def __init__(self, fsm: MealyMachine):
self.fsm = fsm
self.A = fsm.get_alphabet()
self.states = self.fsm.get_states()
self.root = PartitionNode(self.states, self.A, self)
self.nodes = [self.root]
self.wanted = set([state.name for state in fsm.get_states()])
self.closed = set()
self.solution = set()
class LearnSimpleMealy(unittest.TestCase):
def setUp(self):
s1 = MealyState('1')
s2 = MealyState('2')
s3 = MealyState('3')
s1.add_edge('a', 'nice', s2)
s1.add_edge('b', 'B', s1)
s2.add_edge('a', 'nice', s3)
s2.add_edge('b', 'back', s1)
s3.add_edge('a', 'A', s3)
s3.add_edge('b', 'back', s1)
self.mm = MealyMachine(s1)
def test_lstar_wmethod(self):
eqc = WmethodEquivalenceChecker(self.mm, m=len(self.mm.get_states()))
teacher = Teacher(self.mm, eqc)
learner = LStarMealyLearner(teacher)
hyp = learner.run()
equivalent, _ = eqc.test_equivalence(hyp)
self.assertTrue(equivalent)
self.assertEqual(
len(self.mm.get_states()),
len(hyp.get_states()),
)
def test_lstar_bruteforce(self):
eqc = BFEquivalenceChecker(self.mm,
max_depth=len(self.mm.get_states()))
teacher = Teacher(self.mm, eqc)
learner = LStarMealyLearner(teacher)
hyp = learner.run()
equivalent, _ = WmethodEquivalenceChecker(
self.mm, m=len(self.mm.get_states())).test_equivalence(hyp)
self.assertTrue(equivalent)
self.assertEqual(
len(self.mm.get_states()),
len(hyp.get_states()),
)
def _render(fsm: MealyMachine, filename):
states = sorted(fsm.get_states(), key=lambda x: int(x.name.strip('s')))
alphabet = sorted(fsm.get_alphabet())
g = Digraph('G', filename=filename)
g.attr(rankdir='LR')
# Add states
for state in states:
g.node(state.name)
# Add transitions:
for state in states:
for action, (other_state, output) in sorted(state.edges.items(), key=lambda x: x[0]):
g.edge(state.name, other_state.name, label=f'{action}/{output}')
g.save()
def _minimize(mm: MealyMachine):
dset = get_distinguishing_set(mm)
dset_outputs = get_dset_outputs(mm, dset)
# Find non-unique states:
state_map = {}
for state, outputs in dset_outputs.items():
if outputs not in state_map:
state_map[outputs] = [state]
else:
state_map[outputs].append(state)
for outputs, states in state_map.items():
if len(states) > 1:
og_state = states[0]
rest_states = states[1:]
states = mm.get_states()
for state in states:
for action, (other_state, output) in state.edges.items():
if other_state in rest_states:
state.edges[action] = og_state, output
return mm
from stmlearn.equivalencecheckers import WmethodEquivalenceChecker
from stmlearn.learners import TTTMealyLearner
from stmlearn.suls import MealyState, MealyMachine
from stmlearn.teachers import Teacher
# Set up an example mealy machine
s1 = MealyState('1')
s2 = MealyState('2')
s3 = MealyState('3')
s1.add_edge('a', 'nice', s2)
s1.add_edge('b', 'B', s1)
s2.add_edge('a', 'nice', s3)
s2.add_edge('b', 'back', s1)
s3.add_edge('a', 'A', s3)
s3.add_edge('b', 'back', s1)
mm = MealyMachine(s1)
# Use the W method equivalence checker
eqc = WmethodEquivalenceChecker(mm, m=len(mm.get_states()))
teacher = Teacher(mm, eqc)
# We are learning a mealy machine
learner = TTTMealyLearner(teacher)
hyp = learner.run()
hyp.render_graph(tempfile.mktemp('.gv'))
learner.DTree.render_graph(tempfile.mktemp('.gv'))