dpb/backend/tests/test_stat_analysis_tc.py

# -*- coding: utf-8 -*-
"""统计引擎正式 tc 套件（ssGBLUP/OCS/MABC/孟德尔 + 补齐 5 套）的 pytest 包装层。

每套 tc = 一个独立脚本（backend/scripts/test_bre_analysis_*_tc.py），走真实 API
（TestClient + 真实 PostgreSQL 16 + Redis），脚本自带 fixture 清理并 sys.exit(0/1)。
本包装层以 subprocess 顺序调用这些脚本，断言 rc==0 —— 与 run_regression.py 的执行契约一致。

关键点：
- 全部基于真实 PG16 + Redis，仅当探测到 PG/Redis 可达才跑，否则 skipif 跳过。
- 不 import 任何 app 模块（测试库生命周期由 conftest 管理，与统计引擎 tc 独立）；subprocess env 显式构建：
  清空可能残留的 DATABASE_*/REDIS_*/POOL_*/CAPTCHA_ENABLE 等连接变量，
  置 ENVIRONMENT=dev / DATABASE_TYPE=postgres / PYTHONPATH=backend / PYTHONUTF8=1，
  再并入从 backend/env/.env.dev 解析的真实连接值（可被 TC_CI_* 覆盖，用于验证 skipif 路径）。
- 标记 pg（pyproject 已注册），`pytest -m pg` 只跑本组；run_ci.py 的验收门即此命令。

运行：cd backend && C:/ai/miniconda3/envs/dpb/python.exe -X utf8 -m pytest -m pg -v
"""
import asyncio
import os
import re
import subprocess
import sys
from pathlib import Path

import pytest

BACKEND = Path(__file__).resolve().parents[1]
SCRIPTS = BACKEND / "scripts"
ENV_FILE = BACKEND / "env" / ".env.dev"

PY = os.environ.get("DPB_PYTHON", sys.executable)

# (套件名, 脚本文件名) —— 10 套：4 套既有 + 5 套补齐 + O2–O4 方法学广度
TC_SCRIPTS = [
    ("ssgblup", "test_bre_analysis_ssgblup_tc.py"),
    ("ocs", "test_bre_analysis_ocs_tc.py"),
    ("mabc", "test_bre_analysis_mabc_tc.py"),
    ("mendelian", "test_bre_analysis_mendelian_tc.py"),
    ("genetic_corr", "test_bre_analysis_genetic_corr_tc.py"),
    ("type_b_heredity", "test_bre_analysis_type_b_tc.py"),
    ("inbreeding_depression", "test_bre_analysis_inbreeding_tc.py"),
    ("trial_design", "test_bre_analysis_trial_design_tc.py"),
    ("stability", "test_bre_analysis_stability_tc.py"),
    ("o234", "test_bre_analysis_o234_tc.py"),
]

_TC_OVERRIDES = {
    "db_host": "TC_CI_DB_HOST", "db_port": "TC_CI_DB_PORT", "db_user": "TC_CI_DB_USER",
    "db_password": "TC_CI_DB_PASSWORD", "db_name": "TC_CI_DB_NAME",
    "redis_host": "TC_CI_REDIS_HOST", "redis_port": "TC_CI_REDIS_PORT",
    "redis_db": "TC_CI_REDIS_DB",
}


def _env_cfg() -> dict:
    """解析 backend/env/.env.dev（简单 key=value），返回真实连接配置，可被 TC_CI_* 覆盖。"""
    cfg = {}
    if ENV_FILE.exists():
        for line in ENV_FILE.read_text(encoding="utf-8").splitlines():
            line = line.strip()
            if not line or line.startswith("#") or "=" not in line:
                continue
            k, v = line.split("=", 1)
            cfg[k.strip()] = v.strip().strip("'\"")
    for key, envvar in _TC_OVERRIDES.items():
        if envvar in os.environ and os.environ[envvar] != "":
            cfg[key] = os.environ[envvar]
    return {
        "db_host": cfg.get("DATABASE_HOST", "localhost"),
        "db_port": int(cfg.get("DATABASE_PORT", "5432")),
        "db_user": cfg.get("DATABASE_USER", "dpb"),
        "db_password": cfg.get("DATABASE_PASSWORD", ""),
        "db_name": cfg.get("DATABASE_NAME", "dpb"),
        "redis_host": cfg.get("REDIS_HOST", "localhost"),
        "redis_port": int(cfg.get("REDIS_PORT", "6379")),
        "redis_db": int(cfg.get("REDIS_DB_NAME", "1")),
    }


async def _probe() -> tuple[bool, bool, str]:
    import asyncpg
    import redis.asyncio as aioredis

    cfg = _env_cfg()
    db_ok, redis_ok = False, False
    db_err, redis_err = "", ""
    try:
        conn = await asyncpg.connect(
            host=cfg["db_host"], port=cfg["db_port"], user=cfg["db_user"],
            password=cfg["db_password"] or None, database=cfg["db_name"],
            timeout=5)
        await conn.close()
        db_ok = True
    except Exception as exc:  # noqa: BLE001
        db_err = str(exc)[:100]
    try:
        # protocol=2 必须与 app.core.database.redis_connect 一致：本机 WSL redis-server 版本较老，
        # 不支持 redis-py 7.x 默认 RESP3，需走 RESP2。
        r = aioredis.Redis(host=cfg["redis_host"], port=cfg["redis_port"],
                           db=cfg["redis_db"], protocol=2, socket_connect_timeout=5)
        await r.ping()
        await r.aclose()
        redis_ok = True
    except Exception as exc:  # noqa: BLE001
        redis_err = str(exc)[:100]
    detail = (f"PG={'ok' if db_ok else db_err} Redis={'ok' if redis_ok else redis_err}")
    return db_ok, redis_ok, detail


_PG: tuple | None = None


def pg_ready() -> tuple[bool, bool, str]:
    global _PG
    if _PG is None:
        _PG = asyncio.run(_probe())
    return _PG


def _sub_env() -> dict:
    """显式构建子进程环境：清空可能残留的连接变量，置真实 PG/Redis 连接。"""
    env = dict(os.environ)
    for k in ("DATABASE_TYPE", "DATABASE_NAME", "POOL_SIZE", "MAX_OVERFLOW",
              "CAPTCHA_ENABLE", "ENVIRONMENT", "PYTHONPATH", "PYTHONIOENCODING"):
        env.pop(k, None)
    cfg = _env_cfg()
    env["ENVIRONMENT"] = "dev"
    env["DATABASE_TYPE"] = "postgres"
    env["PYTHONPATH"] = str(BACKEND)
    env["PYTHONUTF8"] = "1"
    env["PYTHONIOENCODING"] = "utf-8"
    env["DATABASE_HOST"] = cfg["db_host"]
    env["DATABASE_PORT"] = str(cfg["db_port"])
    env["DATABASE_USER"] = cfg["db_user"]
    env["DATABASE_PASSWORD"] = cfg["db_password"]
    env["DATABASE_NAME"] = cfg["db_name"]
    env["REDIS_HOST"] = cfg["redis_host"]
    env["REDIS_PORT"] = str(cfg["redis_port"])
    env["REDIS_DB_NAME"] = str(cfg["redis_db"])
    return env


def _run_tc(name: str, script: Path) -> str:
    """顺序串跑单个 tc 脚本；返回 '' 表示通过，否则返回失败摘要（尾部日志）。"""
    try:
        proc = subprocess.run(
            [PY, "-X", "utf8", str(script)],
            cwd=str(SCRIPTS), env=_sub_env(), capture_output=True, text=True,
            encoding="utf-8", errors="replace", timeout=1800)
    except subprocess.TimeoutExpired:
        return f"[{name}] 超时（>30min）"
    out = proc.stdout or ""
    summary = ""
    for line in out.splitlines():
        if "===== " in line and "tc 套件" in line:
            summary = line.strip()
            break
    if proc.returncode == 0:
        print(f"  -> {name}: EXIT=0 {summary}")
        return ""
    tail = "\n".join(out.splitlines()[-25:])
    return f"[{name}] EXIT={proc.returncode} {summary}\n{tail}"


@pytest.mark.pg
@pytest.mark.parametrize("name,script", TC_SCRIPTS)
def test_analysis_tc(name: str, script: str):
    db_ok, redis_ok, detail = pg_ready()
    if not (db_ok and redis_ok):
        pytest.skip(f"需要真实 PostgreSQL16 + Redis（{detail}）")
    script_path = SCRIPTS / script
    assert script_path.exists(), f"脚本缺失: {script_path}"
    err = _run_tc(name, script_path)
    assert err == "", f"\n{err}"