def transform_to_nfa(self, regex: str) -> Optional[NFA]:
"""
transform_to_nfa
Main entry point to convert regex to NFA by parsing through the regex.
Returns either the NFA or None if error occurred.
:param regex: The regular expression to convert.
:return: An equivalent NFA.
"""
if not regex:
logging.critical('Empty regex error')
return
# Make an NFA and initialize_nfa
nfa = NFA()
nfa.initialize_nfa(regex)
self.last_state = nfa.initial_state
self.open_groups.append(self.last_state)
# Parse regex:
# If I read an open grouping char (i.e. '(' or '['), add it as an open group.
# If I read a closed grouping char (i.e. ')' or ']'), move the last open state to closed.
# If I read an operator, apply the operator to the last closed group.
# If there are no closed groups, there is an error.
# If I read a character, process it and set it as the last closed group.
for char in regex:
# Check if it is a group
if char in RegexChar.opening_group():
self.open_groups.append(self.last_state)
elif char in RegexChar.closing_group():
# close the group
self.last_closed_group = self.open_groups[-1]
for union in self.union_in_progress[::-1]:
if union[0] == self.open_groups[-1]:
self.close_union(nfa)
self.open_groups = self.open_groups[:-1]
# check if the char is an operator
elif char in RegexChar.operators():
if char == RegexChar.UNION.value:
self.union_nfa(nfa)
# apply the operator to the last closed group
elif self.last_closed_group is None:
# Error in regex
logging.critical('Error applying operator: ' + str(char))
return
else:
if char == RegexChar.STAR.value:
self.star_nfa(nfa)
elif char == RegexChar.OPTIONAL.value:
self.option_nfa(nfa)
elif char == RegexChar.PLUS.value:
self.plus_nfa(nfa)
# build the state
elif char.isalnum():
# if alphanumeric, concatenate
nfa = self.concatenate_nfa(nfa, char)
else:
# error
logging.critical('Error reading character: ' + str(char))
return
# post processing
self.last_closed_group = self.open_groups[-1]
if self.union_in_progress:
for union in self.union_in_progress[::-1]:
self.close_union(nfa)
self.open_groups = self.open_groups[:-1]
return nfa
