buses.py

#!/usr/bin/env python3
import enum, json
from sys import stderr
from datetime import date, time, datetime, timedelta
from copy import copy
from misc import *
from geometry import *

def transform_trip_reference(reference):
	return reference

class BusTrip:
	def __init__(self, reference, route, service, length):
		self.reference, self.route, self.service = reference, route, service
		self.length = length
		self.schedule = []
		self.name = transform_trip_reference(reference)
	def __repr__(self):
		return 'all_trips[%r]' % self.name
	def contains_stop(self, stop):
		for halt in self.schedule:
			if halt.stop is stop:
				return halt
		else:
			return None
	def is_served_at(self, day):
		try:
			return self.service in services_for_day[day]
		except KeyError:
			return False
	def concise_schedule(self, starting_stop = None):
		if starting_stop and starting_stop in self.schedule:
			schedule = copy(self.schedule)
			schedule = schedule[schedule.index(starting_stop):]
		else:
			schedule = self.schedule
		used_areas = set()
		result = []
		for halt in schedule:
			stop = halt.stop
			if stop.region and stop.region not in used_areas:
				used_areas.add(stop.region)
				result.append(stop.region)
		return result

class BusRoute:
	def __init__(self, entry):
		self.id = entry['route_id']
		self.reference = entry['route_short_name']
		self.trips = set()
	def __repr__(self):
		return 'routes[%r]' % self.reference

class BusService:
	def __init__(self, reference):
		self.reference = reference
		self.dates = set()
	def __repr__(self):
		return 'services[%r]' % self.reference

class BusStop:
	def __init__(self, reference, name, location, code = None):
		self.reference, self.name, self.location = reference, name, location
		self.code = code or reference
		self.cluster = None
		self.pairs = set() # samannimiset lähellä olevat pysäkit
		self.involved_trips = set()
	def __repr__(self):
		return 'bus_stops[%r]' % self.reference
	def schedule(self, *, max_amount = 50, arrivals = False):
		'''
			Hakee tämän pysäkin seuraavat `määrä` lähtöä. Päätepysäkille saapuvia busseja ei
			lasketa. Palauttaa pysähdykset listana jossa alkiot ovat muotoa (aika, halt),
			jossa:
			- `aika` on saapumishetki muotoa datetime ja
			- `halt` on vastaava BusHalt olio.
			
			Mikäli pysäkille ei ole määrätty riittävästi pysähdyksiä kalenterissa, tuloslista
			jää alimittaiseksi, mahdollisesti jopa tyhjäksi.
		'''
		result = []
		# -1 päivää yövuoroja varten
		date = today() - timedelta(days = 1)
		# Niin kauan kuin aikatauluja ei ole vielä tarpeeksi,
		while len(result) < max_amount:
			try:
				# hae nykyisen päivän aikataulut ja lisää ne,
				result += self.schedule_for_day(date, arrivals = arrivals)
			except ValueError:
				# paitsi jos mentiin kalenterin ulkopuolelle, jolloin lopetetaan,
				break
			# ja siirry seuraavaan päivään.
			date += timedelta(1)
		# Typistä lopputulos haluttuun tulosmäärään.
		return result[:max_amount]
	def schedule_for_day(self, date, *, arrivals = False):
		'''
			Hakee pysäkin aikataulut tiettynä päivänä.
		'''
		# Jos päädyttiin aikataulukalenterin ulkopuolelle, niin tuotetaan virhe. Jos vain
		# palautettaisiin tyhjä result, niin algoritmi jatkaisi etsintää loputtomiin.
		if date > viimeinen_käyttöpäivä:
			raise ValueError('tried to retrieve schedule for date %s which is outside schedule data' % date)
		result = []
		# Jokaiselle ajovuorolle,
		for trip in self.involved_trips:
			# jos tämä ajovuoro ajetaan tänä päivänä
			if trip.is_served_at(date):
				# ja jos tämä trip pysähtyy tällä pysäkillä, ei kuitenkaan saapuen
				# päätepysäkille,
				stop = trip.contains_stop(self)
				if stop and (arrivals or not stop.is_arrival) and stop is not trip.schedule[-1]:
					# ja jos tämä halt on tulevaisuudessa,
					stop_time = datetime.combine(date, time()) + stop.arrival_time
					if stop_time >= now():
						# lisää halt listaan.
						result.append({
							'time': stop_time,
							'trip': trip,
							'stop': stop,
						})
		# Lajittele lopputulos saapumisajan mukaan.
		result.sort(key = lambda schedule_entry: schedule_entry['time'])
		return result

class BusHalt:
	def __init__(self, arrival_time, departure_time, stop, trip, traveled_distance):
		self.arrival_time, self.departure_time, self.stop, self.trip = arrival_time, departure_time, \
			stop, trip
		self.traveled_distance = traveled_distance
	@property
	def is_arrival(self):
		if not hasattr(self, 'cachedIsArrival'):
			if self.stop.region:
				iterator = iter(self.trip.schedule)
				stop = next(iterator)
				while stop is not self:
					stop = next(iterator)
				for stop in iterator:
					if stop.stop.region != self.stop.region:
						self.cachedIsArrival = False
						break
				else:
					self.cachedIsArrival = True
			else:
				self.cachedIsArrival = False
		return self.cachedIsArrival
	def __repr__(self):
		return 'BusHalt(%r, %r, %r, %r)' % (self.arrival_time, self.departure_time, self.stop, self.trip)

routes = {}
routes_per_id = {}
all_trips = {}
services = {}
bus_stops = {}
all_clusters = set()
viimeinen_käyttöpäivä = None
clusters_by_name = {}
services_for_day = {}

def load_buses(gtfs_zip_path, profile):
	global viimeinen_käyttöpäivä
	from zipfile import ZipFile
	with ZipFile(gtfs_zip_path) as gtfs_zip:
		print('Ladataan linjat... ', file = stderr, end = '', flush = True)
		with gtfs_zip.open('routes.txt') as file:
			for row in read_csv(map(bytes.decode, file)):
				route = BusRoute(row)
				routes[route.reference] = route
				routes_per_id[route.id] = route
		print('%d linjaa' % len(routes), file = stderr)

		print('Ladataan ajovuorot... ', file = stderr, end = '', flush = True)

		shape_distances = {}
		with gtfs_zip.open('shapes.txt') as file:
			for row in read_csv(map(bytes.decode, file)):
				shape_distances[row['shape_id']] = max(shape_distances.get(row['shape_id'], 0), float(row['shape_dist_traveled']))

		with gtfs_zip.open('trips.txt') as file:
			for row in read_csv(map(bytes.decode, file)):
				if row['service_id'] not in services:
					services[row['service_id']] = BusService(row['service_id'])
				route = routes_per_id[row['route_id']]
				trip = BusTrip(
					reference = row['trip_id'],
					route = route,
					service = services[row['service_id']],
					length = shape_distances[row['shape_id']] * float(profile['metrics']['shape-modifier'])
				)
				route.trips.add(trip)
				assert trip.name not in all_trips
				all_trips[trip.name] = trip
		print('%d ajoa' % len(all_trips), file = stderr)

		def read_date(teksti):
			return date(int(teksti[:4]), int(teksti[4:6]), int(teksti[6:]))

		def read_time(teksti):
			tunti, minuutti, sekunti = map(int, teksti.split(':'))
			return timedelta(hours = tunti, minutes = minuutti, seconds = sekunti)

		print('Ladataan päiväykset... ', file = stderr, flush = True)

		viimeinen_käyttöpäivä = date.today()

		def date_range(start_date, end_date, *, include_end = False):
			''' Generates date from start_date to end_date. If include_end is True, then end_date will be yielded. '''
			current_date = start_date
			while current_date < end_date:
				yield current_date
				current_date += timedelta(1)
			if include_end:
				yield end_date

		def add_day_to_service(service_name, day):
			try:
				service = services[service_name]
			except KeyError:
				return
			else:
				service.dates.add(day)
				if day not in services_for_day:
					services_for_day[day] = set()
				services_for_day[day].add(service)
				global viimeinen_käyttöpäivä
				viimeinen_käyttöpäivä = max(day, viimeinen_käyttöpäivä)

		def filter_day(row, day):
			day_names = ['monday', 'tuesday', 'wednesday', 'thursday', 'friday', 'saturday', 'sunday']
			return int(row[day_names[day.isoweekday() - 1]])

		with gtfs_zip.open('calendar.txt') as file:
			for row in read_csv(map(bytes.decode, file)):
				for day in date_range(read_date(row['start_date']), read_date(row['end_date']), include_end = True):
					if filter_day(row, day):
						add_day_to_service(service_name = row['service_id'], day = day)

		with gtfs_zip.open('calendar_dates.txt') as file:
			for row in read_csv(map(bytes.decode, file)):
				add_day_to_service(service_name = row['service_id'], day = read_date(row['date']))

		def services_available_at(day):
			for service in services.values():
				if day in service.dates:
					yield service

		print('Ladataan pysäkit... ', file = stderr, end = '', flush = True)
		with gtfs_zip.open('stops.txt') as file:
			for row in read_csv(map(bytes.decode, file)):
				location = Sijainti(float(row['stop_lat']), float(row['stop_lon']))
				stop = BusStop(
					reference = row['stop_id'],
					name = row['stop_name'],
					location = location, 
					code = row['stop_code'],
				)
				bus_stops[stop.reference] = stop
		with open('regions-per-stop.json') as file:
			for stop_reference, region in json.load(file).items():
				bus_stops[stop_reference].region = region
		print('%d pysäkkiä' % len(bus_stops), file = stderr)


		class BusStopCluster:
			def __init__(self):
				self.stops = set()
				self._center = None
				self.name = None
			@property
			def url_name(self):
				return self.name.lower().replace('(', '').replace(')', '').replace(' ', '-')
			def add_stop(self, stop):
				assert not stop.cluster
				stop.cluster = self
				self.stops.add(stop)
				self._center = None
			@property
			def center(self):
				if not self._center:
					if self.stops:
						from statistics import median
						pointtype = type(next(iter(self.stops)).location)
						self._center = pointtype(
							median(stop.location.x for stop in self.stops),
							median(stop.location.y for stop in self.stops),
						)
					else:
						raise ValueError('an empty cluster has no center point')
				return self._center
			def merge(self, other):
				for bus_stop in other.stops:
					bus_stop.cluster = self
				self.stops |= other.stops
				other.stops = set()
				other._center = None
			def schedule(self, *, max_amount = 50):
				result = []
				for stop in self.stops:
					result += stop.schedule(max_amount = max_amount)
				result.sort(key = lambda schedule_entry: schedule_entry['time'])
				return result[:max_amount]

		from collections import defaultdict
		bus_stops_by_name = defaultdict(set)
		for bus_stop in bus_stops.values():
			bus_stops_by_name[bus_stop.name].add(bus_stop)
		bus_stops_by_name = dict(bus_stops_by_name)

		# ryhmittele bus_stops nimen mukaan
		global all_clusters
		all_clusters = []
		def cluster_bus_stops():
			sorted_bus_stops = sorted(bus_stops.values(), key = lambda bus_stop: bus_stop.name)
			for bus_stop in sorted_bus_stops:
				if not bus_stop.cluster:
					stops_to_cluster = {bus_stop}
					# etsi pysäkin samannimiset vastaparit
					for pair_candidate in bus_stops_by_name[bus_stop.name]:
						distance = pair_candidate.location.etäisyys(bus_stop.location)
						if pair_candidate is not bus_stop and distance <= 0.4:
							stops_to_cluster.add(pair_candidate)
					for stop_to_cluster in stops_to_cluster:
						if stop_to_cluster.cluster:
							cluster = stop_to_cluster.cluster
							break
					else:
						cluster = BusStopCluster()
						all_clusters.append(cluster)
					for stop_to_cluster in stops_to_cluster:
						if not stop_to_cluster.cluster:
							cluster.add_stop(stop_to_cluster)
			# Merkitse muistiin pysäkkien vastaparit käyttäen hyväksi tämänhetkistä ryhmittelytietoa
			for bus_stop in bus_stops.values():
				if bus_stop.cluster:
					bus_stop.pairs = bus_stop.cluster.stops - {bus_stop}
			# Ryhmitä ne bus_stops, joilla ei ollut omaa vastaparia, muiden pysäkkien kanssa
			for bus_stop in sorted_bus_stops:
				if len(bus_stop.cluster.stops) == 1:
					possibilities = set()
					for cluster in all_clusters:
						if cluster is not bus_stop.cluster:
							distance = cluster.center.etäisyys(bus_stop.location)
							if distance <= 0.4:
								possibilities.add((distance, cluster))
					if possibilities:
						best = min(possibilities)[1]
						all_clusters.remove(bus_stop.cluster)
						best.merge(bus_stop.cluster)

		def shared_elements_in_n_sets(sets):
			from itertools import combinations
			result = set()
			for pair in combinations(sets, 2):
				result |= pair[0] & pair[1]
			return result

		def name_clusters():
			from collections import defaultdict
			from pprint import pprint
			clusters_per_name = defaultdict(set)
			for cluster in all_clusters:
				name_representing_stop = min((len(stop.reference), stop.reference, stop) for stop in cluster.stops)[2]
				clusters_per_name[name_representing_stop.name].add(cluster)
			for name, clusters in clusters_per_name.items():
				if len(clusters) == 1:
					# Ryhmä on ainoa jolla on varaus tälle nimelle. Sen kuin vaan.
					next(iter(clusters)).name = name
				else:
					# Olisiko kaikki klusterit eri alueilla?
					common_regions = shared_elements_in_n_sets({stop.region for stop in cluster.stops} for cluster in clusters)
					# Esitys: ryhmä -> ne alueet jotka ovat tälle ryhmälle ainutlaatuisia
					proposal = {
						cluster: {stop.region for stop in cluster.stops} - common_regions - {None}
						for cluster in clusters
					}
					# Jos enintään yksi klusteri tässä esityksessä on kokonaan ilman omaa aluetta, jolla se voisi eritellä,
					# niin nimetään klusterit näiden alueiden mukaan.
					# Se klusteri jolla ei ole omaa aluetta (jos on) jätetään ilman aluepäätettä.
					if sum([1 for unique_areas in proposal.values() if not unique_areas]) <= 1:
						for cluster, unique_areas in proposal.items():
							individual_cluster_name = name
							if unique_areas:
								individual_cluster_name += ' (' + min(unique_areas) + ')'
							cluster.name = individual_cluster_name
					else:
						# Typerä reunatapaus. Indeksoidaan numeroin...
						for n, (_, cluster) in enumerate(sorted(
							min((stop.reference.lower(), cluster) for stop in cluster.stops)
							for cluster in clusters
						), 1):
							individual_cluster_name = name + '-' + str(n)
							cluster.name = individual_cluster_name

		print('Ryhmitellään pysäkit...')
		cluster_bus_stops()
		name_clusters()

		for cluster in all_clusters:
			if cluster.url_name in clusters_by_name:
				print('Warning: Clusters %r and %r share the same URL name: %r' % (cluster.name, clusters_by_name[cluster.url_name].name, cluster.url_name))
			else:
				clusters_by_name[cluster.url_name] = cluster

		print('Ladataan aikataulut... ', end = '', flush = True, file = stderr)
		with gtfs_zip.open('stop_times.txt') as file:
			row_count = sum(line.count(b'\n') for line in file)
		with gtfs_zip.open('stop_times.txt') as file:
			progress = 0
			for row in read_csv(map(bytes.decode, file)):
				trip = all_trips[transform_trip_reference(row['trip_id'])]
				arrival_time = read_time(row['arrival_time'])
				departure_time = read_time(row['departure_time'])
				stop = bus_stops[row['stop_id']]
				traveled_distance = float(row['shape_dist_traveled']) * float(profile['metrics']['shape-modifier'])
				trip.schedule.append(BusHalt(arrival_time, departure_time, stop, trip, traveled_distance))
				stop.involved_trips.add(trip)
				progress += 1
				if progress % 1000 == 0:
					print('\rLadataan aikataulut... %.1f%%' % (progress * 100 / row_count), end = ' ', file = stderr)
		print('\rLadataan aikataulut... ladattu', file = stderr)

		for trip in all_trips.values():
			from busroute import simplify_name
			schedule = trip.concise_schedule()
			try:
				trip.from_place = simplify_name(schedule[0])
				trip.to_place = simplify_name(schedule[-1])
			except IndexError:
				trip.from_place = ''
				trip.to_place = ''

		for route in routes.values():
			from collections import Counter
			from busroute import simplify_name
			tally = Counter()
			for trip in route.trips:
				schedule = trip.concise_schedule()
				places = set(schedule)
				do_add = True
				assert type(schedule) is list
				for candidate in tally:
					if places.issubset(set(candidate)):
						do_add = False
						tally.update({tuple(candidate)})
				if do_add:
					tally.update({tuple(schedule)})
			try:
				most_common_route = tally.most_common(1)[0][0]
				route.description = simplify_name(most_common_route[0]) + ' - ' + simplify_name(most_common_route[-1])
			except:
				route.description = ''
			route.trips = sorted(route.trips, key = lambda trip: trip.schedule[0].departure_time)

		# Fölin datassa on jotain tosi kummaa. Ilmeisesti ajovuoron viimeisen pysähdyksen saapumisaika on ihan täysin
		# väärin. Arvaan että se on seuraavan lähdön aika, mutta joka tapauksessa se on väärin.
		# Arvataan mikä se todellinen saapumisaika on. Se ei voi mennä kauhean paljon pahemmin vikaan kuin alkuperäinen
		# väärin oleva data.
		for trip in all_trips.values():
			bus_speed_coefficient = 750 # metriä minuutissa
			last_leg_distance = trip.schedule[-1].traveled_distance - trip.schedule[-2].traveled_distance
			trip.schedule[-1].arrival_time = trip.schedule[-2].departure_time + timedelta(minutes = last_leg_distance / bus_speed_coefficient)

if __name__ == '__main__':
	from configparser import ConfigParser
	profile = ConfigParser()
	profile.read('profiles/föli.ini')
	load_buses('gtfs.zip', profile)