def test_broken(self):
exception = False
try:
clear_mappers()
dao = Dao("")
dao.load_gtfs(BROKEN_GTFS, lenient=False)
except KeyError:
exception = True
self.assertTrue(exception)
clear_mappers()
dao = Dao("")
dao.load_gtfs(BROKEN_GTFS, lenient=True)
# The following are based on BROKEN GTFS content,
# that is the entities count minus broken ones.
self.assertTrue(len(dao.routes()) == 4)
self.assertTrue(len(list(dao.stops())) == 12)
self.assertTrue(len(dao.calendars()) == 2)
self.assertTrue(len(list(dao.trips())) == 104)
self.assertTrue(len(dao.stoptimes()) == 500)
self.assertTrue(len(dao.fare_attributes()) == 2)
self.assertTrue(len(dao.fare_rules()) == 4)
# This stop has missing coordinates in the broken file
stop00 = dao.stop('FUR_CREEK_RES3')
self.assertAlmostEqual(stop00.stop_lat, 0.0, 5)
self.assertAlmostEqual(stop00.stop_lon, 0.0, 5)
def test_gtfs_data(self):
dao = Dao(DAO_URL, sql_logging=SQL_LOG)
dao.load_gtfs(MINI_GTFS)
# Check feed
feed = dao.feed()
self.assertTrue(feed.feed_id == "")
self.assertTrue(feed.feed_publisher_name is None)
self.assertTrue(feed.feed_publisher_url is None)
self.assertTrue(feed.feed_contact_email is None)
self.assertTrue(feed.feed_contact_url is None)
self.assertTrue(feed.feed_start_date is None)
self.assertTrue(feed.feed_end_date is None)
self.assertTrue(len(dao.agencies()) == 1)
self.assertTrue(len(dao.routes()) == 1)
self.assertTrue(len(feed.agencies) == 1)
self.assertTrue(len(feed.routes) == 1)
# Check if optional route agency is set
a = dao.agency("A")
self.assertTrue(a is not None)
self.assertTrue(len(a.routes) == 1)
# Check for frequency-generated trips
# They should all have the same delta
trips = dao.trips(fltr=(Trip.frequency_generated == True),
prefetch_stop_times=True)
n_trips = 0
deltas = {}
for trip in trips:
original_trip_id = trip.trip_id.rsplit('@', 1)[0]
delta1 = []
for st1, st2 in trip.hops():
delta1.append(st2.arrival_time - st1.departure_time)
delta2 = deltas.get(original_trip_id)
if delta2 is not None:
self.assertTrue(delta1 == delta2)
else:
deltas[original_trip_id] = delta1
n_trips += 1
self.assertTrue(n_trips == 8)
def test_non_overlapping_feeds(self):
dao = Dao(DAO_URL, sql_logging=SQL_LOG)
# Load twice the same data under two distinct namespaces
dao.load_gtfs(DUMMY_GTFS, feed_id='A')
dao.load_gtfs(DUMMY_GTFS, feed_id='B')
# Check that each feed only return it's own data
feed_a = dao.feed('A')
self.assertTrue(feed_a.feed_id == 'A')
feed_b = dao.feed('B')
self.assertTrue(feed_b.feed_id == 'B')
self.assertTrue(len(dao.agencies()) == 4)
self.assertTrue(len(feed_a.agencies) == 2)
self.assertTrue(len(feed_b.agencies) == 2)
self.assertTrue(len(feed_a.routes) * 2 == len(dao.routes()))
self.assertTrue(len(feed_b.routes) * 2 == len(dao.routes()))
self.assertTrue(len(feed_a.stops) * 2 == len(list(dao.stops())))
self.assertTrue(len(feed_b.stops) * 2 == len(list(dao.stops())))
self.assertTrue(len(feed_a.calendars) * 2 == len(dao.calendars()))
self.assertTrue(len(feed_b.calendars) * 2 == len(dao.calendars()))
self.assertTrue(len(feed_a.trips) * 2 == len(list(dao.trips())))
self.assertTrue(len(feed_b.trips) * 2 == len(list(dao.trips())))
class TestProviders(unittest.TestCase):
def setUp(self):
unittest.TestCase.setUp(self)
clear_mappers()
self.dao_object = Dao()
@mock.patch("exporter.api.radcom.Request")
def test_radcom_provider(self, mocked_request):
mocked_request.side_effect = _request_mock
radcomProvider = RadcomApiDataProvider("", feed_id="1")
assert radcomProvider is not None
# load data from provider
load_result = radcomProvider.load_data_source(self.dao_object)
assert load_result is True
# check calendars
assert len(self.dao_object.calendars()) == 2
# check routes
routes_len = 0
for route in self.dao_object.routes():
routes_len = routes_len + 1
assert route.route_id == '43'
assert route.route_short_name == '1'
assert route.route_long_name == 'Bulevardul Tomis - Sere (C.L.)'
assert routes_len == 1
# check stops
stops_len = 0
for stop in self.dao_object.stops():
stops_len = stops_len + 1
assert len(stop.stop_times) > 0
assert stop.stop_lat != 0.0
assert stop.stop_lon != 0.0
assert stops_len == 35
# check trips
trips_len = 0
for trip in self.dao_object.trips():
trips_len = trips_len + 1
assert trip.shape != None
expected_shape_id = "shp1_{0}_{1}".format(trip.route_id,
trip.direction_id)
assert trip.shape_id == expected_shape_id
assert trips_len == 14
# cleanup
self.dao_object.delete_feed('1')
self.dao_object.flush()
def test_brasov_export(self):
provider = BrasovApiDataProvider("2")
provider.dao = self.dao_object
provider._load_agencies()
assert provider.agency_id is not None
assert len(self.dao_object.agencies()) == 1
provider._load_service()
assert len(self.dao_object.calendars()) == 2
assert len(provider.routes) == 0
provider._load_routes()
assert len(provider.routes) > 0
assert len(self.dao_object.routes()) > 0
assert len(provider.stop_map) == 0
provider._load_schedule()
assert len(provider.stop_map) > 0
assert len(list(self.dao_object.stops())) > 0
assert len(list(self.dao_object.trips())) > 0
# stops_visited_by_route_trips = len(list(dao_fixture.stops()))
# provider._load_stops()
# assert len(list(dao_fixture.stops())) == stops_visited_by_route_trips
# cleanup
self.dao_object.delete_feed('2')
self.dao_object.flush()
def test_gtfs_data(self):
dao = Dao(DAO_URL, sql_logging=False)
dao.load_gtfs(DUMMY_GTFS)
# Check feed
feed = dao.feed()
self.assertTrue(feed.feed_id == "")
self.assertTrue(feed.feed_publisher_name == "Mecatran")
self.assertTrue(feed.feed_publisher_url == "http://www.mecatran.com/")
self.assertTrue(feed.feed_contact_email == "*****@*****.**")
self.assertTrue(feed.feed_lang == "fr")
self.assertTrue(len(dao.agencies()) == 2)
self.assertTrue(len(dao.routes()) == 3)
self.assertTrue(len(feed.agencies) == 2)
self.assertTrue(len(feed.routes) == 3)
# Check agencies
at = dao.agency("AT")
self.assertTrue(at.agency_name == "Agency Train")
self.assertTrue(len(at.routes) == 1)
ab = dao.agency("AB")
self.assertTrue(ab.agency_name == "Agency Bus")
self.assertTrue(len(ab.routes) == 2)
# Check calendars
week = dao.calendar("WEEK")
self.assertTrue(len(week.dates) == 253)
summer = dao.calendar("SUMMER")
self.assertTrue(len(summer.dates) == 42)
mon = dao.calendar("MONDAY")
self.assertTrue(len(mon.dates) == 49)
sat = dao.calendar("SAT")
self.assertTrue(len(sat.dates) == 53)
for date in mon.dates:
self.assertTrue(date.dow() == 0)
for date in sat.dates:
self.assertTrue(date.dow() == 5)
for date in week.dates:
self.assertTrue(date.dow() >= 0 and date.dow() <= 4)
for date in summer.dates:
self.assertTrue(date >= CalendarDate.ymd(2016, 7, 1)
and date <= CalendarDate.ymd(2016, 8, 31))
empty = dao.calendars(
func.date(CalendarDate.date) == datetime.date(2016, 5, 1))
# OR USE: empty = dao.calendars(CalendarDate.date == "2016-05-01")
self.assertTrue(len(empty) == 0)
july4 = CalendarDate.ymd(2016, 7, 4)
summer_mon = dao.calendars(func.date(CalendarDate.date) == july4.date)
n = 0
for cal in summer_mon:
self.assertTrue(july4 in cal.dates)
n += 1
self.assertTrue(n == 3)
# Check stops
sbq = dao.stop("BQ")
self.assertAlmostEqual(sbq.stop_lat, 44.844, places=2)
self.assertAlmostEqual(sbq.stop_lon, -0.573, places=2)
self.assertTrue(sbq.stop_name == "Bordeaux Quinconces")
n = 0
for stop in dao.stops(Stop.stop_name.like("Gare%")):
self.assertTrue(stop.stop_name.startswith("Gare"))
n += 1
self.assertTrue(n == 7)
n = 0
for stop in dao.stops(
fltr=dao.in_area(RectangularArea(44.7, -0.6, 44.9, -0.4))):
self.assertTrue(stop.stop_lat >= 44.7 and stop.stop_lat <= 44.9
and stop.stop_lon >= -0.6
and stop.stop_lon <= -0.4)
n += 1
self.assertTrue(n == 16)
for station in dao.stops(Stop.location_type == Stop.TYPE_STATION):
self.assertTrue(station.location_type == Stop.TYPE_STATION)
self.assertTrue(len(station.sub_stops) >= 2)
for stop in station.sub_stops:
self.assertTrue(stop.parent_station == station)
# Check zones
z_inexistant = dao.zone("ZX")
self.assertTrue(z_inexistant is None)
z1 = dao.zone("Z1")
self.assertEqual(16, len(z1.stops))
z2 = dao.zone("Z2")
self.assertEqual(4, len(z2.stops))
# Check transfers
transfers = dao.transfers()
self.assertTrue(len(transfers) == 3)
transfers = dao.transfers(
fltr=(dao.transfer_from_stop().stop_id == 'GBSJB'))
self.assertTrue(len(transfers) == 1)
self.assertTrue(transfers[0].from_stop.stop_id == 'GBSJB')
# Check routes
tgv = dao.route("TGVBP")
self.assertTrue(tgv.agency == at)
self.assertTrue(tgv.route_type == 2)
r1 = dao.route("BR")
self.assertTrue(r1.route_short_name == "R1")
self.assertTrue(r1.route_long_name == "Bus Red")
n = 0
for route in dao.routes(Route.route_type == 3):
self.assertTrue(route.route_type == 3)
n += 1
self.assertTrue(n == 2)
# Check trip for route
n = 0
trips = dao.trips(fltr=Route.route_type == Route.TYPE_BUS)
for trip in trips:
self.assertTrue(trip.route.route_type == Route.TYPE_BUS)
n += 1
self.assertTrue(n > 20)
# Check trips on date
trips = dao.trips(fltr=func.date(CalendarDate.date) == july4.date,
prefetch_calendars=True)
n = 0
for trip in trips:
self.assertTrue(july4 in trip.calendar.dates)
n += 1
self.assertTrue(n > 30)
def test_complex_queries(self):
dao = Dao(DAO_URL, sql_logging=SQL_LOG)
dao.load_gtfs(DUMMY_GTFS)
# Get the list of departures:
# 1) from "Porte de Bourgogne"
# 2) on 4th July
# 3) between 10:00 and 14:00
# 4) on route type BUS
# 5) not the last of trip (only departing)
porte_bourgogne = dao.stop("BBG")
july4 = CalendarDate.ymd(2016, 7, 4)
from_time = gtfstime(10, 00)
to_time = gtfstime(14, 00)
departures = dao.stoptimes(
fltr=(StopTime.stop == porte_bourgogne) &
(StopTime.departure_time >= from_time) &
(StopTime.departure_time <= to_time)
& (Route.route_type == Route.TYPE_BUS) &
(func.date(CalendarDate.date) == july4.date),
prefetch_trips=True)
n = 0
for dep in departures:
self.assertTrue(dep.stop == porte_bourgogne)
self.assertTrue(july4 in dep.trip.calendar.dates)
self.assertTrue(dep.trip.route.route_type == Route.TYPE_BUS)
self.assertTrue(dep.departure_time >= from_time
and dep.departure_time <= to_time)
n += 1
self.assertTrue(n > 10)
# Plage is a stop that is used only in summer (hence the name!)
plage = dao.stop("BPG")
# Get the list of stops used by some route:
# 1) All-year round
route_red = dao.route("BR")
stoplist_all = list(dao.stops(fltr=Trip.route == route_red))
# 2) Only in january
from_date = CalendarDate.ymd(2016, 1, 1)
to_date = CalendarDate.ymd(2016, 1, 31)
stoplist_jan = list(
dao.stops(fltr=(Trip.route == route_red)
& (func.date(CalendarDate.date) >= from_date.date)
& (func.date(CalendarDate.date) <= to_date.date)))
# Now, make some tests
self.assertTrue(len(stoplist_all) > 5)
self.assertTrue(plage in stoplist_all)
self.assertFalse(plage in stoplist_jan)
stoplist = list(stoplist_all)
stoplist.remove(plage)
self.assertTrue(set(stoplist) == set(stoplist_jan))
# Get all routes passing by the set of stops
routes = dao.routes(fltr=or_(StopTime.stop == stop
for stop in stoplist_jan))
stopset = set()
for route in routes:
for trip in route.trips:
for stoptime in trip.stop_times:
stopset.add(stoptime.stop)
self.assertTrue(set(stoplist_jan).issubset(stopset))
def test_demo(self):
dao = Dao(DAO_URL, sql_logging=False)
dao.load_gtfs(DUMMY_GTFS)
print("List of stops named '...Bordeaux...':")
stops_bordeaux = list(
dao.stops(fltr=(Stop.stop_name.ilike('%Bordeaux%'))
& (Stop.location_type == Stop.TYPE_STOP)))
for stop in stops_bordeaux:
print(stop.stop_name)
print("List of routes passing by those stops:")
routes_bordeaux = dao.routes(fltr=or_(StopTime.stop == stop
for stop in stops_bordeaux))
for route in routes_bordeaux:
print("%s - %s" % (route.route_short_name, route.route_long_name))
july4 = CalendarDate.ymd(2016, 7, 4)
print("All departures from those stops on %s:" % (july4.as_date()))
departures = list(
dao.stoptimes(fltr=(or_(StopTime.stop == stop
for stop in stops_bordeaux))
& (StopTime.departure_time != None)
& (func.date(CalendarDate.date) == july4.date)))
print("There is %d departures" % (len(departures)))
for departure in departures:
print("%30.30s %10.10s %-20.20s > %s" %
(departure.stop.stop_name, fmttime(departure.departure_time),
departure.trip.route.route_long_name,
departure.trip.trip_headsign))
print("Number of departures and time range per stop on %s:" %
(july4.as_date()))
departure_by_stop = defaultdict(list)
for departure in departures:
departure_by_stop[departure.stop].append(departure)
for stop, deps in departure_by_stop.items():
min_dep = min(d.departure_time for d in deps)
max_dep = max(d.departure_time for d in deps)
print("%30.30s %3d departures (from %s to %s)" %
(stop.stop_name, len(deps), fmttime(min_dep),
fmttime(max_dep)))
# Compute the average distance and time to next stop by route type
ntd = [[0, 0, 0.0] for type in range(0, Route.TYPE_FUNICULAR + 1)]
for departure in departures:
# The following is guaranteed to succeed as we have departure_time == Null for last stop time in trip
next_arrival = departure.trip.stop_times[departure.stop_sequence +
1]
hop_dist = next_arrival.shape_dist_traveled - departure.shape_dist_traveled
hop_time = next_arrival.arrival_time - departure.departure_time
route_type = departure.trip.route.route_type
ntd[route_type][0] += 1
ntd[route_type][1] += hop_time
ntd[route_type][2] += hop_dist
for route_type in range(0, len(ntd)):
n, t, d = ntd[route_type]
if n > 0:
print(
"The average distance to the next stop on those departures for route type %d is %.2f meters"
% (route_type, d / n))
print(
"The average time in sec to the next stop on those departures for route type %d is %s"
% (route_type, fmttime(t / n)))
class Exporter:
def __init__(self, arguments):
self.logger = logging.getLogger('gtfsexporter')
self._arguments = arguments
if arguments['--id'] is None:
arguments['--id'] = "default"
database_path = os.path.join(__cwd_path__,
arguments['--id'] + ".sqlite")
self._dao = Dao(database_path,
sql_logging=arguments['--logsql'],
schema=arguments['--schema'])
if arguments['--list']:
for feed in self._dao.feeds():
print(feed.feed_id if feed.feed_id != "" else "(default)")
if arguments['--delete']:
feed_id = arguments['--id']
existing_feed = self._dao.feed(feed_id)
if existing_feed:
self.logger.warning("Deleting existing feed ID '%s'" % feed_id)
self._dao.delete_feed(feed_id)
self._dao.commit()
@property
def dao(self):
return self._dao
def load(self, provider: DataProvider):
self.logger.info("Importing data from provided source")
provider.load_data_source(self._dao)
def process(self, processor: Processor = None):
self.logger.info("Processing data from provided source")
# Here we should use a rule processor to have more flexibility when processing data
# processor.process(ruleset)
for route in self._dao.routes():
print("updating route [%s] setting correct color" %
route.route_long_name)
route.route_text_color = "FFFFFF"
if route.route_type == Route.TYPE_BUS:
route.route_color = "195BAD"
elif route.route_type == Route.TYPE_TRAM:
route.route_color = "FFAD33"
elif route.route_type == Route.TYPE_RAIL:
route.route_color = "FF5B33"
elif route.route_type == Route.TYPE_CABLECAR:
route.route_color = "FF8433"
elif route.route_type == Route.TYPE_SUBWAY:
route.route_color = "D13333"
elif route.route_type == Route.TYPE_FERRY:
route.route_color = "62A9DD"
self._dao.commit()
def export(self, bundle=False, out_path: str = __output_path__) -> str:
self.logger.info(
f"Generating archive with name gtfs-{self._arguments['--id']}.zip")
class __Args:
filter = None
context = Context(self._dao, __Args(), out_path)
if bundle:
w = Writer(context, bundle=f"gtfs-{self._arguments['--id']}.zip")
else:
w = Writer(context)
w.run()
return out_path