route_id = args.route
stop_id = args.stop
d = date.fromisoformat(args.date)
comparison = args.comparison
early_min = args.early_min
late_min = args.late_min
agency = config.get_agency(args.agency)
agency_id = agency.id
start_time = datetime.now()
print(f"Start: {start_time}")
timetable = timetables.get_by_date(agency_id, route_id, d)
route_config = agency.get_route_config(route_id)
tz = agency.tz
direction_id = args.dir
timetable_df = timetable.get_data_frame(
stop_id=stop_id, direction_id=direction_id).sort_values('TIME')
early_sec = early_min * 60
late_sec = late_min * 60
timetable_df['scheduled_headway'] = np.r_[
np.nan,
metrics.compute_headway_minutes(timetable_df['TIME'].values)]
dates = util.get_dates_in_range(args.start_date, args.end_date)
else:
raise Exception('missing date, start-date, or end-date')
print(f"Date: {', '.join([str(date) for date in dates])}")
print(f"Local Time Range: [{start_time_str}, {end_time_str})")
print(f"Route: {route_id} ({route_config.title})")
print(f"From: {s1} ({s1_info.title})")
print(f"To: {s2} ({s2_info.title})")
print(f"Direction: {','.join(common_dirs)} ({dir_info.title})")
completed_trips_arr = []
for d in dates:
if show_scheduled:
history = timetables.get_by_date(agency.id, route_id, d, version)
else:
history = arrival_history.get_by_date(agency.id, route_id, d,
version)
start_time = util.get_timestamp_or_none(d, start_time_str, tz)
end_time = util.get_timestamp_or_none(d, end_time_str, tz)
s1_df = history.get_data_frame(stop_id=s1,
start_time=start_time,
end_time=end_time)
s2_df = history.get_data_frame(stop_id=s2, start_time=start_time)
s1_df['trip_min'], s1_df[
'dest_arrival_time'] = trip_times.get_matching_trips_and_arrival_times(
s1_df['TRIP'].values, s1_df['DEPARTURE_TIME'].values,
def compute_stats(d: date, agency: config.Agency, routes, save_to_s3=True):
tz = agency.tz
stat_ids = all_stat_ids
print(d)
time_str_intervals = constants.DEFAULT_TIME_STR_INTERVALS.copy()
time_str_intervals.append(('07:00','19:00'))
timestamp_intervals = [(
int(util.get_localized_datetime(d, start_time_str, tz).timestamp()),
int(util.get_localized_datetime(d, end_time_str, tz).timestamp())
) for start_time_str, end_time_str in time_str_intervals
]
timestamp_intervals.append((None, None))
time_str_intervals.append((None, None))
all_stats = {}
for stat_id in stat_ids:
all_stats[stat_id] = {}
for interval_index, _ in enumerate(timestamp_intervals):
all_stats[stat_id][interval_index] = {}
for route in routes:
route_id = route.id
print(route_id)
t1 = time.time()
route_config = agency.get_route_config(route_id)
try:
history = arrival_history.get_by_date(agency.id, route_id, d)
except FileNotFoundError as ex:
print(ex)
continue
try:
timetable = timetables.get_by_date(agency.id, route_id, d)
except FileNotFoundError as ex:
print(ex)
continue
timetable_df = timetable.get_data_frame()
history_df = history.get_data_frame()
for stat_id in stat_ids:
for interval_index, _ in enumerate(timestamp_intervals):
all_stats[stat_id][interval_index][route_id] = {'directions':{}}
for dir_info in route_config.get_direction_infos():
dir_id = dir_info.id
all_stats[stat_id][interval_index][route_id]['directions'][dir_id] = collections.defaultdict(dict)
add_trip_time_stats_for_route(all_stats, timestamp_intervals, route_config, history_df)
add_wait_time_stats_for_route(all_stats, timestamp_intervals, route_config, history_df)
add_schedule_adherence_stats_for_route(all_stats, timestamp_intervals, route_config, history_df, timetable_df)
t2 = time.time()
print(f' {round(t2-t1, 2)} sec')
for stat_id in stat_ids:
for interval_index, (start_time, end_time) in enumerate(timestamp_intervals):
start_time_str, end_time_str = time_str_intervals[interval_index]
data = {
'routes': all_stats[stat_id][interval_index],
}
precomputed_stats.save_stats(agency.id, stat_id, d, start_time_str, end_time_str, data, save_to_s3)
elif args.start_date is not None and args.end_date is not None:
dates = util.get_dates_in_range(args.start_date, args.end_date)
else:
raise Exception('missing date, start-date, or end-date')
# print results for each direction
for stop_dir in stop_dirs:
dir_info = route_config.get_direction_info(stop_dir)
interval_stats_arr = []
first_bus_date_times = []
last_bus_date_times = []
for d in dates:
if show_scheduled:
hist = timetables.get_by_date(agency.id, route_id, d)
else:
hist = arrival_history.get_by_date(agency.id, route_id, d)
arrivals = hist.get_data_frame(stop_id=stop, direction_id=stop_dir)
start_time = util.get_timestamp_or_none(d, start_time_str, tz)
end_time = util.get_timestamp_or_none(d, end_time_str, tz)
departure_times = np.sort(arrivals['DEPARTURE_TIME'].values)
if len(departure_times) == 0:
continue
first_bus_date_times.append(
datetime.fromtimestamp(departure_times[0], tz))
last_bus_date_times.append(