def test4():
k = 8
A = ks.shingleset_k("hamsterkäufe", k)
A = A.union(ks.wildcard_shinglesets(A, n_max_wildcards=1))
B = ks.shingleset_k("hamsterkauf", k)
B = B.union(ks.wildcard_shinglesets(B, n_max_wildcards=1))
score = ks.jaccard(A, B)
assert 0.412 < score < 0.413
def test2():
s1 = ks.shingleset_k("abc", k=3)
s2 = ks.shingleset_range("abc", 1, 3)
s3 = ks.shingleset_list("abc", [1, 2, 3])
assert s1 == set(["a", "b", "c", "ab", "bc", "abc"])
assert s2 == s1
assert s3 == s2
def test3():
k = 8
A = ks.shingleset_k("hamsterkäufe", k)
B = ks.shingleset_k("hamsterkauf", k)
score = ks.jaccard(A, B)
assert 0.448 < score < 0.449
def test2():
k = 1
A = ks.shingleset_k("hamsterkäufe", k)
B = ks.shingleset_k("hamsterkauf", k)
score = ks.jaccard(A, B)
assert 0.909 < score < 0.910
def test1():
shingles = ks.shingleset_k("abc", k=3)
assert shingles == set(["a", "b", "c", "ab", "bc", "abc"])