Feature Request: Vectorized API

[codingl2k1]

I have tested rbloom, and it is really fast. However, it would be beneficial if it provided a vectorized insert and query API. For example, it should accept an array of NumPy arrays or PyArrow arrays and return an array as well.

import time
import uuid

from rbloom import Bloom

print("generating data")
N = 1000000
data = [uuid.uuid4() for i in range(N)]
testdata = [uuid.uuid4() for i in range(N)]

print("Number of keys", len(data))

bf = Bloom(len(data), 0.00001)
for d in data:
    bf.add(d)

for d in data:
    assert d in bf

count = 0
start = time.time()
for x in testdata:
    count += x in bf
end = time.time()
querytime = end - start
fpp = count / N * 100.0
print(
    "false positive rate",
    "{:.5f}".format(fpp),
    "%",
    ", memory per key",
    "{:.1f}".format(bf.size_in_bits / N),
    "bits",
    ", millions of queries per second: ",
    "{:.2f}".format(N / querytime / 1000000),
    ", total memory",
    "{:.2f}".format(bf.size_in_bits / 8 / 1024.0 / 1024.0),
    "MiB",
)

Output

generating data
Number of keys 1000000
false positive rate 0.00100 % , memory per key 24.0 bits , millions of queries per second:  8.89 , total memory 2.86 MiB

My test env is Apple M4 Pro.