def parse_schedules_trolley(trolley_list):
for trolley in trolley_list:
logging.info("Parsing schedule for trolley with id: %s", trolley["id"])
print "Current trolley_id is:", trolley["id"]
html = parser_utils.download_page(
TransportParser.TRANSPORT_PAGE_URL,
trolley["post_data"]
)
schedule = ScheduleParser.parse_html(html)
json_trolley_object = {
'type': 'trolley',
'id': trolley["id"],
'name': trolley["name"],
'workdays': False,
'weekend': False,
'everyday': False
}
# in trolley case ['workdays'] always True
if trolley['workdays'] is True:
# case, when trolley works only on workdays
if schedule['workdays'] is True:
print "Workdays only"
json_trolley_object['workdays'] = True
json_trolley_object['schedule_workdays'] = (
schedule['schedule_table']
)
json_trolley_object['stations_workdays'] = (
schedule['stations_list']
)
# regular case, when trolley works both on workdays and weekend
else:
json_trolley_object['workdays'] = True
json_trolley_object['schedule_workdays'] = (
schedule['schedule_table']
)
json_trolley_object['stations_workdays'] = (
schedule['stations_list']
)
if schedule['weekend'] is True:
json_trolley_object['weekend'] = True
html = parser_utils.download_page(
schedule['weekend_link']
)
schedule_weekend = ScheduleParser.parse_html(html)
json_trolley_object['schedule_weekend'] = (
schedule_weekend['schedule_table']
)
json_trolley_object['stations_weekend'] = (
schedule_weekend['stations_list']
)
parser_utils.save_json_file(
(parser_configs.directories["TROLLEYS_DIR"] +
trolley["id"] + ".json"),
json_trolley_object
)
def parse_schedules_suburban():
html = parser_utils.download_page(
SuburbanParser.SUBURBAN_TRANSPORT_PAGE_URL
)
suburban_buses = SuburbanParser.parse_html(html)
for bus in suburban_buses:
from_city = in_dict(bus, 'from_city')
if from_city:
# print bus['number']
from_city = no_whitespaces(from_city)
schedule_from_city = SuburbanScheduleParser.parse_schedule(
from_city
)
bus['schedule_from_city'] = schedule_from_city
if bus['number'] == '159':
print schedule_from_city
to_city = in_dict(bus, 'to_city')
if to_city:
# print bus['number']
to_city = no_whitespaces(to_city)
schedule_to_city = SuburbanScheduleParser.parse_schedule(
to_city
)
bus['schedule_to_city'] = schedule_to_city
if bus['number'] == '159':
print schedule_to_city
bus_id = in_dict(bus, 'number')
bus_id = no_whitespaces(bus_id)
bus_id = bus_id.replace(',', '_')
postfix = unidecode(bus['station']).lower().replace('/', '_')
postfix = postfix.replace('.', '')
postfix = postfix.replace('\'', '')
postfix = postfix.replace(' ', '_')
bus_id = unidecode(bus_id) + '_' + postfix
bus_id = "sub_" + bus_id
# print bus_id
bus['id'] = bus_id
parser_utils.save_json_file(
(parser_configs.directories["JSON_DIR"] +
"suburban_transport.json"),
suburban_buses
)
station_schedule_buses = [x[0] for x in station_schedule]
station_schedule_times = [x[1] for x in station_schedule]
station_schedule_weekend_buses = station_schedule_buses
station_schedule_weekend_times = station_schedule_times
station_schedule_weekend = station_schedule
station_schedule = {
"name": u"Балтийский банк",
"schedule_workdays_buses": station_schedule_workdays_buses,
"schedule_workdays_times": station_schedule_workdays_times,
"schedule_weekend_buses": station_schedule_weekend_buses,
"schedule_weekend_times": station_schedule_weekend_times,
}
save_json_file("station_balt_bank.json", station_schedule)
""" DERZHAVINA"""
derzhavina_bus_list = ["bus_20", "bus_4", "bus_19"]
derzhavina_position = [1, 0, 0]
derzhavina_bus_dict = {}
for bus_id in derzhavina_bus_list:
derzhavina_bus_dict[bus_id] = load_json_file(transport_dir + bus_id + ".json")
station_schedule = []
for i in range(len(derzhavina_bus_list)):
bus_id = derzhavina_bus_list[i]
sch = derzhavina_bus_dict[bus_id]["schedule_workdays"][derzhavina_position[i]]
bus_id_list = [bus_id] * len(sch)
sch = zip(bus_id_list, sch)
# print station['name'] + "|"
# if 'stations_weekend' in transport:
# for station in transport['stations_weekend']:
# print station['name'] + "|"
do_replace(transport, replace)
if "stations_weekend" in transport and "stations_workdays" in transport:
if transport["stations_weekend"] == transport["stations_workdays"]:
transport["stations"] = transport["stations_weekend"]
transport.pop("stations_workdays")
transport.pop("stations_weekend")
print "stations is the same"
print "##############################"
# serialization of web version JSON
parser_utils.save_json_file(filename, transport)
# transpose schedules, serializating phone version JSON
if "schedule_workdays" in transport:
schedule = [list(i) for i in zip(*transport["schedule_workdays"])]
transport["schedule_workdays"] = schedule
if "schedule_weekend" in transport:
schedule = [list(i) for i in zip(*transport["schedule_weekend"])]
transport["schedule_weekend"] = schedule
if not os.path.exists(IPHONE_DIR):
os.makedirs(IPHONE_DIR)
filename = IPHONE_DIR + transport["id"] + ".json"
parser_utils.save_json_file(filename, transport)
logging.info("Parsing trolleys schedules started.")
parse_schedules_trolley(transport["trolley_list"])
parse_schedules_suburban()
buses_trolleys = transport["trolley_list"] + transport["bus_list"]
regular_buses = [
elem for elem in transport["bus_list"] if elem['id'] not in MIX_LIST
]
mixed_buses = [
elem for elem in transport["bus_list"] if elem['id'] in MIX_LIST
]
buses = get_transport_json(regular_buses, 'bus')
trolleys = get_transport_json(transport["trolley_list"], 'trolley')
mixed = get_transport_json(mixed_buses, 'mixed')
transport_dict = {
'buses': buses,
'trolleys': trolleys,
'mixed': mixed
}
parser_utils.save_json_file(
directories["JSON_DIR"] + "transport.json",
transport_dict
)
def parse_schedules_bus(bus_list):
for bus in bus_list:
logging.info("Parsing schedule for bus with id: %s", bus["id"])
print "Current bus_id is:", bus["id"]
html = parser_utils.download_page(
TransportParser.TRANSPORT_PAGE_URL,
bus["post_data"]
)
schedule = ScheduleParser.parse_html(html)
json_bus_object = {
'type': 'bus',
'id': bus["id"],
'name': bus["name"],
'everyday': False
}
if bus['id'] in MIX_LIST:
type_ = 'mixed'
else:
type_ = 'bus'
json_bus_object['type'] = type_
# Case when bus have only weekend schedule
if bus['weekend'] is True:
json_bus_object['schedule_weekend'] = schedule['schedule_table']
json_bus_object['stations_weekend'] = schedule['stations_list']
# Case when bus have workdays, and maybe weekend schedule
else:
# Case of '7a' bus (everyday)
# @TODO: could be others buses like this
if schedule['everyday'] is True:
json_bus_object['schedule_everyday'] = (
schedule['schedule_table']
)
json_bus_object['stations'] = schedule['stations_list']
else:
json_bus_object['schedule_workdays'] = (
schedule['schedule_table']
)
json_bus_object['stations_workdays'] = schedule['stations_list']
if schedule['weekend'] is True:
html = parser_utils.download_page(
schedule['weekend_link']
)
schedule_weekend = ScheduleParser.parse_html(html)
json_bus_object['schedule_weekend'] = (
schedule_weekend['schedule_table']
)
json_bus_object['stations_weekend'] = (
schedule_weekend['stations_list']
)
if 'stations_weekend' in json_bus_object and 'stations_workdays' in json_bus_object:
if json_bus_object['stations_weekend'] == json_bus_object['stations_workdays']:
json_bus_object['stations'] = json_bus_object['stations_weekend']
json_bus_object.pop('stations_workdays')
json_bus_object.pop('stations_weekend')
parser_utils.save_json_file(
parser_configs.directories["BUSES_DIR"] + bus["id"] + ".json",
json_bus_object
)
