def test_simple_casefold(self):
"""Test for simple casefolding."""
self.assertEqual(u"weiß", casefold(u"weiß", fullcasefold=False))
self.assertEqual(u"weiß", casefold(u"WEIẞ", fullcasefold=False))
# GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
self.assertEqual(u"\u0390", casefold(u"\u0390", fullcasefold=False))
# GREEK CAPITAL LETTER ALPHA WITH PSILI AND PROSGEGRAMMENI
self.assertEqual(u"\u1f80", casefold(u"\u1f88", fullcasefold=False))
def test_full_casefold(self):
"""Test for full casefolding (explicitly specifying fullcasefold=True)."""
self.assertEqual(u"weiss", casefold(u"weiß", fullcasefold=True))
self.assertEqual(u"weiss", casefold(u"WEIẞ", fullcasefold=True))
# GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
self.assertEqual(u"\u03b9\u0308\u0301",
casefold(u"\u0390", fullcasefold=True))
# GREEK CAPITAL LETTER ALPHA WITH PSILI AND PROSGEGRAMMENI
self.assertEqual(u"\u1f00\u03b9", casefold(u"\u1f88",
fullcasefold=True))
def test_default_casefold(self):
input_string = u"WEIẞ"
resp = requests.post(base_url + "casefold",
json={"input_string": input_string})
resp_str = self.leading_trailing_quotes.sub(
'',
resp.content.strip().decode('unicode-escape'))
self.assertEqual(casefold(input_string), resp_str)
def test_turkic_casefold(self):
input_string = u"DİYARBAKIR"
resp = requests.post(base_url + "casefold",
json={"input_string": input_string})
resp_str = self.leading_trailing_quotes.sub(
'',
resp.content.strip().decode('unicode-escape'))
self.assertEqual(casefold(input_string), resp_str)
resp = requests.post(base_url + "casefold",
json={
"input_string": input_string,
"useturkicmapping": True
})
resp_str = self.leading_trailing_quotes.sub(
'',
resp.content.strip().decode('unicode-escape'))
self.assertEqual(casefold(input_string, useturkicmapping=True),
resp_str)
def test_turkic_casefold(self):
"""Test for casefolding with the useturkicmapping parameter."""
s1 = u"DİYARBAKIR"
s2 = u"Diyarbakır"
self.assertNotEqual(casefold(s1), casefold(s2))
self.assertEqual(casefold(s1, useturkicmapping=True),
casefold(s2, useturkicmapping=True))
s1 = u"MISSISSIPPI"
s2 = u"mississippi"
self.assertEqual(casefold(s1), casefold(s2))
self.assertNotEqual(casefold(s1, useturkicmapping=True),
casefold(s2, useturkicmapping=True))
def do_casefold():
fullcasefold = request.json[
'fullcasefold'] if 'fullcasefold' in request.get_json() else True
useturkicmapping = request.json[
'useturkicmapping'] if 'useturkicmapping' in request.get_json(
) else False
return jsonify(
casefold(request.json['input_string'],
fullcasefold=fullcasefold,
useturkicmapping=useturkicmapping))
def test_default_casefold(self):
"""
This test is based on the str.casefold() tests for python 3. Source:
https://github.com/python/cpython/blob/6f870935c2e024cbd1cc379f85e6a66d7711dcc7/Lib/test/test_unicode.py#L770
"""
self.assertEqual(casefold(u"hello"), u"hello")
self.assertEqual(casefold(u"hELlo"), u"hello")
self.assertEqual(casefold(u"ß"), u"ss")
self.assertEqual(casefold(u"fi"), u"fi")
self.assertEqual(casefold(u"\u03a3"),
u"\u03c3") # GREEK CAPITAL LETTER SIGMA
self.assertEqual(casefold(u"A\u0345\u03a3"),
u"a\u03b9\u03c3") # COMBINING GREEK YPOGEGRAMMENI
self.assertEqual(casefold(u"\u00b5"), u"\u03bc") # MICRO SIGN
def test_full_casefold_surrogate_pair(self):
"""Test for full casefolding of a string with surrogate pairs."""
# A; B; OSAGE CAPITAL LETTER A; C; D; WARANG CITI CAPITAL LETTER NGAA; E; F
self.assertEqual(
u"ab\U000104D8cd\U000118C0ef",
casefold(u"AB\U000104B0CD\U000118A0EF", fullcasefold=True))