def test_address_normalization_nbsp(self):
# Since beautiful soup keeps in the output, it screws up google
# so it should be removed
test = u'WEST HALL ARCH \xa0 1085 SOUTH UNIVERSITY'
actual = parser.normalize_address(test)
expected = u'1085 SOUTH UNIVERSITY, Ann Arbor, MI'
self.assertEqual(expected, actual)
def test_address_normalization_nbsp(self):
# Since beautiful soup keeps in the output, it screws up google
# so it should be removed
test = u'WEST HALL ARCH \xa0 1085 SOUTH UNIVERSITY'
actual = parser.normalize_address(test)
expected = u'1085 SOUTH UNIVERSITY, Ann Arbor, MI'
self.assertEqual(expected, actual)
with open(file, 'r') as f:
data.extend(parser.parse_page(f.read()))
with open('.database', 'r') as f:
connection = sqlite3.connect(f.read().strip())
# Keep track of requests made, if we have made 2000 - stop because google will
# start denying at 2500
reqs = 0
# Sort and group by location so we can reduce the number of requests made to
# the database and insert in bulk
data = sorted(data, key=lambda x: x[2])
for location, group in groupby(data, key=lambda x: x[2]):
group = list(group)
location = parser.normalize_address(location)
print location, len(group)
# Test to see if the location is already in the database, if it is use that
# data, else make a request to google and save the response
query = 'SELECT Latitude, Longitude FROM Locations WHERE Location = ?'
cur = connection.execute(query, (location, ))
rv = cur.fetchall()
if rv:
lat, lng = float(rv[0][0]), float(rv[0][1])
else:
url = 'http://maps.googleapis.com/maps/api/geocode/json?sensor=false'
r = requests.get(url, params={'address': location})
results = r.json()['results']
reqs += 1
print "reqs: ", reqs
def test_empty_address(self):
test = u''
actual = parser.normalize_address(test)
expected = 'Ann Arbor, MI'
self.assertEqual(expected, actual)
def test_address_normalization(self):
test = u'1300 BLOCK CATHERINE STREET'
actual = parser.normalize_address(test)
expected = u'1300 BLOCK CATHERINE STREET, Ann Arbor, MI'
self.assertEqual(expected, actual)
def test_empty_address(self):
test = u''
actual = parser.normalize_address(test)
expected = 'Ann Arbor, MI'
self.assertEqual(expected, actual)
def test_address_normalization(self):
test = u'1300 BLOCK CATHERINE STREET'
actual = parser.normalize_address(test)
expected = u'1300 BLOCK CATHERINE STREET, Ann Arbor, MI'
self.assertEqual(expected, actual)
with open(file, 'r') as f:
data.extend(parser.parse_page(f.read()))
with open('.database', 'r') as f:
connection = sqlite3.connect(f.read().strip())
# Keep track of requests made, if we have made 2000 - stop because google will
# start denying at 2500
reqs = 0
# Sort and group by location so we can reduce the number of requests made to
# the database and insert in bulk
data = sorted(data, key=lambda x: x[2])
for location, group in groupby(data, key=lambda x: x[2]):
group = list(group)
location = parser.normalize_address(location)
print location, len(group)
# Test to see if the location is already in the database, if it is use that
# data, else make a request to google and save the response
query = 'SELECT Latitude, Longitude FROM Locations WHERE Location = ?'
cur = connection.execute(query, (location,))
rv = cur.fetchall()
if rv:
lat, lng = float(rv[0][0]), float(rv[0][1])
else:
url = 'http://maps.googleapis.com/maps/api/geocode/json?sensor=false'
r = requests.get(url, params={'address': location})
results = r.json()['results']
reqs += 1
print "reqs: ", reqs
