155 lines
6.4 KiB
Python
155 lines
6.4 KiB
Python
"""AI 策略生成器 — 读取策略开发文档 + 调用 LLM 生成策略代码。
|
|
|
|
职责: 接收用户自然语言描述 → 读取 docs/strategy-guide.md → 调用 LLM → 返回策略代码。
|
|
不知道: 引擎内部、API、前端、配置持久化、回测。
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
import ast
|
|
import logging
|
|
import re
|
|
import tempfile
|
|
from pathlib import Path
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
# 策略开发文档路径
|
|
GUIDE_PATH = Path(__file__).resolve().parent.parent.parent.parent / "docs" / "strategy-guide.md"
|
|
|
|
_SYSTEM_PREFIX = """你是A股量化策略设计专家。根据用户描述的需求，参考下方的《策略开发指南》生成一个完整的策略Python文件。
|
|
|
|
核心约束:
|
|
- 只创建这一个 .py 文件，不要修改任何现有文件，不要跨文件引用
|
|
- 只 import polars as pl，不 import 其他模块
|
|
|
|
要求:
|
|
1. 用户可能调整的策略阈值通过 META["params"] 暴露；公式常数、固定窗口边界、布尔开关不必强行参数化
|
|
2. 遵循指南中的文件结构，但优先贴合用户规则，不要为了套模板歪曲策略含义
|
|
3. ENTRY_SIGNALS/EXIT_SIGNALS 根据策略逻辑自行选择匹配的信号列，不要照搬示例
|
|
4. scoring 权重根据策略核心逻辑定制，总和 = 1.0
|
|
5. 优先使用 Polars 表达式、窗口函数、聚合和 with_columns/filter 实现，避免逐行/逐股 Python 循环；只有表达式难以描述的复杂状态机才使用 partition_by/to_dicts
|
|
6. 直接输出Python代码，不要输出其他内容
|
|
|
|
--- 策略开发指南 ---
|
|
|
|
"""
|
|
|
|
|
|
class AIStrategyGenerator:
|
|
"""AI 策略生成器"""
|
|
|
|
def __init__(self) -> None:
|
|
self._guide_cache: str | None = None
|
|
|
|
def _get_guide(self) -> str:
|
|
if self._guide_cache is None:
|
|
if GUIDE_PATH.exists():
|
|
self._guide_cache = GUIDE_PATH.read_text(encoding="utf-8")
|
|
else:
|
|
logger.warning("strategy-guide.md not found at %s", GUIDE_PATH)
|
|
self._guide_cache = ""
|
|
return self._guide_cache
|
|
|
|
async def generate(self, user_prompt: str) -> dict:
|
|
"""根据用户描述生成策略代码
|
|
|
|
Returns: {"code": str, "meta": dict, "valid": bool, "error": str | None}
|
|
"""
|
|
guide = self._get_guide()
|
|
|
|
# 调用 LLM
|
|
code = await self._call_llm(user_prompt, guide)
|
|
|
|
# 验证
|
|
try:
|
|
self._validate_safety(code)
|
|
except ValueError as e:
|
|
return {"code": code, "meta": {}, "valid": False, "error": str(e)}
|
|
|
|
# 试加载获取 META
|
|
try:
|
|
meta = self._extract_meta(code)
|
|
except Exception as e:
|
|
return {"code": code, "meta": {}, "valid": False, "error": f"解析META失败: {e}"}
|
|
|
|
return {"code": code, "meta": meta, "valid": True, "error": None}
|
|
|
|
async def _call_llm(self, user_prompt: str, guide: str) -> str:
|
|
"""调用 OpenAI 兼容 API（流式，避免 CDN 长连接超时）"""
|
|
from openai import AsyncOpenAI
|
|
from app import secrets_store
|
|
|
|
ai_key = secrets_store.get_ai_key()
|
|
if not ai_key:
|
|
raise RuntimeError("AI API Key 未配置，请在设置页面配置")
|
|
|
|
client = AsyncOpenAI(
|
|
api_key=ai_key,
|
|
base_url=secrets_store.get_ai_config("ai_base_url", "https://api.alysc.top"),
|
|
timeout=180.0,
|
|
max_retries=2,
|
|
)
|
|
# 使用流式请求：CDN 收到首个 token 后会持续转发，不会因等待超时
|
|
stream = await client.chat.completions.create(
|
|
model=secrets_store.get_ai_config("ai_model", "gpt-5.5"),
|
|
messages=[
|
|
{"role": "system", "content": _SYSTEM_PREFIX + guide},
|
|
{"role": "user", "content": user_prompt},
|
|
],
|
|
temperature=0.3,
|
|
max_tokens=3000,
|
|
stream=True,
|
|
)
|
|
chunks: list[str] = []
|
|
async for chunk in stream:
|
|
delta = chunk.choices[0].delta if chunk.choices else None
|
|
if delta and delta.content:
|
|
chunks.append(delta.content)
|
|
content = "".join(chunks).strip()
|
|
# 提取代码块
|
|
if "```python" in content:
|
|
content = content.split("```python", 1)[1].split("```", 1)[0].strip()
|
|
elif "```" in content:
|
|
content = content.split("```", 1)[1].split("```", 1)[0].strip()
|
|
return content
|
|
|
|
@staticmethod
|
|
def _validate_safety(code: str) -> None:
|
|
"""AST 级安全检查"""
|
|
tree = ast.parse(code)
|
|
|
|
forbidden_modules = {"os", "sys", "subprocess", "socket", "shutil",
|
|
"pathlib", "http", "urllib", "requests", "httpx"}
|
|
forbidden_calls = {"open", "exec", "eval", "compile", "__import__",
|
|
"globals", "locals", "vars", "dir", "getattr",
|
|
"setattr", "delattr", "type", "input"}
|
|
|
|
for node in ast.walk(tree):
|
|
if isinstance(node, ast.Import):
|
|
for alias in node.names:
|
|
if alias.name.split(".")[0] not in ("polars",):
|
|
if alias.name.split(".")[0] in forbidden_modules:
|
|
raise ValueError(f"禁止 import {alias.name}")
|
|
if isinstance(node, ast.ImportFrom):
|
|
if node.module and node.module.split(".")[0] not in ("polars",):
|
|
if node.module.split(".")[0] in forbidden_modules:
|
|
raise ValueError(f"禁止 from {node.module} import")
|
|
if isinstance(node, ast.Call):
|
|
if isinstance(node.func, ast.Name) and node.func.id in forbidden_calls:
|
|
raise ValueError(f"禁止调用 {node.func.id}()")
|
|
|
|
@staticmethod
|
|
def _extract_meta(code: str) -> dict:
|
|
"""从代码字符串中提取 META 字典（不执行代码）"""
|
|
tree = ast.parse(code)
|
|
for node in ast.walk(tree):
|
|
if isinstance(node, ast.Assign):
|
|
for target in node.targets:
|
|
if isinstance(target, ast.Name) and target.id == "META":
|
|
# 找到 META 赋值，用 compile+eval 安全提取
|
|
# 只允许字面量
|
|
meta_node = node.value
|
|
code_obj = compile(ast.Expression(meta_node), "<meta>", "eval")
|
|
return eval(code_obj, {"__builtins__": {}}) # noqa: S307
|
|
return {}
|