Tiny JS values canonicalization for hashing.

It uses a simple serialization algorithm, generating a consistent string representation of JS values. It is built to use with Smol xxHash.

It is tiny and efficient. It is just 185B and 30%+ faster than other stable serialization libraries.

Unlike alternatives, it is focused on hashing and doesn't produce valid JSON, making it more efficient, and also supports more value types, i.e., undefined.

It features dual CJS/ESM support and built-in TypeScript definitions.

The package is available on npm:

npm install smolcanon

Pass any JS value to the canonize function to get its string representation:

import { canonize } from "smolcanon";

const canon = canonize({ foo: "bar", baz: "qux" });
// => '{foo:"bar";baz:"qux"}'

You can use it with Smol xxHash to produce consistent hashes for your data:

import { canonize } from "smolcanon";
import { xxh32 } from "smolxxh";

const canon = canonize({ foo: "bar", baz: "qux" });
const hash = xxh32(Buffer.from("hello world", "utf8")).toString(16);
//=> "ed4e5029"

The benchmark shows that Smol Canon is significantly faster than other popular libraries for canonicalizing JavaScript values:

canonicalize:
  4 197 ops/s, ±0.28%   | 43.59% slower

json-canon:
  5 171 ops/s, ±2.11%   | 30.5% slower

fast-json-stable-stringify:
  4 548 ops/s, ±2.32%   | 38.87% slower

fast-safe-stringify:
  5 310 ops/s, ±2.30%   | 28.63% slower

fast-stable-stringify:
  4 973 ops/s, ±1.56%   | 33.16% slower

json-stable-stringify:
  3 648 ops/s, ±1.93%   | 50.97% slower

json-stringify-deterministic:
  3 054 ops/s, ±1.80%   | slowest, 58.95% slower

safe-stable-stringify:
  5 345 ops/s, ±2.17%   | 28.16% slower

smolcanon:
  7 440 ops/s, ±1.68%   | fastest

See the changelog.

MIT © Sasha Koss