python日常工作处理-文件按比例分割数据

python日常工作处理-文件按比例分割数据

把一个保存用户id文本进行比例分割，比例为50%，分别另存为另外两个文件

代码

import random

input_file = '/Users/Desktop/2023-03-28.txt'
group1_file = '/Users/Desktop/group1_2023-03-28.txt'
group2_file = '/Users/Desktop/group2_2023-03-28.txt'

if __name__ == '__main__':
    output_files = [
        open(group1_file, 'w'),
        open(group2_file, 'w')
    ]
    counts = [0, 0]

    with open(input_file, 'r') as input:
        total_count = 0
        for _ in input:
            total_count += 1
        print(f'{input_file} size:{total_count}')
        total_count //= 2
        input.seek(0)
        for line in input:
            index = random.randint(0, 1)
            if counts[index] >= total_count:
                index = 1 - index
            output_files[index].write(line)
            counts[index] += 1
    print(f'{group1_file} size:{counts[0]}')
    print(f'{group2_file} size:{counts[1]}')
    output_files[0].close()
    output_files[1].close()