def benchmark_fst():
from rust_fst import Set
start = time.time()
lines = Set.from_iter(SENTENCES, path="temp.fst")
for key in KEYS:
matches = list(lines.search(key, max_dist=0))
print(time.time() - start)
def test_isdisjoint(tmpdir, fst_set):
oth_path = tmpdir.join('other.fst')
do_build(str(oth_path), keys=[u'ene', u'mene'])
other_set = Set(str(oth_path))
assert fst_set.isdisjoint(other_set)
assert other_set.isdisjoint(fst_set)
assert not fst_set.isdisjoint(fst_set)
assert not fst_set.issuperset(other_set)
assert not fst_set.issubset(other_set)
def do_build(path, keys=TEST_KEYS, sorted_=True):
with Set.build(path) as builder:
for key in (sorted(keys) if sorted_ else keys):
builder.insert(key)
def test_issuperset(tmpdir, fst_set):
oth_path = tmpdir.join('other.fst')
do_build(str(oth_path), keys=TEST_KEYS[:-2])
other_set = Set(str(oth_path))
assert fst_set.issuperset(other_set)
assert fst_set.issuperset(fst_set)
def test_load_badfile(tmpdir):
bad_path = tmpdir.join("bad.fst")
with bad_path.open('wb') as fp:
fp.write(b'\xFF' * 16)
with pytest.raises(lib.TransducerError):
Set(str(bad_path))
def test_build_memory():
memset = Set.from_iter(sorted(TEST_KEYS))
assert len(memset) == 4
def test_build_baddir():
fst_path = "/guaranteed-to-not-exist/set.fst"
with pytest.raises(OSError):
with Set.build(fst_path) as builder:
for key in sorted(TEST_KEYS):
builder.insert(key)
def fst_set(tmpdir):
fst_path = str(tmpdir.join('test.fst'))
do_build(fst_path)
return Set(fst_path)
def test_intersection():
a = Set.from_iter(["bar", "foo"])
b = Set.from_iter(["baz", "foo"])
assert list(a.intersection(b)) == ["foo"]
def test_symmetric_difference():
a = Set.from_iter(["bar", "foo"])
b = Set.from_iter(["baz", "foo"])
assert list(a.symmetric_difference(b)) == ["bar", "baz"]
def test_difference():
a = Set.from_iter(["bar", "foo"])
b = Set.from_iter(["baz", "foo"])
assert list(a.difference(b)) == ["bar"]
def test_union():
a = Set.from_iter(["bar", "foo"])
b = Set.from_iter(["baz", "foo"])
assert list(a.union(b)) == ["bar", "baz", "foo"]