# tests/factor/test_fundamental_library.py
"""财务因子表达式库: 32 个 P0 因子注册 + 表达式↔adapter 特征列契约锁定."""
import re
import sys, os
sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..")))
sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..", "vnpy_v4.4.0")))

import pytest

from sanguo_factor import fundamental_library  # noqa: F401  import 即注册
from sanguo_factor.fundamental_adapter import FEATURE_COLUMNS
from sanguo_factor.registry import list_factors, get_factor

@pytest.fixture(autouse=True)
def _ensure_fundamental_registered():
    """其它测试模块清空 _REGISTRY 后只重挂 alpha/builtin(顺序依赖前科),
    这里逐测试幂等重注册财务因子,保证本模块与顺序无关."""
    fundamental_library._register_all()


# 表达式可引用的列 = adapter 特征列 + 行情列 close(估值类 ÷ close×share_capital)
# + 表达式函数名(P1 起用到 log)
_ALLOWED = set(FEATURE_COLUMNS) | {"close", "cs_rank", "log"}


def _fundamental_factors() -> list[dict]:
    return list_factors("fundamental")


def test_p0_32_factors_registered():
    facs = _fundamental_factors()
    names = {f["name"] for f in facs}
    # P0 32 + P1 37(35 因子 + 2 互评变体;P1 名单见 test_fundamental_p1_library)
    assert len(facs) == 75, f"P0+P1+P1-B 应为 75 个,实际 {len(facs)}"
    # 六族代表抽查(全部名单见 fundamental_library 注释)
    expect = {
        "fund_roe_ttm", "fund_gp_over_assets",          # A
        "fund_tacc", "fund_nonrec_ratio",               # B
        "fund_rev_q_yoy", "fund_np_q_yoy", "fund_asset_growth",  # C
        "fund_ep_ttm", "fund_bp", "fund_cp",            # D
        "fund_nsi", "fund_ibd_ratio",                   # E
        "fund_sue_np", "fund_forecast_type",            # F
    }
    assert expect <= names


def test_expressions_only_reference_feature_columns():
    """契约: 表达式裸标识符 ⊆ adapter 特征列 + close/cs_rank(漏加列=拼写错)."""
    for f in _fundamental_factors():
        idents = set(re.findall(r"[A-Za-z_][A-Za-z0-9_]*", f["expression"]))
        bad = idents - _ALLOWED
        assert not bad, f"{f['name']} 引用了未产出列: {bad} in {f['expression']}"


def test_all_factors_are_cross_sectional_rank():
    """P0 设计: 因子 = cs_rank(基础指标) 一层(负 IC 因子取负定向,高=好)."""
    for f in _fundamental_factors():
        assert f["expression"].startswith("cs_rank("), f["name"]
        assert f["expression"].endswith(")")


def test_negative_ic_factors_flipped():
    """文档负 IC 因子(B01/B04/B05/B07/B08/B11/C15/E01/E02/E05/E09)表达式含负号."""
    flipped = {"fund_tacc", "fund_nonrec_ratio", "fund_impairment_ratio",
               "fund_invest_income_dep", "fund_receivables_anomaly",
               "fund_other_rece_ratio", "fund_asset_growth",
               "fund_nsi", "fund_equity_fin_intensity", "fund_ibd_ratio",
               "fund_goodwill_ratio"}
    for name in flipped:
        expr = get_factor(name)["expression"]
        assert "(-" in expr or expr.startswith("cs_rank(-"), f"{name} 应翻转: {expr}"


def test_valuation_factors_use_close_times_share_capital():
    """估值族市值 = close × share_capital 自算(规避 valuation 中文列名表)."""
    for name in ("fund_ep_ttm", "fund_ep_deduct_ttm", "fund_bp", "fund_cp",
                 "fund_equity_fin_intensity"):
        expr = get_factor(name)["expression"]
        assert "close * share_capital" in expr, f"{name}: {expr}"


def test_registration_idempotent():
    """重复 import 不炸(注册表防重入,同 library.py 模式)."""
    import importlib
    importlib.reload(fundamental_library)
    assert len(_fundamental_factors()) == 75