def test_new_set_methods(a, b):
# A whole bunch of new methods were added to the FSM module to enable FSMs to
# function exactly as if they were sets of strings (symbol lists), see:
# https://docs.python.org/3/library/stdtypes.html#set-types-set-frozenset
# But do they work?
assert len(a) == 1
assert len((a | b) * 4) == 16
try:
len(a.star())
assert False
except OverflowError:
pass
# "in"
assert "a" in a
assert not "a" in b
assert "a" not in b
# List comprehension!
four = (a | b) * 2
for string in four:
assert string == ["a", "a"]
break
assert [s for s in four] == [["a", "a"], ["a", "b"], ["b", "a"], ["b", "b"]]
# set.union() imitation
assert FSM.union(a, b) == a.union(b)
assert len(FSM.union()) == 0
assert FSM.intersection(a, b) == a.intersection(b)
# This takes a little explaining. In general, `a & b & c` is equivalent to
# `EVERYTHING & a & b & c` where `EVERYTHING` is an FSM accepting every
# possible string. Similarly `a` is equivalent to `EVERYTHING & a`, and the
# intersection of no sets at all is... `EVERYTHING`.
# However, since we compute the union of alphabets, and there are no
# alphabets, the union is the empty set. So the only string which `EVERYTHING`
# actually recognises is the empty string, [] (or "" if you prefer).
int_none = FSM.intersection()
assert len(int_none) == 1
assert [] in int_none
assert (a | b).difference(a) == FSM.difference((a | b), a) == (a | b) - a == b
assert (a | b).difference(a, b) == FSM.difference((a | b), a, b) == (a | b) - a - b == null("ab")
assert a.symmetric_difference(b) == FSM.symmetric_difference(a, b) == a ^ b
assert a.isdisjoint(b)
assert a <= (a | b)
assert a < (a | b)
assert a != (a | b)
assert (a | b) > a
assert (a | b) >= a
assert list(a.concatenate(a, a).strings()) == [["a", "a", "a"]]
assert list(a.concatenate().strings()) == [["a"]]
assert list(FSM.concatenate(b, a, b).strings()) == [["b", "a", "b"]]
assert list(FSM.concatenate().strings()) == []
assert not a.copy() is a
def to_fsm(self, alphabet=None, prefix_postfix=None, flags=None) -> FSM:
if alphabet is None:
alphabet = self.alphabet
if prefix_postfix is None:
prefix_postfix = self.prefix_postfix
all = _ALL.to_fsm(alphabet)
all_star = all.star()
fsm_parts = []
current = [all.times(prefix_postfix[0])]
for part in self.parts:
if isinstance(part, _NonCapturing):
inner = part.inner.to_fsm(alphabet, (0, 0), flags)
if part.backwards:
raise NotImplementedError("lookbacks are not implemented")
else:
# try:
# inner.cardinality()
# except OverflowError:
# raise NotImplementedError("Can not deal with infinite length lookaheads")
fsm_parts.append((None, current))
fsm_parts.append((part, inner))
current = []
else:
current.append(part.to_fsm(alphabet, (0, 0), flags))
current.append(all.times(prefix_postfix[1]))
result = FSM.concatenate(*current)
for m, f in reversed(fsm_parts):
if m is None:
result = FSM.concatenate(*f, result)
else:
assert isinstance(m, _NonCapturing) and not m.backwards
if m.negate:
result = result.difference(f)
else:
result = result.intersection(f + all_star)
return result
def test_concatenate_bug(a):
# This exposes a defect in fsm.concatenate.
assert FSM.concatenate(a, epsilon({"a"}), a).accepts("aa")
assert FSM.concatenate(a, epsilon({"a"}), epsilon({"a"}), a).accepts("aa")
