hadoop学习-1 获取测试数据

2019独角兽企业重金招聘Python工程师标准>>>

制造数据:  每行101个随机数字。

import random


def get_file():
    """
    获得一个文件，如果没有，则；创建并返回，如果有，则直接返回。
    :return: file
    """
    file = open('data.txt', 'a', encoding='utf-8')
    return file


def append_data(file, data_num):
    """
    获取数据
    :param file:
    :return:
    """
    for i in range(data_num):
        tmp_data = random.randrange(0, data_num)
        tmp_data = str(tmp_data) + ' '
        file.write(tmp_data)
        if i != 0 and i % 100 == 0:
            file.write('\n')
            print(i)


if __name__ == '__main__':
    file = get_file()
    data_num = 100000000000000000000
    append_data(file, data_num)
    file.close()

获取文件中的单个行：

if __name__ == '__main__':
    file = open('data.txt', 'r', encoding='utf-8')
    try:
        while True:
            line = file.readline()
            if line:
                words = line.split(' ')
                print("line=", len(words))
            else:
                break
    finally:
        file.close()

 

转载于:https://my.oschina.net/marjeylee/blog/1559781