def test_year_notyear_number(self):
try:
expected = 999
actual = TextToNumEng.convertTryYear('nine hundred ninety nine')
self.assertEqual(actual, expected)
except Exception:
self.assertTrue(False, traceback.format_exc())
def test_year_invalid(self):
try:
actual = TextToNumEng.convertTryYear('nineteen hundred four')
self.assertIsNone(actual)
except Exception as e:
#print type(e), str(e)
self.assertTrue(isinstance(e, Exception))
def test_year_2015(self):
expected = 2015
actual = TextToNumEng.convertTryYear('twenty fifteen')
self.assertEqual(actual, expected)
def test_year_1992(self):
expected = 1992
actual = TextToNumEng.convertTryYear('nineteen ninety two')
self.assertEqual(actual, expected)
def test_year_984(self):
expected = 984
actual = TextToNumEng.convertTryYear("nine eighty four")
self.assertEqual(actual, expected)
def test_year_911(self):
expected = 911
actual = TextToNumEng.convertTryYear("nine 11")
self.assertEqual(actual, expected)
def test_year_1996(self):
expected = 1996
actual = TextToNumEng.convertTryYear("nineteen ninety six")
self.assertEqual(actual, expected)
def getNormOptions(word_string, extended=True):
# Returns possible ways to normalize the string
options_dict = defaultdict(set)
# Number conversion
if word_string.isdigit():
word_number = int(word_string)
# Convert digits to words
try:
num_text = NumToTextEng.convert(word_number)
options_dict['textnum'].add(num_text)
except:
pass
# Convert digits to year
try:
num_text = NumToTextEng.convertTryYear(word_number)
options_dict['textyear'].add(num_text)
except:
pass
else:
if extended and len(word_string.split()) > 1:
ext = set()
# Hyphenation
ext.add('-'.join(word_string.split()))
# All one single word
ext.add(''.join(word_string.split()))
# Strip punctuation (both sides)
ext.add(word_string.strip(string.punctuation))
# Strip left punct only
ext.add(word_string.lstrip(string.punctuation))
# Strip right punct only
ext.add(word_string.rstrip(string.punctuation))
options_dict['extended'].update(ext)
wstr = word_string
# Deyphenation
if '-' in word_string:
wstr = word_string.replace('-', ' ')
options_dict['text'].add(wstr)
# Convert words to digits
try:
num = TextToNumEng.convertTryYear(wstr)
options_dict['num'].add(str(num))
except:
pass
# Contractions
if "'" in word_string:
# Expand contractions
cont_options = ContractionsEng.expandOptions(word_string)
else:
# Apply contractions
cont_options = ContractionsEng.contractOptions(word_string)
if cont_options:
options_dict['text'].update(cont_options)
return options_dict