in cc_net/flat_hash_set.py [0:0]
def compare_load(*filenames):
assert filenames, "No file given"
def load_list():
hashes = []
for f in filenames:
h = FlatHashSet()
h.load(f)
print(f"Loaded {h} from {f}.")
hashes.append(h)
return hashes
def load_all(load, ext):
hashes = FlatHashSet()
for f in filenames:
load(hashes, f + ext)
def dump_all(hashes, dump, ext):
for h, f in zip(hashes, filenames):
dump(h, f + ext)
hashes = load_list()
dump_gp = getattr(FlatHashSet, "dump_gp")
if dump_gp is not None:
timeit("Dumping using gp.dump", dump_all, hashes, dump_gp, ".gp.test")
timeit("Dumping using dump_np", dump_all, hashes, FlatHashSet.dump_np, ".npy.test")
timeit(
"Dumping using dump_np2", dump_all, hashes, FlatHashSet.dump_np2, ".npy2.test"
)
load_gp = getattr(FlatHashSet, "load_gp")
if load_gp is not None:
timeit("Loading using gp.load", load_all, load_gp, ".gp.test")
timeit("Loading using load_np", load_all, FlatHashSet.load_np, ".npy.test")
timeit("Loading using load_np2", load_all, FlatHashSet.load_np2, ".npy2.test")