M1 最小闭环:CLI + FastAPI 网关 + SQL 闸 + 100% 覆盖率验收链

- CLI:init/serve/update/query/export/status(typer + rich),单写者锁三元判活
- 存储:DuckDB schema migrations(14 表 + 8 联邦视图,语句元组逐条执行)
- ETL:A股证券/日历/日K,整区替换幂等;日历全年补全(calendar 语义唯一)
- 网关:POST /v1/query(线程池 + interrupt 超时 504)、/v1/export、meta/stats/state、healthz/ready
- 安全:SQL 闸 sqlglot fail-closed(文件/网络表函数默认拒)、非 loopback 空 token 拒启、导出路径双判
- 验收:254 测试全绿、1265 语句 100% 覆盖、[CI] 真起服验收链、smoke_m1.py 交付自检
- 文档:设计文档 v0.7(圆桌 45 条采纳 + 3 条实施校准),README 重写
- 发布:release.yml Trusted Publishing(OIDC 免密钥),action refs 已核验
This commit is contained in:
openbot committed 2026-09-24 09:06:47 +08:00
commit d37c697bbe
59 files changed
+7586

No files matched your search

+83
View File
@@ -0,0 +1,83 @@
"""[CI] 验收链(设计 §14/§15):init --skip-history + 迷你 seed → serve → POST /v1/query 查日K → export csv。
一条链必绿,全部离线。
"""
from __future__ import annotations
import base64
import pyarrow as pa
from fastapi.testclient import TestClient
from typer.testing import CliRunner
from pyquantdata.cli import app
from pyquantdata.config import default_config
from pyquantdata.serve.app import AppContext, create_app
from pyquantdata.store.db import configure_query_conn
runner = CliRunner()
def test_m1_ci_acceptance_chain(catalog, dbpath, mini_seed, tmp_path):
# 1. init --skip-history + seed(CLI 入口,建目录/config/schema/seed)
result = runner.invoke(app, [
"init", "-d", str(dbpath), "--skip-history", "--seed", str(mini_seed),
])
assert result.exit_code == 0, result.output
# 2. serve 语境(TestClient 同事件循环;真网 uvicorn 按 §14 白名单每周冒烟)
cursor = catalog.cursor()
configure_query_conn(cursor)
ctx = AppContext(dbpath=dbpath, config=default_config(), cursor=cursor, schema_version=1)
with TestClient(create_app(ctx)) as client:
# 3. POST /v1/query 查到日K(参数化 + JSON)
r = client.post("/v1/query", json={
"sql": "SELECT date, close FROM cn_stock_1d WHERE symbol = ? ORDER BY date",
"params": ["sh600000"],
})
assert r.status_code == 200
body = r.json()
assert body["row_count"] == 10
assert body["rows"][0][1] == 10.2 # 9/1 close
# 3b. fmt=arrow → Arrow IPC streaming → DataFrame 生态解码路径(附录 A)
r_arrow = client.post("/v1/query", json={
"sql": "SELECT date, close FROM cn_stock_1d ORDER BY date", "fmt": "arrow",
})
tbl = pa.ipc.open_stream(
pa.BufferReader(base64.b64decode(r_arrow.json()["b64_arrow"]))
).read_all()
df = tbl.to_pandas()
assert len(df) == 10 and float(df["close"].iloc[-1]) == 11.1
# 4. export csv(文件流回传 + exports/{job_id}/ 落盘 + job 记录)
r_exp = client.post("/v1/export", json={
"sql": "SELECT symbol, date, open, close FROM cn_stock_1d ORDER BY date",
"fmt": "csv", "filename": "mini_daily.csv",
})
assert r_exp.status_code == 200
lines = r_exp.content.decode("utf-8").strip().splitlines()
header = [c.strip('"') for c in lines[0].split(",")]
assert header == ["symbol", "date", "open", "close"] and len(lines) == 11
job_row = client.post(
"/v1/query", json={"sql": "SELECT job_id, rows, state FROM export_jobs"}
).json()
assert job_row["rows"][0][1] == 10 and job_row["rows"][0][2] == "done"
# 6. SQL 闸在链路内生效
blocked = client.post("/v1/query", json={"sql": "SELECT read_parquet('x')"})
assert blocked.status_code == 400
# 7. CLI 本地直查(只读连接,serve 之外也可用)
# DuckDB 同进程不允许同一库文件「写连接 + 只读连接」并存(配置不同会直接报错)——
# 先释放本测试持有的写连接,再走 CLI 只读路径。
cursor.close()
catalog.close()
q = runner.invoke(app, [
"query", "-d", str(dbpath), "--fmt", "json",
"SELECT count(*) AS n FROM cn_stock_1d",
])
assert q.exit_code == 0, q.output
assert '"row_count": 1' in q.output