unicode-segmenter

A lightweight implementation of the Unicode Text Segmentation (UAX #29)

• Zero-dependencies: It doesn't bloat node_modules or the networks tab.

• Excellent compatibility: It works well on older browsers, edge runtimes, and embedded JavaScript runtimes like Hermes and QuickJS.

• Small bundle size: It effectively compresses Unicode data and provides a tree-shakeable format, allowing unnecessary codes to be eliminated.

• Extremely efficient: It's carefully optimized for performance, making it the fastest one in the ecosystem—outperforming even the built-in Intl.Segmenter.

• TypeScript: It's fully type-checked, and provides definitions with JSDoc.

• ESM-first: It natively supports ES Modules, also supports CommonJS too.

Unicode® Version

Unicode® 16.0.0

Unicode® Standard Annex #29 - Revision 45 (2024-08-28)

APIs

There are several entries for text segmentation.

• unicode-segmenter/grapheme: Segments and counts extended grapheme clusters
• unicode-segmenter/intl-adapter: Intl.Segmenter adapter
• unicode-segmenter/intl-polyfill: Intl.Segmenter polyfill

And extra utilities for combined use cases.

• unicode-segmenter/emoji: Matches single codepoint emojis
• unicode-segmenter/general: Matches single codepoint alphanumerics
• unicode-segmenter/utils: Handles UTF-16 codepoints

Export unicode-segmenter/grapheme

Utilities for text segmentation by extended grapheme cluster rules.

Example: Count graphemes

import { countGrapheme } from 'unicode-segmenter/grapheme';

'👋 안녕!'.length;
// => 6
countGrapheme('👋 안녕!');
// => 5

'a̐éö̲'.length;
// => 7
countGrapheme('a̐éö̲');
// => 3

Example: Get grapheme segments

import { graphemeSegments } from 'unicode-segmenter/grapheme';

[...graphemeSegments('a̐éö̲\r\n')];
// 0: { segment: 'a̐', index: 0, input: 'a̐éö̲\r\n' }
// 1: { segment: 'é', index: 2, input: 'a̐éö̲\r\n' }
// 2: { segment: 'ö̲', index: 4, input: 'a̐éö̲\r\n' }
// 3: { segment: '\r\n', index: 7, input: 'a̐éö̲\r\n' }

Example: Build an advanced grapheme matcher

graphemeSegments() exposes some knowledge identified in the middle of the process to support some useful cases.

For example, knowing the Grapheme_Cluster_Break category at the beginning and end of a segment can help approximately infer the applied boundary rule.

import { graphemeSegments, GraphemeCategory } from 'unicode-segmenter/grapheme';

function* matchEmoji(str) {
  for (const { segment, _catBegin } of graphemeSegments(input)) {
    // `_catBegin` identified as Extended_Pictographic means the segment is emoji
    if (_catBegin === GraphemeCategory.Extended_Pictographic) {
      yield segment;
    }
  }
}

[...matchEmoji('1🌷2🎁3💩4😜5👍')]
// 0: 🌷
// 1: 🎁
// 2: 💩
// 3: 😜
// 4: 👍

Export unicode-segmenter/intl-adapter

Intl.Segmenter API adapter (only granularity: "grapheme" available yet)

import { Segmenter } from 'unicode-segmenter/intl-adapter';

// Same API with the `Intl.Segmenter`
const segmenter = new Segmenter();

Export unicode-segmenter/intl-polyfill

Intl.Segmenter API polyfill (only granularity: "grapheme" available yet)

// Apply polyfill to the `globalThis.Intl` object.
import 'unicode-segmenter/intl-polyfill';

const segmenter = new Intl.Segmenter();

Export unicode-segmenter/emoji

Utilities for matching emoji-like characters.

Example: Use Unicode emoji property matches

import {
  isEmojiPresentation,    // match \p{Emoji_Presentation}
  isExtendedPictographic, // match \p{Extended_Pictographic}
} from 'unicode-segmenter/emoji';

isEmojiPresentation('😍'.codePointAt(0));
// => true
isEmojiPresentation('♡'.codePointAt(0));
// => false

isExtendedPictographic('😍'.codePointAt(0));
// => true
isExtendedPictographic('♡'.codePointAt(0));
// => true

Export unicode-segmenter/general

Utilities for matching alphanumeric characters.

Example: Use Unicode general property matchers

import {
  isLetter,       // match \p{L}
  isNumeric,      // match \p{N}
  isAlphabetic,   // match \p{Alphabetic}
  isAlphanumeric, // match [\p{N}\p{Alphabetic}]
} from 'unicode-segmenter/general';

Export unicode-segmenter/utils

You can access some internal utilities to deal with JavaScript strings.

Example: Handle UTF-16 surrogate pairs

import {
  isHighSurrogate,
  isLowSurrogate,
  surrogatePairToCodePoint,
} from 'unicode-segmenter/utils';

const u32 = '😍';
const hi = u32.charCodeAt(0);
const lo = u32.charCodeAt(1);

if (isHighSurrogate(hi) && isLowSurrogate(lo)) {
  const codePoint = surrogatePairToCodePoint(hi, lo);
  // => equivalent to u32.codePointAt(0)
}

Example: Determine the length of a character

import { isBMP } from 'unicode-segmenter/utils';

const char = '😍'; // .length = 2
const cp = char.codePointAt(0);

char.length === isBMP(cp) ? 1 : 2;
// => true

Runtime Compatibility

unicode-segmenter uses only fundamental features of ES2015, making it compatible with most browsers.

To ensure compatibility, the runtime should support:

• String.prototype.codePointAt()
• Generators
• Modules

If the runtime doesn't support these features, it can easily be fulfilled with tools like Babel.

React Native Support

Since Hermes doesn't support the Intl.Segmenter API yet, unicode-segmenter is a good alternative.

unicode-segmenter is compiled into small & efficient Hermes bytecode than other JavaScript libraries. See the benchmark for details.

Comparison

unicode-segmenter aims to be lighter and faster than alternatives in the ecosystem while fully spec compliant. So the benchmark is tracking several libraries' performance, bundle size, and Unicode version compliance.

unicode-segmenter/grapheme vs

• graphemer@1.4.0 (16.6M+ weekly downloads on NPM)
• grapheme-splitter@1.0.4 (5.7M+ weekly downloads on NPM)
• @formatjs/intl-segmenter@11.5.7 (5.4K+ weekly downloads on NPM)
• WebAssembly build of unicode-segmentation@1.12.0 with minimum bindings
• Built-in Intl.Segmenter API

JS Bundle Stats

Name	Unicode®	ESM?	Size	Size (min)	Size (min+gzip)	Size (min+br)
unicode-segmenter/grapheme	16.0.0	✔️	17,125	12,720	5,256	3,913
graphemer	15.0.0	✖️ ️	410,435	95,104	15,752	10,660
grapheme-splitter	10.0.0	✖️	122,252	23,680	7,852	4,841
@formatjs/intl-segmenter*	15.0.0	✖️	491,043	318,721	54,248	34,380
unicode-segmentation*	16.0.0	✔️	56,529	52,443	24,110	17,343
Intl.Segmenter*	-	-	0	0	0	0

• @formatjs/intl-segmenter handles grapheme, word, and sentence, but it's not tree-shakable.
• unicode-segmentation size contains only minimum WASM binary and its bindings to execute benchmarking. It will increases to expose more features.
• Intl.Segmenter's Unicode data depends on the host, and may not be up-to-date.
• Intl.Segmenter may not be available in some old browsers, edge runtimes, or embedded environments.

Hermes Bytecode Stats

Name	Bytecode size	Bytecode size (gzip)*
unicode-segmenter/grapheme	23,992	12,533
graphemer	133,949	31,710
grapheme-splitter	63,810	19,125
@formatjs/intl-segmenter*	315,865	99,063

• It would be compressed when included as an app asset.

Runtime Performance

Here is a brief explanation, and you can see archived benchmark results.

Performance in Node.js: unicode-segmenter/grapheme is significantly faster than alternatives.

• 6~15x faster than other JavaScript libraries
• 1.5~3x faster than WASM binding of the Rust's unicode-segmentation
• 1.5~3x faster than built-in Intl.Segmenter

Performance in Bun: unicode-segmenter/grapheme has almost the same performance as the built-in Intl.Segmenter, with no performance degradation compared to other JavaScript libraries.

Performance in Browsers: The performance in browser environments varies greatly due to differences in browser engines and versions, which makes benchmarking less consistent. Despite these variations, unicode-segmenter/grapheme generally outperforms other JavaScript libraries in most environments.

Performance in React Native: unicode-segmenter/grapheme is significantly faster than alternatives when compiled to Hermes bytecode. It's 3~8x faster than graphemer and 20~26x faster than grapheme-splitter, with the performance gap increasing with input size.

Instead of trusting these claims, you can try yarn perf:grapheme directly in your environment or build your own benchmark.

LICENSE

MIT

Note

The initial implementation was ported manually from Rust's unicode-segmentation library, which is licensed under the MIT license.