def test_shared_suffix():
st = gwrite(enlist("blowing blue glowing"))
gr = greader(st)
cur1 = fst.Cursor(gr)
cur2 = fst.Cursor(gr)
cur1.find_path(b("blo"))
cur2.find_path(b("glo"))
assert cur1.stack[-1].target == cur2.stack[-1].target
def test_words():
words = enlist("alfa alpaca amtrak bellow fellow fiona zebulon")
with TempStorage() as st:
gwrite(words, st)
gr = greader(st)
cur = fst.Cursor(gr)
assert list(cur.flatten_strings()) == words
gr.close()
def test_fields():
with TempStorage() as st:
f = st.create_file("test")
gw = fst.GraphWriter(f)
gw.start_field("f1")
gw.insert("a")
gw.insert("aa")
gw.insert("ab")
gw.finish_field()
gw.start_field("f2")
gw.insert("ba")
gw.insert("baa")
gw.insert("bab")
gw.close()
gr = fst.GraphReader(st.open_file("test"))
cur1 = fst.Cursor(gr, gr.root("f1"))
cur2 = fst.Cursor(gr, gr.root("f2"))
assert list(cur1.flatten_strings()) == ["a", "aa", "ab"]
assert list(cur2.flatten_strings()) == ["ba", "baa", "bab"]
gr.close()
def _fst_roundtrip(domain, t):
with TempStorage() as st:
f = st.create_file("test")
gw = fst.GraphWriter(f, vtype=t)
gw.start_field("_")
for key, value in domain:
gw.insert(key, value)
gw.finish_field()
gw.close()
f = st.open_file("test")
gr = fst.GraphReader(f, vtype=t)
cur = fst.Cursor(gr)
assert list(cur.flatten_v()) == domain
f.close()
def test_inactive_raise():
st = gwrite(enlist("alfa bravo charlie"))
cur = fst.Cursor(greader(st))
while cur.is_active():
cur.next_arc()
pytest.raises(fst.InactiveCursor, cur.label)
pytest.raises(fst.InactiveCursor, cur.prefix)
pytest.raises(fst.InactiveCursor, cur.prefix_bytes)
pytest.raises(fst.InactiveCursor, list, cur.peek_key())
pytest.raises(fst.InactiveCursor, cur.peek_key_bytes)
pytest.raises(fst.InactiveCursor, cur.stopped)
pytest.raises(fst.InactiveCursor, cur.value)
pytest.raises(fst.InactiveCursor, cur.accept)
pytest.raises(fst.InactiveCursor, cur.at_last_arc)
pytest.raises(fst.InactiveCursor, cur.next_arc)
pytest.raises(fst.InactiveCursor, cur.follow)
pytest.raises(fst.InactiveCursor, cur.switch_to, b("a"))
pytest.raises(fst.InactiveCursor, cur.skip_to, b("a"))
pytest.raises(fst.InactiveCursor, list, cur.flatten())
pytest.raises(fst.InactiveCursor, list, cur.flatten_v())
pytest.raises(fst.InactiveCursor, list, cur.flatten_strings())
pytest.raises(fst.InactiveCursor, cur.find_path, b("a"))
def test_random():
def randstring():
length = random.randint(1, 5)
a = array("B", (random.randint(0, 255) for _ in xrange(length)))
return array_tobytes(a)
keys = sorted(randstring() for _ in xrange(100))
with TempStorage() as st:
gwrite(keys, st)
gr = greader(st)
cur = fst.Cursor(gr)
s1 = cur.flatten()
s2 = sorted(set(keys))
for i, (k1, k2) in enumerate(zip(s1, s2)):
assert k1 == k2, "%s: %r != %r" % (i, k1, k2)
sample = list(keys)
random.shuffle(sample)
for key in sample:
cur.reset()
cur.find_path(key)
assert cur.prefix_bytes() == key
gr.close()
def test_duplicate_keys():
st = gwrite(enlist("alfa bravo bravo bravo charlie"))
cur = fst.Cursor(greader(st))
assert list(cur.flatten_strings()) == ["alfa", "bravo", "charlie"]