def set_config(base_dir):
config.base = base_dir
config.temp_data = os.path.join(config.base, 'tmp_data')
# gtfs
config.gtfs = os.path.join(config.temp_data, 'gtfs')
config.gtfs_raw_data = os.path.join(config.gtfs, 'data')
config.gtfs_processed_data = os.path.join(config.gtfs, 'processed_data')
mkdir_p(config.gtfs_processed_data)
config.gtfs_stop_file = os.path.join(config.gtfs_processed_data, 'stop.data')
config.gtfs_shape_file = os.path.join(config.gtfs_processed_data, 'shape.data')
# reports
config.output_data = os.path.join(config.temp_data, 'output')
mkdir_p(config.output_data)
# params
config.max_accuracy_radius_meters = 300
config.min_accuracy_radius_meters = 200
config.route_sampling__min_distance_between_points_meters = 10.0
config.station_radius_in_meters = 300
config.early_arrival_max_seconds = 35 * 60 # how early can a train arrive before the actual arrival
config.late_arrival_max_seconds = 35 * 60 # how late can a train arrive after the actual arrival
config.early_departure_max_seconds = 15 * 60 # how early can a train depart before the actual departure
config.late_departure_max_seconds = 35 * 60 # how late can a train depart after the actual departure
config.shape_probability_threshold= 0.80
config.stop_discovery_location_timeout_seconds = 60
config.stop_discovery_probability_thresh = 0.90
config.stop_discovery_count_thresh = 3
# if the trip list is longer than the threshold, we do not
# have a match to GTFS
config.trip_list_length_thresh = 3
config.no_report_timegap = datetime.timedelta(minutes = 60)
config.no_stop_timegap = datetime.timedelta(minutes = 60)
def download_gtfs_file():
""" download gtfs zip file from mot, and put it in DATA_DIR in its own subfolder """
local_dir = os.path.join(GTFS_DATA_DIR,ot_utils.get_utc_time_underscored())
ot_utils.mkdir_p(local_dir)
local_path = os.path.join(local_dir,FILE_NAME)
ot_utils.ftp_get_file(MOT_FTP,FILE_NAME,local_path)
ot_utils.unzip_file(local_path,local_dir)
def set_config(base_dir):
config.base = base_dir
config.temp_data = os.path.join(config.base, 'tmp_data')
# gtfs
config.gtfs = os.path.join(config.temp_data, 'gtfs')
config.gtfs_raw_data = os.path.join(config.gtfs, 'data')
config.gtfs_processed_data = os.path.join(config.gtfs, 'processed_data')
mkdir_p(config.gtfs_processed_data)
config.gtfs_stop_file = os.path.join(config.gtfs_processed_data, 'stop.data')
config.gtfs_shape_file = os.path.join(config.gtfs_processed_data, 'shape.data')
# reports
config.output_data = os.path.join(config.temp_data, 'output')
mkdir_p(config.output_data)
config.output_shelve_file = os.path.join(config.output_data, 'shelve.data')
# params
config.max_accuracy_radius_meters = 300
config.min_accuracy_radius_meters = 200
config.route_sampling__min_distance_between_points_meters = 10.0
config.station_radius_in_meters = 300
config.early_arrival_max_seconds = 35 * 60 # how early can a train arrive before the actual arrival
config.late_arrival_max_seconds = 35 * 60 # how late can a train arrive before the actual arrival
config.early_departure_max_seconds = 15 * 60 # how early can a train depart before the actual departure
config.late_departure_max_seconds = 35 * 60 # how late can a train depart before the actual departure
config.shape_probability_threshold= 0.80
config.stop_discovery_location_timeout_seconds = 60
config.stop_discovery_probability_thresh = 0.95
config.stop_discovery_count_thresh = 10
def download_gtfs_file(force=False,gtfs_url=None):
""" download gtfs zip file from mot, and put it in DATA_DIR in its own subfolder """
import shutil
time_suffix = ot_utils.get_utc_time_underscored()
if not os.path.exists(GTFS_DATA_DIR):
ot_utils.mkdir_p(GTFS_DATA_DIR)
tmp_file = '/tmp/%s_tmp.zip' % (time_suffix)
print 'downloading GTFS to tmp file'
if not gtfs_url:
ot_utils.ftp_get_file(MOT_FTP,FILE_NAME,tmp_file)
else:
ot_utils.download_url(gtfs_url,tmp_file)
if not force:
tmp_md5 = ot_utils.md5_for_file(tmp_file)
last_dir = ot_utils.find_lastest_in_dir(GTFS_DATA_DIR)
if last_dir:
was_success = os.path.exists(os.path.join(last_dir,'success'))
if not was_success:
print 'Last time was not success'
else:
last_file = os.path.join(last_dir,FILE_NAME)
try:
last_md5 = ot_utils.md5_for_file(last_file)
except Exception,e:
print e
last_md5 = 'error_in_md5'
if last_md5 == tmp_md5:
print 'Checksum is identical - removing tmp file'
os.remove(tmp_file)
return None
def download_gtfs_file(force=False):
""" download gtfs zip file from mot, and put it in DATA_DIR in its own subfolder """
import shutil
time_suffix = ot_utils.get_utc_time_underscored()
if not os.path.exists(GTFS_DATA_DIR):
ot_utils.mkdir_p(GTFS_DATA_DIR)
tmp_file = '/tmp/{0}_tmp.zip'.format(time_suffix)
LOGGER.info('downloading GTFS to tmp file')
ot_utils.ftp_get_file(MOT_FTP, FILE_NAME, tmp_file)
if not force:
tmp_md5 = ot_utils.md5_for_file(tmp_file)
last_dir = ot_utils.find_lastest_in_dir(GTFS_DATA_DIR)
if last_dir:
was_success = os.path.exists(os.path.join(last_dir, 'success'))
if not was_success:
LOGGER.info('Last time was not success')
else:
last_file = os.path.join(last_dir, FILE_NAME)
try:
last_md5 = ot_utils.md5_for_file(last_file)
except Exception as e:
LOGGER.exception('failed in md5 for last file - ignoring')
last_md5 = 'error_in_md5'
if last_md5 == tmp_md5:
LOGGER.info('Checksum is identical - removing tmp file')
os.remove(tmp_file)
return None
LOGGER.info('Checksum is different or force -- copying')
local_dir = os.path.join(GTFS_DATA_DIR, time_suffix)
ot_utils.mkdir_p(local_dir)
try:
os.remove(os.path.join(GTFS_DATA_DIR,'latest'))
except (IOError, OSError):
pass
os.symlink(local_dir, os.path.join(GTFS_DATA_DIR,'latest'))
local_file = os.path.join(local_dir, FILE_NAME)
shutil.move(tmp_file, local_file)
ot_utils.unzip_file(local_file, local_dir)
LOGGER.info('All gtfs files are in %s' % local_dir)
return local_dir
basedir = GTFS_ZIP_DIR
local_dir = os.path.join(basedir,time_suffix)
tmp_file = '/tmp/%s_tmp.zip' % (time_suffix)
ot_utils.ftp_get_file(MOT_FTP,FILE_NAME,tmp_file)
tmp_md5 = ot_utils.md5_for_file(tmp_file)
last_dir = ot_utils.find_lastest_in_dir(basedir)
last_file = os.path.join(last_dir,FILE_NAME)
try:
last_md5 = ot_utils.md5_for_file(last_file)
except Exception,e:
print e
last_md5 = 'error_in_md5'
if last_md5 != tmp_md5:
print 'Checksum is different- copying'
ot_utils.mkdir_p(local_dir)
local_file = os.path.join(local_dir,FILE_NAME)
shutil.move(tmp_file,local_file)
else:
print 'Checksum is identical - removing tmp file'
os.remove(tmp_file)
return
if not download_only:
ot_utils.unzip_file(local_file,local_dir)
def find_gtfs_data_dir():
""" returns the lastest subfolder in DATA_DIR """
dirnames = glob.glob("%s/*" % (GTFS_DATA_DIR))
if not dirnames:
