ens-normalize.js

0-dependancy Ethereum Name Service (ENS) Name Normalizer.

• Follows ENSIP-15: ENS Name Normalization Standard
  • Other implementations:
    • Python — namehash/ens-normalize-python
    • Javascript — ensdomains/eth-ens-namehash
• Passes 100% Validation Tests
• Custom NFC Implementation (or use native)
  • Passes 100% Unicode 15.0.0 Normalization Tests
• Minified File Sizes:
  • 27KB — native NFC via nf-native.js using String.normalize()
  • 36KB Default — custom NFC via nf.js
  • 42KB Everything! — custom NFC + sub-libraries: parts.js, utils.js
• Included Apps:
  • Resolver Demo ⭐
  • Character Viewer
  • Supported Emoji
  • Confused Explainer
• Related Projects:
  • Recent Registrations
  • Emoji Frequency Explorer
  • ENS+NFT Matcher
  • Batch Resolver
  • Punycode Coder
  • Label Database
• Breakdown Reports:
  • Disallowed Characters
  • Different Norm
  • Illegal Mixtures
  • Whole-script Confusables
  • Illegal Placement
  • Non-spacing Marks

import {ens_normalize} from '@adraffy/ens-normalize'; // or require()
// npm i @adraffy/ens-normalize
// browser: https://cdn.jsdelivr.net/npm/@adraffy/ens-normalize@latest/dist/index.min.{m|c}js

// *** ALL errors thrown by this library are safe to print ***
// - characters are shown as {HEX} if should_escape()
// - potentially different bidi directions inside "quotes"
// - 200E is used near "quotes" to prevent spillover
// - an "error type" can be extracted by slicing up to the first (:)

// string -> string
// throws on invalid names
// output ready for namehash
let normalized = ens_normalize('RaFFY🚴‍♂️.eTh');
// => "raffy🚴‍♂.eth"

// note: does not enforce .eth TLD 3-character minimum

Format names with fully-qualified emoji:

// works like ens_normalize()
// output ready for display
let pretty = ens_beautify('1⃣2⃣.eth'); 
// => "1️⃣2️⃣.eth"

// note: normalization is unchanged:
// ens_normalize(ens_beautify(x)) == ens_normalize(x)

Normalize name fragments for substring search:

// these fragments fail ens_normalize() 
// but will normalize fine as fragments
let frag1 = ens_normalize_fragment('AB--');    // expected error: label ext
let frag2 = ens_normalize_fragment('\u{303}'); // expected error: leading cm
let frag3 = ens_normalize_fragment('οо');      // expected error: mixture

Input-based tokenization:

// string -> Token[]
// never throws
let tokens = ens_tokenize('_R💩\u{FE0F}a\u{FE0F}\u{304}\u{AD}./');
// [
//     { type: 'valid', cp: [ 95 ] }, // valid (as-is)
//     {
//         type: 'mapped', 
//         cp: 82,         // input
//         cps: [ 114 ]    // output
//     }, 
//     { 
//         type: 'emoji',
//         input: Emoji(2) [ 128169, 65039 ],  // input 
//         emoji: [ 128169, 65039 ],           // fully-qualified
//         cps: Emoji(1) [ 128169 ]            // output (normalized)
//     },
//     {
//         type: 'nfc',
//         input: [ 97, 772 ],  // input  (before nfc)
//         tokens0: [           // tokens (before nfc)
//             { type: 'valid', cps: [ 97 ] },
//             { type: 'ignored', cp: 65039 },
//             { type: 'valid', cps: [ 772 ] }
//         ],
//         cps: [ 257 ],        // output (after nfc)
//         tokens: [            // tokens (after nfc)
//             { type: 'valid', cps: [ 257 ] }
//         ]
//     },
//     { type: 'ignored', cp: 173 },
//     { type: 'stop', cp: 46 },
//     { type: 'disallowed', cp: 47 }
// ]

// note: if name is normalizable, then:
// ens_normalize(ens_tokenize(name).map(token => {
//     ** convert valid/mapped/nfc/stop to string **
// }).join('')) == ens_normalize(name)

Output-based tokenization:

// string -> Label[]
// never throws
let labels = ens_split('💩Raffy.eth_');
// [
//   {
//     input: [ 128169, 82, 97, 102, 102, 121 ],  
//     offset: 0, // index of codepoint, not substring index!
//                // (corresponding length can be inferred from input)
//     tokens: [
//       Emoji(2) [ 128169, 65039 ],   // emoji
//       [ 114, 97, 102, 102, 121 ]    // nfc-text
//     ],
//     output: [ 128169, 114, 97, 102, 102, 121 ],
//     emoji: true,
//     type: 'Latin'
//   },
//   {
//     input: [ 101, 116, 104, 95 ],
//     offset: 7,
//     tokens: [ [ 101, 116, 104, 95 ] ],
//     output: [ 101, 116, 104, 95 ],
//     error: Error('underscore allowed only at start')
//   }
// ]

Generate a sorted array of supported emoji codepoints:

// () -> number[][]
console.log(ens_emoji());
// [
//     [ 2764 ],
//     [ 128169, 65039 ],
//     [ 128105, 127997, 8205, 9877, 65039 ],
//     ...
// ]

Determine if a character shouldn't be printed directly:

// number -> bool
console.log(should_escape(0x202E)); // eg. RIGHT-TO-LEFT OVERRIDE => true

Determine if a character is a combining mark:

// number -> bool
console.log(is_combining_mark(0x20E3)); // eg. COMBINING ENCLOSING KEYCAP => true

Build

• git clone this repo, then npm install
• Follow instructions in /derive/ to generate data files
  • npm run derive
    • spec.json
    • nf.json
    • nf-tests.json
• npm run make — compress data files from /derive/output/
  • include-ens.js
  • include-nf.js
  • include-versions.js
• Follow instructions in /validate/ to generate validation tests
  • npm run validate
    • tests.json
• npm run test — perform validation tests
• npm run build — create /dist/
• npm run rebuild — run all the commands above
• npm run order — create optimal group ordering and rebuild again

Publishing to NPM

This project uses .js instead of .mjs so package.json uses type: module. To avoid bundling issues, type is dropped during packing. pre/post hooks aren't used because they're buggy.

• npm run pack instead of npm pack
• npm run pub instead of npm publish

Security

• Build and compare against include-versions.js
  • "spec_hash" — SHA-256 of spec.json bytes
  • "base64_ens_hash" — SHA-256 of include-ens.js base64 literal
  • "base64_nf_hash" — SHA-256 of include-nf.js base64 literal