# -*- coding: utf-8 -*-
# Time : 2023/12/8 9:11
# Author : lirunsheng
# User : l'r's
# Software: PyCharm
# File : read_data.py

import os
import glob
import random
import pandas as pd

def read_folders(folder_path):
    # 遍历文件夹中的子文件夹
    subfolders = []
    for root, dirs, files in os.walk(folder_path):
        for subdir in dirs:
            subfolders.append(subdir)
    # print(subfolders)

    # 提取子文件夹中的图像路径
    image_paths = []
    for root, dirs, files in os.walk(folder_path):
        for file in files:
            if file.endswith('.png') or file.endswith('.jpg'):
                image_path = os.path.join(root, file)
                image_path = image_path.replace("\\", "/")  # 将路径中的反斜杠替换为斜杠
                index_name = image_path.split('/')[-2]
                image_paths.append([image_path, subfolders.index(index_name)])
    # 打乱列表中的元素
    random.shuffle(image_paths)
    # 打印图像路径列表
    path_list = []
    label_list = []
    for path in image_paths:
        # print(path)
        path_list.append(path[0])
        label_list.append(path[1])
    # print(label_list)
    # 将列表转换为DataFrame
    df = pd.DataFrame(label_list, columns=['label'])
    # print(df)
    label_name = folder_path.split('/')[-2]
    df.to_csv(label_name+'.csv', header=False, index=False)
    return subfolders, path_list  # 标签，图片路径集