"""
读取数据
数据清洗
"""
import os
import imghdr
import cv2
from progressbar import ProgressBar

# error_file_name = 'check_error.txt'
path = r'E:\pythonProject\ftc\images'
# path = r'E:\pythonProject\ftc\images_v2'


def error_file():
    original_images = []
    for root, dirs, filenames in os.walk(path):
        for filename in filenames:
            original_images.append(os.path.join(root, filename))
    original_images = sorted(original_images)
    print('total_file_num:', len(original_images))
    error_images = []
    progress = ProgressBar()
    for filename in progress(original_images):
        check = imghdr.what(filename)
        if check is None:
            error_images.append(filename)
            os.remove(filename)
            continue
        img = cv2.imread(filename)
        if img is None:
            error_images.append(filename)
            os.remove(filename)
    print('error_file_num:', len(error_images))


error_file()
# E:\pythonProject\ftc\images_v1\mulberry\1706603229570325.jpg
# v1
# first
# error_file_num: 654
# second
# error_file_num: 44
# small_file_num: 0
# third
# error_file_num: 0
# small_file_num: 0

# v2
# first
# error_file_num: 1144
# second
# error_file_num: 64
# small_file_num: 0
# third
# error_file_num: 0
# small_file_num: 0