timekeeper_ical.py

#!/usr/bin/env python3

import datetime
from io import BytesIO
from sys import argv
import requests
from PyPDF2 import PdfFileReader
from icalendar import Calendar, Event, Alarm
from icalendar.prop import vRecur
from pytz import UTC
import argparse


def parse_pdf(pdata, period):
    pdata = BytesIO(pdata)
    pdf = PdfFileReader(pdata)
    extract = []
    for page_num in range(pdf.numPages):
        page = pdf.getPage(page_num)
        extract.extend(page.extractText().split())

    pos = 0
    due_dates = []
    while pos < len(extract):
        if extract[pos] == period:
            due_dates.append(extract[pos+5])
            pos += 5
        else:
            pos += 1

    return due_dates

def create_ics(dates, url, period):
    reminder_delta = 24  # hours
    cal = Calendar()
    cal.add('prodid', '-//TimeKeeper Calendar//mxm.dk//')
    cal.add('version', '2.0')
    counter = 0

    for date in dates:
        month, day, year = date.split("/")
        date = datetime.date(int(year), int(month), int(day))
        event = Event()
        event.add('SUMMARY', "Payroll Time Keeper Due Date")
        event.add('DTSTART', date)
        event.add('DTEND', date)
        event.add('DTSTAMP', datetime.datetime.now(tz=UTC))
        event.add('LOCATION', "HRMS")
        event.add('URL', url)
        d = datetime.datetime.now(tz=UTC)
        ds = str(d)
        ds = ds.replace('+00:00', 'Z')
        ds = ds.replace(' ', 'T')
        ds = ds.replace('-', '')
        ds = ds.replace(':', '')
        event['uid'] = '{0}-{1}@pas.rochester.edu'.format(
            ds,
            counter)
        alarm = Alarm()
        alarm.add("action", "DISPLAY")
        alarm.add('description', "Reminder")
        #alarm.add("trigger", dt.timedelta(hours=-reminderHours))
        # The only way to convince Outlook to do it correctly
        alarm.add("TRIGGER;RELATED=START", "-PT{0}H".format(reminder_delta))

        event.add_component(alarm)
        cal.add_component(event)
        counter += 1

    with open(f"timekeeper_{period.lower()}.ics", 'wb') as f:
        f.write(cal.to_ical())


def main():
    parser = argparse.ArgumentParser()

    parser.add_argument("-u", "--url",
                        type=str,
                        nargs="?",
                        default=None,
                        help="provide a Timekeeper PDF URL",
                        required=True,
                        action='store')

    parser.add_argument("-p", "--period",
                        type=str,
                        nargs="?",
                        default=None,
                        help="paygroup code, e.g. \"MTH\"",
                        required=True,
                        action='store')

    args = parser.parse_args()
    data = requests.get(args.url)
    due_dates = parse_pdf(data.content, args.period)
    create_ics(due_dates, args.url, args.period)


if __name__ == '__main__':
    main()