def query_address_api(df, address_api='openmapquest', address_api_key=None):
logger.info("Querying address API")
# Create the address data frame and cache file if it doesn't exist already
if not os.path.exists(ADDRESS_CACHE):
logger.info(
"Converting {} geometries to EPSG 4326, this may take awhile ".
format(len(df)))
df_address = geopandas.GeoDataFrame(df.to_crs(epsg='4326').geometry,
crs='epsg:4326')
df_address['address'] = None
logger.info("Creating new address cache file {}".format(ADDRESS_CACHE))
df_address.to_file(ADDRESS_CACHE, driver='ESRI Shapefile')
else:
logger.info(
"Reading in previous address cache file {}".format(ADDRESS_CACHE))
df_address = geopandas.read_file(ADDRESS_CACHE)
empty_address = df_address.loc[pd.isna(df_address['address'])]
logger.info("Querying for {} addresses".format(len(empty_address)))
for row in tqdm(empty_address.itertuples(), total=empty_address.shape[0]):
try:
address = reverse_geocode(row.geometry.centroid,
user_agent='caladrius',
provider=address_api,
api_key=address_api_key)
df_address.loc[row.Index, 'address'] = address['address'][0]
df_address.to_file(ADDRESS_CACHE, driver='ESRI Shapefile')
except Exception as e:
logger.exception('Geocoding failed for {latlon}'.format(
latlon=row.geometry.centroid))
continue
def get_point_location(points, provider='nominatim', user_agent='petk'):
centroid = MultiPoint(points).centroid
if importlib.util.find_spec('geopy') is not None:
from geopandas.tools import reverse_geocode
return reverse_geocode(centroid,
provider=provider,
user_agent=user_agent)['address'][0]
else:
return ', '.join([str(x) for x in mapping(centroid)['coordinates']])
def test_reverse(self):
with mock.patch('geopy.geocoders.googlev3.GoogleV3.reverse',
ReverseMock()) as m:
g = reverse_geocode(self.points, provider='googlev3', timeout=2)
self.assertEqual(len(self.points), m.call_count)
self.assertIsInstance(g, gpd.GeoDataFrame)
expected = GeoSeries(self.points, crs=from_epsg(4326))
assert_geoseries_equal(expected, g['geometry'])
tm.assert_series_equal(
g['address'],
pd.Series('address' + str(x) for x in range(len(self.points))))
def test_reverse(self):
with mock.patch('geopy.geocoders.googlev3.GoogleV3.reverse',
ReverseMock()) as m:
g = reverse_geocode(self.points, provider='googlev3', timeout=2)
self.assertEqual(len(self.points), m.call_count)
self.assertIsInstance(g, gpd.GeoDataFrame)
expected = GeoSeries(self.points, crs=from_epsg(4326))
assert_geoseries_equal(expected, g['geometry'])
address = pd.Series(['address' + str(x) for x in range(len(self.points))],
name='address')
tm.assert_series_equal(g['address'], address)
def test_reverse(self):
with mock.patch('geopy.geocoders.googlev3.GoogleV3.reverse',
ReverseMock()) as m:
g = reverse_geocode(self.points, provider='googlev3', timeout=2)
assert len(self.points) == m.call_count
assert isinstance(g, GeoDataFrame)
expected = GeoSeries(self.points, crs=from_epsg(4326))
assert_geoseries_equal(expected, g['geometry'])
address = pd.Series(
['address' + str(x) for x in range(len(self.points))],
name='address')
assert_series_equal(g['address'], address)
def test_reverse(locations, points):
from geopy.geocoders import GeocodeFarm
for provider in ['geocodefarm', GeocodeFarm]:
with mock.patch('geopy.geocoders.GeocodeFarm.reverse',
ReverseMock()) as m:
g = reverse_geocode(points, provider=provider, timeout=2)
assert len(points) == m.call_count
assert isinstance(g, GeoDataFrame)
expected = GeoSeries(points, crs=from_epsg(4326))
assert_geoseries_equal(expected, g['geometry'])
address = pd.Series(['address' + str(x) for x in range(len(points))],
name='address')
assert_series_equal(g['address'], address)
def test_reverse(locations, points):
from geopy.geocoders import Photon
for provider in ["photon", Photon]:
with mock.patch("geopy.geocoders.Photon.reverse", ReverseMock()) as m:
g = reverse_geocode(points, provider=provider, timeout=2)
assert len(points) == m.call_count
assert isinstance(g, GeoDataFrame)
expected = GeoSeries(points, crs="EPSG:4326")
assert_geoseries_equal(expected, g["geometry"])
address = pd.Series(["address" + str(x) for x in range(len(points))],
name="address")
assert_series_equal(g["address"], address)
def test_reverse(locations, points):
from geopy.geocoders import GoogleV3
for provider in ['googlev3', GoogleV3]:
with mock.patch('geopy.geocoders.googlev3.GoogleV3.reverse',
ReverseMock()) as m:
g = reverse_geocode(points, provider=provider, timeout=2)
assert len(points) == m.call_count
assert isinstance(g, GeoDataFrame)
expected = GeoSeries(points, crs=from_epsg(4326))
assert_geoseries_equal(expected, g['geometry'])
address = pd.Series(
['address' + str(x) for x in range(len(points))],
name='address')
assert_series_equal(g['address'], address)
def test_bad_provider_reverse():
from geopy.exc import GeocoderNotFound
with pytest.raises(GeocoderNotFound):
reverse_geocode(["cambridge, ma"], "badprovider")
def test_bad_provider_reverse():
from geopy.exc import GeocoderNotFound
with pytest.raises(GeocoderNotFound):
reverse_geocode(['cambridge, ma'], 'badprovider')
def test_bad_provider_reverse(self):
with self.assertRaises(ValueError):
reverse_geocode(['cambridge, ma'], 'badprovider')
def test_bad_provider_reverse(self):
from geopy.exc import GeocoderNotFound
with pytest.raises(GeocoderNotFound):
reverse_geocode(['cambridge, ma'], 'badprovider')
def test_bad_provider_reverse(self):
with self.assertRaises(ValueError):
reverse_geocode(['cambridge, ma'], 'badprovider')
def test_googlev3_reverse(self):
g = reverse_geocode(self.points, provider='googlev3', timeout=2)
self.assertIsInstance(g, gpd.GeoDataFrame)
def test_bad_provider_reverse():
from geopy.exc import GeocoderNotFound
with pytest.raises(GeocoderNotFound):
reverse_geocode([Point(0, 0)], "badprovider")
def apply(data, options=default_options, config=default_config, warning=print):
"""Perform conversion between latitude,longitude and address
ARGUMENTS:
data (pandas.DataFrame)
The data frame must contain either an `address` field or `latitude` and `longitude`
fields, according to whether `options["reverse"]` is `True` or `False`.
options (dict)
"reverse" (bool) specifies the direction of resolution. Forward resolves `address`
from `latitude,longitude` from . Reverse resolves `latitude,longitude`
`address` from. The default is `options["reverse"] = False`, which provides
forward resolution.
config (dict)
"provider" (str) specifies the providers of the address resolution algorithm.
The default is "nominatim".
"user_agent" (str) specifies the user agent for the address resolution algorithm.
The default is "csv_user_ht".
RETURNS:
pandas.DataFrame
The first (and only) return value is the `data` data frame with either the
`address` or `latitude,longitude` fields updated/added.
"""
if options["reverse"]:
# convert address to lat,lon
if not "address" in list(data.columns):
raise Exception(
"reserve address resolution requires 'address' field")
data.reset_index(inplace=True) # index is not meaningful
for retries in range(config["retries"]):
try:
pos = geocode(
data["address"],
provider=config["provider"],
user_agent=config["user_agent"],
timeout=config["timeout"],
)
break
except Exception as err:
pos = err
import time
time.sleep(config["sleep"])
if type(pos) is Exception:
raise pos
data["longitude"] = list(map(lambda p: p.x, pos["geometry"]))
data["latitude"] = list(map(lambda p: p.y, pos["geometry"]))
return data
else:
# convert lat,lon to address
try:
lats = list(map(lambda x: float(x), data["latitude"]))
lons = list(map(lambda x: float(x), data["longitude"]))
pos = list(map(lambda xy: Point(xy), list(zip(lons, lats))))
except:
pos = None
if type(pos) == type(None):
raise Exception(
"address resolution requires 'latitude' and 'longitude' fields"
)
for retries in range(config["retries"]):
try:
addr = reverse_geocode(
pos,
provider=config["provider"],
user_agent=config["user_agent"],
timeout=config["timeout"],
)
break
except Exception as err:
addr = err
import time
time.sleep(config["sleep"])
if type(addr) is Exception:
raise addr
data["address"] = Series(addr["address"], dtype="string").tolist()
return data