def testHighwayAbbreviations(self):
'hwy, expy'
# 3 IN 1 KITCHEN: 4902 FORT HAMILTON PARKWAY BROOKLYN, NY
text = "238 KINGS HWY BROOKLYN, NY"
expected = "238 KINGS Highway BROOKLYN, NY"
got = parser.parse(text)[0]
self.assertEqual(got, expected)
text = "3050 WHITESTONE EXPY QUEENS, NY"
expected = "3050 WHITESTONE Expressway QUEENS, NY"
got = parser.parse(text)[0]
self.assertEqual(got, expected)
def testAptAndSuite(self):
'handle ste and apt'
text = "35 WEST 89TH STREET APT. 1A NEW YORK, NY"
expected = "35 WEST 89TH STREET Apt 1A Manhattan, NY"
got = parser.parse(text)[0]
self.assertEqual(got, expected)
text = "35 WEST 89TH STREET STE. 1A NEW YORK, NY"
expected = "35 WEST 89TH STREET Suite 1A Manhattan, NY"
got = parser.parse(text)[0]
self.assertEqual(got, expected)
def __init__(self, source, verbose=False):
self.source = source
self.address = parser.parse(source, verbose=verbose)
if self.address:
self.address = self.address[0]
else:
self.address = None
def testAddressWithMultipleCity(self):
'test ... Queens, NY NY, NY finds first address'
text = "11 W. 19th Street, NY, NY 10011 , New York, NY"
expected = "11 W 19th Street, Manhattan, NY"
got = parser.parse(text)[0]
self.assertEqual(expected, got)
def testStreetNamePreTypeAveOfAmericas(self):
'find Avenue of the Americas'
expected = "131 Avenue Of The Americas Manhattan, NY"
text = 'blab blah bleu %s foo fe hu' % expected
got = parser.parse(text)
self.assertIn(got[0], [expected])
def checkExpectation(self, sample, expect, verbose=False):
source = os.path.join(self.datadir, sample)
expectation = os.path.join(self.datadir, expect)
expected = open(expectation).readlines()
expected = [e.strip() for e in expected]
text = codecs.open(source, 'r', encoding='utf8') \
.read().encode('ascii', 'ignore')
addresses = parser.parse(text, verbose)
if verbose:
print 'expect:\t'
if isinstance(expected, list):
print
for e in expected:
print '\t%s' % e
else:
print '%s' % expected
print
print 'got \t:'
if isinstance(addresses, list):
print
for e in addresses:
print '\t%s' % e
else:
print '%s' % addresses
for loc in addresses:
self.assertIn(loc, expected)
expected.remove(loc)
self.assertEqual(expected, [])
def checkExpectation(self, source, expected, verbose=False):
addresses = parser.parse(source, verbose)
if verbose:
print 'source: %s' % source
print 'expected: %s' % expected
print 'got: %s' % addresses
for loc in addresses:
self.assertIn(loc, expected)
expected.remove(loc)
self.assertEqual(expected, [])
def testSaintNotStreet(self):
#This is unimplemented
'701 St. Anns should resolve to Saint Anns instead of Street Anns'
expected = ['701 St. Anns Avenue Bronx, NY']
for text in expected:
print text
text = 'blab blah bleu %s foo fe hu' % text
got = parser.parse(text, verbose=True)[0]
self.assertIn(got, expected)
def testInitials(self):
text = '''
1180 Reverend J.A. Polite Ave. Bronx, NY.
'''
expected = [
'1180 Reverend J A Polite Avenue Bronx, NY'
]
got = parser.parse(text)[0]
self.assertIn(got, expected)
def testStreetNamePreTypes(self):
'test Avenue xxx'
expected = [
"1600 Avenue L Brooklyn, NY",
"3000 Avenue X Brooklyn, NY",
"50 Avenue X Brooklyn, NY"
]
for text in expected:
text = 'blab blah bleu %s foo fe hu' % text
got = parser.parse(text)[0]
self.assertIn(got, expected)
def testPeriodBetweenDirectionAndStreet(self):
'period handled between direction and street'
text = "Decker Design: 14W.23rd Street 3rd Floor, New York, NY"
expected = "14 W 23rd Street 3rd Floor, Manhattan, NY"
got = parser.parse(text)[0]
self.assertEqual(got, expected)
def testInferredStreet(self):
'infer street in manhattan'
text = "10 W 15th , New York, NY"
expected = "10 W 15th Street, Manhattan, NY"
got = parser.parse(text)[0]
self.assertEqual(got, expected)
def testSaintAnneAvenue(self):
'name with apostrophe'
expected = [u"600 Saint Ann's Avenue Bronx, NY"]
text = "Academy of Science: 600 Saint Ann's Avenue Bronx, NY"
address = parser.parse(text)[0]
self.assertIn(address, expected)