def import_to_datastore(directory, provider_code, batch_size):
"""
goes through all the files in a given directory, parses and commits them
"""
try:
files_from_lms = dict(get_files(directory))
if len(files_from_lms) == 0:
return 0
logging.info("Importing '{}'".format(directory))
started = datetime.now()
accidents = list(import_accidents(provider_code=provider_code, **files_from_lms))
db.session.execute(Marker.__table__.insert(), accidents)
db.session.commit()
involved = list(import_involved(provider_code=provider_code, **files_from_lms))
db.session.execute(Involved.__table__.insert(), involved)
db.session.commit()
vehicles = list(import_vehicles(provider_code=provider_code, **files_from_lms))
db.session.execute(Vehicle.__table__.insert(), vehicles)
db.session.commit()
total = len(accidents) + len(involved) + len(vehicles)
logging.info("\t{0} items in {1}".format(total, time_delta(started)))
return total
except ValueError as e:
failed_dirs[directory] = e.message
return 0
def import_to_datastore(directory, provider_code, batch_size):
"""
goes through all the files in a given directory, parses and commits them
"""
try:
files_from_lms = dict(get_files(directory))
if len(files_from_lms) == 0:
return 0
logging.info("Importing '{}'".format(directory))
started = datetime.now()
accidents = list(import_accidents(provider_code=provider_code, **files_from_lms))
new_ids = [m["id"] for m in accidents
if 0 == Marker.query.filter(and_(Marker.id == m["id"],
Marker.provider_code == m["provider_code"])).count()]
if not new_ids:
logging.info("\t\tNothing loaded, all accidents already in DB")
return 0
db.session.execute(Marker.__table__.insert(), [m for m in accidents if m["id"] in new_ids])
db.session.commit()
involved = list(import_involved(provider_code=provider_code, **files_from_lms))
db.session.execute(Involved.__table__.insert(), [i for i in involved if i["accident_id"] in new_ids])
db.session.commit()
vehicles = list(import_vehicles(provider_code=provider_code, **files_from_lms))
db.session.execute(Vehicle.__table__.insert(), [v for v in vehicles if v["accident_id"] in new_ids])
db.session.commit()
total = len(accidents) + len(involved) + len(vehicles)
logging.info("\t{0} items in {1}".format(total, time_delta(started)))
return total
except ValueError as e:
failed_dirs[directory] = e.message
return 0
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--specific_folder',
dest='specific_folder',
action='store_true',
default=False)
parser.add_argument('--delete_all',
dest='delete_all',
action='store_true',
default=True)
parser.add_argument('--path', type=str, default="static/data/lms")
parser.add_argument('--batch_size', type=int, default=100)
parser.add_argument('--provider_code', type=int)
args = parser.parse_args()
if args.specific_folder:
if fileDialog:
dir_name = tkFileDialog.askdirectory(
initialdir=os.path.abspath(args.path),
title='Please select a directory')
else:
dir_name = raw_input('Please provide the directory path: ')
dir_list = [dir_name]
if args.delete_all:
confirm_delete_all = raw_input(
"Are you sure you want to delete all the current data? (y/n)\n"
)
if confirm_delete_all.lower() == 'n':
args.delete_all = False
else:
dir_list = glob.glob("{0}/*/*".format(args.path))
# wipe all the Markers and Involved data first
if args.delete_all:
tables = (Vehicle, Involved, Marker)
logging.info("Deleting tables: " + ", ".join(table.__name__
for table in tables))
for table in tables:
db.session.query(table).delete()
db.session.commit()
started = datetime.now()
total = 0L
for directory in dir_list:
parent_directory = os.path.basename(
os.path.dirname(os.path.join(os.pardir, directory)))
provider_code = args.provider_code if args.provider_code else get_provider_code(
parent_directory)
total += import_to_datastore(directory, provider_code, args.batch_size)
delete_invalid_entries()
failed = [
"\t'{0}' ({1})".format(directory, fail_reason)
for directory, fail_reason in failed_dirs.iteritems()
]
logging.info("Finished processing all directories{0}{1}".format(
", except:\n" if failed else "", "\n".join(failed)))
logging.info("Total: {0} items in {1}".format(total, time_delta(started)))
def import_to_datastore(directory, provider_code, batch_size):
"""
goes through all the files in a given directory, parses and commits them
"""
try:
files_from_lms = dict(get_files(directory))
if len(files_from_lms) == 0:
return 0
logging.info("Importing '{}'".format(directory))
started = datetime.now()
accidents = list(
import_accidents(provider_code=provider_code, **files_from_lms))
db.session.execute(Marker.__table__.insert(), accidents)
db.session.commit()
involved = list(
import_involved(provider_code=provider_code, **files_from_lms))
db.session.execute(Involved.__table__.insert(), involved)
db.session.commit()
vehicles = list(
import_vehicles(provider_code=provider_code, **files_from_lms))
db.session.execute(Vehicle.__table__.insert(), vehicles)
db.session.commit()
total = len(accidents) + len(involved) + len(vehicles)
logging.info("\t{0} items in {1}".format(total, time_delta(started)))
return total
except ValueError as e:
failed_dirs[directory] = e.message
return 0
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--specific_folder', dest='specific_folder', action='store_true', default=False)
parser.add_argument('--delete_all', dest='delete_all', action='store_true', default=True)
parser.add_argument('--path', type=str, default="static/data/lms")
parser.add_argument('--batch_size', type=int, default=100)
parser.add_argument('--provider_code', type=int)
args = parser.parse_args()
if args.specific_folder:
if fileDialog:
dir_name = tkFileDialog.askdirectory(initialdir=os.path.abspath(args.path),
title='Please select a directory')
else:
dir_name = raw_input('Please provide the directory path: ')
dir_list = [dir_name]
if args.delete_all:
confirm_delete_all = raw_input("Are you sure you want to delete all the current data? (y/n)\n")
if confirm_delete_all.lower() == 'n':
args.delete_all = False
else:
dir_list = glob.glob("{0}/*/*".format(args.path))
# wipe all the Markers and Involved data first
if args.delete_all:
tables = (Vehicle, Involved, Marker)
logging.info("Deleting tables: " + ", ".join(table.__name__ for table in tables))
for table in tables:
db.session.query(table).delete()
db.session.commit()
started = datetime.now()
total = 0L
for directory in dir_list:
parent_directory = os.path.basename(os.path.dirname(os.path.join(os.pardir, directory)))
provider_code = args.provider_code if args.provider_code else get_provider_code(parent_directory)
total += import_to_datastore(directory, provider_code, args.batch_size)
delete_invalid_entries()
failed = ["\t'{0}' ({1})".format(directory, fail_reason) for directory, fail_reason in
failed_dirs.iteritems()]
logging.info("Finished processing all directories{0}{1}".format(", except:\n" if failed else "",
"\n".join(failed)))
logging.info("Total: {0} items in {1}".format(total, time_delta(started)))
def import_to_datastore(directory, provider_code, batch_size):
"""
goes through all the files in a given directory, parses and commits them
"""
try:
files_from_lms = dict(get_files(directory))
if len(files_from_lms) == 0:
return 0
logging.info("Importing '{}'".format(directory))
started = datetime.now()
accidents = list(
import_accidents(provider_code=provider_code, **files_from_lms))
new_ids = [
m["id"] for m in accidents if 0 == Marker.query.filter(
and_(Marker.id == m["id"], Marker.provider_code ==
m["provider_code"])).count()
]
if not new_ids:
logging.info("\t\tNothing loaded, all accidents already in DB")
return 0
db.session.execute(Marker.__table__.insert(),
[m for m in accidents if m["id"] in new_ids])
db.session.commit()
involved = list(
import_involved(provider_code=provider_code, **files_from_lms))
db.session.execute(
Involved.__table__.insert(),
[i for i in involved if i["accident_id"] in new_ids])
db.session.commit()
vehicles = list(
import_vehicles(provider_code=provider_code, **files_from_lms))
db.session.execute(
Vehicle.__table__.insert(),
[v for v in vehicles if v["accident_id"] in new_ids])
db.session.commit()
total = len(accidents) + len(involved) + len(vehicles)
logging.info("\t{0} items in {1}".format(total, time_delta(started)))
return total
except ValueError as e:
failed_dirs[directory] = e.message
return 0
def main(username=None, password=None, lastmail=False):
username = username or os.environ.get("MAILUSER")
password = password or os.environ.get("MAILPASS")
if not username:
logging.error("Username not set. Please set env var MAILUSER or use the --username argument")
if not password:
logging.error("Password not set. Please set env var MAILPASS or use the --password argument")
if not username or not password:
exit()
imapsession = imaplib.IMAP4_SSL("imap.gmail.com")
try:
imapsession.login(username, password)
except imaplib.IMAP4.error:
logging.error("Bad credentials, unable to sign in!")
exit()
try:
imapsession.select(mail_dir)
typ, data = imapsession.search(None, "ALL")
except imaplib.IMAP4.error:
logging.error("Error searching given mailbox: %s" % mail_dir)
exit()
file_found = False
listdir = os.listdir(detach_dir)
is_empty = len(listdir) <= 1 or not lastmail
total = 0
# Iterating over all emails
started = datetime.now()
logging.info("Login successful! Importing files, please hold...")
for msgId in data[0].split():
typ, message_parts = imapsession.fetch(msgId, "(RFC822)")
if typ != "OK":
logging.error("Error fetching mail.")
raise
email_body = message_parts[0][1]
mail = email.message_from_string(email_body)
mtime = datetime.strptime(mail["Date"][:-6], "%a, %d %b %Y %H:%M:%S")
if not is_empty:
# Accident folder is not empty, we only need the latest
if datetime.now() - mtime < timedelta(hours=4):
file_found = True
else:
continue
# Handles Gmail bug which hasn't physically removed some of the deleted files
mail_date = datetime(2015, 10, 06, 10)
if mtime < mail_date:
continue
for part in mail.walk():
if part.get_content_maintype() == "multipart" or part.get("Content-Disposition") is None:
continue
filename = part.get_filename()
if bool(filename) and filename.endswith(".csv"):
filename = "UH-{0}_{1}-{2}.csv".format(mtime.date(), mtime.hour, mtime.minute)
filepath = os.path.join(detach_dir, filename)
if os.path.isfile(filepath):
break
total += 1
print "Currently loading: " + filename + " "
sys.stdout.write("\033[F")
time.sleep(0.1)
with open(filepath, "wb") as fp:
fp.write(part.get_payload(decode=True))
if file_found:
break
logging.info("Imported {0} file(s) in {1}".format(total, time_delta(started)))
imapsession.close()
imapsession.logout()
def main(username=None, password=None, lastmail=False):
username = username or os.environ.get('MAILUSER')
password = password or os.environ.get('MAILPASS')
if not username:
logging.error(
"Username not set. Please set env var MAILUSER or use the --username argument"
)
if not password:
logging.error(
"Password not set. Please set env var MAILPASS or use the --password argument"
)
if not username or not password:
exit()
imapsession = imaplib.IMAP4_SSL('imap.gmail.com')
try:
imapsession.login(username, password)
except imaplib.IMAP4.error:
logging.error('Bad credentials, unable to sign in!')
exit()
try:
imapsession.select(mail_dir)
typ, data = imapsession.search(None, 'ALL')
except imaplib.IMAP4.error:
logging.error('Error searching given mailbox: %s' % mail_dir)
exit()
file_found = False
listdir = os.listdir(detach_dir)
is_empty = len(listdir) <= 1 or not lastmail
total = 0
# Iterating over all emails
started = datetime.now()
logging.info("Login successful! Importing files, please hold...")
for msgId in data[0].split():
typ, message_parts = imapsession.fetch(msgId, '(RFC822)')
if typ != 'OK':
logging.error('Error fetching mail.')
raise
email_body = message_parts[0][1]
mail = email.message_from_string(email_body)
mtime = datetime.strptime(mail['Date'][:-6], '%a, %d %b %Y %H:%M:%S')
if not is_empty:
# Accident folder is not empty, we only need the latest
if datetime.now() - mtime < timedelta(hours=4):
file_found = True
else:
continue
# Handles Gmail bug which hasn't physically removed some of the deleted files
mail_date = datetime(2015, 10, 06, 10)
if mtime < mail_date:
continue
for part in mail.walk():
if part.get_content_maintype(
) == 'multipart' or part.get('Content-Disposition') is None:
continue
filename = part.get_filename()
if bool(filename) and filename.endswith(".csv"):
filename = 'UH-{0}_{1}-{2}.csv'.format(mtime.date(),
mtime.hour,
mtime.minute)
filepath = os.path.join(detach_dir, filename)
if os.path.isfile(filepath):
break
total += 1
print 'Currently loading: ' + filename + ' '
sys.stdout.write("\033[F")
time.sleep(0.1)
with open(filepath, 'wb') as fp:
fp.write(part.get_payload(decode=True))
if file_found:
break
logging.info("Imported {0} file(s) in {1}".format(total,
time_delta(started)))
imapsession.close()
imapsession.logout()
continue
self.cache[prop] = [response.get('daily').get('data')[0].get(prop)]
if __name__ == '__main__':
temperature_client = TemperatureClient('https://api.darksky.net/forecast')
properties = [
'time', 'sunriseTime', 'sunsetTime', 'temperatureHigh', 'dewPoint',
'humidity', 'windSpeed', 'cloudCover'
]
start_date = starting_date()
end_date = start_date + year()
# locations = ['USA_AK_FAIRBANKS', 'USA_CA_LOS_ANGELES', 'USA_IL_CHICAGO-OHARE', 'USA_MN_MINNEAPOLIS', 'USA_TX_HOUSTON', 'USA_WA_SEATTLE']
# locations = ['USA_AK_FAIRBANKS','USA_CA_LOS_ANGELES', 'USA_IL_CHICAGO-OHARE', 'USA_MN_MINNEAPOLIS', 'USA_TX_HOUSTON', 'USA_WA_SEATTLE']
locations = [
'USA_NV_LAS_VEGAS', 'USA_CA_SAN_FRANCISCO', 'USA_AZ_PHOENIX',
'USA_GA_ATLANTA', 'USA_MD_BALTIMORE', 'USA_CO_BOULDER'
]
for location in locations:
temperature_client.cache = {}
coordinates = get_lat_lng(location)
current_date = start_date
while current_date <= end_date:
response = temperature_client.get_data(coordinates + ',' +
str(current_date))
temperature_client.parse_data(response, properties)
current_date = current_date + time_delta()
data = temperature_client.get_cache()
temperature_data = pd.DataFrame.from_dict(data)
temperature_data.to_csv('data/' + location + '_temperature_usage.csv')