def test_random_big(self, _, build, algorithm):
T, n, m, A = 100, 1000, 100, ['a', 'b', 'c']
for _ in range(T):
t = random_word(n, A)
W = {random_word(random.randint(2, 5), A) for _ in range(m)}
reference = [(w[1:], i) for w in W
for i in forward.brute_force(t, w, n,
len(w) - 1)]
self.check_all_matches(t, n, W, reference, build, algorithm)
def test_all_exact_string_matching(self):
N, M, A = 7, 3, ['a', 'b']
for n in range(2, N + 1):
for m in range(1, M + 1):
for t in itertools.product(A, repeat=n):
t = '#' + ''.join(t)
for w in itertools.product(A, repeat=m):
w = '#' + ''.join(w)
reference = list(forward.brute_force(t, w, n, m))
self.check_get_all_exact_matches(t, w, n, m, reference)
def test_random_small(self):
n, m, A = 100, 25, ['a', 'b', 'c']
for _ in range(100):
t = random_word(n, A)
patterns = {(random_word(randint(2, 5), A)) for _ in range(m)}
cw_automat = commentz_walter_build(patterns)
expected = set()
for p in patterns:
starts = brute_force(t, p, n, len(p) - 1)
indices = [(p, i) for i in starts]
expected.union(set(indices))
found = set(commentz_walter_search(t[1::], n, cw_automat))
self.assertSetEqual(expected, found)
def test_random(self):
n, m, A = 500, 30, ['a', 'b', 'c']
for _ in range(100):
t = random_word(n, A)
patterns = {random_word(randint(2, 10), A) for _ in range(m)}
automaton = self._create_from(patterns)
expected = set()
for p in patterns:
starts = brute_force(t, f'#{p}', n, len(p) + 1)
expected.union({(p, i) for i in starts})
found = set(find_occurrences(t, n, automaton))
self.assertSetEqual(expected, found)
def test_random_small(self):
n, m, A = 100, 25, ['a', 'b', 'c']
for _ in range(100):
t = random_word(n, A)
patterns = {(random_word(randint(2, 5), A)) for _ in range(m)}
cw_automat = fast_practical_multi_string_matching_build(patterns)
expected = set()
for p in patterns:
starts = brute_force(t, p, n, len(p) - 1)
indices = [(p, i) for i in starts]
expected.union(set(indices))
found = set(
fast_practical_multi_string_matching(t[1::], n, cw_automat))
self.assertSetEqual(expected, found)
def test_random_exact_string_matching(self):
T, n, m, A = 100, 500, 10, ['a', 'b']
for _ in range(T):
t, w = rand.random_word(n, A), rand.random_word(m, A)
reference = list(forward.brute_force(t, w, n, m))
self.check_get_all_exact_matches(t, w, n, m, reference)
