🌸 Bloomz

Fast, flexible Bloom filter for Rust with pluggable hashers and parallel operations.

Features

• Fast: Optimized bit operations with efficient double hashing
• Flexible: Pluggable hash builders (SipHash, AHash, xxHash, etc.)
• Parallel: Batch operations with Rayon for multi-core performance
• Serializable: JSON and binary serialization with Serde
• Safe: No unsafe code, extensive testing

Quick Start

Add to your Cargo.toml:

[dependencies]

bloomz = "0.1"



# Enable optional features

bloomz = { version = "0.1", features = ["serde", "rayon"] }

Basic Usage

use bloomz::BloomFilter;

// Create a filter for ~1000 items with 1% false positive rate
let mut filter = BloomFilter::new_for_capacity(1000, 0.01);

// Insert items
filter.insert(&"hello");
filter.insert(&42);

// Check membership
assert!(filter.contains(&"hello"));
assert!(!filter.contains(&"world"));

Parallel Operations (with rayon feature)

use bloomz::BloomFilter;
use rayon::prelude::*;
use std::collections::hash_map::RandomState;

let rs = RandomState::new();
let mut filter = BloomFilter::with_hasher(10000, 7, rs);

// Parallel batch insert
let items: Vec<i32> = (0..1000).collect();
filter.insert_batch(items.par_iter().cloned());

// Parallel batch contains
let test_items: Vec<i32> = (500..600).collect();
let results = filter.contains_batch(test_items.par_iter().cloned());

// Check if all items are present
let all_present = filter.contains_all(test_items.par_iter().cloned());

Serialization (with serde feature)

use bloomz::BloomFilter;

let mut filter = BloomFilter::new_for_capacity(100, 0.01);
filter.insert(&"data");

// JSON serialization
let json = serde_json::to_string(&filter)?;
let restored: BloomFilter = serde_json::from_str(&json)?;

// Binary serialization  
let bytes = filter.to_bytes();
let restored = BloomFilter::from_bytes(&bytes).unwrap();

Custom Hash Builders

use bloomz::BloomFilter;
use std::collections::hash_map::RandomState;

// Default SipHash (secure)
let filter1 = BloomFilter::new(1000, 5);

// Custom RandomState
let rs = RandomState::new();
let filter2 = BloomFilter::with_hasher(1000, 5, rs);

// Fast hashers (requires feature flags)
#[cfg(feature = "fast-ahash")]
{
    use ahash::AHasher;
    let filter3 = BloomFilter::with_hasher(1000, 5, 
        ahash::RandomState::new());
}

Performance

Bloomz uses several optimizations:

• Double Hashing: Generate k hash functions from just 2 base hashes
• Efficient Bit Operations: Word-aligned bit manipulation with u64
• Parallel Processing: Multi-threaded batch operations with Rayon
• Zero-Copy Serialization: Direct bit vector serialization

Benchmarks

Run benchmarks to compare hashers and parallel vs sequential operations:

# Compare different hash builders

cargo bench --features "fast-ahash,fast-xxh3" bloom_hashers


# Compare parallel vs sequential operations  

cargo bench --features rayon parallel_bloom

API Reference

Core Types

• BloomFilter<S> - Main bloom filter with hasher type S
• BitSet - Underlying bit storage with optimized operations

Key Methods

Insertion

• insert(&item) - Insert a single item
• insert_batch(items) - Parallel batch insert (rayon feature)

Membership

• contains(&item) - Check if item is probably in set
• contains_batch(items) - Parallel batch check (rayon feature)
• contains_all(items) - Check if all items are present (rayon feature)

Set Operations

• union_inplace(&other) - Merge with another filter
• intersect_inplace(&other) - Keep only common elements
• clear() - Remove all items

Serialization

• to_bytes() / from_bytes() - Binary format
• Serde support for JSON/other formats

Mathematical Functions

use bloomz::math;

// Calculate optimal parameters
let m = math::optimal_m(n_items, false_positive_rate);
let k = math::optimal_k(m, n_items);

let filter = BloomFilter::new(m, k);

Feature Flags

Examples

See src/main.rs for a complete web crawler URL filter demo:

# Basic demo

cargo run


# With all features

cargo run --features "rayon,serde,fast-ahash"

Use Cases

• Web Crawlers: Avoid revisiting URLs
• Caching: Quick "not in cache" checks
• Databases: Reduce disk lookups
• Networking: Packet deduplication
• Analytics: Unique visitor tracking

Contributing

Contributions welcome! Please check:

• Run cargo test --all-features
• Run cargo bench --all-features
• Add tests for new features
• Update documentation

License

MIT License - see LICENSE file.

🌸 Bloomz: Where speed meets flexibility in Rust Bloom filters!