This is a set of Python utilities that I use to deal with real-time NYC subway data.
I usually want to know when trains are going to depart a specific stop along a specific train line, so right now the tools are mostly for that. But I tried to write them to support arbitrary functionality.
pip install underground
Or if you'd like to live dangerously:
pip install git+https://github.com/nolanbconaway/underground.git#egg=underground
To request data from the MTA, you'll also need a free API key. Register here.
Once you have your API key, use the Python API like:
import os
from underground import metadata, SubwayFeed
API_KEY = os.getenv('MTA_API_KEY')
ROUTE = 'Q'
feed = SubwayFeed.get(ROUTE, api_key=API_KEY)
# request will read from $MTA_API_KEY if a key is not provided
feed = SubwayFeed.get(ROUTE)
# under the hood, the Q route is mapped to a URL. This call is equivalent:
URL = 'https://api-endpoint.mta.info/Dataservice/mtagtfsfeeds/nyct%2Fgtfs-nqrw'
feed = SubwayFeed.get(URL)
# or
URL = metadata.resolve_url(ROUTE)
feed = SubwayFeed.get(URL)
feed.extract_stop_dict
will return a dictionary of dictionaries, like:
>>> feed.extract_stop_dict()
{
"route_1": {
"stop_1": [datetime.datetime(...), datetime.datetime(...)],
"stop_2": [datetime.datetime(...), datetime.datetime(...)],
...
},
"route_2": {
"stop_1": [datetime.datetime(...), datetime.datetime(...)],
"stop_2": [datetime.datetime(...), datetime.datetime(...)],
...
}
}
The underground
command line tool is also installed with the package.
$ underground feed --help
Usage: underground feed [OPTIONS] ROUTE_OR_URL
Request an MTA feed via a route or URL.
ROUTE_OR_URL may be either a feed URL or a route (which will be used to
look up the feed url).
Examples (both access the same feed):
underground feed Q --json > feed_nrqw.json
URL='https://api-endpoint.mta.info/Dataservice/mtagtfsfeeds/nyct%2Fgtfs-nqrw' &&
underground feed $URL --json > feed_nrqw.json
Options:
--api-key TEXT MTA API key. Will be read from $MTA_API_KEY if not
provided.
--json Option to output the feed data as JSON. Otherwise
output will be bytes.
-r, --retries INTEGER Retry attempts in case of API connection failure.
Default 100.
--help Show this message and exit.
$ underground stops --help
Usage: underground stops [OPTIONS] [H|M|D|1|Z|A|N|GS|SI|J|G|Q|L|B|R|F|E|2|7|W|
6|4|C|5|FS]
Print out train departure times for all stops on a subway line.
Options:
-f, --format TEXT strftime format for stop times. Use `epoch` for a
unix timestamp.
-r, --retries INTEGER Retry attempts in case of API connection failure.
Default 100.
--api-key TEXT MTA API key. Will be read from $MTA_API_KEY if not
provided.
-t, --timezone TEXT Output timezone. Ignored if --epoch. Default to NYC
time.
-s, --stalled-timeout INTEGER Number of seconds between the last movement
of a train and the API update before
considering a train stalled. Default is 90 as
recommended by the MTA. Numbers less than 1
disable this check.
--help Show this message and exit.
Stops are printed to stdout in the format stop_id t1 t2 ... tn
.
$ export MTA_API_KEY='...'
$ underground stops Q | tail -2
Q05S 19:01 19:09 19:16 19:25 19:34 19:44 19:51 19:58
Q04S 19:03 19:11 19:18 19:27 19:36 19:46 19:53 20:00
If you know your stop id (stop IDs can be found in stops.txt), you can grep the results:
$ underground stops Q | grep Q05S
Q05S 19:09 19:16 19:25 19:34 19:44 19:51 19:58
If you don't know your stop, see below for a handy tool!
$ underground findstops --help
Usage: underground findstops [OPTIONS] QUERY...
Find your stop ID.
Query a location and look for your stop ID, like:
$ underground findstops parkside av
Options:
--json Option to output the data as JSON. Otherwise will be human readable
table.
--help Show this message and exit.
Enter the name of your stop and a table of stops with matching names will be returned.
$ underground findstops parkside
ID: D27N Direction: NORTH Lat/Lon: 40.655292, -73.961495 Name: PARKSIDE AV
ID: D27S Direction: SOUTH Lat/Lon: 40.655292, -73.961495 Name: PARKSIDE AV
Some names are ambiguous (try "fulton st"), for these you'll have to dig into the metadata more carefully.