Skip to content

A fast bloom filter implemented by Rust for Python! 10x faster than pybloom!

License

Notifications You must be signed in to change notification settings

yankun1992/fastbloom

Repository files navigation

fastbloom

OSCS Status docs.rs Test Rust Test Python Benchmark Crates Latest Release PyPI Latest Release Sonatype Nexus (Snapshots)

A fast bloom filter | counting bloom filter implemented by Rust for Rust and Python!

Language: 简体中文

setup

Python

requirements

Python >= 3.7

install

Install the latest fastbloom version with:

pip install fastbloom-rs

Rust

fastbloom-rs = "{latest}"

Java

maven

<dependency>
    <groupId>io.github.yankun1992</groupId>
    <artifactId>fastbloom</artifactId>
    <version>{latest-version}</version>
</dependency>

Examples

BloomFilter

A Bloom filter is a space-efficient probabilistic data structure, conceived by Burton Howard Bloom in 1970, that is used to test whether an element is a member of a set. False positive matches are possible, but false negatives are not.

Reference: Bloom, B. H. (1970). Space/time trade-offs in hash coding with allowable errors. Communications of the ACM, 13(7), 422-426. Full text article

Python

basic usage

from fastbloom_rs import BloomFilter

bloom = BloomFilter(100_000_000, 0.01)

bloom.add_str('hello')
bloom.add_bytes(b'world')
bloom.add_int(9527)

assert bloom.contains('hello')
assert bloom.contains(b'world')
assert bloom.contains(9527)

assert not bloom.contains('hello world')

build bloom filter from bytes or list

from fastbloom_rs import BloomFilter

bloom = BloomFilter(100_000_000, 0.01)
bloom.add_str('hello')
assert bloom.contains('hello')

bloom2 = BloomFilter.from_bytes(bloom.get_bytes(), bloom.hashes())
assert bloom2.contains('hello')

bloom3 = BloomFilter.from_int_array(bloom.get_int_array(), bloom.hashes())
assert bloom3.contains('hello')

there are some bulk api for python to reduce ffi cost between python and rust

bloom = BloomFilter(100_000_000, 0.01)
inserts = [1, 2, 3, 4, 5, 6, 7, 9, 18, 68, 90, 100]
checks = [1, 2, 3, 4, 5, 6, 7, 9, 18, 68, 90, 100, 190, 290, 390]
results = [True, True, True, True, True, True, True, True, True, True, True, True, False, False, False]

bloom.add_int_batch(inserts)
contains = bloom.contains_int_batch(checks)
assert contains == results

bloom.add_str_batch(list(map(lambda x: str(x), inserts)))
assert bloom.contains_str_batch(list(map(lambda x: str(x), checks))) == results

bloom.add_bytes_batch(list(map(lambda x: bytes(x), inserts)))
assert bloom.contains_bytes_batch(list(map(lambda x: bytes(x), checks))) == results

more examples at py_tests.

Rust

use fastbloom_rs::{BloomFilter, FilterBuilder};

let mut bloom = FilterBuilder::new(100_000_000, 0.01).build_bloom_filter();

bloom.add(b"helloworld");
assert_eq!(bloom.contains(b"helloworld"), true);
assert_eq!(bloom.contains(b"helloworld!"), false);

more examples at docs.rs

CountingBloomFilter

A Counting Bloom filter works in a similar manner as a regular Bloom filter; however, it is able to keep track of insertions and deletions. In a counting Bloom filter, each entry in the Bloom filter is a small counter associated with a basic Bloom filter bit.

Reference: F. Bonomi, M. Mitzenmacher, R. Panigrahy, S. Singh, and G. Varghese, “An Improved Construction for Counting Bloom Filters,” in 14th Annual European Symposium on Algorithms, LNCS 4168, 2006

Python

from fastbloom_rs import CountingBloomFilter

cbf = CountingBloomFilter(1000_000, 0.01)
cbf.add('hello')
cbf.add('hello')
assert 'hello' in cbf
cbf.remove('hello')
assert 'hello' in cbf  # because 'hello' added twice. 
# If add same element larger than 15 times, then remove 15 times the filter will not contain the element.
cbf.remove('hello')
assert 'hello' not in cbf

A CountingBloomFilter has a four bits counter to save hash index, so when insert an element repeatedly, the counter will spill over quickly. So, you can set enable_repeat_insert to False to check whether the element has added. if it has added, it will not add again. enable_repeat_insert default set to True.

from fastbloom_rs import CountingBloomFilter

cbf = CountingBloomFilter(1000_000, 0.01, False)
cbf.add('hello')
cbf.add('hello')  # because enable_repeat_insert=False, this addition will not take effect. 
assert 'hello' in cbf
cbf.remove('hello')
assert 'hello' not in cbf 

more examples at py_tests.

Rust

use fastbloom_rs::{CountingBloomFilter, FilterBuilder};

let mut builder = FilterBuilder::new(100_000, 0.01);
let mut cbf = builder.build_counting_bloom_filter();
cbf.add(b"helloworld");
assert_eq!(bloom.contains(b"helloworld"), true);

benchmark

computer info

CPU Memory OS
AMD Ryzen 7 5800U with Radeon Graphics 16G Windows 10

add one str to bloom filter

Benchmark insert one str to bloom filter:

bloom_add_test          time:   [41.168 ns 41.199 ns 41.233 ns]
                        change: [-0.4891% -0.0259% +0.3417%] (p = 0.91 > 0.05)
                        No change in performance detected.
Found 13 outliers among 100 measurements (13.00%)
  1 (1.00%) high mild
  12 (12.00%) high severe

add one million to bloom filter

Benchmark loop insert (1..1_000_000).map(|n| { n.to_string() }) to bloom filter:

bloom_add_all_test      time:   [236.24 ms 236.86 ms 237.55 ms]
                        change: [-3.4346% -2.9050% -2.3524%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 5 outliers among 100 measurements (5.00%)
  4 (4.00%) high mild
  1 (1.00%) high severe

check one contains in bloom filter

bloom_contains_test     time:   [42.065 ns 42.102 ns 42.156 ns]
                        change: [-0.7830% -0.5901% -0.4029%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 15 outliers among 100 measurements (15.00%)
  1 (1.00%) low mild
  5 (5.00%) high mild
  9 (9.00%) high severe

check one not contains in bloom filter

bloom_not_contains_test time:   [22.695 ns 22.727 ns 22.773 ns]
                        change: [-3.1948% -2.9695% -2.7268%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 12 outliers among 100 measurements (12.00%)
  4 (4.00%) high mild
  8 (8.00%) high severe

add one str to counting bloom filter

counting_bloom_add_test time:   [60.822 ns 60.861 ns 60.912 ns]
                        change: [+0.2427% +0.3772% +0.5579%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 10 outliers among 100 measurements (10.00%)
  1 (1.00%) low severe
  4 (4.00%) low mild
  1 (1.00%) high mild
  4 (4.00%) high severe

add one million to counting bloom filter

Benchmark loop insert (1..1_000_000).map(|n| { n.to_string() }) to counting bloom filter:

counting_bloom_add_million_test
                        time:   [272.48 ms 272.58 ms 272.68 ms]
Found 2 outliers among 100 measurements (2.00%)
  1 (1.00%) low mild
  1 (1.00%) high mild