OSM2GTFS - a simple scripts to get GTFS from OSM

Some scripts to get GTFS from OSM in area id. Make sure that routes are in OSM PTv2 schema:

• each route has an public_transport:version=2, ref, name tag
• in each route relation the participants ordered like:

stop_entry_only: stop_position
platform_entry_only: platform
stop: stop_position
platform: platform
...
stop_exit_only: stop_position
platform_exit_only: platform
   : road segment 1 by trip continuality
   : road segment 2 by trip continuality
...
   : road segment N by trip continuality

How it works:

1. Get all routes of specified type within area relation id from OSM by overpass-turbo API
2. If subway or commuter specified, get platforms of each route. Else - get stop_positions
3. Rebuild responses to json by schema below.

Note: the procedure can take a very long time due to the shortcomings of the script (fetching route masters, building stop2stop connections)

4. Assembling base GTFS from JSON's

Base GTFS - a GTFS dataset which has no schedules for trips but only their base stop_times from 00:00:00. You need to add or model schedules for public transport by yourself.

osm_grabber

Download the script into your working directory:

from osm_grabber import OSM_Grabber
# Available types are bus, tram, trolleybus, subway, commuter 
# Each type of transport should be initialized separately 
# Enable stop2stop connections if you want to get L-space
trips, stops = OSM_Grabber(type='bus', network=None, operator=None, area=1430616).fetch(s2s=True)
# or save json's in directory
OSM_Grabber(type='bus', network=None, operator=None, area=1430616).fetch(s2s=True, out_dir='kja')

By default it highlights potentially invalid route relation ids if there could be mismatches between stops and platforms counts

The JSON Schema of trips:

[
        {
            'stop_sequence': list, 
            'shape': wkt, 
            'colour': str, 
            'ref': str, 
            'route_id': <relation id>,
            'route_name': str,
            'route_master': <relation id> / NONE
            'route_master_name': str / NONE
            'route_master_ref': str / NONE
        }
]

The JSON Schema of stops:

[
        {
            'stop_id': <nwr id>, 
            'stop_name': str, 
            'stop_shape': wkt, 
            'wheelchair': str yes/no
        }
]

The JSON Schema of s2s:

[
    {
        'from': <stop id>,
        'to': <stop_id>,
        'shape': wkt,
        'length': float,
        'trip_ref': str,
        'trip_id': <relation id>
    }
]

These JSON's can be simply rebuilded into networkx DiGraph, because stops and s2s can be implemented as L-space of transit network.

osm2gtfs

After grabbing OSM data in trips.json, s2s.json, stops.json, download the script into your working directory and run:

from osm2gtfs import OSM2GTFS

tram=OSM2GTFS(trips_f=trips_f, 
            stops_f=stops_f, 
            s2s_f=s2s_f, 
            type='tram', 
            speed=25)
tram.add_stops() # Add stops from tram_stops.json
# /!\ Currently, each trip is an unique route id. It is not properly correct. Trying to solve it
tram.add_trips() # Add trips from tram_trips.json
tram.add_routes() # Add routes from tram_trips.json
# Export GTFS to directory
tram.export(gtfs_dir=r"path/to/gtfs_base_tram")

To validate and combine GTFS datasets, use class GTFS_Dataset:

from osm2gtfs import GTFS_Dataset, combine_gtfs_datasets

tram_feed = GTFS_Dataset('gtfs_base_tram')
metro_feed = GTFS_Dataset('gtfs_base_subway')
commuter_feed = GTFS_Dataset('gtfs_base_comm')
bus_feed = GTFS_Dataset('gtfs_base_bus')

combine_gtfs_datasets([tram_feed, metro_feed, commuter_feed, bus_feed], 'ru-mow')