We use almost all of the GTFS files we obtain from PVTA, and at the moment we do expect them to have certain fields.
We look for a service_id column, and for its entries to contain the letters UMTS. Currently, Bojangles is only interested in UMass Transit Services's bus data. We place no other constraints on this ID, as long the identifiers match those found in trips.txt.
We also look for columns monday, tuesday, etc., which we expect to have boolean values representing whether a vehicle schedule does or does not operate on the given weekday.
Finally, we use the start_date and end_date columns to determine whether a particular vehicle schedule runs on today's example of today's weekday. We expect these dates to be formatted year-month-day, as in 20161027.
We use two columns in this file: stop_name and stop_id. We expect the stop IDs given here to match the stop IDs given in stop_times.txt.
We expect the service_id column to match the service_id column from calendar.txt.
We then use the trip_id column to identify trips, which we expect to match the trip IDs given in stop_times.txt.
We expect the route_id to match the ShortName of the route at the Avail endpoint.
The direction_id and headsign must be present. We have no other constraints on their format.
We expect trip_id and stop_id to match the IDs from trips.txt and stops.txt respectively.
We expect departure_time to be a time in the format H:M:S, as in 16:36:00.
We expect the object which we receive at the configured endpoint to have a RouteDirections object.
We expect each of those RouteDirections to have a route_id matching the route ID at the routes endpoint, and to have a Departures object.
We expect those Departures to be ordered by estimated departure time.
Within each departure we expect an SDT Unix timestamp with a timezone of +0400 or +0500, and we expect a Trip object.
We expect that Trip object to have an InternetServiceDesc which should match the headsign we got from trips.txt.