hashkit-match

hashkit-match is an in-memory multi-index Hamming matcher for PDQ hashes. It pairs with hashkit: once you can compute PDQ hashes, this layer matches incoming hashes against a known-bad hash set.

Install

cargo add hashkit-match

What it does

• Provides a multi-index Hamming (MIH) matcher over the 256-bit PDQ hash space, avoiding naive linear comparison at scale.
• Matches incoming hashes against caller-supplied hash sets you've received from sources such as NCMEC, IWF, or Project Arachnid.
• Uses a configurable distance threshold, defaulting to 31/256 (the PhotoDNA-equivalent threshold).
• Is a pure data structure that ships no hash lists.
• Offers bindings parallel to hashkit across Rust, WASM, Node, Deno, Bun, and Python.

Quickstart

use hashkit_match::Matcher;

// Build an index from a caller-supplied set of known PDQ hashes.
let mut matcher = Matcher::new();
matcher.insert(known_hash);

// Query an incoming hash; the default threshold is 31/256.
if let Some(hit) = matcher.query(incoming_hash) {
    println!("matched within distance {}", hit.distance);
}

Status

Pre-release: the first crates.io publish is still pending, and the API is planned to ship alongside hashkit. Treat names and signatures as subject to change until the initial release.

Source

packages/hashkit-match