faker_diary/bin/faker_diary.py

# -*- coding: utf-8 -*-
import jinja2
import datetime
import subprocess
import random
import shutil

from pathlib import Path
from faker import Faker
from chinese_calendar import is_workday


NOW = Path(__file__).parent
ETC = NOW.parent / "etc"
OUTPUT = NOW.parent / "output"


def faker_datas(number=7):
    """
    生成日志模板所需要的内容,一天一个默认7天,排除掉工作日（排除掉的也算一天）
    """
    f = Faker(locale="zh_CN")
    # 用来存所有的
    datas = []
    # 现在的时间
    today = datetime.datetime.now().date()
    # 默认生成前7天的数据
    for n in range(0, number):
        # 前N天
        date = today + datetime.timedelta(days=-n)
        # 判断是不是工作日
        if not is_workday(date):
            continue
        # 用来保存1天所需要的内容
        diary_log = {}
        # 创建日期
        diary_log["create_date"] = date
        # 创建时间，控制在8点~11点之间
        str_time1 = date.strftime("%Y-%m-%d ") + "08:00:00"
        start_time = datetime.datetime.strptime(str_time1, "%Y-%m-%d %H:%M:%S")
        str_time2 = date.strftime("%Y-%m-%d ") + "11:00:00"
        end_time = datetime.datetime.strptime(str_time2, "%Y-%m-%d %H:%M:%S")

        CreationTime = f.date_time_between(start_date=start_time, end_date=end_time)
        diary_log["CreationTime"] = CreationTime
        # 修改时间，比创建时间晚6~10小时
        LastWriteTime = CreationTime + datetime.timedelta(
            hours=f.pyint(min_value=6, max_value=10),
            seconds=f.pyint(min_value=1, max_value=300),
        )
        diary_log["LastWriteTime"] = LastWriteTime

        # 访问时间，比修改时间晚1~n天
        LastAccessTime = LastWriteTime + datetime.timedelta(
            days=f.pyint(min_value=0, max_value=n),
            seconds=f.pyint(min_value=1, max_value=300),
        )
        diary_log["LastAccessTime"] = LastAccessTime
        # 随机的状态，成功失败比 8：2
        status_list = {"sucess": 80, "failed": 20}
        diary_log["status1"] = random_weight(status_list)

        datas.append(diary_log)

    return datas


def random_weight(weight_data):
    """
    按照权重，随机取值
    eg:
    weight_data = {'a': 10, 'b': 40, 'c': 50}
    random_weight(weight_data)
    """
    total = sum(weight_data.values())  # 权重求和
    ra = random.uniform(0, total)  # 在0与权重和之前获取一个随机数
    curr_sum = 0
    ret = None
    # keys = weight_data.iterkeys()    # 使用Python2.x中的iterkeys
    keys = weight_data.keys()  # 使用Python3.x中的keys
    for k in keys:
        curr_sum += weight_data[k]  # 在遍历中，累加当前权重值
        if ra <= curr_sum:  # 当随机数<=当前权重和时，返回权重key
            ret = k
            break
    return ret


def main(number):
    # 获取数据
    datas = faker_datas(number)
    # 初始化输出目录
    if OUTPUT.exists():
        shutil.rmtree(OUTPUT)
    OUTPUT.mkdir()

    # 根据模板来生成数据
    template_file = ETC / "template.md.j2"
    for day_log in datas:
        # 读取模板
        jinja2_env = jinja2.Environment()
        tempate_data = open(template_file, encoding="utf-8").read()
        # 渲染模板
        temp_out = jinja2_env.from_string(tempate_data).render({"datas": day_log})

        # 输出结果
        file_name = day_log["create_date"].strftime("%Y-%m-%d") + ".md"
        output_file = OUTPUT / file_name
        with open(output_file, "w", encoding="utf-8") as f:
            f.writelines(temp_out)

        # 修改生成文本的创建时间、修改时间和访问时间
        # ls 'sysin.txt' | foreach-object { $_.CreationTime = '02/02/2021 01:01:01'; $_.LastWriteTime = '01/01/2021 01:01:01'; $_.LastAccessTime = '03/03/2021 01:01:01' }
        args = (
            "powershell.exe"
            + ' "ls '
            + "'"
            + str(output_file)
            + "'"
            + " | foreach-object { $_.CreationTime = '"
            + day_log["CreationTime"].strftime("%m/%d/%Y %H:%M:%S")
            + "'; $_.LastWriteTime = '"
            + day_log["LastWriteTime"].strftime("%m/%d/%Y %H:%M:%S")
            + "'; $_.LastAccessTime = '"
            + day_log["LastAccessTime"].strftime("%m/%d/%Y %H:%M:%S")
            + "' }\""
        )
        subprocess.call(args, shell=True)


if __name__ == "__main__":
    # 生成30天的数据
    main(30)