import os
import re
import shutil


def process_files(cif_folder,input_folder,output_folder, anion):

    # 确保输出文件夹存在
    os.makedirs(output_folder, exist_ok=True)

    # 获取 txt 文件夹中的所有 txt 文件
    txt_files = [f for f in os.listdir(cif_folder) if f.endswith('.txt')]


    # 遍历 txt 文件
    for txt_file in txt_files:
        txt_path = os.path.join(cif_folder, txt_file)

        # 打开并读取 txt 文件内容
        with open(txt_path, 'r', encoding='utf-8') as file:
            content = file.read()
        matches = re.findall(r"the minium of d\s+([\d\.]+)\s*#", content)
        # 使用正则表达式查找符合条件的内容
        if matches:

            # 提取文件名（去掉.txt后缀）
            base_name = os.path.splitext(txt_file)[0]
            check = False
            if anion == "O":
                print(f"{base_name}的最短距离为{matches[0]}A")
                if float(matches[0]) < 3:
                    check = True
                    print(f"符合要求")
                else:
                    print("不符合要求")
            elif anion == "S":
                print(f"{base_name}的最短距离为{matches[0]}A")
                if float(matches[0]) < 3:
                    check = True
                    print(f"符合要求")
                else:
                    print("不符合要求")
            elif anion == "Cl":
                print(f"{base_name}的最短距离为{matches[0]}A")
                if float(matches[0]) < 3:
                    check = True
                    print(f"符合要求")
                else:
                    print("不符合要求")
            elif anion == "Br":
                print(f"{base_name}的最短距离为{matches[0]}A")
                if float(matches[0]) < 3:
                    check = True
                    print(f"符合要求")
                else:
                    print("不符合要求")
            if check:
                # 查找与 txt 文件同名的 cif 文件
                cif_path = os.path.join(input_folder, base_name)

                # 如果对应的 cif 文件存在，复制到 output_folder
                if os.path.exists(cif_path):
                    shutil.copy(cif_path, os.path.join(output_folder, base_name))
                    print(f"Copied {base_name} to {output_folder}")



if __name__ == '__main__':
    # process_files("../data/after_step1/O","../data/after_step2/O", "../data/after_step3/O", "O")
    # process_files("../data/after_step1/S", "../data/after_step2/S","../data/after_step3/S", "S")
    process_files("../data/after_step1/Cl", "../data/after_step2/Cl","../data/after_step3/Cl", "Cl")
    process_files("../data/after_step1/Br", "../data/after_step2/Br","../data/after_step3/Br", "Br")