Qiita Advent Calendar 2016 で削除されたカレンダー

deleted_calendar.py

import re

import requests
import time


def main():

    url = "http://qiita.com/sudoyu/items/0a51593b257f419aeb20.md"
    print(url)
    res = requests.get(url)
    print("status code: {}".format(res.status_code))
    if res.status_code != 200:
        return

    r = re.compile("^\* (.+)$")
    calendar_urls = [m.group(1) for m in map(r.match, res.text.split("\n")) if m]

    deleted_url = []
    for idx, url in enumerate(calendar_urls):
        time.sleep(0.2)
        while True:
            print("({:3d}/{:3d}) {} ... ".format(idx, len(calendar_urls), url), end="", flush=True)
            status_code = requests.head(url).status_code
            print(status_code, flush=True)

            if status_code == 404:
                deleted_url.append(url)
            if status_code in [200, 404]:
                break

            print("retry...")
            time.sleep(30)

    print("--------\ndeleted calendars")
    for url in deleted_url:
        print(url)


if __name__ == "__main__":
    main()

結果.txt

http://qiita.com/advent-calendar/2016/give_me_kareshi
http://qiita.com/advent-calendar/2016/gpd-win
http://qiita.com/advent-calendar/2016/it_event_de_jgc
http://qiita.com/advent-calendar/2016/holydisupute
http://qiita.com/advent-calendar/2016/rite
http://qiita.com/advent-calendar/2016/free_zankoku
http://qiita.com/advent-calendar/2016/test
http://qiita.com/advent-calendar/2016/enough
http://qiita.com/advent-calendar/2016/dena2
http://qiita.com/advent-calendar/2016/job2
http://qiita.com/advent-calendar/2016/yes-girlfriend
http://qiita.com/advent-calendar/2016/chat-bot
http://qiita.com/advent-calendar/2016/dena
http://qiita.com/advent-calendar/2016/muscle
http://qiita.com/advent-calendar/2016/poet
http://qiita.com/advent-calendar/2016/no-girlfriend-ml
http://qiita.com/advent-calendar/2016/no-girlfriend
http://qiita.com/advent-calendar/2016/job