myagent-ai 1.0.0

package/config.py

@@ -0,0 +1,464 @@
+"""
+config.py - 全局配置管理模块
+=============================
+集中管理所有配置项，支持从环境变量、配置文件、默认值三级加载。
+配置文件默认路径: ~/.myagent/config.json
+"""
+from __future__ import annotations
+
+import json
+import os
+import sys
+import platform
+from pathlib import Path
+from dataclasses import dataclass, field, asdict
+from typing import Optional, Dict, Any, List
+
+
+# ==============================================================================
+# 配置数据类
+# ==============================================================================
+
+@dataclass
+class LLMConfig:
+    """LLM 大模型配置"""
+    provider: str = "openai"           # openai | anthropic | ollama | custom
+    api_key: str = ""
+    base_url: str = "https://api.openai.com/v1"
+    model: str = "gpt-4"
+    temperature: float = 0.1
+    max_tokens: int = 4096
+    timeout: int = 120                 # 请求超时(秒)
+    max_retries: int = 3               # 最大重试次数
+    # Anthropic 专用
+    anthropic_api_key: str = ""
+    # Ollama 专用
+    ollama_base_url: str = "http://localhost:11434"
+    ollama_model: str = "llama3"
+
+
+@dataclass
+class MemoryConfig:
+    """记忆系统配置"""
+    db_path: str = ""                  # SQLite 数据库路径，默认 ~/.myagent/memory.db
+    max_short_term: int = 50           # 短期记忆最大轮数
+    max_working: int = 100             # 工作记忆最大条数
+    auto_summarize: bool = True        # 自动总结开关
+    summarize_threshold: int = 20      # 触发总结的对话轮数
+
+
+@dataclass
+class ExecutorConfig:
+    """执行引擎配置"""
+    timeout: int = 300                 # 默认执行超时(秒)
+    max_retries: int = 2               # 自动重试次数
+    auto_fix: bool = True              # 自动修复错误
+    max_output_length: int = 50000     # 输出最大长度
+    execution_mode: str = "local"      # 执行模式: local(本机) | sandbox(沙盒)
+    sandbox_image: str = "python:3.12-slim"  # 沙盒 Docker 镜像
+    sandbox_network: bool = False      # 沙盒是否允许网络
+    sandbox_memory: str = "512m"       # 沙盒内存限制
+    allowed_dirs: List[str] = field(default_factory=list)  # 允许访问的目录(空=全部)
+    blocked_commands: List[str] = field(default_factory=lambda: [
+        "rm -rf /", "format", "del /f /s /q C:\\", "mkfs", "dd if=/dev/zero"
+    ])
+
+
+@dataclass
+class AgentConfig:
+    """Agent 配置"""
+    max_iterations: int = 30           # 单任务最大迭代次数
+    max_parallel: int = 3              # 最大并行任务数
+    verbose: bool = True               # 详细日志
+
+
+@dataclass
+class TrayConfig:
+    """系统托盘配置"""
+    auto_start: bool = False           # 开机自启
+    show_notifications: bool = True    # 显示通知
+    icon_path: str = ""                # 托盘图标路径
+
+
+@dataclass
+class ModelEntry:
+    """模型库中的单个模型定义"""
+    id: str = ""                 # 唯一标识符，如 "gpt-4o", "claude-3.5-sonnet"
+    name: str = ""               # 显示名称
+    provider: str = ""            # openai | anthropic | ollama | zhipu | custom
+    model: str = ""               # API 调用使用的实际模型字符串
+    base_url: str = ""            # 自定义 Base URL（空=使用 provider 默认值）
+    api_key: str = ""             # 专用 API Key（空=使用全局默认值）
+    max_tokens: int = 4096
+    temperature: float = 0.1
+    enabled: bool = True
+
+
+@dataclass
+class ChatPlatformConfig:
+    """单个聊天平台配置"""
+    enabled: bool = False
+    platform: str = ""                 # telegram | discord | feishu | qq | wechat
+    token: str = ""                    # Bot Token
+    app_id: str = ""                   # App ID (某些平台需要)
+    app_secret: str = ""               # App Secret
+    webhook_url: str = ""              # Webhook URL
+    allowed_users: List[str] = field(default_factory=list)  # 允许的用户白名单(空=全部)
+    extra: Dict[str, Any] = field(default_factory=dict)     # 平台特有配置
+
+
+@dataclass
+class OrganizationConfig:
+    """组织设置配置"""
+    enabled: bool = False              # 默认不启用
+    # 组织知识库管理员（agent path），只有指定的agent才能修改组织知识库
+    knowledge_admin: str = ""          # 如 "default" 或 "manager"
+
+
+@dataclass
+class CommunicationConfig:
+    """Agent 间通信配置"""
+    enabled: bool = False              # 默认不启用
+    server_url: str = "wss://aicq.online/ws"  # AICQ 中继服务器
+    agent_id: str = ""                 # 本机 Agent ID (Ed25519 公钥，空=自动生成)
+    private_key: str = ""              # 本机私钥 (Ed25519，空=自动生成)
+    max_friends: int = 200
+    auto_accept: bool = False          # 自动接受好友请求
+
+
+@dataclass
+class AppConfig:
+    """应用总配置"""
+    llm: LLMConfig = field(default_factory=LLMConfig)
+    memory: MemoryConfig = field(default_factory=MemoryConfig)
+    executor: ExecutorConfig = field(default_factory=ExecutorConfig)
+    agent: AgentConfig = field(default_factory=AgentConfig)
+    tray: TrayConfig = field(default_factory=TrayConfig)
+    organization: OrganizationConfig = field(default_factory=OrganizationConfig)
+    communication: CommunicationConfig = field(default_factory=CommunicationConfig)
+    chat_platforms: List[ChatPlatformConfig] = field(default_factory=list)
+    models_library: List[ModelEntry] = field(default_factory=list)
+    log_level: str = "INFO"
+    data_dir: str = ""                 # 数据目录，默认 ~/.myagent/
+    language: str = "zh-CN"
+
+
+# ==============================================================================
+# 配置管理器
+# ==============================================================================
+
+class ConfigManager:
+    """
+    配置管理器 - 三级加载策略
+    优先级: 环境变量 > 配置文件 > 默认值
+    """
+
+    CONFIG_DIR_NAME = ".myagent"
+    CONFIG_FILE_NAME = "config.json"
+
+    def __init__(self):
+        self._config = AppConfig()
+        self._config_dir = Path.home() / self.CONFIG_DIR_NAME
+        self._config_file = self._config_dir / self.CONFIG_FILE_NAME
+        self._data_dir = self._config_dir
+        self._ensure_dirs()
+
+    def _ensure_dirs(self):
+        """确保必要目录存在"""
+        self._config_dir.mkdir(parents=True, exist_ok=True)
+        (self._config_dir / "data").mkdir(exist_ok=True)
+        (self._config_dir / "logs").mkdir(exist_ok=True)
+
+    @property
+    def config(self) -> AppConfig:
+        return self._config
+
+    @property
+    def config_dir(self) -> Path:
+        return self._config_dir
+
+    @property
+    def data_dir(self) -> Path:
+        return self._data_dir / "data"
+
+    @property
+    def logs_dir(self) -> Path:
+        return self._config_dir / "logs"
+
+    def load(self) -> AppConfig:
+        """加载配置(配置文件 + 环境变量覆盖)"""
+        self._load_from_file()
+        self._load_from_env()
+        self._apply_defaults()
+        return self._config
+
+    def save(self) -> None:
+        """保存当前配置到文件"""
+        config_dict = self._to_dict(self._config)
+        with open(self._config_file, "w", encoding="utf-8") as f:
+            json.dump(config_dict, f, ensure_ascii=False, indent=2)
+
+    def _load_from_file(self):
+        """从配置文件加载"""
+        if not self._config_file.exists():
+            self.save()  # 创建默认配置
+            return
+        try:
+            with open(self._config_file, "r", encoding="utf-8") as f:
+                data = json.load(f)
+            self._apply_dict(self._config, data)
+        except (json.JSONDecodeError, IOError) as e:
+            print(f"[CONFIG] 配置文件读取失败，使用默认值: {e}")
+
+    def _load_from_env(self):
+        """从环境变量加载(前缀 MYAGENT_)"""
+        env_mapping = {
+            "MYAGENT_LLM_PROVIDER": ("llm", "provider"),
+            "MYAGENT_LLM_API_KEY": ("llm", "api_key"),
+            "MYAGENT_LLM_BASE_URL": ("llm", "base_url"),
+            "MYAGENT_LLM_MODEL": ("llm", "model"),
+            "MYAGENT_LLM_TEMPERATURE": ("llm", "temperature", float),
+            "MYAGENT_LLM_MAX_TOKENS": ("llm", "max_tokens", int),
+            "MYAGENT_ANTHROPIC_API_KEY": ("llm", "anthropic_api_key"),
+            "MYAGENT_OLLAMA_BASE_URL": ("llm", "ollama_base_url"),
+            "MYAGENT_OLLAMA_MODEL": ("llm", "ollama_model"),
+            "MYAGENT_LOG_LEVEL": ("log_level", None, str),
+            "MYAGENT_LANGUAGE": ("language", None, str),
+        }
+        for env_key, mapping in env_mapping.items():
+            value = os.environ.get(env_key)
+            if value is None:
+                continue
+            if len(mapping) == 3:
+                group, attr, converter = mapping
+            elif len(mapping) == 2:
+                group, attr = mapping
+                converter = str
+            else:
+                continue
+            target = getattr(self._config, group) if group else self._config
+            try:
+                setattr(target, attr, converter(value))
+            except (ValueError, AttributeError):
+                pass
+
+        # 聊天平台 Token 从环境变量加载
+        platform_env = {
+            "MYAGENT_TELEGRAM_TOKEN": "telegram",
+            "MYAGENT_DISCORD_TOKEN": "discord",
+            "MYAGENT_FEISHU_APP_ID": "feishu",
+            "MYAGENT_FEISHU_APP_SECRET": "feishu",
+            "MYAGENT_QQ_TOKEN": "qq",
+            "MYAGENT_WECHAT_TOKEN": "wechat",
+        }
+        for env_key, platform_name in platform_env.items():
+            token = os.environ.get(env_key)
+            if token:
+                self._ensure_chat_platform(platform_name, token)
+
+    def _ensure_chat_platform(self, platform: str, token: str):
+        """确保聊天平台配置存在"""
+        for cp in self._config.chat_platforms:
+            if cp.platform == platform:
+                cp.token = token or cp.token
+                cp.enabled = True
+                return
+        cp = ChatPlatformConfig(platform=platform, token=token, enabled=True)
+        self._config.chat_platforms.append(cp)
+
+    def _apply_defaults(self):
+        """应用平台相关的默认值"""
+        if not self._config.memory.db_path:
+            self._config.memory.db_path = str(self.data_dir / "memory.db")
+        if not self._config.data_dir:
+            self._config.data_dir = str(self._config_dir)
+
+        # 设置 Ollama base_url
+        if self._config.llm.provider == "ollama" and not self._config.llm.base_url:
+            self._config.llm.base_url = self._config.llm.ollama_base_url
+            self._config.llm.model = self._config.llm.ollama_model
+
+    def _to_dict(self, obj) -> dict:
+        """递归 dataclass 转 dict"""
+        if hasattr(obj, '__dataclass_fields__'):
+            return {k: self._to_dict(v) for k, v in asdict(obj).items()}
+        elif isinstance(obj, list):
+            return [self._to_dict(i) for i in obj]
+        return obj
+
+    def _apply_dict(self, target, data: dict):
+        """递归将 dict 应用到 dataclass"""
+        for key, value in data.items():
+            if not hasattr(target, key):
+                continue
+            current = getattr(target, key)
+            if isinstance(value, dict) and hasattr(current, '__dataclass_fields__'):
+                self._apply_dict(current, value)
+            elif isinstance(value, list) and key == "chat_platforms":
+                platforms = []
+                for item in value:
+                    cp = ChatPlatformConfig(**{
+                        k: v for k, v in item.items() if k in ChatPlatformConfig.__dataclass_fields__
+                    })
+                    platforms.append(cp)
+                setattr(target, key, platforms)
+            elif isinstance(value, list) and key == "models_library":
+                models = []
+                for item in value:
+                    me = ModelEntry(**{
+                        k: v for k, v in item.items() if k in ModelEntry.__dataclass_fields__
+                    })
+                    models.append(me)
+                setattr(target, key, models)
+            elif key in getattr(type(target), '__dataclass_fields__', {}):
+                setattr(target, key, value)
+
+    def get_chat_platform(self, platform: str) -> Optional[ChatPlatformConfig]:
+        """获取指定聊天平台配置"""
+        for cp in self._config.chat_platforms:
+            if cp.platform == platform:
+                return cp
+        return None
+
+    def get_enabled_platforms(self) -> List[ChatPlatformConfig]:
+        """获取所有启用的聊天平台"""
+        return [cp for cp in self._config.chat_platforms if cp.enabled]
+
+    # ── 热重载 ──
+    def reload(self) -> AppConfig:
+        """从配置文件重新加载配置(热重载，不重启服务)"""
+        old_config = self._config
+        self._config = AppConfig()
+        self._load_from_file()
+        self._load_from_env()
+        self._apply_defaults()
+        return self._config
+
+    def update_llm(self, **kwargs) -> AppConfig:
+        """更新 LLM 配置并立即生效"""
+        for k, v in kwargs.items():
+            if hasattr(self._config.llm, k):
+                setattr(self._config.llm, k, v)
+        self.save()
+        return self._config
+
+    def update_executor(self, **kwargs) -> AppConfig:
+        """更新执行引擎配置并立即生效"""
+        for k, v in kwargs.items():
+            if hasattr(self._config.executor, k):
+                setattr(self._config.executor, k, v)
+        self.save()
+        return self._config
+
+    # ── 导入 / 导出 ──
+    def export_config(self, include_secrets: bool = False) -> dict:
+        """导出完整配置为字典（含备份元数据）"""
+        cfg = self._to_dict(self._config)
+        if not include_secrets:
+            # 脱敏处理
+            if 'api_key' in cfg.get('llm', {}):
+                key = cfg['llm']['api_key']
+                if key:
+                    cfg['llm']['api_key'] = key[:6] + '****' + key[-4:] if len(key) > 10 else '****'
+            if 'anthropic_api_key' in cfg.get('llm', {}):
+                key = cfg['llm']['anthropic_api_key']
+                if key:
+                    cfg['llm']['anthropic_api_key'] = key[:6] + '****' + key[-4:] if len(key) > 10 else '****'
+            # 脱敏聊天平台 token
+            for p in cfg.get('chat_platforms', []):
+                if p.get('token'):
+                    t = p['token']
+                    p['token'] = t[:6] + '****' + t[-4:] if len(t) > 10 else '****'
+                if p.get('app_secret'):
+                    s = p['app_secret']
+                    p['app_secret'] = s[:6] + '****' + s[-4:] if len(s) > 10 else '****'
+
+        result = {
+            '_meta': {
+                'version': '1.0',
+                'exported_at': __import__('datetime').datetime.now().isoformat(),
+                'hostname': platform.node(),
+                'include_secrets': include_secrets,
+ },
+            'config': cfg,
+        }
+        return result
+
+    def import_config(self, data: dict, overwrite: bool = False) -> dict:
+        """
+        导入配置。
+
+        Args:
+            data: 导入的配置字典（含 _meta 和 config）
+            overwrite: 是否完全覆盖（否则合并）
+
+        Returns:
+            dict: {ok, message, changed_keys}
+        """
+        if isinstance(data, dict) and 'config' in data:
+            cfg_data = data['config']
+        elif isinstance(data, dict):
+            cfg_data = data
+        else:
+            return {'ok': False, 'message': '无效的配置格式'}
+
+        # 验证基本结构
+        if not isinstance(cfg_data, dict):
+            return {'ok': False, 'message': '配置数据必须是字典'}
+
+        changed = []
+        if overwrite:
+            # 完全覆盖
+            self._config = AppConfig()
+            self._apply_dict(self._config, cfg_data)
+            changed = list(cfg_data.keys())
+        else:
+            # 合并模式：只覆盖有意义的字段
+            for section, fields in cfg_data.items():
+                if section.startswith('_'):
+                    continue
+                current = getattr(self._config, section, None)
+                if current is None or not hasattr(current, '__dataclass_fields__'):
+                    continue
+                if isinstance(fields, dict):
+                    for k, v in fields.items():
+                        if hasattr(current, k):
+                            old_val = getattr(current, k)
+                            if old_val != v:
+                                setattr(current, k, v)
+                                changed.append(f"{section}.{k}")
+
+        # 保存并应用默认值
+        self._apply_defaults()
+        self.save()
+        return {
+            'ok': True,
+            'message': f'配置已导入（{len(changed)} 项变更）',
+            'changed_keys': changed,
+        }
+
+    def get_full_config(self) -> dict:
+        """获取完整配置字典（敏感字段脱敏）"""
+        return self.export_config(include_secrets=False)
+
+
+# ==============================================================================
+# 全局配置实例
+# ==============================================================================
+
+_global_config: Optional[ConfigManager] = None
+
+
+def get_config() -> ConfigManager:
+    """获取全局配置管理器实例"""
+    global _global_config
+    if _global_config is None:
+        _global_config = ConfigManager()
+        _global_config.load()
+    return _global_config
+
+
+def reset_config():
+    """重置全局配置(测试用)"""
+    global _global_config
+    _global_config = None

package/core/__init__.py

@@ -0,0 +1,10 @@
+"""
+core/__init__.py - 核心模块初始化
+"""
+from core.logger import setup_logger, get_logger
+from core.utils import (
+    timestamp, generate_id, truncate_str,
+    safe_json_parse, validate_json_schema
+)
+from core.llm import LLMClient, get_llm_client
+from core.task_queue import TaskQueue, TaskItem, TaskStatus