def test_div_zero_contained(track_abundance):
# verify that empty MHs do not yield divide by zero errors for contained_by
mh = MinHash(1, 4, track_abundance=track_abundance)
mh2 = mh.copy_and_clear()
mh.add_sequence('ATGC')
assert mh.contained_by(mh2) == 0
assert mh2.contained_by(mh) == 0
def test_div_zero_contained(track_abundance):
# verify that empty MHs do not yield divide by zero errors for contained_by
mh = MinHash(1, 4, track_abundance=track_abundance)
mh2 = mh.copy_and_clear()
mh.add_sequence('ATGC')
assert mh.contained_by(mh2) == 0
assert mh2.contained_by(mh) == 0
def test_mh_copy_and_clear_with_max_hash(track_abundance):
# test basic creation of new, empty MinHash w/max_hash param set
a = MinHash(0, 10, track_abundance=track_abundance, max_hash=20)
for i in range(0, 40, 2):
a.add_hash(i)
b = a.copy_and_clear()
assert a.ksize == b.ksize
assert b.num == a.num
assert b.max_hash == 20
assert not b.is_protein
assert b.track_abundance == track_abundance
assert b.seed == a.seed
assert len(b.get_mins()) == 0
assert a.scaled == b.scaled
assert b.scaled != 0
def test_mh_copy_and_clear_with_max_hash(track_abundance):
# test basic creation of new, empty MinHash w/max_hash param set
a = MinHash(0, 10, track_abundance=track_abundance, max_hash=20)
for i in range(0, 40, 2):
a.add_hash(i)
b = a.copy_and_clear()
assert a.ksize == b.ksize
assert b.num == a.num
assert b.max_hash == 20
assert not b.is_protein
assert b.track_abundance == track_abundance
assert b.seed == a.seed
assert len(b.get_mins()) == 0
assert a.scaled == b.scaled
assert b.scaled != 0
