Test a GTFS file's calendar and calendar_dates tables for a specific date (like today)

valid_calendar_dates.py

#!/usr/bin/python3

"""
Check a GTFS file's calendar and calendar_dates tables.
Return True if the dates are current or future.
Return False if all dates are in the past and the GTFS file
  can be safely deleted.

This is a working example. Try it. Edit the path to match your GTFS file.
"""

import zipfile


def valid_dates(gtfs_path_and_file, date_string):
    """
    Check a GTFS file's calendar and calendar_dates tables.
    If all valid dates are in the past, return False.
    If any valid dates include the date string or the future, return True.
    Date string is format YYYYMMDD, just like GTFS dates.
    """
    gtfs_zipfile = zipfile.ZipFile(gtfs_path_and_file, mode='r')
    valid = False

    if 'calendar.txt' in gtfs_zipfile.namelist():
        cal_string = gtfs_zipfile.read('calendar.txt').decode('utf-8')
        cal_lines  = cal_string.split('\r\n')
        del cal_lines[0]       # Delete the header row

        for line in cal_lines:
            if len(line.split(',')) < 9:
                continue      
            start_date = int(line.split(',')[8])
            end_date   = int(line.split(',')[9])
            if start_date >= int(date_string) \
            or end_date   >= int(date_string):
                valid = True
                break 

    elif 'calendar_dates.txt' in gtfs_zipfile.namelist():
        cal_string = gtfs_zipfile.read('calendar_dates.txt').decode('utf-8')
        cal_lines  = cal_string.split('\r\n')
        del cal_lines[0]         # Delete the header row

        for line in cal_lines:
            if len(line.split(',')) < 3:
                continue
            date = line.split(',')[1]
            if date >= date_string:
                valid = True
                break

    gtfs_zipfile.close()
    return valid

if __name__ == "__main__":
    GTFS_FILE = '/var/cache/gtfs/Milwaukee County Transit System/20140401.gtfs'
    TODAY     = '20140331'
    print(valid_dates(GTFS_FILE, TODAY))