def test_lsm_tree_large_inserts(test_folder):
n = 30000
tree = LSMTree(test_folder)
for i in range(n):
tree.insert(i), i
for i in range(n):
assert tree.query(i), i
assert not tree.query(n)
def test_lsm_tree_random_inserts(test_folder):
n = 3000
arr = np.arange(n)
np.random.shuffle(arr)
tree = LSMTree(test_folder)
for i in arr:
tree.insert(i), i
for i in arr:
assert tree.query(i), i
assert not tree.query(n)
def test_lsm_tree_large_random_inserts(test_folder):
n = 100000
arr = np.arange(n)
rng = np.random.RandomState(42)
rng.shuffle(arr)
tree = LSMTree(test_folder)
for i in arr:
tree.insert(i), i
for i in np.arange(n):
assert tree.query(i), i
assert not tree.query(n)
def get_wods(wods_type, block_size, n_blocks, n_input_data):
path = f'./experiments/{wods_type}'
shutil.rmtree(path, ignore_errors=True)
params = {
'block_size': block_size,
'n_blocks': n_blocks,
'n_input_data': n_input_data,
}
if wods_type == 'b_epsilon_tree':
return BEpsilonTree(disk_filepath=path, **params, b=8)
elif wods_type == 'b_tree':
return BTree(disk_filepath=path, **params, b=8)
elif wods_type == 'lsm_tree':
return LSMTree(disk_filepath=path, enable_bloomfilter=False, growth_factor=16, **params)
elif wods_type == 'lsm_bf_tree':
return LSMTree(disk_filepath=path, enable_bloomfilter=True, growth_factor=16, **params)
elif wods_type == 'basic_cola':
return BasicCola(disk_filepath=path, growth_factor=16, **params)
elif wods_type == 'fractional_cola':
return FractionalCola(disk_filepath=path, growth_factor=16, **params)
elif wods_type == 'basic_bloom_cola':
return BasicBloomCola(disk_filepath=path, growth_factor=16, **params)
def test_level_metadata():
metadata = LSMTree.LevelMetadata()
metadata.insert(3, 3)
metadata.insert(1, 1)
metadata.insert(2, 2)
metadata.insert(4, 4)
metadata.insert(5, 5)
assert len(metadata) == 5
assert metadata.uuids == [1, 2, 3, 4, 5]
assert metadata.first_indices == [1, 2, 3, 4, 5]
metadata.clear(2, 4)
assert len(metadata) == 3
assert metadata.uuids == [1, 2, 5]
assert metadata.first_indices == [1, 2, 5]
metadata.insert(3, 3)
assert len(metadata) == 4
assert metadata.uuids == [1, 2, 3, 5]
assert metadata.first_indices == [1, 2, 3, 5]
metadata.clear()
assert len(metadata) == 0
assert metadata.uuids == []
assert metadata.first_indices == []
def test_lsm_tree_reverse_inserts(test_folder):
n = 1000000
tree = LSMTree(test_folder)
for i in reversed(range(n)):
tree.insert(i), i
assert not tree.query(n)