manager import TodoManager
|
||
from modules.sub_agent import SubAgentManager
|
||
from modules.webpage_extractor import extract_webpage_content, tavily_extract
|
||
from modules.multi_agent.prompts import build_available_agents_prompt
|
||
from modules.ocr_client import OCRClient
|
||
from modules.easter_egg_manager import EasterEggManager
|
||
from modules.personalization_manager import (
|
||
load_personalization_config,
|
||
build_personalization_prompt,
|
||
RECENT_CONVERSATIONS_PROMPT_LIMIT_MIN,
|
||
RECENT_CONVERSATIONS_PROMPT_LIMIT_MAX,
|
||
RECENT_CONVERSATIONS_PROMPT_LIMIT_DEFAULT,
|
||
)
|
||
from modules.skills_manager import (
|
||
get_skills_catalog,
|
||
build_skills_list,
|
||
merge_enabled_skills,
|
||
build_skills_prompt,
|
||
infer_private_skills_dir,
|
||
)
|
||
from modules.custom_tool_registry import CustomToolRegistry, build_default_tool_category
|
||
from modules.custom_tool_executor import CustomToolExecutor
|
||
|
||
|
||
from modules.container_monitor import collect_stats, inspect_state
|
||
from core.tool_config import TOOL_CATEGORIES
|
||
from utils.api_client import APIClient
|
||
from utils.context_manager import ContextManager, AUTO_SHALLOW_PLACEHOLDER
|
||
from utils.host_workspace_debug import write_host_workspace_debug
|
||
from utils.tool_result_formatter import format_tool_result_for_context
|
||
from utils.logger import setup_logger
|
||
from config.model_profiles import (
|
||
get_model_profile,
|
||
get_model_prompt_replacements,
|
||
get_model_context_window,
|
||
model_supports_image,
|
||
model_supports_video,
|
||
)
|
||
|
||
logger = setup_logger(__name__)
|
||
DISABLE_LENGTH_CHECK = True
|
||
|
||
|
||
class MessagesMixin:
|
||
"""MainTerminalContextMixin messages 能力 mixin。"""
|
||
|
||
def build_messages(self, context: Dict, user_input: str) -> List[Dict]:
|
||
"""构建消息列表（添加终端内容注入）"""
|
||
try:
|
||
file_tree_preview = (context.get("project_info", {}).get("file_tree") or "").splitlines()
|
||
write_host_workspace_debug(
|
||
"main_terminal.build_messages.context_snapshot",
|
||
terminal_id=id(self),
|
||
terminal_project_path=str(getattr(self, "project_path", "")),
|
||
context_project_path=str(getattr(getattr(self, "context_manager", None), "project_path", "")),
|
||
project_info_path=str(context.get("project_info", {}).get("path", "")),
|
||
file_tree_first_line=file_tree_preview[0] if file_tree_preview else "",
|
||
current_conversation_id=getattr(getattr(self, "context_manager", None), "current_conversation_id", None),
|
||
)
|
||
except Exception:
|
||
pass
|
||
# 根据当前模型多模态能力选择系统提示。system prompt 按对话冻结；
|
||
# 工具 schema 保持动态，由 define_tools 每轮根据真实状态生成。
|
||
current_model = getattr(self, "model_key", None)
|
||
prompt_name = "main_system_vl" if (model_supports_image(current_model) or model_supports_video(current_model)) else "main_system"
|
||
model_key = getattr(self, "model_key", None)
|
||
prompt_replacements = get_model_prompt_replacements(model_key)
|
||
|
||
is_multi_agent_mode = bool(getattr(self, "multi_agent_mode", False))
|
||
|
||
def _build_main_system_prompt() -> str:
|
||
if is_multi_agent_mode:
|
||
# 多智能体模式下主 prompt 使用 Team Leader 专属模板
|
||
try:
|
||
from modules.multi_agent.prompts import _load_template
|
||
return _load_template("master")
|
||
except Exception as exc:
|
||
logger.warning(f"[messages] 加载多智能体主 prompt 失败: {exc}")
|
||
system_prompt_template = self.load_prompt(prompt_name)
|
||
# main_system.txt / main_system_vl.txt 仅使用 {model_description}
|
||
base_prompt = system_prompt_template.format(
|
||
model_description=prompt_replacements.get("model_description", "")
|
||
)
|
||
# 修改留痕附言：开关开启且 conversation_id 可用时追加独立 prompt 段落
|
||
# （含本对话专属留痕目录；冻结在 prompt 中，与该对话终身一致）
|
||
try:
|
||
from modules.modify_history import build_modify_history_prompt_note
|
||
note = build_modify_history_prompt_note(
|
||
project_path=getattr(self, "project_path", None),
|
||
data_dir=getattr(self, "data_dir", None),
|
||
conversation_id=getattr(getattr(self, "context_manager", None), "current_conversation_id", None),
|
||
)
|
||
if note:
|
||
base_prompt = f"{base_prompt}\n\n{note}"
|
||
except Exception as exc:
|
||
logger.warning(f"[messages] 构建修改留痕 prompt 失败: {exc}")
|
||
return base_prompt
|
||
|
||
main_system_frozen_key = (
|
||
"frozen_main_system_prompt_multi_agent"
|
||
if is_multi_agent_mode
|
||
else "frozen_main_system_prompt"
|
||
)
|
||
system_prompt = self._get_or_init_frozen_prompt(
|
||
main_system_frozen_key,
|
||
_build_main_system_prompt,
|
||
)
|
||
|
||
messages = [
|
||
{"role": "system", "content": system_prompt}
|
||
]
|
||
|
||
personalization_config = getattr(self.context_manager, "custom_personalization_config", None) or load_personalization_config(self.data_dir)
|
||
shallow_replace_enabled = bool(personalization_config.get("auto_shallow_compress_enabled", False)) if isinstance(personalization_config, dict) else False
|
||
|
||
# 顺序：主prompt → 权限模式 → 执行环境 → 最近对话 → 个性化配置 → 工作区信息 → AGENTS.md → skills → 记忆 → 自定义 → 禁用提示
|
||
|
||
# 权限模式
|
||
permission_mode_message = self._get_or_init_frozen_mode_prompt(
|
||
"frozen_permission_prompt",
|
||
self._build_permission_mode_message,
|
||
)
|
||
if permission_mode_message:
|
||
messages.append({"role": "system", "content": permission_mode_message})
|
||
|
||
# 执行环境
|
||
execution_mode_message = self._get_or_init_frozen_mode_prompt(
|
||
"frozen_execution_prompt",
|
||
self._build_execution_mode_message,
|
||
)
|
||
if execution_mode_message:
|
||
messages.append({"role": "system", "content": execution_mode_message})
|
||
|
||
# 最近对话
|
||
def _build_recent_conversations_prompt() -> str:
|
||
recent_conversations_enabled = (
|
||
bool(personalization_config.get("recent_conversations_prompt_enabled", False))
|
||
if isinstance(personalization_config, dict)
|
||
else False
|
||
)
|
||
if not recent_conversations_enabled:
|
||
return ""
|
||
try:
|
||
recent_limit = int(
|
||
personalization_config.get(
|
||
"recent_conversations_prompt_limit",
|
||
RECENT_CONVERSATIONS_PROMPT_LIMIT_DEFAULT,
|
||
)
|
||
)
|
||
except Exception:
|
||
recent_limit = RECENT_CONVERSATIONS_PROMPT_LIMIT_DEFAULT
|
||
recent_limit = max(
|
||
RECENT_CONVERSATIONS_PROMPT_LIMIT_MIN,
|
||
min(RECENT_CONVERSATIONS_PROMPT_LIMIT_MAX, recent_limit),
|
||
)
|
||
return self._build_recent_conversations_message(limit=recent_limit) or ""
|
||
|
||
recent_conversations_prompt = self._get_or_init_frozen_prompt(
|
||
"frozen_recent_conversations_prompt",
|
||
_build_recent_conversations_prompt,
|
||
)
|
||
if recent_conversations_prompt:
|
||
messages.append({"role": "system", "content": recent_conversations_prompt})
|
||
|
||
# 个性化配置
|
||
def _build_personalization_system_prompt() -> str:
|
||
personalization_block = build_personalization_prompt(personalization_config, include_header=False)
|
||
if not personalization_block:
|
||
return ""
|
||
personalization_template = self.load_prompt("personalization").strip()
|
||
if personalization_template and "{personalization_block}" in personalization_template:
|
||
return personalization_template.format(personalization_block=personalization_block)
|
||
if personalization_template:
|
||
return f"{personalization_template}\n{personalization_block}"
|
||
return personalization_block
|
||
|
||
personalization_text = self._get_or_init_frozen_prompt(
|
||
"frozen_personalization_prompt",
|
||
_build_personalization_system_prompt,
|
||
)
|
||
if personalization_text:
|
||
messages.append({"role": "system", "content": personalization_text})
|
||
|
||
# 工作区信息
|
||
workspace_system = self._get_or_init_frozen_prompt(
|
||
"frozen_workspace_prompt",
|
||
lambda: self.context_manager._build_workspace_system_message(context) or "",
|
||
)
|
||
if workspace_system:
|
||
messages.append({"role": "system", "content": workspace_system})
|
||
|
||
# AGENTS.md
|
||
def _build_agents_md_prompt() -> str:
|
||
agents_md_inject_enabled = bool(personalization_config.get("agents_md_auto_inject", False)) if isinstance(personalization_config, dict) else False
|
||
if not agents_md_inject_enabled:
|
||
return ""
|
||
agents_md_content = self._load_agents_md_content()
|
||
if not agents_md_content:
|
||
return ""
|
||
agents_md_template = self.load_prompt("agents_md_inject").strip()
|
||
if agents_md_template and "{{AGENTS_MD_CONTENT}}" in agents_md_template:
|
||
return agents_md_template.replace("{{AGENTS_MD_CONTENT}}", agents_md_content)
|
||
return f"【AGENTS.md 项目规范】\n\n{agents_md_content}\n\n---\n请注意：以上规范来自工作区根目录的 AGENTS.md 文件，若有冲突请以 AGENTS.md 为准。"
|
||
|
||
# AGENTS.md 注入开关可能变化，只在开启时生成并缓存，避免关闭时把空字符串冻住
|
||
agents_md_inject_enabled = (
|
||
bool(personalization_config.get("agents_md_auto_inject", False))
|
||
if isinstance(personalization_config, dict)
|
||
else False
|
||
)
|
||
if agents_md_inject_enabled:
|
||
agents_md_text = self._get_or_init_frozen_prompt(
|
||
"frozen_agents_md_prompt",
|
||
_build_agents_md_prompt,
|
||
)
|
||
if agents_md_text:
|
||
messages.append({"role": "system", "content": agents_md_text})
|
||
|
||
# skills 列表
|
||
skills_catalog = get_skills_catalog(private_dir=infer_private_skills_dir(self.data_dir))
|
||
enabled_skills = merge_enabled_skills(
|
||
personalization_config.get("enabled_skills") if isinstance(personalization_config, dict) else None,
|
||
skills_catalog,
|
||
personalization_config.get("skills_catalog_snapshot") if isinstance(personalization_config, dict) else None,
|
||
)
|
||
def _build_skills_system_prompt() -> str:
|
||
skills_template = self.load_prompt("skills_system").strip()
|
||
skills_list = build_skills_list(skills_catalog, enabled_skills)
|
||
return build_skills_prompt(skills_template, skills_list) or ""
|
||
|
||
skills_prompt = self._get_or_init_frozen_prompt(
|
||
"frozen_skills_prompt",
|
||
_build_skills_system_prompt,
|
||
)
|
||
if skills_prompt:
|
||
messages.append({"role": "system", "content": skills_prompt})
|
||
|
||
# 记忆系统（总体长期记忆 + 项目记忆）
|
||
def _build_memory_system_prompt() -> str:
|
||
return self._build_memory_system_content()
|
||
|
||
memory_prompt = self._get_or_init_frozen_prompt(
|
||
"frozen_memory_prompt",
|
||
_build_memory_system_prompt,
|
||
)
|
||
if memory_prompt:
|
||
messages.append({"role": "system", "content": memory_prompt})
|
||
|
||
# API 自定义 system prompt
|
||
custom_system_prompt = self._get_or_init_frozen_prompt(
|
||
"frozen_custom_system_prompt",
|
||
lambda: (
|
||
getattr(self.context_manager, "custom_system_prompt", "").strip()
|
||
if isinstance(getattr(self.context_manager, "custom_system_prompt", None), str)
|
||
else ""
|
||
),
|
||
)
|
||
if custom_system_prompt:
|
||
messages.append({"role": "system", "content": custom_system_prompt})
|
||
|
||
# 禁用工具提示
|
||
disabled_notice = self._get_or_init_frozen_prompt(
|
||
"frozen_disabled_tools_prompt",
|
||
lambda: self._format_disabled_tool_notice() or "",
|
||
)
|
||
|
||
# 多智能体模式额外提示词：可用的子智能体角色（动态 prompt，第一个用户消息后冻结）
|
||
if getattr(self, "multi_agent_mode", False):
|
||
try:
|
||
available_agents_prompt = self._get_or_init_frozen_prompt(
|
||
"frozen_available_agents_prompt",
|
||
lambda: build_available_agents_prompt() or "",
|
||
)
|
||
if available_agents_prompt:
|
||
messages.append({"role": "system", "content": available_agents_prompt})
|
||
except Exception as exc:
|
||
logger.warning(f"[messages] 注入可用子智能体 prompt 失败: {exc}")
|
||
if disabled_notice:
|
||
messages.append({
|
||
"role": "system",
|
||
"content": disabled_notice
|
||
})
|
||
|
||
# 添加对话历史（保留完整结构，包括tool_calls和tool消息）
|
||
conversation = context["conversation"]
|
||
replaced_tool_count = 0
|
||
deep_compacted_skipped = 0
|
||
orphan_tool_skipped = 0
|
||
# 已随 assistant 消息发出的 tool_call id 集合：用于剥离孤儿 tool 消息。
|
||
# 防御线：历史数据一旦被并发写入/截断损坏（assistant.tool_calls 与其 tool
|
||
# 响应错位），孤儿 tool 消息会让 API 直接 400（tool_call_id is not found），
|
||
# 对话永久不可用。这里保证即使数据损坏，请求也始终合法（内容降级但不崩）。
|
||
emitted_tool_call_ids = set()
|
||
for idx, conv in enumerate(conversation):
|
||
metadata = conv.get("metadata") or {}
|
||
# 深压缩：被标记为 deep_compacted 的消息（整段已压缩前缀）原文保留在历史中用于展示，
|
||
# 但构建请求时整体跳过，不纳入上下文。由于按连续前缀整体标记，
|
||
# 不会出现 assistant.tool_calls 与其 tool 响应被拆散导致的配对悬空。
|
||
if metadata.get("deep_compacted"):
|
||
deep_compacted_skipped += 1
|
||
continue
|
||
if conv["role"] == "assistant":
|
||
# Assistant消息可能包含工具调用
|
||
message = {
|
||
"role": conv["role"],
|
||
"content": conv["content"]
|
||
}
|
||
# 对于思考模式（如 DeepSeek thinking），assistant 历史消息中的
|
||
# reasoning_content 即使为空字符串也需要原样回传，否则下一轮可能被
|
||
# API 判定为“未回传 reasoning_content”并返回 400。
|
||
if "reasoning_content" in conv:
|
||
message["reasoning_content"] = conv.get("reasoning_content", "")
|
||
# 如果有工具调用信息，添加到消息中
|
||
tool_calls = conv.get("tool_calls") or []
|
||
if tool_calls and self._tool_calls_followed_by_tools(conversation, idx, tool_calls):
|
||
message["tool_calls"] = tool_calls
|
||
for _tc in tool_calls:
|
||
_tcid = (_tc or {}).get("id")
|
||
if _tcid:
|
||
emitted_tool_call_ids.add(_tcid)
|
||
messages.append(message)
|
||
|
||
elif conv["role"] == "tool":
|
||
# 孤儿 tool 消息（父 assistant 被跳过/未携带 tool_calls/数据损坏）
|
||
# 直接剥离，避免 API 400；正常情况下不会发生。
|
||
_tool_call_id = conv.get("tool_call_id") or ""
|
||
if not _tool_call_id or _tool_call_id not in emitted_tool_call_ids:
|
||
orphan_tool_skipped += 1
|
||
continue
|
||
if shallow_replace_enabled and metadata.get("auto_shallow_compacted"):
|
||
messages.append({
|
||
"role": "tool",
|
||
"content": AUTO_SHALLOW_PLACEHOLDER,
|
||
"tool_call_id": conv.get("tool_call_id", ""),
|
||
"name": conv.get("name", "")
|
||
})
|
||
replaced_tool_count += 1
|
||
continue
|
||
# Tool消息需要保留完整结构
|
||
images = conv.get("images") or metadata.get("images") or []
|
||
videos = conv.get("videos") or metadata.get("videos") or []
|
||
media_refs = conv.get("media_refs") or metadata.get("media_refs") or []
|
||
content_value = conv.get("content")
|
||
if isinstance(content_value, list):
|
||
content_payload = content_value
|
||
elif images or videos or media_refs:
|
||
content_payload = self.context_manager._build_content_with_images(
|
||
content_value,
|
||
images,
|
||
videos,
|
||
media_refs=media_refs,
|
||
)
|
||
else:
|
||
content_payload = content_value
|
||
message = {
|
||
"role": "tool",
|
||
"content": content_payload,
|
||
"tool_call_id": conv.get("tool_call_id", ""),
|
||
"name": conv.get("name", "")
|
||
}
|
||
messages.append(message)
|
||
|
||
else:
|
||
# User 或普通 System 消息
|
||
images = conv.get("images") or metadata.get("images") or []
|
||
videos = conv.get("videos") or metadata.get("videos") or []
|
||
media_refs = conv.get("media_refs") or metadata.get("media_refs") or []
|
||
content_payload = (
|
||
self.context_manager._build_content_with_images(
|
||
conv["content"],
|
||
images,
|
||
videos,
|
||
media_refs=media_refs,
|
||
)
|
||
if (images or videos or media_refs) else conv["content"]
|
||
)
|
||
# 调试：记录所有 system 消息
|
||
if conv["role"] == "system":
|
||
logger.info(f"[DEBUG build_messages] 添加 system 消息: content前50字={conv['content'][:50]}")
|
||
# 调试：记录多智能体子智能体输出消息是否进入上下文
|
||
if metadata.get("multi_agent_message"):
|
||
logger.info(
|
||
f"[DEBUG build_messages] 添加多智能体子智能体消息到上下文: "
|
||
f"role={conv['role']}, subtype={metadata.get('multi_agent_subtype')}, "
|
||
f"display_name={metadata.get('multi_agent_display_name')}, "
|
||
f"content前80字={str(conv.get('content', ''))[:80]}"
|
||
)
|
||
messages.append({
|
||
"role": conv["role"],
|
||
"content": content_payload
|
||
})
|
||
|
||
# 当前用户输入已经在conversation中了，不需要重复添加
|
||
if shallow_replace_enabled:
|
||
print(f"[ContextCompression] build_messages 替换tool占位符: {replaced_tool_count} 条")
|
||
if deep_compacted_skipped:
|
||
print(f"[ContextCompression] build_messages 跳过已深压缩消息: {deep_compacted_skipped} 条")
|
||
if orphan_tool_skipped:
|
||
logger.warning(
|
||
f"[messages] build_messages 剥离孤儿 tool 消息 {orphan_tool_skipped} 条 "
|
||
f"（对话历史疑似损坏，建议检查 tool_call 配对）"
|
||
)
|
||
return messages
|