import os
import re
from pathlib import Path

def read_end_indices(filename='a.txt'):
    """读取 end.txt，返回整数列表"""
    with open(filename, 'r', encoding='utf-8') as f:
        content = f.read()
    return [int(x) for x in re.findall(r'-?\d+', content)]

def delete_extra_npz():
    root = Path("put_toy_in_cabinet")
    end_indices = read_end_indices()
    print(f"读取到 {len(end_indices)} 个 episode 的保留上限: {end_indices}")

    for ep_idx, max_keep in enumerate(end_indices):
        ep_dir = root / str(ep_idx)
        if not ep_dir.exists():
            print(f"⚠️  Episode {ep_idx} 不存在，跳过")
            continue

        # 获取该目录下所有 .npz 文件
        npz_files = list(ep_dir.glob("*.npz"))
        deleted_count = 0

        for npz in npz_files:
            # 提取文件名中的数字（如 "5.npz" → 5）
            stem = npz.stem
            if not stem.isdigit():
                print(f"❓ 跳过非标准命名文件: {npz}")
                continue

            idx = int(stem)
            if idx > max_keep:
                npz.unlink()  # 删除文件
                deleted_count += 1

        print(f"✅ Episode {ep_idx}: 删除了 {deleted_count} 个多余的 .npz 文件")

if __name__ == "__main__":
    delete_extra_npz()