フットサルの大会見つけるやつ

footsal_tourment_list.py

import re
import datetime
import locale

import urllib3
import urllib.parse
from bs4 import BeautifulSoup

def safe_cast(val, to_type, default=None):
    try:
        return to_type(val)
    except (ValueError, TypeError):
        return default

def extract_place(title):

	keywords = ["原宿", "渋谷", "池袋", "新宿"]

	for key in keywords:		
		pattern = re.compile(rf'.*{key}.*')		
		if pattern.match(title) is not None:				
				return key

locale.setlocale(locale.LC_TIME, 'ja_JP.UTF-8')
now = datetime.datetime.now()
search_date_str = now.strftime("%Y-%m-%d")

def get_list(base_url):
	
	outputs = []

	for page_num in range(10):
				
		url = base_url + "/?page=" + str(page_num + 1)+ "&hold_on_from=" + search_date_str + "&embed=normal"

		http = urllib3.PoolManager()
		r = http.request('GET', url)
		soup = BeautifulSoup(r.data, "html.parser")

		ul = soup.select_one('#event_list')
		children = ul.findAll("li")

		for num in range(len(children) - 1):
			
			body = soup.select_one("#event_list > li:nth-child(" + str(num + 1) + ")")

			try:
				date_str = body.select_one("div.date").text
			except Exception as e:
				continue			

			title = body.select_one("h2").text		
			place = extract_place(title + date_str)

			rest_raw_str = body.select_one("div.capacity > span").text		
			rest_num_str = re.sub(r'\（|\）|(残り)|\s', "", rest_raw_str)
			rest_num = safe_cast(rest_num_str, int)
			
			if rest_num is not None and rest_num > 0 and place is not None:	
										
				datetime_str = re.sub(r'\｜.*$', "", date_str)
				date = datetime_str[:13]
				start_time = datetime_str[13:18]
				end_time = datetime_str[19:]

				start_datetime = datetime.datetime.strptime('{} {}'.format(date, start_time), "%Y/%m/%d（%a） %H:%M")
				end_datetime = datetime.datetime.strptime('{} {}'.format(date, end_time), "%Y/%m/%d（%a） %H:%M")

				link_raw = body.select_one("a")["href"]			
				link = re.sub(r'\/$|\/\?.*$|(https://labola.jp)', "", link_raw)
				formed_url = "https://labola.jp" + link				

				out_raw = {
					"start_datetime": start_datetime,
					"end_datetime": end_datetime,
					"rest": rest_num,
					"url": formed_url,
					"place": place,
				}

				outputs.append(out_raw)

	return outputs

# 大会
total_list = get_list("https://labola.jp/reserve/events/search/tournament/area-13")
total_list.extend(get_list("https://labola.jp/r/event/3010"))

# 個サル
total_list = get_list("https://labola.jp/reserve/events/search/personal/area-13")

format_list = map(lambda x: "{} {} {}h @{} (残{}) {}".format(
	x['start_datetime'].strftime("%m/%d (%a)"),
	x['start_datetime'].strftime("%H:%M~"),
	((x['end_datetime'] - x['start_datetime']).seconds/60/60),
	x['place'],
	x['rest'],
	x['url'],
) , total_list)

for a in sorted(set(format_list)):
	print(a)