def process_file(fn):
_, yymm = fn[:-len('.csv')].split('-')
print 'handle the file; %s' % yymm
logging_msg('handle the file; %s' % yymm)
with open('%s/%s' % (merged_trip_dir, fn), 'rb') as r_csvfile:
reader = csv.reader(r_csvfile)
headers = reader.next()
#
id_did = headers.index('driver-id')
id_vid = headers.index('vehicle-id')
id_st, id_et = headers.index('start-time'), headers.index('end-time')
id_dur, id_fare = headers.index('duration'), headers.index('fare')
id_s_long, id_s_lat = headers.index('start-long'), headers.index('start-lat')
id_e_long, id_e_lat = headers.index('end-long'), headers.index('end-lat')
#
vehicle_prev_trip_position_time = {}
with open('%s/full-drivers-trips-%s.csv' % (for_full_driver_dir, yymm), 'wt') as w_csvfile:
writer = csv.writer(w_csvfile)
new_headers = ['did', 'prev-trip-end-time', 'prev-trip-end-location', 'start-time', 'start-location', 'end-time', 'end-location', 'duration', 'fare']
writer.writerow(new_headers)
count = 0
for row in reader:
did = int(row[id_did])
if did > len(driver_full_or_not) or did < 0:
continue
if not driver_full_or_not[did]:
continue
vid = row[id_vid]
start_time, end_time = eval(row[id_st]), eval(row[id_et]),
s_long, s_lat = eval(row[id_s_long]), eval(row[id_s_lat])
e_long, e_lat = eval(row[id_e_long]), eval(row[id_e_lat])
s_location, e_location = is_in_airport(s_long, s_lat), is_in_airport(e_long, e_lat)
#
if not vehicle_prev_trip_position_time.has_key(vid):
# ASSUMPTION
# If this trip is the driver's first trip in a month,
# let's assume that the previous trip occurred out of the airport
# and also assume that the previous trip's end time is the current trip's start time
vehicle_prev_trip_position_time[vid] = (OUT_AP, start_time)
prev_trip_end_location, prev_trip_end_time = vehicle_prev_trip_position_time[vid]
#
new_row = [did, prev_trip_end_time, prev_trip_end_location,
start_time, s_location,
end_time, e_location,
row[id_dur], row[id_fare]]
writer.writerow(new_row)
#
vehicle_prev_trip_position_time[vid] = (e_location, end_time)
#
count +=1
if count % check_progress == 0:
print '%s-----%d' % (yymm, count)
logging_msg('%s-----%d' % (yymm, count))
print 'end the file; %s' % yymm
logging_msg('end the file; %s' % yymm)
def process_file(path_to_csv_file):
print path_to_csv_file
ori_log_fn = path_to_csv_file.split('/')[-1]
_, yymm, _ = ori_log_fn.split('-')
print 'handle the file; %s' % yymm
logging_msg('handle the file; %s' % yymm)
with open(path_to_csv_file, 'rb') as r_csvfile:
reader = csv.reader(r_csvfile)
headers = reader.next()
id_time, id_vid, id_did = headers.index('time'), headers.index(
'vehicle-id'), headers.index('driver-id')
index_long, index_lat = headers.index('longitude'), headers.index(
'latitude')
with open('%s/log-%s.csv' % (logs_dir, yymm), 'wt') as w_csvfile:
writer = csv.writer(w_csvfile)
new_headers = ['time', 'vid', 'did', 'ap-or-not', 'np-or-not']
writer.writerow(new_headers)
#
for row in reader:
ap_or_not = is_in_airport(eval(row[index_long]),
eval(row[index_lat]))
np_or_not = is_in_night_safari(eval(row[index_long]),
eval(row[index_lat]))
new_row = [
row[id_time], row[id_vid], row[id_did], ap_or_not,
np_or_not
]
writer.writerow(new_row)
print 'end the file; %s' % yymm
logging_msg('end the file; %s' % yymm)
def process_file(path_to_csv_file):
print path_to_csv_file
ori_log_fn = path_to_csv_file.split('/')[-1]
_, yymm, _ = ori_log_fn.split('-')
print 'handle the file; %s' % yymm
logging_msg('handle the file; %s' % yymm)
with open(path_to_csv_file, 'rb') as r_csvfile:
reader = csv.reader(r_csvfile)
headers = reader.next()
id_time, id_vid, id_did = headers.index('time'), headers.index('vehicle-id'), headers.index('driver-id')
index_long, index_lat = headers.index('longitude'), headers.index('latitude')
with open('%s/log-%s.csv' % (logs_dir, yymm), 'wt') as w_csvfile:
writer = csv.writer(w_csvfile)
new_headers = ['time', 'vid', 'did', 'ap-or-not', 'np-or-not']
writer.writerow(new_headers)
#
for row in reader:
ap_or_not = is_in_airport(eval(row[index_long]), eval(row[index_lat]))
np_or_not = is_in_night_safari(eval(row[index_long]), eval(row[index_lat]))
new_row = [row[id_time], row[id_vid], row[id_did], ap_or_not, np_or_not]
writer.writerow(new_row)
print 'end the file; %s' % yymm
logging_msg('end the file; %s' % yymm)