fontmetrics

FontMetrics

A lightweight JavaScript library for computing accurate font metrics such as x-height, cap height, ascent, descent and tittle for any loaded web font.

Demo

You can see it in action here.

Installation

Install via npm

npm install fontmetrics

Or include the script in your page

Usage

Make sure your font is loaded first! A good way to do this is by using WebFontLoader.

// If installed via npm...
import FontMetrics from 'fontmetrics'

const metrics = FontMetrics({
  fontFamily: 'Roboto',
  // Optional (defaults)
  fontWeight: 'normal',
  fontSize: 200,
  origin: 'baseline'
})

This will return an object containing values that are normalised to fontSize and relative to origin, for example:

{
  capHeight: -0.73,
  baseline: 0,
  xHeight: -0.54,
  descent: 0.195,
  bottom: 0.5,
  ascent: -0.76,
  tittle: -0.73,
  top: -0.935,
  fontFamily: 'Roboto',
  fontWeight: 'normal',
  fontSize: 200
}

How to use metrics

As mentioned above, the values returned are normalised to fontSize and relative to origin. This is useful because you should then be able to use the returned values regardless of your display font size, simply by scaling them.

For example, if the value returned for ascent is -0.695 with origin set to 'baseline', then you can get the pixel value for ascent by using yourBaselinePosition + metrics.ascent * fontSize.

Origin

You will likely need different origins based on how you're wanting to use the metrics. 'baseline' is the default, but you can pass the name of any metric returned from FontMetrics to be used as the origin (e.g. 'top'.) This will mean that all returned metrics will now be relative to top and of course top will be 0.

Settings

You can define the characters used to test certain metrics via FontMetrics.settings.chars. For example, FontMetrics.settings.chars.xHeight defines the character used to measure the xHeight metric (defaults to 'x'.)

How it works

FontMetrics works by rendering your font to a CanvasRenderingContext2D and then measuring the pixel bounds of the output. This is a fairly well documented technique and seems to provide pretty decent results.