noble-hashes

This fork adds support for legacy browsers without BigInt (e.g. Safari 13 or less), and only includes hash algorithms needed by openpgpjs: SHA1, SHA2, SHA3, RIPEMD.
We recommend you use the upstream repo. The rest of the README refers to the upstream library.

---

Audited & minimal JS implementation of SHA, RIPEMD, BLAKE, HMAC, HKDF, PBKDF, Scrypt & Argon2.

• 🔒 Audited by an independent security firm
• 🔻 Tree-shaking-friendly: use only what's necessary, other code won't be included
• 🏎 Ultra-fast, hand-optimized for caveats of JS engines
• 🔍 Unique tests ensure correctness: chained tests, sliding window tests, DoS tests, fuzzing
• 🔁 No unrolled loops: makes it easier to verify and reduces source code size up to 5x
• 🐢 Scrypt supports N: 2**22, while other implementations are limited to 2**20
• 🦘 SHA3 supports Keccak, cSHAKE, KangarooTwelve, MarsupilamiFourteen and TurboSHAKE
• 🪶 Just 3.4k lines / 17KB gzipped. SHA256-only is 240 lines / 3KB gzipped

The library's initial development was funded by Ethereum Foundation.

This library belongs to noble crypto

noble-crypto — high-security, easily auditable set of contained cryptographic libraries and tools.

• Zero or minimal dependencies
• Highly readable TypeScript / JS code
• PGP-signed releases and transparent NPM builds
• All libraries: ciphers, curves, hashes
• Check out homepage for reading resources, documentation and apps built with noble

Usage

npm install @openpgp/noble-hashes

We support all major platforms and runtimes. For Deno, ensure to use npm specifier. For React Native, you may need a polyfill for getRandomValues. A standalone file noble-hashes.js is also available.

1// import * from '@noble/hashes'; // Error: use sub-imports, to ensure small app size
2import { sha256 } from '@noble/hashes/sha256'; // ECMAScript modules (ESM) and Common.js
3// import { sha256 } from 'npm:@noble/hashes@1.3.0/sha256'; // Deno
4console.log(sha256(new Uint8Array([1, 2, 3]))); // Uint8Array(32) [3, 144, 88, 198, 242...]
5// you could also pass strings that will be UTF8-encoded to Uint8Array
6console.log(sha256('abc')); // == sha256(new TextEncoder().encode('abc'))

• Implementations
  • sha2: sha256, sha384, sha512, sha512_256
  • sha3: FIPS, SHAKE, Keccak
  • sha3-addons: cSHAKE, KMAC, K12, M14, TurboSHAKE
  • ripemd160
  • blake2b, blake2s, blake3
  • sha1: legacy hash
  • hmac
  • hkdf
  • pbkdf2
  • scrypt
  • argon2
  • utils
  • All available imports
• Security
  • Constant-timeness
  • Memory dumping
  • Supply chain security
  • Randomness
• Speed
• Contributing & testing
• Resources
• License

Implementations

All hash functions:

• can be called directly, with Uint8Array.
• return Uint8Array
• can receive string, which is automatically converted to Uint8Array via utf8 encoding (not hex)
• support hashing 4GB of data per update on 64-bit systems (unlimited with streaming)

1function hash(message: Uint8Array | string): Uint8Array;
2hash(new Uint8Array([1, 3]));
3hash('string') == hash(new TextEncoder().encode('string'));

All hash functions can be constructed via hash.create() method:

• the result is Hash subclass instance, which has update() and digest() methods
• digest() finalizes the hash and makes it no longer usable

1hash
2  .create()
3  .update(new Uint8Array([1, 3]))
4  .digest();

Some hash functions can also receive options object, which can be either passed as a:

• second argument to hash function: blake3('abc', { key: 'd', dkLen: 32 })
• first argument to class initializer: blake3.create({ context: 'e', dkLen: 32 })

sha2: sha256, sha384, sha512, sha512_256

1import { sha256 } from '@noble/hashes/sha256';
2const h1a = sha256('abc');
3const h1b = sha256
4  .create()
5  .update(Uint8Array.from([1, 2, 3]))
6  .digest();

1import { sha512 } from '@noble/hashes/sha512';
2const h2a = sha512('abc');
3const h2b = sha512
4  .create()
5  .update(Uint8Array.from([1, 2, 3]))
6  .digest();
7
8// SHA512/256 variant
9import { sha512_256 } from '@noble/hashes/sha512';
10const h3a = sha512_256('abc');
11const h3b = sha512_256
12  .create()
13  .update(Uint8Array.from([1, 2, 3]))
14  .digest();
15
16// SHA384
17import { sha384 } from '@noble/hashes/sha512';
18const h4a = sha384('abc');
19const h4b = sha384
20  .create()
21  .update(Uint8Array.from([1, 2, 3]))
22  .digest();

See RFC 4634 and the paper on SHA512/256.

sha3: FIPS, SHAKE, Keccak

1import {
2  sha3_224,
3  sha3_256,
4  sha3_384,
5  sha3_512,
6  keccak_224,
7  keccak_256,
8  keccak_384,
9  keccak_512,
10  shake128,
11  shake256,
12} from '@noble/hashes/sha3';
13const h5a = sha3_256('abc');
14const h5b = sha3_256
15  .create()
16  .update(Uint8Array.from([1, 2, 3]))
17  .digest();
18const h6a = keccak_256('abc');
19const h7a = shake128('abc', { dkLen: 512 });
20const h7b = shake256('abc', { dkLen: 512 });

See FIPS PUB 202, Website.

Check out the differences between SHA-3 and Keccak

sha3-addons: cSHAKE, KMAC, K12, M14, TurboSHAKE

1import {
2  cshake128,
3  cshake256,
4  kmac128,
5  kmac256,
6  k12,
7  m14,
8  turboshake128,
9  turboshake256,
10  tuplehash128,
11  tuplehash256,
12  parallelhash128,
13  parallelhash256,
14  keccakprg,
15} from '@noble/hashes/sha3-addons';
16const h7c = cshake128('abc', { personalization: 'def' });
17const h7d = cshake256('abc', { personalization: 'def' });
18const h7e = kmac128('key', 'message');
19const h7f = kmac256('key', 'message');
20const h7h = k12('abc');
21const h7g = m14('abc');
22const h7t1 = turboshake128('abc');
23const h7t2 = turboshake256('def', { D: 0x05 });
24const h7i = tuplehash128(['ab', 'c']); // tuplehash(['ab', 'c']) !== tuplehash(['a', 'bc']) !== tuplehash(['abc'])
25// Same as k12/blake3, but without reduced number of rounds. Doesn't speedup anything due lack of SIMD and threading,
26// added for compatibility.
27const h7j = parallelhash128('abc', { blockLen: 8 });
28// pseudo-random generator, first argument is capacity. XKCP recommends 254 bits capacity for 128-bit security strength.
29// * with a capacity of 254 bits.
30const p = keccakprg(254);
31p.feed('test');
32const rand1b = p.fetch(1);

• Full NIST SP 800-185: cSHAKE, KMAC, TupleHash, ParallelHash + XOF variants
• Reduced-round Keccak:
  • 🦘 K12 aka KangarooTwelve
  • M14 aka MarsupilamiFourteen
  • TurboSHAKE
• KeccakPRG: Pseudo-random generator based on Keccak

ripemd160

1import { ripemd160 } from '@noble/hashes/ripemd160';
2// function ripemd160(data: Uint8Array): Uint8Array;
3const hash8 = ripemd160('abc');
4const hash9 = ripemd160
5  .create()
6  .update(Uint8Array.from([1, 2, 3]))
7  .digest();

See RFC 2286, Website

blake2b, blake2s, blake3

1import { blake2b } from '@noble/hashes/blake2b';
2import { blake2s } from '@noble/hashes/blake2s';
3const h10a = blake2s('abc');
4const b2params = { key: new Uint8Array([1]), personalization: t, salt: t, dkLen: 32 };
5const h10b = blake2s('abc', b2params);
6const h10c = blake2s
7  .create(b2params)
8  .update(Uint8Array.from([1, 2, 3]))
9  .digest();
10
11import { blake3 } from '@noble/hashes/blake3';
12// All params are optional
13const h11 = blake3('abc', { dkLen: 256, key: 'def', context: 'fji' });

See RFC 7693, Website.

sha1: legacy hash

SHA1 was cryptographically broken, however, it was not broken for cases like HMAC.

See RFC4226 B.2.

Don't use it for a new protocol.

1import { sha1 } from '@noble/hashes/sha1';
2const h12 = sha1('def');

hmac

1import { hmac } from '@noble/hashes/hmac';
2import { sha256 } from '@noble/hashes/sha256';
3const mac1 = hmac(sha256, 'key', 'message');
4const mac2 = hmac
5  .create(sha256, Uint8Array.from([1, 2, 3]))
6  .update(Uint8Array.from([4, 5, 6]))
7  .digest();

Matches RFC 2104.

hkdf

1import { hkdf } from '@noble/hashes/hkdf';
2import { sha256 } from '@noble/hashes/sha256';
3import { randomBytes } from '@noble/hashes/utils';
4const inputKey = randomBytes(32);
5const salt = randomBytes(32);
6const info = 'abc';
7const dkLen = 32;
8const hk1 = hkdf(sha256, inputKey, salt, info, dkLen);
9
10// == same as
11import * as hkdf from '@noble/hashes/hkdf';
12import { sha256 } from '@noble/hashes/sha256';
13const prk = hkdf.extract(sha256, inputKey, salt);
14const hk2 = hkdf.expand(sha256, prk, info, dkLen);

Matches RFC 5869.

pbkdf2

1import { pbkdf2, pbkdf2Async } from '@noble/hashes/pbkdf2';
2import { sha256 } from '@noble/hashes/sha256';
3const pbkey1 = pbkdf2(sha256, 'password', 'salt', { c: 32, dkLen: 32 });
4const pbkey2 = await pbkdf2Async(sha256, 'password', 'salt', { c: 32, dkLen: 32 });
5const pbkey3 = await pbkdf2Async(sha256, Uint8Array.from([1, 2, 3]), Uint8Array.from([4, 5, 6]), {
6  c: 32,
7  dkLen: 32,
8});

Matches RFC 2898.

scrypt

1import { scrypt, scryptAsync } from '@noble/hashes/scrypt';
2const scr1 = scrypt('password', 'salt', { N: 2 ** 16, r: 8, p: 1, dkLen: 32 });
3const scr2 = await scryptAsync('password', 'salt', { N: 2 ** 16, r: 8, p: 1, dkLen: 32 });
4const scr3 = await scryptAsync(Uint8Array.from([1, 2, 3]), Uint8Array.from([4, 5, 6]), {
5  N: 2 ** 22,
6  r: 8,
7  p: 1,
8  dkLen: 32,
9  onProgress(percentage) {
10    console.log('progress', percentage);
11  },
12  maxmem: 2 ** 32 + 128 * 8 * 1, // N * r * p * 128 + (128*r*p)
13});

Conforms to RFC 7914, Website

• N, r, p are work factors. To understand them, see the blog post.
• dkLen is the length of output bytes
• It is common to use N from 2**10 to 2**22 and {r: 8, p: 1, dkLen: 32}
• onProgress can be used with async version of the function to report progress to a user.

Memory usage of scrypt is calculated with the formula N * r * p * 128 + (128 * r * p), which means {N: 2 ** 22, r: 8, p: 1} will use 4GB + 1KB of memory. To prevent DoS, we limit scrypt to 1GB + 1KB of RAM used, which corresponds to {N: 2 ** 20, r: 8, p: 1}. If you want to use higher values, increase maxmem using the formula above.

Note: noble supports 2**22 (4GB RAM) which is the highest amount amongst JS libs. Many other implementations don't support it. We cannot support 2**23, because there is a limitation in JS engines that makes allocating arrays bigger than 4GB impossible, but we're looking into other possible solutions.

argon2

Experimental Argon2 RFC 9106 implementation. It may be removed at any time.

1import { argon2d, argon2i, argon2id } from '@noble/hashes/argon2';
2const result = argon2id('password', 'salt', { t: 2, m: 65536, p: 1 });

utils

1import { bytesToHex as toHex, randomBytes } from '@noble/hashes/utils';
2console.log(toHex(randomBytes(32)));

• bytesToHex will convert Uint8Array to a hex string
• randomBytes(bytes) will produce cryptographically secure random Uint8Array of length bytes

All available imports

1// sha384 is here, because it uses same internals as sha512
2import { sha512, sha512_256, sha384 } from '@noble/hashes/sha512';
3// prettier-ignore
4import {
5  sha3_224, sha3_256, sha3_384, sha3_512,
6  keccak_224, keccak_256, keccak_384, keccak_512,
7  shake128, shake256
8} from '@noble/hashes/sha3';
9// prettier-ignore
10import {
11  cshake128, cshake256,
12  k12, m14,
13  turboshake128, turboshake256,
14  kmac128, kmac256,
15  tuplehash256, parallelhash256, keccakprg
16} from '@noble/hashes/sha3-addons';
17import { ripemd160 } from '@noble/hashes/ripemd160';
18import { blake3 } from '@noble/hashes/blake3';
19import { blake2b } from '@noble/hashes/blake2b';
20import { blake2s } from '@noble/hashes/blake2s';
21import { hmac } from '@noble/hashes/hmac';
22import { hkdf } from '@noble/hashes/hkdf';
23import { pbkdf2, pbkdf2Async } from '@noble/hashes/pbkdf2';
24import { scrypt, scryptAsync } from '@noble/hashes/scrypt';
25
26import { sha1 } from '@noble/hashes/sha1'; // legacy
27
28// small utility method that converts bytes to hex
29import { bytesToHex as toHex } from '@noble/hashes/utils';
30console.log(toHex(sha256('abc'))); // ba7816bf8f01cfea414140de5dae2223b00361a396177a9cb410ff61f20015ad

Security

The library has been independently audited:

• at version 1.0.0, in Jan 2022, by cure53
  • PDFs: online, offline, in-repo
  • Changes since audit.
  • Scope: everything, besides blake3, sha3-addons, sha1 and argon2, which have not been audited
  • The audit has been funded by Ethereum Foundation with help of Nomic Labs

It is tested against property-based, cross-library and Wycheproof vectors, and has fuzzing by Guido Vranken's cryptofuzz.

If you see anything unusual: investigate and report.

Constant-timeness

JIT-compiler and Garbage Collector make "constant time" extremely hard to achieve timing attack resistance in a scripting language. Which means any other JS library can't have constant-timeness. Even statically typed Rust, a language without GC, makes it harder to achieve constant-time for some cases. If your goal is absolute security, don't use any JS lib — including bindings to native ones. Use low-level libraries & languages. Nonetheless we're targetting algorithmic constant time.

Memory dumping

The library shares state buffers between hash function calls. The buffers are zeroed-out after each call. However, if an attacker can read application memory, you are doomed in any case:

• At some point, input will be a string and strings are immutable in JS: there is no way to overwrite them with zeros. For example: deriving key from scrypt(password, salt) where password and salt are strings
• Input from a file will stay in file buffers
• Input / output will be re-used multiple times in application which means it could stay in memory
• await anything() will always write all internal variables (including numbers) to memory. With async functions / Promises there are no guarantees when the code chunk would be executed. Which means attacker can have plenty of time to read data from memory
• There is no way to guarantee anything about zeroing sensitive data without complex tests-suite which will dump process memory and verify that there is no sensitive data left. For JS it means testing all browsers (incl. mobile), which is complex. And of course it will be useless without using the same test-suite in the actual application that consumes the library

Supply chain security

• Commits are signed with PGP keys, to prevent forgery. Make sure to verify commit signatures.
• Releases are transparent and built on GitHub CI. Make sure to verify provenance logs
• Rare releasing is followed to ensure less re-audit need for end-users
• Dependencies are minimized and locked-down:
  • If your app has 500 dependencies, any dep could get hacked and you'll be downloading malware with every install. We make sure to use as few dependencies as possible
  • We prevent automatic dependency updates by locking-down version ranges. Every update is checked with npm-diff
• Dev Dependencies are only used if you want to contribute to the repo. They are disabled for end-users:
  • scure-base, scure-bip32, scure-bip39, micro-bmark and micro-should are developed by the same author and follow identical security practices
  • prettier (linter), fast-check (property-based testing) and typescript are used for code quality, vector generation and ts compilation. The packages are big, which makes it hard to audit their source code thoroughly and fully

Randomness

We're deferring to built-in crypto.getRandomValues which is considered cryptographically secure (CSPRNG).

In the past, browsers had bugs that made it weak: it may happen again. Implementing a userspace CSPRNG to get resilient to the weakness is even worse: there is no reliable userspace source of quality entropy.

Speed

Benchmarks measured on Apple M1 with macOS 12. Note that PBKDF2 and Scrypt are tested with extremely high work factor. To run benchmarks, execute npm run bench:install and then npm run bench

SHA256 32B x 1,219,512 ops/sec @ 820ns/op ± 2.58% (min: 625ns, max: 4ms)
SHA384 32B x 512,032 ops/sec @ 1μs/op
SHA512 32B x 509,943 ops/sec @ 1μs/op
SHA3-256, keccak256, shake256 32B x 199,600 ops/sec @ 5μs/op
Kangaroo12 32B x 336,360 ops/sec @ 2μs/op
Marsupilami14 32B x 298,418 ops/sec @ 3μs/op
BLAKE2b 32B x 379,794 ops/sec @ 2μs/op
BLAKE2s 32B x 515,995 ops/sec @ 1μs/op ± 1.07% (min: 1μs, max: 4ms)
BLAKE3 32B x 588,235 ops/sec @ 1μs/op ± 1.36% (min: 1μs, max: 5ms)
RIPEMD160 32B x 1,140,250 ops/sec @ 877ns/op ± 3.12% (min: 708ns, max: 6ms)
HMAC-SHA256 32B x 377,358 ops/sec @ 2μs/op

HKDF-SHA256 32B x 108,377 ops/sec @ 9μs/op
PBKDF2-HMAC-SHA256 262144 x 3 ops/sec @ 326ms/op
PBKDF2-HMAC-SHA512 262144 x 1 ops/sec @ 970ms/op
Scrypt r: 8, p: 1, n: 262144 x 1 ops/sec @ 616ms/op

Compare to native node.js implementation that uses C bindings instead of pure-js code:

SHA256 32B node x 1,302,083 ops/sec @ 768ns/op ± 10.54% (min: 416ns, max: 7ms)
SHA384 32B node x 975,609 ops/sec @ 1μs/op ± 11.32% (min: 625ns, max: 8ms)
SHA512 32B node x 983,284 ops/sec @ 1μs/op ± 11.24% (min: 625ns, max: 8ms)
SHA3-256 32B node x 910,746 ops/sec @ 1μs/op ± 12.19% (min: 666ns, max: 10ms)
keccak, k12, m14 are not implemented
BLAKE2b 32B node x 967,117 ops/sec @ 1μs/op ± 11.26% (min: 625ns, max: 9ms)
BLAKE2s 32B node x 1,055,966 ops/sec @ 947ns/op ± 11.07% (min: 583ns, max: 7ms)
BLAKE3 is not implemented
RIPEMD160 32B node x 1,002,004 ops/sec @ 998ns/op ± 10.66% (min: 625ns, max: 7ms)
HMAC-SHA256 32B node x 919,963 ops/sec @ 1μs/op ± 6.13% (min: 833ns, max: 5ms)
HKDF-SHA256 32 node x 369,276 ops/sec @ 2μs/op ± 13.59% (min: 1μs, max: 9ms)
PBKDF2-HMAC-SHA256 262144 node x 25 ops/sec @ 39ms/op
PBKDF2-HMAC-SHA512 262144 node x 7 ops/sec @ 132ms/op
Scrypt r: 8, p: 1, n: 262144 node x 1 ops/sec @ 523ms/op

It is possible to make this library 4x+ faster by doing code generation of full loop unrolls. We've decided against it. Reasons:

• the library must be auditable, with minimum amount of code, and zero dependencies
• most method invocations with the lib are going to be something like hashing 32b to 64kb of data
• hashing big inputs is 10x faster with low-level languages, which means you should probably pick 'em instead

The current performance is good enough when compared to other projects; SHA256 takes only 900 nanoseconds to run.

Contributing & testing

1. Clone the repository
2. npm install to install build dependencies like TypeScript
3. npm run build to compile TypeScript code
4. npm run test will execute all main tests. See our approach to testing
5. npm run test:dos will test against DoS; by measuring function complexity. Takes ~20 minutes
6. npm run test:big will execute hashing on 4GB inputs, scrypt with 1024 different N, r, p combinations, etc. Takes several hours. Using 8-32+ core CPU helps.

Resources

Check out paulmillr.com/noble for useful resources, articles, documentation and demos related to the library.

License

The MIT License (MIT)

Copyright (c) 2022 Paul Miller (https://paulmillr.com)

See LICENSE file.