Scrapes the data for EuroPython talks and writes to a CSV file to upload to a calendar

europython_schedule.py

from pandas import DataFrame
from bs4 import BeautifulSoup
import requests
from calendar import month_abbr
import datetime

base = 'https://ep2021.europython.eu'
sessions_url = f'{base}/events/sessions/'
resp = requests.get(sessions_url)
soup = BeautifulSoup(resp.text, 'html.parser')
anchors = soup.select('#content ul li a', href=True)
talks = [a['href'] for a in anchors if a['href'].startswith('/talks/')]

# this talk doesn't seem to have a date assigned
talks.remove('/talks/BjExfG9-python-security-best-practises/')

res = []
month_map = {month: idx for idx, month in enumerate(month_abbr) if month}
for talk in talks:
    resp = requests.get(f'{base}{talk}')
    soup = BeautifulSoup(resp.text, 'html.parser')
    
    title = soup.select('h1')[0].text
    speaker = soup.select('h5')[0].text
    
    date = soup.select('#talk_page .talk a.btn')[0].text
    _, day_and_month, time_range = date.split(', ')
    month, day = day_and_month.split()
    month = month_map[month]
    day = int(day)
    year = 2021
    
    time_start, *rest = time_range.partition('-')
    hour, minute = time_start.split(':')
    hour = int(hour)
    minute = int(minute)
    
    time_start = datetime.datetime(year, month, day, hour, minute)
    
    details = soup.select('#talk_page p code')[0].text
    talk_type, python_level, domain_level = details.split('; ')
    duration = int(talk_type[talk_type.find('(') + 1:talk_type.find(' mins')])
    duration = duration if duration != 180 else 180 + 15
    
    td = datetime.timedelta(minutes=duration)
    
    time_end = time_start + td
    
    res.append({
        'Subject': title,
        'Start': time_start,
        'End': time_end,
        'Description': f'{speaker} \n{base}{talk}\n{details}'
    })
    
df = DataFrame(res)
df['Start'] = df['Start'].dt.tz_localize('CET').dt.tz_convert('UTC')
df['End'] = df['End'].dt.tz_localize('CET').dt.tz_convert('UTC')
df['Start Date'] = df['Start'].dt.strftime('%Y-%m-%d')
df['Start Time'] = df['Start'].dt.strftime('%H:%M')
df['End Date'] = df['End'].dt.strftime('%Y-%m-%d')
df['End Time'] = df['End'].dt.strftime('%H:%M')

cols = [
    'Subject', 
    'Start Date', 
    'Start Time', 
    'End Date', 
    'End Time', 
    'Description',
]
df[cols].to_csv('schedule.csv', index=False)

Google Calendar: https://calendar.google.com/calendar/embed?src=o5ie1enl7dsatv0dr99rejhlp0%40group.calendar.google.com
iCal: https://calendar.google.com/calendar/ical/o5ie1enl7dsatv0dr99rejhlp0%40group.calendar.google.com/public/basic.ics