def test_simple_case(self):
# construct nfa
pattern = "abc|bc|ad" # "a(b|c)*" | ""
lexer = Lexer(pattern)
lexer.analyse()
parser = Parser(lexer)
postfix_expr = parser.parse()
nfa_stack = TokenToNFA(postfix_expr).translate()
if len(nfa_stack) != 1:
raise PatternSyntaxError
nfa = nfa_stack[0]
# nfa.plot()
online = Online(nfa)
dfa = online.construct()
dfa.plot()
simplified_dfa = dfa.simplify()
simplified_dfa.plot("simplified_dfa")
self.assertTrue(0)
def test_simple_case(self):
pattern = "abc"
print("Pattern: {}".format(pattern))
lexer = Lexer(pattern)
lexer.analyse()
lexemes = lexer.tokens
for i in lexemes:
print(i)
def test_complicate_case(self):
pattern = r"(((d|e)?)|(a+))+"
print("Pattern: {}".format(pattern))
lexer = Lexer(pattern)
lexer.analyse()
lexemes = lexer.tokens
for i in lexemes:
print(i)
def test_middle_case(self):
pattern = r"a?b+c*(d|e)\\"
print("Pattern: {}".format(pattern))
lexer = Lexer(pattern)
lexer.analyse()
lexemes = lexer.tokens
for i in lexemes:
print(i)
def test_simple(self):
pattern = "abc"
print("Pattern: {}".format(pattern))
lexer = Lexer(pattern)
lexer.analyse()
parser = Parser(lexer)
postfix_expr = parser.parse()
for i in postfix_expr:
print(i)
self.assertEqual(1, 2)
def test_complicate_case(self):
pattern = r"(((d|e)?)|(a+))+"
print("Pattern: {}".format(pattern))
lexer = Lexer(pattern)
lexer.analyse()
parser = Parser(lexer)
postfix_expr = parser.parse()
for i in postfix_expr:
print(i)
self.assertEqual(1, 2)
def test_middle_case(self):
pattern = r"(ab|a)(bc|c)"
print("Pattern: {}".format(pattern))
lexer = Lexer(pattern)
lexer.analyse()
parser = Parser(lexer)
postfix_expr = parser.parse()
for i in postfix_expr:
print(i)
self.assertEqual(1, 2)
def compile(self, pattern, use_dfa=False):
lexer = Lexer(pattern)
lexer.analyse()
parser = Parser(lexer)
postfix_expr = parser.parse()
nfa_stack = TokenToNFA(postfix_expr).translate()
if len(nfa_stack) != 1:
raise PatternSyntaxError
automaton = nfa_stack[0]
if use_dfa:
nfa2dfa = NFA2DFA(automaton)
automaton = nfa2dfa.convert()
return automaton
def test_simple_case(self):
# construct nfa
# "a(b|c)*" | "abc|bc|ad" | "r0|r1|r2|r3|r4|r5|r6|r7|r8|r9" | "(a|b)*abb"
pattern = "a(b|c*)"
lexer = Lexer(pattern)
lexer.analyse()
parser = Parser(lexer)
postfix_expr = parser.parse()
nfa_stack = TokenToNFA(postfix_expr).translate()
if len(nfa_stack) != 1:
raise PatternSyntaxError
nfa = nfa_stack[0]
# nfa.plot()
online = Online(nfa)
dfa = online.construct()
# dfa.plot()
simplified_dfa = dfa.simplify()
simplified_dfa.plot("simplified_dfa")
brzozowski = Brzozowski(simplified_dfa)
reversed_nfa = brzozowski.reverse()
# reversed_nfa.plot("reversed_nfa")
postfix_minimal_dfa = brzozowski._construct_postfix_minimal_dfa()
# postfix_minimal_dfa.plot("postfix_minimal_dfa")
minimal_dfa = brzozowski.construct()
minimal_dfa.simplify().plot("minimal_dfa")