82b734d1a8
Summary is the title of the event in your calendar, so we want it to be a little less verbose.
107 lines
3.3 KiB
Python
107 lines
3.3 KiB
Python
import logging
|
|
import logging.config
|
|
from typing import List
|
|
|
|
import pendulum
|
|
import requests
|
|
from bs4 import BeautifulSoup
|
|
from bs4.element import Tag
|
|
|
|
from autopilot import config
|
|
from autopilot.reservation import Reservation
|
|
|
|
|
|
LOGGER = logging.getLogger('autopilot.scrape')
|
|
|
|
|
|
def init_session() -> requests.Session:
|
|
headers = {
|
|
# required for some requests or else it errors
|
|
'user-agent': ('Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHT'
|
|
'ML, like Gecko) Chrome/71.0.3578.98 Safari/537.36'),
|
|
}
|
|
session = requests.Session()
|
|
session.headers.update(headers)
|
|
return session
|
|
|
|
|
|
def authenticate(session: requests.Session):
|
|
credentials = {
|
|
'txtUserID': config.USERNAME,
|
|
'txtPassword': config.PASSWORD,
|
|
'mode': 'process', # set by client-side JS; required to auth
|
|
}
|
|
response = session.post(f'{config.BASE_URL}/login.asp', data=credentials)
|
|
if "Welcome" not in response.text:
|
|
raise ValueError('failed to authenticate')
|
|
|
|
|
|
def format_day(day: pendulum.DateTime) -> str:
|
|
return day.format('M/D/YYYY') # standard format throughout aerocalendar
|
|
|
|
|
|
def raw_schedule_html(session: requests.Session) -> BeautifulSoup:
|
|
today = pendulum.today(config.TIME_ZONE)
|
|
params = (
|
|
('date', format_day(today)),
|
|
('location', '1'), # Bedford
|
|
)
|
|
url = f'{config.BASE_URL}/MySchedule.asp'
|
|
data = {
|
|
'rdoDays': 56, # Look ahead 56 days
|
|
'txtStartDate': format_day(today),
|
|
'txtEndDate': format_day(today.end_of('week')),
|
|
'txtSaveLocation': 1,
|
|
'txtSaveDate': format_day(today),
|
|
'submittype': 'notes',
|
|
'reservation': '',
|
|
'cancelcode': '',
|
|
'submitted': 'true',
|
|
'user': config.USER_ID,
|
|
}
|
|
response = session.post(url, params=params, data=data)
|
|
return response.text
|
|
|
|
|
|
def make_reservation_from_tag(tag: Tag) -> Reservation:
|
|
"""Parse attributes from the title attribute of the reservation
|
|
|
|
Example raw string parsed:
|
|
|
|
Pilot : Ian Naval
|
|
Instructor : Agnelo Lopes CFI, CFII
|
|
Aircraft : N222ND
|
|
Location : Bedford
|
|
Start : 1/26/2019 12:00 PM
|
|
End : 1/26/2019 3:00 PM
|
|
Comments -----------------------------
|
|
"
|
|
"""
|
|
raw = tag.get('title')
|
|
lines = raw.replace('\t', '').split('\r\n')
|
|
attributes = {}
|
|
for line in lines:
|
|
if ':' in line:
|
|
key, value = line.split(':', maxsplit=1)
|
|
attributes[key.lower().strip()] = value.strip()
|
|
attributes['comments'] = lines[-2]
|
|
for datetime_attr in {'start', 'end'}:
|
|
parsed_datetime = pendulum.parse(attributes[datetime_attr],
|
|
strict=False,
|
|
tz=config.TIME_ZONE)
|
|
attributes[datetime_attr] = parsed_datetime
|
|
attributes['raw'] = raw.replace('\r', '').replace('\t', '')
|
|
return Reservation(**attributes)
|
|
|
|
|
|
def get_reservations() -> List[Reservation]:
|
|
logging.config.dictConfig(config.LOGGING_CONFIG)
|
|
session = init_session()
|
|
authenticate(session)
|
|
LOGGER.info("Authentication successful")
|
|
html = raw_schedule_html(session)
|
|
soup = BeautifulSoup(html, 'html.parser')
|
|
reservations = [make_reservation_from_tag(tag) for tag in soup.select('td.cR')]
|
|
LOGGER.info("Got %d reservations", len(reservations))
|
|
return reservations
|