def to_valid_latitude(latitude):
'''Convert longitude into the -180 to 180 scale'''
if not is_valid_latitude(latitude):
raise ValueError('Invalid latitude {}'.format(latitude))
if isclose(latitude, 90.0):
latitude = 89.9999
elif isclose(latitude, -90.0):
latitude = -89.9999
return latitude
def near_dupe_hashes(cls,
address,
geohash_precision=DEFAULT_GEOHASH_PRECISION,
use_latlon=True,
use_city=False,
use_postal_code=False):
address_expansions = cls.component_expansions(address)
lat = address.get(Coordinates.LATITUDE)
lon = address.get(Coordinates.LONGITUDE)
postcode = safe_decode(address.get(AddressComponents.POSTAL_CODE,
u'')).strip()
city = safe_decode(address.get(AddressComponents.CITY, u'')).strip()
if not any(address_expansions):
return
if lat and lon and use_latlon and not (
(isclose(lat, 0.0) and isclose(lon, 0.0)) or lat >= 90.0
or lat <= -90.0):
geo = geohash.encode(lat, lon)[:geohash_precision]
geohash_neighbors = [geo] + geohash.neighbors(geo)
base_key = cls.GEOHASH_KEY_PREFIX
for keys in six.itertools.product(geohash_neighbors,
*address_expansions):
yield u'{}|{}'.format(base_key, u'|'.join(keys))
if postcode and use_postal_code:
postcode_expansions = expand_address(
postcode, address_components=ADDRESS_POSTAL_CODE)
base_key = cls.POSTCODE_KEY_PREFIX
for keys in six.itertools.product(postcode_expansions,
*address_expansions):
yield u'{}|{}'.format(base_key, u'|'.join(keys))
if city and use_city:
city_expansions = expand_address(
city, address_components=ADDRESS_TOPONYM)
base_key = cls.CITY_KEY_PREFIX
for keys in six.itertools.product(city_expansions,
*address_expansions):
yield u'{}|{}'.format(base_key, u'|'.join(keys))
def normalized_vector_l2(cls, vector):
n = math.sqrt(sum((s ** 2 for s in vector)))
if isclose(n, 0.0):
n = len(vector)
return [1. / n] * n
return [s / n for s in vector]
def normalized_vector_l1(cls, vector):
n = float(sum(vector))
if isclose(n, 0.0):
n = len(vector)
return [1. / n] * n
return [s / n for s in vector]
def normalized_tfidf_vector(cls, tfidf_vector):
norm = math.sqrt(sum((s ** 2 for w, s in tfidf_vector)))
if isclose(norm, 0.0):
return tfidf_vector
return [(w, s / norm) for w, s in tfidf_vector]
