import warnings
import numpy as np
import pandas as pd
import json

# 抑制numpy的警告（由于数据中包含NaN值，这是正常的）
warnings.filterwarnings("ignore", category=RuntimeWarning, module="numpy")
np.seterr(all="ignore")  # 忽略numpy的浮点错误警告

from data import load_data
from factor_mining.gp_miner import GPMiner, GPConfig

if __name__ == "__main__":
    df = load_data("/Users/aszer/Documents/vscode/factorhack/ETH_USDT-1h.feather")
    # 以4小时为周期重采样K线数据（假定有datetime索引，常见ohlcv列）
    df = (
        df.resample("4h")
        .agg(
            {
                "open": "first",
                "high": "max",
                "low": "min",
                "close": "last",
                "volume": "sum",
            }
        )
        .dropna()
    )
    df = df[df.index < '2023-01-01']
    print("数据加载成功，前5行：")
    print(df.head())
    print(f"\n数据形状: {df.shape}")
    print(f"数据列: {df.columns.tolist()}")
    gp_config = GPConfig(
        ret_horizon=48,
        ic_window=120,
        ic_method="spearman",
        seed=None,
        population_size=200,
        generations=30,
        tournament_size=5,
        crossover_prob=0.9,
        mutation_prob=0.05,
        elitism=5,
        max_depth_init=1,
        max_depth=30,
        complexity_penalty=0.001,
    )
    miner = GPMiner(config=gp_config)
    res = miner.mine(df, ["open", "high", "low", "close", "volume"])
    with open("gp_miner_result.txt", "w") as out_file:
        for formula, ic_tuple in res:
            # ic_tuple 是元组，取第一个元素作为IC值
            ic = ic_tuple[0] if isinstance(ic_tuple, tuple) else ic_tuple
            print(f"{formula.expression}, IC: {ic:.4f}")
            # 将因子公式转换为字典并写入文件
            out_file.write(json.dumps(formula.to_dict(), ensure_ascii=False))
            out_file.write("\n")