def test_length(self):
char_class = CharacterClass('0-9A-Z')
self.assertListEqual(char_class.positive.codepoints, [(48, 58),
(65, 91)])
self.assertListEqual(char_class.negative.codepoints, [])
self.assertEqual(len(char_class), 36)
char_class.complement()
self.assertListEqual(char_class.positive.codepoints, [])
self.assertListEqual(char_class.negative.codepoints, [(48, 58),
(65, 91)])
self.assertEqual(len(char_class), sys.maxunicode + 1 - 36)
char_class.add('k-m')
self.assertListEqual(char_class.positive.codepoints, [(107, 110)])
self.assertListEqual(char_class.negative.codepoints, [(48, 58),
(65, 91)])
self.assertEqual(str(char_class), '[\x00-/:-@\\[-\U0010ffffk-m]')
self.assertEqual(len(char_class), sys.maxunicode + 1 - 36)
char_class.add('K-M')
self.assertListEqual(char_class.positive.codepoints, [(75, 78),
(107, 110)])
self.assertListEqual(char_class.negative.codepoints, [(48, 58),
(65, 91)])
self.assertEqual(len(char_class), sys.maxunicode + 1 - 33)
self.assertEqual(str(char_class), '[\x00-/:-@\\[-\U0010ffffK-Mk-m]')
char_class.clear()
self.assertListEqual(char_class.positive.codepoints, [])
self.assertListEqual(char_class.negative.codepoints, [])
self.assertEqual(len(char_class), 0)
def test_iterate(self):
char_class = CharacterClass('A-Za-z')
self.assertEqual(''.join(chr(c) for c in char_class),
string.ascii_uppercase + string.ascii_lowercase)
char_class.complement()
self.assertEqual(len(''.join(chr(c) for c in char_class)),
sys.maxunicode + 1 - len(string.ascii_letters))
def test_char_class_init(self):
char_class = CharacterClass()
self.assertEqual(char_class.positive, [])
self.assertEqual(char_class.negative, [])
char_class = CharacterClass('a-z')
self.assertEqual(char_class.positive, [(97, 123)])
self.assertEqual(char_class.negative, [])
def test_in_operator(self):
char_class = CharacterClass('A-Za-z')
self.assertIn(100, char_class)
self.assertIn('d', char_class)
self.assertNotIn(49, char_class)
self.assertNotIn('1', char_class)
char_class.complement()
self.assertNotIn(100, char_class)
self.assertNotIn('d', char_class)
self.assertIn(49, char_class)
self.assertIn('1', char_class)
def test_complement(self):
char_class = CharacterClass('a-z')
self.assertListEqual(char_class.positive.codepoints, [(97, 123)])
self.assertListEqual(char_class.negative.codepoints, [])
char_class.complement()
self.assertListEqual(char_class.positive.codepoints, [])
self.assertListEqual(char_class.negative.codepoints, [(97, 123)])
self.assertEqual(str(char_class), '[^a-z]')
char_class = CharacterClass()
char_class.complement()
self.assertEqual(len(char_class), sys.maxunicode + 1)
def character_class_objects():
return [CharacterClass(r'\c') for _ in range(10000)]
def run_timeit(stmt='pass', setup='pass', number=1000):
seconds = timeit(stmt, setup=setup, number=number)
print("{}: {}s".format(stmt, seconds))
@profile
def character_class_objects():
return [CharacterClass(r'\c') for _ in range(10000)]
if __name__ == '__main__':
print('*' * 62)
print("*** Memory and timing profile of CharacterClass class ***")
print("***" + ' ' * 56 + "***")
print("*** Note: save ~15% of memory with __slots__ (from v2.2.3) ***")
print('*' * 62)
print()
character_class_objects()
character_class = CharacterClass(r'\c')
character_class -= CharacterClass(r'\i')
SETUP = 'from __main__ import character_class'
NUMBER = 10000
run_timeit('"9" in character_class # True ', SETUP, NUMBER)
run_timeit('"q" in character_class # False', SETUP, NUMBER)
run_timeit('8256 in character_class # True ', SETUP, NUMBER)
run_timeit('8257 in character_class # False', SETUP, NUMBER)
def test_discard(self):
char_class = CharacterClass('0-9')
char_class.discard('6-9')
self.assertListEqual(char_class.positive.codepoints, [(48, 54)])
self.assertListEqual(char_class.negative.codepoints, [])
self.assertEqual(len(char_class), 6)
char_class.add(r'\p{Nd}')
self.assertEqual(len(char_class), 630)
char_class.discard(r'\p{Nd}')
self.assertEqual(len(char_class), 0)
with self.assertRaises(RegexError):
char_class.discard(r'\p{}')
with self.assertRaises(RegexError):
char_class.discard(r'\p{XYZ}')
char_class.add(r'\P{Nd}')
self.assertEqual(len(char_class), sys.maxunicode + 1 - 630)
char_class.discard(r'\P{Nd}')
self.assertEqual(len(char_class), 0)
char_class = CharacterClass('a-z')
char_class.discard(r'\p{IsFoo}')
self.assertEqual(len(char_class), 0)
char_class = CharacterClass()
char_class.complement()
char_class.discard('\\n')
self.assertListEqual(char_class.positive.codepoints, [(0, 10),
(11, 1114112)])
self.assertListEqual(char_class.negative.codepoints, [])
self.assertEqual(len(char_class), sys.maxunicode)
char_class.discard('\\s')
self.assertListEqual(char_class.positive.codepoints, [(0, 9), (11, 13),
(14, 32),
(33, 1114112)])
self.assertEqual(len(char_class), sys.maxunicode - 3)
char_class.discard('\\S')
self.assertEqual(len(char_class), 0)
char_class.clear()
char_class.negative.codepoints.append(10)
char_class.discard('\\s')
self.assertListEqual(char_class.positive.codepoints, [])
self.assertListEqual(char_class.negative.codepoints, [(9, 11), 13, 32])
char_class = CharacterClass('\t')
char_class.complement()
self.assertListEqual(char_class.negative.codepoints, [9])
char_class.discard('\\n')
self.assertListEqual(char_class.positive.codepoints, [])
self.assertListEqual(char_class.negative.codepoints, [(9, 11)])
self.assertEqual(len(char_class), sys.maxunicode - 1)
def test_add(self):
char_class = CharacterClass()
self.assertListEqual(char_class.positive.codepoints, [])
self.assertListEqual(char_class.negative.codepoints, [])
self.assertEqual(len(char_class), 0)
char_class.add('0-9')
self.assertListEqual(char_class.positive.codepoints, [(48, 58)])
self.assertListEqual(char_class.negative.codepoints, [])
self.assertEqual(len(char_class), 10)
char_class.add(r'\p{Nd}')
self.assertEqual(len(char_class), 630)
with self.assertRaises(RegexError):
char_class.add(r'\p{}')
with self.assertRaises(RegexError):
char_class.add(r'\p{XYZ}')
char_class.add(r'\P{Nd}')
self.assertEqual(len(char_class), sys.maxunicode + 1)
char_class = CharacterClass()
char_class.add(r'\p{IsFoo}')
def test_isub_operator(self):
char_class = CharacterClass('A-Za-z')
char_class -= CharacterClass('a-z')
self.assertEqual(str(char_class), '[A-Z]')
char_class = CharacterClass('a-z')
other = CharacterClass('A-Za-c')
other.complement()
char_class -= other
self.assertEqual(str(char_class), '[a-c]')
char_class = CharacterClass('a-z')
other = CharacterClass('A-Za-c')
other.complement()
other.add('b')
char_class -= other
self.assertEqual(str(char_class), '[ac]')
char_class = CharacterClass('a-c')
char_class.complement()
other = CharacterClass('a-z')
other.complement()
char_class -= other
self.assertEqual(str(char_class), '[d-z]')
def test_char_class_repr(self):
char_class = CharacterClass('a-z')
self.assertEqual(repr(char_class), 'CharacterClass([a-z])')
char_class.complement()
self.assertEqual(repr(char_class), 'CharacterClass([^a-z])')
