Summaery 2023 calender event scraper

summaery_scrape.py

"""visits the event list of the summaery 2023 and scrapes all events into one big icalendar file. 
update the url and year in the future"""

from datetime import datetime
import pytz
from icalendar import Calendar, Event

from selenium.webdriver.common.by import By
from selenium import webdriver
from selenium.webdriver.chrome.options import Options


def get_event_start_end_place(summaery_event):
	"""read the start, end time and possible place from the same calender info element"""
	date_string = get_event_date(summaery_event)
	time_info = summaery_event.find_element(By.CLASS_NAME, "cal_maininfo")
	infos = time_info.find_elements(By.CSS_SELECTOR, "li")

	# filter event start and end for only the datetime string
	start = date_string + " " + infos[0].text[8:-4]
	end = infos[1].text[6:-4]
	
	# add the event date to event end if it's not in the string yet
	if len(end) < 10:
		end = date_string + " " + end

	place = infos[2].text[5:] if len(infos) > 2 else None
	
	time_format = "%d.%m.%Y %H.%M"
	germany = pytz.timezone('Europe/Berlin')
	start = datetime.strptime(start, time_format)
	end = datetime.strptime(end, time_format)
	
	return germany.localize(start), germany.localize(end), place


def get_event_date(summaery_event):
	"""read the event date string from the big date element to the left. and append a year"""
	month = summaery_event.find_element(By.CLASS_NAME, "month").text
	day = summaery_event.find_element(By.CLASS_NAME, "day_of_month").text
	return day.zfill(2) + "." + month.zfill(2) + ".2023"


def get_event_title_url(summaery_event):
	"""read the event title and url for more info from the desciption to the right"""
	title_elem = summaery_event.find_element(By.CSS_SELECTOR, "h2").find_element(By.CSS_SELECTOR, "a")
	title = title_elem.text
	url = title_elem.get_attribute("href")
	return title, url


def get_event_teaser(summaery_event):
	"""read the description text of the event on the right"""
	return summaery_event.find_element(By.CLASS_NAME, "teasertext").text[:-4]


def add_event_to_cal(cal, start, end, place, title, url, teaser):
	"""create and add a calendar event with all the info to the calendar"""
	event = Event()
	event.add("summary", title)
	event.add("dtstart", start)
	event.add("dtend", end)
	
	if place:
		event.add("location", place)
	
	event.add("url", url)
	event.add("description", teaser)
	cal.add_component(event)


if __name__ == "__main__":
	opts = Options()
	opts.add_argument("user-agent=firefox")
	driver = webdriver.Chrome(options=opts)

	page_url = "https://www.uni-weimar.de/de/universitaet/aktuell/jaehrliche-veranstaltungen/jaehrliche-veranstaltungen-2023/summaery/veranstaltungen/"
	driver.get(page_url)

	# find all events listed on the page
	summaery_events = driver.find_elements(By.CLASS_NAME, "summaeryLiveEvent")
	cal = Calendar()
		
	for i, event in enumerate(summaery_events):
		add_event_to_cal(
			cal,
			*get_event_start_end_place(event),
			*get_event_title_url(event),
			get_event_teaser(event))
		print(i + 1, "/", len(summaery_events))

	# write them into one big icalendar file
	with open("summaery.ics", "wb") as f:
		f.write(cal.to_ical())
	
	driver.quit()