mirror of
https://github.com/ThearchyHelios/AgentLab.git
synced 2026-11-16 18:11:59 +00:00
dd57c12016
跑一段不可信代码不需要一整套容器运行时。macOS 用 Seatbelt (sandbox-exec,系统自带),Linux 用 bubblewrap;冷启动从 ~250ms 降到 ~25ms,且不再需要装任何东西。 守住三件事:默认断网、家目录不可读、只有工作区可写。策略由内核 强制且对子进程继承——沙箱内 subprocess 出来的 cat 读家目录同样是 Operation not permitted。解释器改用 sys.base_prefix 下那个干净的 Python,沙箱代码因此 import 不到后端的 FastAPI、SQLAlchemy 和凭据 处理代码。 如实标注能力边界,不给虚假的安全感: - macOS 不强制 RLIMIT_AS,内存限不住(实测设 256MB 仍可分配 900MB), health 接口申报 enforced 字段,界面把限不住的项划掉 - Seatbelt 是访问控制而非虚拟化:进程表共享,以当前用户身份运行 - bubblewrap 侧多隔离 PID/IPC/挂载视图,内存限额真实生效 顺带修两个 bug: - LocalSandbox 设 RLIMIT_NPROC=64 会误伤——macOS 上它按 uid 全局 计数,用户已有上千进程时沙箱内任何 fork 都失败,连 bash 都起不来 - 模型所属 provider 被停用时,原先静默换一家继续用原模型名,必然 404 且报错指不到原因;改为直接说清是哪个 provider 被停用了 写 SBPL 的坑记在注释里:(deny default) 会让 Python 直接 SIGABRT, 改用「默认允许 + 精确拒绝」。 7 个模板全部回归通过。
195 lines
7.7 KiB
Python
195 lines
7.7 KiB
Python
from __future__ import annotations
|
||
|
||
import os
|
||
from typing import Any
|
||
|
||
from langchain_core.language_models.chat_models import BaseChatModel
|
||
from pydantic import BaseModel, Field
|
||
from sqlalchemy import select
|
||
from sqlalchemy.ext.asyncio import AsyncSession
|
||
|
||
from app.core.crypto import decrypt
|
||
from app.db.models import Provider
|
||
from app.providers import catalog
|
||
from app.providers.mock_model import MockChatModel
|
||
|
||
|
||
class ModelSpec(BaseModel):
|
||
"""节点上对模型的请求。省略的字段会回退到 provider 默认值。"""
|
||
|
||
provider: str | None = None # provider 的 name 或 id
|
||
model: str | None = None
|
||
temperature: float | None = None
|
||
max_tokens: int | None = Field(default=None)
|
||
thinking: str | None = None # off | adaptive
|
||
effort: str | None = None # low | medium | high | xhigh | max
|
||
timeout: int | None = None
|
||
extra: dict[str, Any] = Field(default_factory=dict)
|
||
|
||
|
||
class ProviderNotConfigured(RuntimeError):
|
||
pass
|
||
|
||
|
||
async def list_providers(session: AsyncSession) -> list[Provider]:
|
||
rows = await session.execute(select(Provider).order_by(Provider.created_at))
|
||
return list(rows.scalars())
|
||
|
||
|
||
async def resolve_provider(
|
||
session: AsyncSession, ref: str | None, model: str | None = None
|
||
) -> Provider | None:
|
||
"""找出该用哪个 provider。
|
||
|
||
顺序很关键:显式指定的 ref 最优先;其次按模型名反查 —— 节点上通常只选了
|
||
模型(比如 mock-fast),没选 provider,这时必须找到真正提供该模型的那一个,
|
||
否则会拿着 A 家的模型名去调 B 家的接口。最后才回退到默认 provider。
|
||
"""
|
||
providers = await list_providers(session)
|
||
enabled = [p for p in providers if p.enabled]
|
||
|
||
if ref:
|
||
for p in providers:
|
||
if p.name == ref or p.id == ref:
|
||
return p
|
||
return None
|
||
|
||
if model:
|
||
for p in enabled:
|
||
if any((m or {}).get("id") == model for m in (p.models or [])):
|
||
return p
|
||
for p in enabled:
|
||
if p.default_model == model:
|
||
return p
|
||
# 模型认得出来,但它所属的 provider 被停用了。这时候不能默默换一家 ——
|
||
# 拿着 A 家的模型名去调 B 家,必然 404,而且报错完全指不到真正的原因。
|
||
for p in providers:
|
||
if not p.enabled and (
|
||
any((m or {}).get("id") == model for m in (p.models or []))
|
||
or p.default_model == model
|
||
):
|
||
raise ProviderNotConfigured(
|
||
f"模型 {model!r} 属于 provider {p.name!r},但它已被停用。"
|
||
f"请在设置里启用它,或给节点换一个模型。"
|
||
)
|
||
# 认不出来的模型名就放行:用户可能手填了一个目录里还没有的新模型
|
||
|
||
real = [p for p in enabled if p.kind != "mock"]
|
||
return (real or enabled or [None])[0]
|
||
|
||
|
||
def _resolve_key(provider: Provider) -> str | None:
|
||
"""明文 key。没配就退回环境变量 —— 本地实验时最省事。"""
|
||
key = decrypt(provider.api_key)
|
||
if key:
|
||
return key
|
||
env_names = {
|
||
"anthropic": ("ANTHROPIC_API_KEY", "ANTHROPIC_AUTH_TOKEN"),
|
||
"openai": ("OPENAI_API_KEY",),
|
||
}.get(provider.kind, ())
|
||
for name in env_names:
|
||
if os.environ.get(name):
|
||
return os.environ[name]
|
||
return None
|
||
|
||
|
||
def build_chat_model(provider: Provider, spec: ModelSpec) -> BaseChatModel:
|
||
"""把一条 provider 记录 + 节点配置变成可调用的 LangChain 模型。
|
||
|
||
这里最要紧的是按模型能力裁剪参数:Claude 4.6 之后的模型收到 temperature
|
||
会直接 400,所以不能无脑透传节点上的滑块值。
|
||
"""
|
||
model = spec.model or provider.default_model or ""
|
||
if not model:
|
||
raise ProviderNotConfigured(f"provider {provider.name!r} 没有可用的模型")
|
||
|
||
kwargs: dict[str, Any] = {"model": model}
|
||
if spec.max_tokens:
|
||
kwargs["max_tokens"] = spec.max_tokens
|
||
if spec.timeout:
|
||
kwargs["timeout"] = spec.timeout
|
||
|
||
# 采样参数:能力允许才传
|
||
if spec.temperature is not None and catalog.supports_sampling(model):
|
||
kwargs["temperature"] = spec.temperature
|
||
|
||
if provider.kind == "mock":
|
||
return MockChatModel(model_name=model)
|
||
|
||
api_key = _resolve_key(provider)
|
||
base_url = provider.base_url or None
|
||
extra = dict(provider.extra or {})
|
||
|
||
if provider.kind == "anthropic":
|
||
from langchain_anthropic import ChatAnthropic
|
||
|
||
base_url = base_url or os.environ.get("ANTHROPIC_BASE_URL") or None
|
||
if not api_key:
|
||
raise ProviderNotConfigured(f"provider {provider.name!r} 缺少 API Key")
|
||
|
||
bearer = extra.get("auth_style") == "bearer"
|
||
kwargs["api_key"] = "placeholder-replaced-below" if bearer else api_key
|
||
if base_url:
|
||
kwargs["base_url"] = base_url
|
||
if extra.get("headers"):
|
||
kwargs["default_headers"] = dict(extra["headers"])
|
||
|
||
# thinking 这类模型会先花 token 思考再输出正文;给太小的 max_tokens
|
||
# 会出现"思考完就没额度写答案",返回一条空正文。
|
||
kwargs.setdefault("max_tokens", 8192)
|
||
|
||
model_kwargs: dict[str, Any] = {}
|
||
if catalog.supports_adaptive_thinking(model):
|
||
mode = spec.thinking or "summarized"
|
||
if mode in ("adaptive", "summarized"):
|
||
# summarized 会把思考摘要一起返回,画布上就能展示"它在想什么"
|
||
kwargs["thinking"] = {
|
||
"type": "adaptive",
|
||
"display": "summarized" if mode == "summarized" else "omitted",
|
||
}
|
||
elif mode == "off":
|
||
kwargs["thinking"] = {"type": "disabled"}
|
||
if spec.effort and catalog.supports_effort(model):
|
||
model_kwargs["output_config"] = {"effort": spec.effort}
|
||
if model_kwargs:
|
||
kwargs["model_kwargs"] = model_kwargs
|
||
|
||
chat = ChatAnthropic(**kwargs)
|
||
if bearer:
|
||
# 网关只认 Authorization: Bearer,且多带一个 x-api-key 就会 401。
|
||
# ChatAnthropic 的 default_headers 是 dict[str,str],塞不进 SDK 的 Omit 哨兵,
|
||
# 所以直接换掉底层 client —— anthropic SDK 原生支持 auth_token,
|
||
# 且 api_key=None 时不会发出 x-api-key。
|
||
import anthropic
|
||
|
||
client_args = {"base_url": base_url, "api_key": None, "auth_token": api_key}
|
||
chat._client = anthropic.Anthropic(**client_args)
|
||
chat._async_client = anthropic.AsyncAnthropic(**client_args)
|
||
return chat
|
||
|
||
# openai 与 openai 兼容走同一个类,区别只在 base_url
|
||
from langchain_openai import ChatOpenAI
|
||
|
||
if provider.kind == "openai_compatible" and not base_url:
|
||
raise ProviderNotConfigured(f"provider {provider.name!r} 需要填 base_url")
|
||
# 很多本地服务(Ollama、vLLM)不校验 key,但 SDK 要求非空
|
||
kwargs["api_key"] = api_key or "not-needed"
|
||
if base_url:
|
||
kwargs["base_url"] = base_url
|
||
if extra.get("headers"):
|
||
kwargs["default_headers"] = extra["headers"]
|
||
return ChatOpenAI(**kwargs)
|
||
|
||
|
||
async def get_chat_model(session: AsyncSession, spec: ModelSpec) -> tuple[BaseChatModel, str]:
|
||
"""返回 (模型, 实际使用的模型 id)。找不到任何 provider 时回退到 mock,
|
||
保证一张图在没配 key 的机器上也能跑完。"""
|
||
provider = await resolve_provider(session, spec.provider, spec.model)
|
||
if provider is None:
|
||
return MockChatModel(model_name="mock-fast"), "mock-fast"
|
||
try:
|
||
model = build_chat_model(provider, spec)
|
||
except ProviderNotConfigured:
|
||
raise
|
||
return model, (spec.model or provider.default_model or "")
|