get_pt_vacc_schedules.py

from selenium import webdriver
from collections import deque

browser = webdriver.Chrome()
try:
    browser.get('http://www.portugal-vacc.org/')

    # just chrome devtools'd this
    # TODO: make sure this doesn't break when theres no scheduled ATC
    elem = browser.find_element_by_css_selector('\
        body > table > tbody > tr:nth-child(3) > td > table:nth-child(1) > \
        tbody > tr:nth-child(2) > td:nth-child(2) > p:nth-child(5) > table')

    rows = elem.find_elements_by_tag_name('tr')

    data = deque([])
    for row in rows:
        row = row.text
        if (row != ''):
            data.insert(0, row)
            continue

        if len(data) > 0:

            # split the parts
            where, who = data.pop().split(' ', 1)

            # lookup for training sessions
            mentor = None
            if '\n' in who:
                who, mentor = who.split('\n', 1)
                who = who[:len(who) - len(' [Training]')]
                mentor = mentor[len('Mentor:') + 1:]

            # TODO: clear this
            if mentor:
                print('%s training at %s' % (who, where))
            else:
                print('%s controlling at %s' % (who, where))
        data = []
finally:
    browser.quit()