tests/test_formula_integrity.py

import os
import sys
import importlib.util
import numpy as np
import pandas as pd
import pytest

NMSE_THRESHOLD = 3e-5

# 读取所有需要测试的formula.py路径
with open(os.path.join(os.path.dirname(__file__), './filtered_formula_py.txt')) as f:
    formula_py_list = [line.strip() for line in f if line.strip()]

def dynamic_import_formula(formula_path):
    # 动态导入formula.py，并注入np
    spec = importlib.util.spec_from_file_location('formula', formula_path)
    module = importlib.util.module_from_spec(spec)
    setattr(module, 'np', np)
    sys.modules['formula'] = module
    spec.loader.exec_module(module)
    # 取第一个非__开头的函数
    func = None
    for name in dir(module):
        if not name.startswith('__') and callable(getattr(module, name)):
            func = getattr(module, name)
            break
    return func

def find_all_csv_files(formula_path):
    # 返回同级目录下所有csv文件
    base_dir = os.path.dirname(formula_path)
    csv_files = []
    for fname in os.listdir(base_dir):
        if fname.endswith('.csv'):
            fpath = os.path.join(base_dir, fname)
            # 排除空文件
            if os.path.getsize(fpath) > 0:
                csv_files.append(fpath)
    return csv_files

def calc_nmse(y_true, y_pred):
    y_true = np.asarray(y_true)
    y_pred = np.asarray(y_pred)
    mse = np.mean((y_true - y_pred) ** 2)
    var = np.mean((y_true - np.mean(y_true)) ** 2)
    if var == 0:
        return np.inf if mse > 0 else 0
    return mse / var

@pytest.mark.parametrize('formula_path', formula_py_list)
def test_formula(formula_path):
    func = dynamic_import_formula(formula_path)
    csv_files = find_all_csv_files(formula_path)
    assert csv_files, f'No csv files found for {formula_path}'
    for csv_file in csv_files:
        df = pd.read_csv(csv_file)
        if df.shape[1] < 2:
            continue  # 跳过只有一列的csv
        target_col = df.columns[-1]  # 只取最后一列为target
        input_cols = df.columns[:-1]  # 前面所有列为输入
        # 只测前1000行，防止超大
        df = df.iloc[:1000]
        y_true = df[target_col].values
        y_pred = []
        for _, row in df.iterrows():
            try:
                y_pred.append(func(*(row[col] for col in input_cols)))
            except Exception as e:
                pytest.fail(f'Error in {formula_path} {csv_file} row: {e}')
        nmse = calc_nmse(y_true, y_pred)
        assert nmse < NMSE_THRESHOLD, f'{formula_path} {csv_file} nmse={nmse} >= {NMSE_THRESHOLD}'