Extract font weight (in steps of spec-valid 100s) and style (italic) data from the name of a font in a more robust way than most libraries do.

I'll be honest – font names are a bitch since they follow no actual standard at all and sometimes want to be creative in their naming which is not helpful for developers. Lots of implementations out there only catch the most regular font names, but I wanted to have something that catches more obscure and rare cases. Also anything that is not a multiple of 100 between 100 and 900 is simply not valid, so an attempt was made to clamp weights into these values. You can supply font-filenames with or without extensions, it doesn't matter.

We try to find direct matches of keywords in your filenames and then pick the best match using the Sørensen–Dice coefficient, if there are no string matches, but there is already a valid X00 numeric value in the filename, we return that, otherwise we fall back to a default value. The reason we prefer string matches over numeric values is that it's easy to have a typo on a single digit and more unlikely to accidentally have "Regular" in a filename. Also there could be triple digits in the actual font family name like "Nimbus500-Regular.ttf".

Recognizing weights and styles of 100% of font names is obviously impossible, but the heart of this repo, the numericFontWeightMap has a lot more than you'd typically find out there in the internet. Feel free to open issues or direct PRs for more cases you found!

• zero dependencies
• fully typed in typescript
• 100% code test coverage (100% Statements 28/28, 100% Branches 36/36, 100% Functions 7/7, 100% Lines 27/27)
• small footprint (2.2kb minified)
• CommonJS bundle, .mjs bundle, .d.ts bundle + type-checking & Source maps
• written in an actually readable way

npm i parse-font-name or yarn add parse-font-name

import { parseNumericWeightFromName, parseStyleFromName, numericFontWeightMap } from 'parse-font-name'

// parse numeric weights
parseNumericWeightFromName('Helvetica Neue Light.ttf') // returns 300
parseNumericWeightFromName(['Akkurat Pro Regular', 'Akkurat Pro Fett Kursiv']) // returns [400, 700]
parseNumericWeightFromName('Bebas Neue') // returns 400 (fallbackValue)
parseNumericWeightFromName('Bebas Neue 500') // returns 500 (no string, but valid numeric value found)
parseNumericWeightFromName('Bebas Neue', 700) // returns 700 (fallbackValue)

// parse font styles
parseStyleFromName('Akkurat Pro Fett Italic') // returns 'italic'
parseStyleFromName('Akkurat Pro Fett Kursiv', 'boolean') // returns true
parseStyleFromName('Akkurat-Pro-Fett-Oblique.otf') // returns 'oblique'
parseStyleFromName(['Akkurat Pro Fett Kursiv', 'Helvetica Neue Light'], 'boolean') // returns [true, false]

// or just toy around with the hardcoded map, which is also exported as a type
console.log(numericFontWeightMap)

parseNumericWeightFromName(
  fontStrings: string | string[],
  fallbackValue?: number // fallbackValue has a default of 400
) => number | number[] // providing strings gives you strings and is properly typed to do so

parseStyleFromName(
  fontStrings: string | string[],
  format?: 'cssString' | 'boolean', // format has a default of 'cssString'
  fallbackValue?: boolean | ('normal' | 'oblique' | 'italic') // fallbackValue defaults to false or 'normal' depending on format
) => (boolean | ('normal' | 'oblique' | 'italic'))[]

See src/index.test.js for more examples and src/index.ts for types.

parseNumericWeightFromName from 100 font strings 605 ops/s, ±0.51% parseStyle from 100 font strings 109 111 ops/s, ±0.23%

Which is something that could be improved, but so far I didn't need to parse 100 fonts more than a couple of hundred times a second, so whatever.