def verify_popcount_indices(arena):
assert len(arena.popcount_indices) % 4 == 0
format = "i"*(len(arena.popcount_indices)//4)
values = struct.unpack(format, arena.popcount_indices)
assert len(values) == arena.metadata.num_bits+2, (len(values), arena.metadata.num_bits)
assert values[-1] == len(arena), (values[-1], len(arena), values)
for i in range(len(values)-1):
start = values[i]
end = values[i+1]
for j in range(start, end):
fp = arena[j][1]
assert byte_popcount(fp) == i, (byte_popcount(fp), i, start, end)
def test_arena_is_ordered_by_popcount(self):
arena = self._open(CHEBI_TARGETS)
prev = 0
for id, fp in arena:
popcount = bitops.byte_popcount(fp)
self.assertTrue(prev <= popcount, (prev, popcount))
prev = popcount
def test_every_bit_sorted(self):
# This is a bit tricker since there's no guaranteed order
# of the contents of the arena
all_bytes = [chr(i) for i in range(256)]
expected = sorted(byte_popcount(fp) for fp in all_bytes)
arena = _load(all_bytes, True)
popcounts = map(byte_popcount, arena.arena)
self.assertEquals(popcounts, expected)
verify_popcount_indices(arena)
arena = _load(all_bytes[::-1], True)
popcounts = map(byte_popcount, arena.arena)
self.assertEquals(popcounts, expected)
verify_popcount_indices(arena)
def test_iter_arenas_select_size(self):
arena = self._open(CHEBI_TARGETS)
ids = [id for (id, fp) in arena]
prev = 0
for subarena in arena.iter_arenas(100):
self._check_target_metadata(subarena.metadata)
self.assertEqual(len(subarena), 100)
subids = []
for id, fp in subarena:
subids.append(id)
popcount = bitops.byte_popcount(fp)
self.assertTrue(prev <= popcount, (prev, popcount))
prev = popcount
self.assertEquals(ids[:100], subids)
del ids[:100]
self.assertFalse(ids)