def __init__(self, recreate=False):
datamodel = Datamodel()
self.table, self.table_config = datamodel.population()
self.postgres = Postgresql(user_name='postgres',
password='******',
host='localhost',
port='5432',
db='TestProject')
self.postgres.initialize_table(self.table,
recreate=False,
**self.table_config)
self.googlegeo = GoogleGeo()
# myan: only get major cities data once per request
self.major_cities_postgres = Postgresql(user_name='postgres',
password='******',
host='localhost',
port='5432',
db='TestProject')
self.major_cities = self.major_cities_postgres.get(
"select * from TestMajorCities")
self.all_states = self.major_cities['state'].values
self.all_cities = self.major_cities['city'].values
self.all_lats = self.major_cities['lat'].values
self.all_lngs = self.major_cities['lng'].values
self.all_population = self.major_cities['population'].values
def _get_city_info(major_cities):
place_ids = []
lats = []
lngs = []
states = []
# base_url = 'http://maps.googleapis.com/maps/api/geocode/json?address='
progress = 0.0
google_geo = GoogleGeo()
for i, city in enumerate(major_cities.city.values):
sleep(0.5)
output = dict(place_id=None, state=None, lat=None, lng=None)
try:
# response = requests.get(base_url + '{string}'.format(string=city))
# output = _parse_google_geo_response(response.json())
output = google_geo.get(city,
fields_to_get=('place_id', 'lat', 'lng',
'state'))
if float(i) / len(major_cities) - progress >= 0.1:
print "Processed {pct}%".format(
pct=str(float(i) / len(major_cities) * 100))
progress = float(i) / len(major_cities)
except:
pass
place_ids.append(output['place_id'])
states.append(output['state'])
lats.append(output['lat'])
lngs.append(output['lng'])
major_cities['place_id'] = place_ids
major_cities['state'] = states
major_cities['lat'] = lats
major_cities['lng'] = lngs
major_cities.drop_duplicates(subset='place_id', inplace=True)
major_cities = major_cities.loc[np.logical_not(
major_cities.place_id.isnull())]
return major_cities
