Subsetting

Modern PDF documents usually do not embed entire fonts but only those parts that are needed to render the text written with that font. This is called a font subset.

Incomplete Subsets

@pdfa-lab/fontkit is able to create font subsets for embedding in PDF documents but they are currently missing some required tables for being universally usable. They can only be embedded as CID fonts.

If you embed fonts via @cantoo/pdf-lib or the original, now unmaintained pdf-lib, this happens automatically, and the resulting PDFs are fully usable.

Creating Subsets

import * as fs from 'node:fs';
import { TrueTypeFont } from '@pdfa-lab/fontkit';

const filename = 'OpenSans/OpenSans-Regular.ttf';
const bytes = fs.readFileSync(filename);
const font = new TrueTypeFont(bytes);

const subset = font.createSubset();

const glyphs = font.glyphsForString('Hello, world!');
for (let i = 0; i < glyphs.length; ++i) {
	const glyph = glyphs[i];
	if (!glyph) {
		throw new Error('Font is missing glyphs!');
	}
	subset.includeGlyph(glyph);
}

const outBytes = subset.encode();

fs.writeFileSync('subset.ttf', outBytes);
console.log('Wrote subset.ttf');

You start by creating an empty subset with createSubset().

Next, you have to create an array of Glyph objects, that the subset should contain. In the example code, this is achieved with `glyphsForString(), but there are other ways to achieve this.

Each of the required glyphs is then passed to the method includeGlyph() of the Subset class. The method automatically filters out duplicate glyphs.

Finally, the subset is serialised with encode() into a buffer.