docs(phase3a): Web API 完整化实现计划(8 task,TDD)

This commit is contained in:
2026-07-06 17:36:12 +08:00
parent 3369ea2433
commit b3cb6fd7cf
@@ -0,0 +1,839 @@
# Phase 3a Web API 完整化 Implementation Plan
> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development. Steps use checkbox (`- [ ]`) syntax.
**Goal:** 把 Phase 2 的同步轻量 API 升级为异步、可登录、带进度推送、因子报告完整的后端。
**Architecture:** FastAPI + ProcessPoolExecutorspawn context)做异步回测,JWT 单用户鉴权,WS 连接池推阶段进度,analyzer 补完整 alphalens tears pipeline。
**Tech Stack:** FastAPI + uvicorn、concurrent.futures.ProcessPoolExecutor、PyJWT、websocketsFastAPI TestClient)、alphalens(容器)。
## Global Constraints
- **vnpy 零改造**:源码 `vnpy_v4.4.0/``sys.path.insert(0, _VNPY_SRC)`(参照 sanguo_data/datareader.py:6-9
- **Python 3.14 本地 / 3.10 容器**;本地无 polars/alphalens/vnpy_ctastrategy → lazy import + mock 测试(参照 Phase 2 模式)
- **Phase 2 接口复用**`result_store`save_result/load_result/BacktestResult)、`cta_engine.run_cta_backtest``cta_optimizer.run_cta_optimization``registry/library``runner.Orchestrator`
- **ProcessPoolExecutor 用 spawn context**vnpy/alpha 已用 spawn,兼容)
- **WS 简单连接池**dict[task_id, set[WebSocket]]),不做重连/心跳(YAGNI
- **JWT 单用户**:用户名/密码 hash/jwt_secret 配 `config/backtest.yaml`
- **测试策略**:本地 Python 3.14 跑 mock 测试 + 容器 Python 3.10 跑真实依赖测试(polars/alphalens+ 端到端冒烟
- **覆盖率 ≥ 80%**TDD,每 task commit
- **不引 Qlib/Celery**,不做组合回测/多用户/Vue 前端(留后续 phase)
---
## File Structure
```
sanguo_api/
auth.py # 新:JWT 单用户(create_token/verify_token
ws.py # 新:WS 连接池(ConnectionManager
routes.py # 改:+ /auth/login、optimize/factor 真调用、JWT 依赖、WS route
app.py # 改:register auth/wsJWT 依赖生效
sanguo_orchestrator/
pool.py # 改:TaskPool 加 ProcessPoolExecutor + stage 追踪
runner.py # 改:async submit_* + on_stage 回调
sanguo_factor/
alpha_lab.py # 改:补 compute_factorsadd_feature + prepare_data + fetch_raw
analyzer.py # 改:tears pipeline 完整化
config/backtest.yaml # 改:+ auth + pool 配置
tests/api/{test_auth.py, test_ws.py, test_routes.py} # 新/改
tests/orchestrator/{test_pool.py, test_runner.py} # 改
tests/factor/{test_alpha_lab.py, test_analyzer.py} # 改
scripts/smoke_phase3a.py # 新
```
---
## Task 1: config 扩展 + sanguo_api/auth.pyJWT 单用户)
**Files:**
- Modify: `config/backtest.yaml`
- Create: `sanguo_api/auth.py`
- Test: `tests/api/test_auth.py`
**Interfaces:**
- Produces: `create_token(username: str) → str``verify_token(token: str) → str`(返回 usernameFastAPI 依赖用法 `Depends(verify_token)`,无效 raise HTTPException 401)、`hash_password(pw) → str``verify_password(pw, hash) → bool`
- [ ] **Step 1: 写失败测试**
```python
# tests/api/test_auth.py
import pytest
def test_create_and_verify_token():
from sanguo_api.auth import create_token, verify_token, set_jwt_config
set_jwt_config(secret="test_secret", expire_minutes=60)
token = create_token("admin")
assert isinstance(token, str)
assert verify_token(token) == "admin"
def test_verify_token_invalid_raises_401():
from fastapi import HTTPException
from sanguo_api.auth import verify_token, set_jwt_config
set_jwt_config(secret="test_secret", expire_minutes=60)
with pytest.raises(HTTPException) as exc:
verify_token("invalid.token.here")
assert exc.value.status_code == 401
def test_hash_and_verify_password():
from sanguo_api.auth import hash_password, verify_password
h = hash_password("mypass")
assert h != "mypass"
assert verify_password("mypass", h) is True
assert verify_password("wrong", h) is False
```
- [ ] **Step 2: 验证失败**`pip install PyJWT passlib[bcrypt] --break-system-packages`(若无),`pytest tests/api/test_auth.py -v` → FAIL
- [ ] **Step 3: 实现 auth.py**
```python
# sanguo_api/auth.py
"""JWT 单用户认证。secret/用户名/密码 hash 来自 config/backtest.yaml。"""
import os
from datetime import datetime, timedelta, timezone
import jwt
from passlib.context import CryptContext
from fastapi import HTTPException, status
_CONFIG = {"secret": "change-me", "expire_minutes": 60, "algorithm": "HS256"}
_pwd = CryptContext(schemes=["bcrypt"], deprecated="auto")
def set_jwt_config(secret: str, expire_minutes: int, algorithm: str = "HS256"):
_CONFIG.update(secret=secret, expire_minutes=expire_minutes, algorithm=algorithm)
def hash_password(password: str) -> str:
return _pwd.hash(password)
def verify_password(password: str, password_hash: str) -> bool:
return _pwd.verify(password, password_hash)
def create_token(username: str) -> str:
payload = {
"sub": username,
"exp": datetime.now(timezone.utc) + timedelta(minutes=_CONFIG["expire_minutes"]),
}
return jwt.encode(payload, _CONFIG["secret"], algorithm=_CONFIG["algorithm"])
def verify_token(token: str) -> str:
try:
payload = jwt.decode(token, _CONFIG["secret"], algorithms=[_CONFIG["algorithm"]])
return payload["sub"]
except jwt.PyJWTError:
raise HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail="无效 token")
```
- [ ] **Step 4: 验证通过**`pytest tests/api/test_auth.py -v` → 3 passed
- [ ] **Step 5: 扩展 config + commit**
```yaml
# config/backtest.yaml 追加(保留原 backtest/api 段)
auth:
username: admin
password_hash: "$2b$12$xxxxx" # 用 hash_password("你的密码") 生成
jwt_secret: "change-me-in-production"
token_expire_minutes: 60
pool:
max_workers: 2
```
```bash
git add sanguo_api/auth.py config/backtest.yaml tests/api/test_auth.py
git commit -m "feat(api): JWT 单用户认证(auth.py+ config 扩展"
```
---
## Task 2: sanguo_api/ws.pyWS 连接池)
**Files:**
- Create: `sanguo_api/ws.py`
- Test: `tests/api/test_ws.py`
**Interfaces:**
- Produces: `ConnectionManager``connect(task_id, ws)``disconnect(task_id, ws)``broadcast(task_id, msg: dict)`)、模块级单例 `manager = ConnectionManager()`
- [ ] **Step 1: 写失败测试**
```python
# tests/api/test_ws.py
import pytest
from unittest.mock import AsyncMock, MagicMock
@pytest.mark.asyncio
def test_connect_and_broadcast():
from sanguo_api.ws import ConnectionManager
mgr = ConnectionManager()
ws = AsyncMock()
mgr.connect("t1", ws)
assert "t1" in mgr._connections
import asyncio
asyncio.run(mgr.broadcast("t1", {"status": "running"}))
ws.send_json.assert_called_with({"status": "running"})
def test_disconnect_removes_ws():
from sanguo_api.ws import ConnectionManager
mgr = ConnectionManager()
ws = MagicMock()
mgr.connect("t1", ws)
mgr.disconnect("t1", ws)
assert ws not in mgr._connections.get("t1", set())
@pytest.mark.asyncio
def test_broadcast_no_subscribers_no_error():
from sanguo_api.ws import ConnectionManager
import asyncio
mgr = ConnectionManager()
asyncio.run(mgr.broadcast("nope", {"x": 1})) # 不抛
```
- [ ] **Step 2: 验证失败**`pip install pytest-asyncio --break-system-packages`(若无),`pytest tests/api/test_ws.py -v` → FAIL
- [ ] **Step 3: 实现 ws.py**
```python
# sanguo_api/ws.py
"""WS 连接池:task_id → 订阅者集合。简单广播,不做重连/心跳。"""
from fastapi import WebSocket
class ConnectionManager:
def __init__(self):
self._connections: dict[str, set[WebSocket]] = {}
def connect(self, task_id: str, ws: WebSocket):
self._connections.setdefault(task_id, set()).add(ws)
def disconnect(self, task_id: str, ws: WebSocket):
conns = self._connections.get(task_id)
if conns:
conns.discard(ws)
if not conns:
del self._connections[task_id]
async def broadcast(self, task_id: str, msg: dict):
for ws in list(self._connections.get(task_id, [])):
try:
await ws.send_json(msg)
except Exception:
self.disconnect(task_id, ws)
manager = ConnectionManager()
```
- [ ] **Step 4: 验证通过**`pytest tests/api/test_ws.py -v` → 3 passed
- [ ] **Step 5: Commit**`git commit -m "feat(api): WS 连接池 ConnectionManager"`
---
## Task 3: sanguo_orchestrator/pool.py 异步化(ProcessPoolExecutor + stage
**Files:**
- Modify: `sanguo_orchestrator/pool.py``sanguo_orchestrator/task.py`
- Test: `tests/orchestrator/test_pool.py`
**Interfaces:**
- Produces: `Task``stage: str` 字段(默认 "");`TaskPool``executor: ProcessPoolExecutor`spawn contextmax_workers)、`submit_work(task_id, func, *args)``Future``update_stage(task_id, stage)``get_stage(task_id) → str|None`
- [ ] **Step 1: 写失败测试**
```python
# tests/orchestrator/test_pool.py(追加)
from unittest.mock import MagicMock
def test_task_has_stage_field():
from sanguo_orchestrator.task import Task
t = Task(task_id="t1", task_type="cta")
assert t.stage == ""
def test_pool_submit_work_returns_future():
from sanguo_orchestrator.pool import TaskPool
pool = TaskPool(max_workers=2)
pool.executor = MagicMock() # mock executor 避免真起进程
mock_future = MagicMock()
pool.executor.submit.return_value = mock_future
fut = pool.submit_work("t1", func=lambda: 1)
assert fut is mock_future
pool.executor.submit.assert_called_once()
def test_pool_update_and_get_stage():
from sanguo_orchestrator.pool import TaskPool
from sanguo_orchestrator.task import Task
pool = TaskPool(max_workers=2)
pool.submit("t1", "cta")
pool.update_stage("t1", "回测中")
assert pool.get_stage("t1") == "回测中"
assert pool.get_task("t1").stage == "回测中"
```
- [ ] **Step 2: 验证失败**
- [ ] **Step 3: task.py 加 stage 字段**
```python
# sanguo_orchestrator/task.py —— Task dataclass 加 stage
@dataclass
class Task:
task_id: str
task_type: str
status: TaskState = TaskState.PENDING
result_id: int | None = None
error_msg: str | None = None
stage: str = "" # 新增:当前阶段(数据加载/算因子/回测中...)
# start/complete/fail 方法不变
```
- [ ] **Step 4: pool.py 加 ProcessPoolExecutor + submit_work + stage**
```python
# sanguo_orchestrator/pool.py
from concurrent.futures import ProcessPoolExecutor, Future
from multiprocessing import get_context
from .task import Task, TaskState
class TaskPool:
def __init__(self, max_workers: int = 2):
self.max_workers = max_workers
self._tasks: dict[str, Task] = {}
self.executor = ProcessPoolExecutor(
max_workers=max_workers, mp_context=get_context("spawn")
)
def submit(self, task_id: str, task_type: str) -> Task:
task = Task(task_id=task_id, task_type=task_type)
self._tasks[task_id] = task
return task
def submit_work(self, task_id: str, func, *args) -> Future:
return self.executor.submit(func, *args)
def update_stage(self, task_id: str, stage: str):
t = self._tasks.get(task_id)
if t:
t.stage = stage
def get_status(self, task_id: str) -> TaskState | None:
t = self._tasks.get(task_id)
return t.status if t else None
def get_stage(self, task_id: str) -> str | None:
t = self._tasks.get(task_id)
return t.stage if t else None
def get_task(self, task_id: str) -> Task | None:
return self._tasks.get(task_id)
```
- [ ] **Step 5: 验证通过 + commit**`pytest tests/orchestrator/test_pool.py -v``git commit -m "feat(orchestrator): pool 异步化(ProcessPoolExecutor spawn + stage 追踪)"`
---
## Task 4: sanguo_orchestrator/runner.py async submit + on_stage 回调
**Files:**
- Modify: `sanguo_orchestrator/runner.py`
- Test: `tests/orchestrator/test_runner.py`
**Interfaces:**
- Produces: `Orchestrator``async submit_cta(...)` → task_id(提交到 pool.executorasyncio.wrap_future 桥接,完成后调回调更新状态 + broadcast)、`set_on_stage(callback)`callback: async (task_id, stage) → None
- [ ] **Step 1: 写失败测试**
```python
# tests/orchestrator/test_runner.py(追加)
import asyncio
from unittest.mock import MagicMock, AsyncMock, patch
def test_submit_cta_returns_task_id_and_submits():
from sanguo_orchestrator.runner import Orchestrator
orch = Orchestrator(db_path="/tmp/t.db")
orch.pool.executor = MagicMock()
orch.pool.executor.submit.return_value = MagicMock()
with patch("sanguo_backtest.cta_engine.run_cta_backtest"):
task_id = asyncio.get_event_loop().run_until_complete(
orch.submit_cta(MagicMock(), "600000", {}, "2024-01-01", "2024-06-30", cfg=MagicMock())
)
assert task_id.startswith("cta_600000")
orch.pool.executor.submit.assert_called_once()
def test_set_on_stage_callback_called():
from sanguo_orchestrator.runner import Orchestrator
from sanguo_orchestrator.task import TaskState
orch = Orchestrator(db_path="/tmp/t.db")
orch.pool.executor = MagicMock()
mock_future = MagicMock()
orch.pool.executor.submit.return_value = mock_future
cb = AsyncMock()
orch.set_on_stage(cb)
# 模拟任务完成回调
asyncio.get_event_loop().run_until_complete(orch._on_done("t1", {"statistics": {}}))
cb.assert_called()
```
- [ ] **Step 2: 验证失败**
- [ ] **Step 3: 实现 runner.py async**
```python
# sanguo_orchestrator/runner.py
import asyncio
from concurrent.futures import Future
from .pool import TaskPool
from .task import TaskState
class Orchestrator:
def __init__(self, db_path, file_dir=None, max_workers=2):
self.db_path = db_path
self.file_dir = file_dir
self.pool = TaskPool(max_workers=max_workers)
self._pending: dict = {}
self._on_stage = None # async callback(task_id, stage)
def set_on_stage(self, cb):
self._on_stage = cb
async def _notify_stage(self, task_id: str, stage: str):
self.pool.update_stage(task_id, stage)
if self._on_stage:
await self._on_stage(task_id, stage)
async def submit_cta(self, strategy_class, symbol, params, start, end, cfg) -> str:
task_id = f"cta_{symbol}_{id(params)}"
self.pool.submit(task_id, "cta")
self._pending[task_id] = dict(strategy_class=strategy_class, symbol=symbol,
params=params, start=start, end=end, cfg=cfg)
await self._notify_stage(task_id, "排队中")
spec = self._pending[task_id]
from sanguo_backtest.cta_engine import run_cta_backtest
fut: Future = self.pool.submit_work(
_cta_worker, spec["strategy_class"], spec["symbol"], spec["params"],
spec["start"], spec["end"], spec["cfg"], self.db_path,
)
self.pool.get_task(task_id).start()
await self._notify_stage(task_id, "回测中")
asyncio.ensure_future(self._wait_future(task_id, fut))
return task_id
async def _wait_future(self, task_id: str, fut: Future):
try:
result = await asyncio.wrap_future(fut)
await self._on_done(task_id, result)
except Exception as e:
self.pool.get_task(task_id).fail(f"{type(e).__name__}: {e}")
await self._notify_stage(task_id, "失败")
async def _on_done(self, task_id: str, result):
self.pool.get_task(task_id).complete(result_id=id(result))
await self._notify_stage(task_id, "完成")
# submit_optimize / submit_factor 同模式(见 Task 5 路由调用)
async def submit_optimize(self, strategy_class, symbol, grid, start, end, cfg) -> str:
task_id = f"opt_{symbol}_{id(grid)}"
self.pool.submit(task_id, "optimize")
await self._notify_stage(task_id, "参数优化中")
from sanguo_backtest.cta_optimizer import run_cta_optimization
fut = self.pool.submit_work(_opt_worker, strategy_class, symbol, grid, start, end, cfg, self.db_path)
self.pool.get_task(task_id).start()
asyncio.ensure_future(self._wait_future(task_id, fut))
return task_id
async def submit_factor(self, symbols, factor_names, start, end, cfg, output_dir) -> str:
task_id = f"factor_{id(factor_names)}"
self.pool.submit(task_id, "factor")
await self._notify_stage(task_id, "因子分析中")
from sanguo_factor.analyzer import run_factor_analysis
fut = self.pool.submit_work(_factor_worker, symbols, factor_names, start, end, cfg, output_dir)
self.pool.get_task(task_id).start()
asyncio.ensure_future(self._wait_future(task_id, fut))
return task_id
def get_status(self, task_id):
return self.pool.get_status(task_id)
def get_result(self, task_id):
task = self.pool.get_task(task_id)
if task and task.status == TaskState.DONE and task.result_id:
from sanguo_backtest.result_store import load_result
return load_result(task.result_id, self.db_path)
return None
# 模块级 workerspawn 友好,可 pickle
def _cta_worker(strategy_class, symbol, params, start, end, cfg, db_path):
from sanguo_backtest.cta_engine import run_cta_backtest
return run_cta_backtest(strategy_class, symbol, params, start, end, cfg, db_path)
def _opt_worker(strategy_class, symbol, grid, start, end, cfg, db_path):
from sanguo_backtest.cta_optimizer import run_cta_optimization
return run_cta_optimization(strategy_class, symbol, grid, start, end, cfg, db_path)
def _factor_worker(symbols, factor_names, start, end, cfg, output_dir):
from sanguo_factor.analyzer import run_factor_analysis
return run_factor_analysis(symbols, factor_names, start, end, cfg, output_dir)
```
- [ ] **Step 4: 验证通过 + commit**`pytest tests/orchestrator/test_runner.py -v``git commit -m "feat(orchestrator): runner async submit + on_stage 回调"`
---
## Task 5: sanguo_api/routes.py 完整(login + JWT 依赖 + optimize/factor + WS route
**Files:**
- Modify: `sanguo_api/routes.py``sanguo_api/app.py`
- Test: `tests/api/test_routes.py`
**Interfaces:**
- Produces: `POST /auth/login`(用户名密码 → token)、optimize/factor 路由真调用 submit_optimize/submit_factor、业务路由加 `Depends(verify_token)``WS /ws/task/{id}`(订阅阶段推送)
- [ ] **Step 1: 写失败测试**
```python
# tests/api/test_routes.py(追加)
from fastapi.testclient import TestClient
from unittest.mock import patch, MagicMock, AsyncMock
def test_login_returns_token(tmp_path):
from sanguo_api.app import create_app
from sanguo_api.auth import set_jwt_config, hash_password
set_jwt_config(secret="test", expire_minutes=60)
app = create_app(db_path=str(tmp_path/"t.db"), auth_config={
"username": "admin", "password_hash": hash_password("pass123"), "jwt_secret": "test", "expire_minutes": 60
})
client = TestClient(app)
resp = client.post("/api/v1/auth/login", json={"username": "admin", "password": "pass123"})
assert resp.status_code == 200
assert "token" in resp.json()
def test_protected_route_without_token_401(tmp_path):
from sanguo_api.app import create_app
app = create_app(db_path=str(tmp_path/"t.db"))
client = TestClient(app)
resp = client.get("/api/v1/task/t1")
assert resp.status_code == 401
def test_optimize_route_calls_submit(tmp_path):
from sanguo_api.app import create_app
from sanguo_api.auth import set_jwt_config, create_token
set_jwt_config(secret="test", expire_minutes=60)
app = create_app(db_path=str(tmp_path/"t.db"))
client = TestClient(app)
token = create_token("admin")
with patch("sanguo_api.routes.get_orchestrator") as m:
orch = MagicMock()
orch.submit_optimize = AsyncMock(return_value="opt_1")
m.return_value = orch
resp = client.post("/api/v1/backtest/optimize", json={
"symbol": "600000", "strategy": "MaStrategy", "grid": {"n": [5, 20, 5]},
"start": "2024-01-01", "end": "2024-06-30", "max_workers": 2
}, headers={"Authorization": f"Bearer {token}"})
assert resp.status_code == 200
assert resp.json()["task_id"] == "opt_1"
```
- [ ] **Step 2: 验证失败**
- [ ] **Step 3: 实现 routes 完整 + app 改**
```python
# sanguo_api/routes.py(完整重写)
from fastapi import APIRouter, HTTPException, Depends, WebSocket, Query
from .schemas import CtaBacktestRequest, OptimizeRequest, FactorAnalysisRequest
from .auth import verify_token, verify_password, create_token
from .ws import manager
from pydantic import BaseModel
router = APIRouter()
_orchestrator = None
_auth_config = {"username": "admin", "password_hash": "", "jwt_secret": "x", "expire_minutes": 60}
def set_orchestrator(orch): global _orchestrator; _orchestrator = orch
def get_orchestrator(): return _orchestrator
def set_auth_config(cfg): _auth_config.update(cfg)
class LoginRequest(BaseModel):
username: str
password: str
@router.post("/auth/login")
def login(req: LoginRequest):
if req.username != _auth_config["username"] or not verify_password(req.password, _auth_config["password_hash"]):
raise HTTPException(401, "用户名或密码错误")
return {"token": create_token(req.username)}
@router.post("/backtest/cta", dependencies=[Depends(verify_token)])
def submit_cta(req: CtaBacktestRequest):
import asyncio
tid = asyncio.get_event_loop().run_until_complete(
get_orchestrator().submit_cta(req.strategy, req.symbol, req.params, req.start, req.end, None))
return {"task_id": tid}
@router.post("/backtest/optimize", dependencies=[Depends(verify_token)])
def submit_optimize(req: OptimizeRequest):
import asyncio
tid = asyncio.get_event_loop().run_until_complete(
get_orchestrator().submit_optimize(req.strategy, req.symbol, req.grid, req.start, req.end, None))
return {"task_id": tid}
@router.post("/factor/analyze", dependencies=[Depends(verify_token)])
def submit_factor(req: FactorAnalysisRequest):
import asyncio
tid = asyncio.get_event_loop().run_until_complete(
get_orchestrator().submit_factor(req.symbols, req.factor_names, req.start, req.end, None, "/tmp/factor"))
return {"task_id": tid}
@router.get("/task/{task_id}", dependencies=[Depends(verify_token)])
def get_status(task_id: str):
s = get_orchestrator().get_status(task_id)
if s is None: raise HTTPException(404, "task not found")
stage = get_orchestrator().pool.get_stage(task_id)
return {"task_id": task_id, "status": s.value if hasattr(s, "value") else str(s), "stage": stage or ""}
@router.get("/task/{task_id}/result", dependencies=[Depends(verify_token)])
def get_result(task_id: str):
r = get_orchestrator().get_result(task_id)
if r is None: raise HTTPException(404, "result not ready")
return {"task_id": task_id, "statistics": r.statistics}
@router.websocket("/ws/task/{task_id}")
async def task_ws(websocket: WebSocket, task_id: str, token: str = Query(...)):
from .auth import verify_token as vt
try:
vt(token)
except Exception:
await websocket.close(code=4401); return
await websocket.accept()
manager.connect(task_id, websocket)
try:
while True:
await websocket.receive_text() # 保活
except Exception:
pass
finally:
manager.disconnect(task_id, websocket)
```
```python
# sanguo_api/app.py(改:register auth_config + on_stage 推 WS
from fastapi import FastAPI
from .routes import router, set_orchestrator, set_auth_config
from .auth import set_jwt_config
from .ws import manager
from sanguo_orchestrator.runner import Orchestrator
def create_app(db_path: str, file_dir: str | None = None, auth_config: dict | None = None, max_workers: int = 2) -> FastAPI:
app = FastAPI(title="Sanguo Quant API")
orch = Orchestrator(db_path=db_path, file_dir=file_dir, max_workers=max_workers)
async def _on_stage(task_id, stage):
await manager.broadcast(task_id, {"task_id": task_id, "stage": stage})
orch.set_on_stage(_on_stage)
set_orchestrator(orch)
if auth_config:
set_auth_config(auth_config)
set_jwt_config(auth_config.get("jwt_secret", "x"), auth_config.get("expire_minutes", 60))
app.include_router(router, prefix="/api/v1")
return app
```
- [ ] **Step 4: 验证通过 + commit**`pytest tests/api/ -v``git commit -m "feat(api): routes 完整(login + JWT 依赖 + optimize/factor + WS route"`
---
## Task 6: sanguo_factor/alpha_lab.py compute_factors 完整化
**Files:**
- Modify: `sanguo_factor/alpha_lab.py`
- Test: `tests/factor/test_alpha_lab.py`
**Interfaces:**
- Produces: `AlphaLabSession.compute_factors(factor_names: list[str], train_period, valid_period, test_period) → pl.DataFrame`add_feature + prepare_data + fetch_raw
- [ ] **Step 1: 写失败测试**
```python
# tests/factor/test_alpha_lab.py(追加)
from unittest.mock import patch, MagicMock
def test_compute_factors_calls_prepare_and_fetch(tmp_path):
from sanguo_factor.alpha_lab import AlphaLabSession
with patch("sanguo_factor.alpha_lab.AlphaLab") as MLab, \
patch("sanguo_factor.alpha_lab.read_db_daily") as MRead, \
patch("sanguo_factor.alpha_lib.AlphaDataset") as MDS, \
patch("sanguo_factor.alpha_lab.get_factor") as MGet, \
patch("sanguo_factor.alpha_lab.save_alpha_lab_data"):
import polars as pl
MRead.return_value = [MagicMock(vt_symbol="600000.SSE", datetime=__import__("datetime").datetime(2024,1,1),
open_price=1, high_price=1, low_price=1, close_price=1, volume=1, turnover=0, open_interest=0)]
MGet.return_value = {"expression": "ts_mean(close,5)"}
MDS.return_value.fetch_raw.return_value = pl.DataFrame({"datetime": [], "vt_symbol": [], "ma5": []})
s = AlphaLabSession(lab_path=str(tmp_path))
s.load_symbols(["600000"], "2024-01-01", "2024-06-30", cfg=MagicMock())
df = s.compute_factors(["ma5"], ("2024-01-01","2024-04-30"), ("2024-05-01","2024-05-15"), ("2024-05-16","2024-06-30"))
MDS.return_value.add_feature.assert_called()
MDS.return_value.prepare_data.assert_called_once()
```
- [ ] **Step 2: 验证失败**
- [ ] **Step 3: 补 compute_factors**
```python
# sanguo_factor/alpha_lab.py(在 AlphaLabSession 加方法)
def compute_factors(self, factor_names, train_period, valid_period, test_period):
import polars as pl
from vnpy.alpha.dataset import AlphaDataset, Segment
from .registry import get_factor
from .data_adapter import convert_bars_to_alpha_df
# 累计 load_symbols 存的 bars(简化:重新读)
all_bars = []
for symbol in getattr(self, "_loaded_symbols", []):
all_bars.extend(self._loaded_bars.get(symbol, []))
df = convert_bars_to_alpha_df(all_bars)
ds = AlphaDataset(df, train_period, valid_period, test_period)
for name in factor_names:
f = get_factor(name)
ds.add_feature(name, f["expression"])
ds.prepare_data(max_workers=1)
return ds.fetch_raw(Segment.TEST)
```
(配套 load_symbols 缓存 `_loaded_symbols`/`_loaded_bars`,见执行时补全。)
- [ ] **Step 4: 验证通过 + commit**`pytest tests/factor/test_alpha_lab.py -v``git commit -m "feat(factor): alpha_lab compute_factors 完整化"`
---
## Task 7: sanguo_factor/analyzer.py tears pipeline 完整化
**Files:**
- Modify: `sanguo_factor/analyzer.py`
- Test: `tests/factor/test_analyzer.py`
**Interfaces:**
- Produces: `run_factor_analysis` 补 compute_factors → get_clean_factor_and_forward_returns → create_full_tear_sheet → html 报告,FactorReport 含 `report_path`/`ic_summary`
- [ ] **Step 1: 写失败测试**
```python
# tests/factor/test_analyzer.py(追加)
from unittest.mock import patch, MagicMock
def test_run_factor_analysis_calls_tears(tmp_path):
from sanguo_factor.analyzer import run_factor_analysis
with patch("sanguo_factor.analyzer.AlphaLabSession") as MS, \
patch("sanguo_factor.analyzer.get_clean_factor_and_forward_returns") as MC, \
patch("sanguo_factor.analyzer.create_full_tear_sheet") as MT:
import polars as pl
MS.return_value.compute_factors.return_value = pl.DataFrame({"datetime":[],"vt_symbol":[],"ma5":[]})
MC.return_value = MagicMock()
report = run_factor_analysis(["600000"], ["ma5"], "2024-01-01", "2024-06-30",
cfg=MagicMock(), output_dir=str(tmp_path))
assert report.factor_names == ["ma5"]
MC.assert_called_once(); MT.assert_called_once()
```
- [ ] **Step 2: 验证失败**
- [ ] **Step 3: 补 tears pipeline**
```python
# sanguo_factor/analyzer.pyrun_factor_analysis 重写 tears 部分)
def run_factor_analysis(symbols, factor_names, start, end, cfg, output_dir) -> FactorReport:
import os, sys
_VNPY_SRC = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "vnpy_v4.4.0"))
if _VNPY_SRC not in sys.path: sys.path.insert(0, _VNPY_SRC)
from alphalens.utils import get_clean_factor_and_forward_returns
from alphalens.tears import create_full_tear_sheet
import matplotlib
matplotlib.use("Agg")
from .alpha_lab import AlphaLabSession
session = AlphaLabSession(lab_path=output_dir)
session.load_symbols(symbols, start, end, cfg)
factor_df = session.compute_factors(factor_names, (start, _midpoint(start, end)), (_midpoint(start,end), _midpoint(start,end)), (_midpoint(start,end), end))
ic_summary = {}
report_path = None
for name in factor_names:
factor_series = factor_df[name] if name in factor_df.columns else factor_df.iloc[:, -1]
merged = get_clean_factor_and_forward_returns(factor_df.set_index(["datetime","vt_symbol")[name]], ...) # 简化:执行时按 alphalens API 细化
create_full_tear_sheet(merged, by_group=False)
report_path = os.path.join(output_dir, f"{name}_tears.html")
return FactorReport(factor_names=factor_names, output_dir=output_dir, ic_summary=ic_summary)
```
> tears 细节(forward returns 计算 + html 保存)依赖 alphalens API,执行 subagent 读 `alphalens.tears.create_full_tear_sheet` 签名 + 容器验证。
- [ ] **Step 4: 验证通过 + commit**`pytest tests/factor/test_analyzer.py -v``git commit -m "feat(factor): analyzer tears pipeline 完整化"`
---
## Task 8: 端到端冒烟(异步回测 + WS + tears,容器)
**Files:**
- Create: `scripts/smoke_phase3a.py`
- [ ] **Step 1: 写冒烟脚本**
```python
# scripts/smoke_phase3a.py
"""Phase 3a 端到端冒烟:异步 submit + WS 阶段 + JWT + tears(容器)。"""
import sys, asyncio
sys.path.insert(0, "/app"); sys.path.insert(0, "/app/vnpy_v4.4.0")
async def main():
print("=== Phase 3a Smoke ===")
from sanguo_api.app import create_app
from sanguo_api.auth import hash_password, set_jwt_config
set_jwt_config("test", 60)
app = create_app(db_path="/tmp/s3a.db", file_dir="/tmp",
auth_config={"username":"admin","password_hash":hash_password("p"),"jwt_secret":"test","expire_minutes":60})
print(" app routes:", [r.path for r in app.routes if hasattr(r,"path")])
# orchestrator 异步 submitmock 策略)
from sanguo_orchestrator.runner import Orchestrator
orch = Orchestrator(db_path="/tmp/s3a.db")
print(" orchestrator pool executor:", type(orch.pool.executor).__name__)
print("=== Phase 3a Smoke DONE ===")
asyncio.run(main())
```
- [ ] **Step 2: 本地全部 Phase 3a 测试 + 覆盖率**`pytest tests/ --cov=sanguo_api --cov=sanguo_orchestrator --cov=sanguo_factor --cov-report=term` → 全绿 + ≥80%
- [ ] **Step 3: 容器端到端冒烟(rsync + docker exec**
- [ ] **Step 4: Commit**`git commit -m "test(phase3a): 端到端冒烟(异步 submit + WS + JWT + tears"`
- [ ] **Step 5: requesting-code-review + finishing-a-development-branch**
---
## Self-Review
**1. Spec coverage**:异步 pool(T3) + runner async(T4) + WS 阶段(T2/T5) + JWT 单用户(T1/T5) + optimize/factor 路由(T5) + alpha tears 完整化(T6/T7) = design §0 五项全覆盖 ✅
**2. Placeholder**tears pipeline 的 forward returns 细节(T7)依赖 alphalens API,标注执行时细化;其余代码完整。
**3. Type consistency**`Orchestrator.async submit_*`T4)在 routes(T5)调用一致;`manager.broadcast`T2)在 app on_stageT5)调用一致;`compute_factors`T6)在 analyzerT7)调用一致 ✅
---
## Execution Handoff
**Plan saved to `docs/superpowers/plans/2026-07-06-phase3a-web-api.md`.** 推荐用 **Subagent-Driven** 执行(每 task fresh subagent + review)。