def build(self):
self._reset_nfa()
self.regex_postfix = prepare_regex(self.regex_infix)
sm_queue = []
for char in self.regex_postfix:
if char == RegexConfig.ZERO_OR_MORE:
if len(sm_queue) < 1:
raise ValueError('Regex parsing error')
sm1 = sm_queue.pop()
sm = self._build_zero_or_more_sm(sm1)
sm_queue.append(sm)
elif char == RegexConfig.ONE_OR_MORE:
if len(sm_queue) < 1:
raise ValueError('Regex parsing error')
sm1 = sm_queue.pop()
sm = self._build_one_or_more_sm(sm1)
sm_queue.append(sm)
elif char == RegexConfig.AND:
if len(sm_queue) < 2:
raise ValueError('Regex parsing error')
sm2 = sm_queue.pop()
sm1 = sm_queue.pop()
sm = self._build_and_sm(sm1, sm2)
sm_queue.append(sm)
elif char == RegexConfig.OR:
if len(sm_queue) < 2:
raise ValueError('Regex parsing error')
sm2 = sm_queue.pop()
sm1 = sm_queue.pop()
sm = self._build_or_sm(sm1, sm2)
sm_queue.append(sm)
# [A-Za-z0-9]
elif char in RegexConfig.VALID_SYMBOLS:
sm = self._build_basic_sm(char)
sm_queue.append(sm)
else:
raise ValueError('Regex parsing error')
self.nfa = sm_queue.pop()
def test_miss_opening_bracket(self):
with self.assertRaises(ValueError):
prepare_regex('a|b)')
def test_unacceptable(self):
with self.assertRaises(ValueError):
prepare_regex('abc123$')
def test(self):
result = ''.join(prepare_regex('(a|b|c)&(a|b)'))
print(result)
def test_concatenation(self):
result = ''.join(prepare_regex('(a|b)*a?b?c'))
self.assertEqual(result, 'ab|*a?&b?&c&')
def test_correct_simple(self):
result = ''.join(prepare_regex('(a|b)*&a?&b?&c'))
self.assertEqual(result, 'ab|*a?&b?&c&')
def test_spaces(self):
result = ''.join(prepare_regex('a | b'))
self.assertEqual(result, 'ab|')
def test_miss_closing_bracket(self):
with self.assertRaises(ValueError):
prepare_regex('(a|b')