import os
from skimage.metrics import structural_similarity as ssim
import cv2
import concurrent.futures


def split_file_name(file_name):
    """
    将文件名拆分为前缀和后缀
    :param file_name: 文件名
    :return: 前缀和后缀
    """
    file_name = os.path.basename(file_name)
    file_name = file_name.split('.')[0]
    file_names = file_name.split('_')
    return file_names[0] + '_' + file_names[1], file_names[2]


def get_sorted_files_group(directory):
    try:
        # 检查目录是否存在
        if not os.path.exists(directory):
            print(f"目录 {directory} 不存在！")
            return {}

        # 获取目录下的所有文件及其完整路径
        files = [
            os.path.join(directory, f) for f in os.listdir(directory)
            if os.path.isfile(os.path.join(directory, f))
        ]

        files_group = {}
        for file in files:
            prefix, suffix = split_file_name(file)
            files_group.setdefault(prefix, []).append(file)
        return files_group

    except Exception as e:
        print(f"发生错误: {e}")
        return {}


def compare_images(image1_path, image2_path):
    try:
        # 读取两张图片
        img1 = cv2.imread(image1_path, cv2.IMREAD_GRAYSCALE)
        img2 = cv2.imread(image2_path, cv2.IMREAD_GRAYSCALE)

        # 检查图片是否成功读取
        if img1 is None or img2 is None:
            print(f"无法读取图片: {image1_path} 或 {image2_path}")
            return None

        # 调整两张图片为相同大小
        img1 = cv2.resize(
            img1, (min(img1.shape[1], img2.shape[1]), min(img1.shape[0], img2.shape[0]))
        )
        img2 = cv2.resize(img2, (img1.shape[1], img1.shape[0]))

        # 计算相似度
        similarity, _ = ssim(img1, img2, full=True)
        return similarity

    except Exception as e:
        print(f"比较图片时发生错误: {e}")
        return None


def del_group(group):
    """
    返回需要删除的图片列表，不在函数里直接执行删除操作。
    """
    del_list = []
    compared_pairs = set()  # 用于记录已经比较过的图片对

    for i in range(len(group)):
        for j in range(len(group)):
            if i != j and (i, j) not in compared_pairs and (j, i) not in compared_pairs:
                # 标记为已比较
                compared_pairs.add((i, j))
                # group[i] 或 group[j] 已经在 del_list 中就没必要比较了
                if group[i] not in del_list and group[j] not in del_list:
                    similarity = compare_images(group[i], group[j])
                    if similarity is not None and similarity > 0.965:
                        # 相似度大于 0.98，判定为重复
                        # 保留 group[j]，删除 group[i]，也可以根据需要调换顺序
                        del_list.append(group[i])

    return del_list


def process_group(prefix, file_list):
    """
    子进程运行函数：
    1. 对指定 group 进行去重计算
    2. 返回 (prefix, del_list)
    """
    print(f'开始处理 {prefix}, 共 {len(file_list)} 张图片')
    del_list = del_group(file_list)
    print(f'需要删除 {len(del_list)}/{len(file_list)} 张图片')
    return prefix, del_list


if __name__ == "__main__":
    directory = r"C:\Users\10561\Desktop\frames"
    files_group = get_sorted_files_group(directory)
    # for e in files_group:
    #     print(e, len(files_group[e]))
    # exit()
    # 使用多进程来处理每一个 group
    results = []
    with concurrent.futures.ProcessPoolExecutor() as executor:
        future_to_prefix = {}
        for prefix, file_list in files_group.items():
            future = executor.submit(process_group, prefix, file_list)
            future_to_prefix[future] = prefix

        for future in concurrent.futures.as_completed(future_to_prefix):
            prefix = future_to_prefix[future]
            try:
                ret_prefix, del_list = future.result()
                # 在主进程里进行删除
                for e in del_list:
                    if os.path.exists(e):
                        # print(f'删除 {e}')
                        os.remove(e)
            except Exception as e:
                print(f"{prefix} 处理时发生错误: {e}")

    print("所有任务已完成！")