⏎ μWrap

A 10x faster and more accurate text wrapping util in < 2KB (min) (MIT Licensed)

---

Introduction

uWrap exists to efficiently predict varying row heights for list and grid virtualization, a technique for UI performance optimization when rendering large, scrollable datasets. Doing this both quickly and accurately turns out to be a non-trivial task since Canvas2D provides no API for text wrapping, and measureText() is quite expensive; measuring via DOM is also a non-starter due to poor performance. Additionally, font size, variable-width kerning, letter-spacing, explicit line breaks, and different white-space choices affect the wrapping locations.

Notes:

• Currently works most accurately with Latin charsets
• Does not yet handle Windows-style \r\n explicit line breaks
• Only pre-line wrapping strategy is implemented so far

---

Performance

uWrap outperforms canvas-hypertxt by a wide margin in both CPU and memory usage while being significantly more accurate.

The benchmark below wraps 100,000 random sentences into boxes of random widths between 50px and 250px. You can see this live in DevTools console of the demo page.

	Chrome 135	Firefox 137	Safari 18.1
uWrap	82ms	90ms	185ms
canvas-hypertxt	770ms	1660ms	1430ms

---

Installation

npm i uwrap

or

<script src="./dist/uWrap.iife.min.js"></script>

---

API

See uWrap.d.ts TypeScript def.

---

Usage

// import fn for wrapping variable-width fonts using pre-line strategy
import { varPreLine } from 'uwrap';

// create a Canvas2D context with desired font settings
let ctx = document.createElement('canvas').getContext('2d');
ctx.font = "14px Inter, sans-serif";
ctx.letterSpacing = '0.15px';

// init util fns
const { count, test, split } = varPreLine(ctx);

// example text
let text = 'The quick brown fox jumps over the lazy dog.';

// count lines
let numLines = count(text, width);

// test if text will wrap
let willWrap = test(text, width);

// split lines (with optional limit)
let lines = split(text, width, 3);