Pseudolocalization inspired by Netflix (https://netflixtechblog.com/pseudo-localization-netflix-12fff76fbcbe)

pseudoLocalize.test.ts

import {pseudoLocalizeString} from 'utils/pseudoLocalize'

describe('pseudoLocalizeString', () => {
  it.each([
    ['Hello, world!', 'Ħḗḗḗŀŀǿǿǿ, ẇǿǿřŀḓ!'],
    ['Get help online.', 'Ɠḗḗḗŧ ħḗḗŀƥ ǿǿƞŀīīƞḗḗ.'],
    [
      'The quick brown fox jumps over the lazy dog',
      'Ŧħḗḗḗ ɋŭŭŭīīīƈķ ƀřǿǿǿẇƞ ƒǿǿǿẋ ĵŭŭḿƥş ǿǿṽḗḗř ŧħḗḗ ŀȧȧẑẏ ḓǿǿɠ',
    ],
  ])(
    'should pseudo-localize strings with +35% longer text',
    (input, expected) => {
      const localizedString = pseudoLocalizeString(input)

      expect(localizedString).toBe(expected)
      expect(localizedString.length).toBeGreaterThanOrEqual(input.length * 1.35)
    }
  )
})

pseudoLocalize.ts

/* eslint-disable id-length */

const ACCENTED_MAP = {
  a: 'ȧ',
  A: 'Ȧ',
  b: 'ƀ',
  B: 'Ɓ',
  c: 'ƈ',
  C: 'Ƈ',
  d: 'ḓ',
  D: 'Ḓ',
  e: 'ḗ',
  E: 'Ḗ',
  f: 'ƒ',
  F: 'Ƒ',
  g: 'ɠ',
  G: 'Ɠ',
  h: 'ħ',
  H: 'Ħ',
  i: 'ī',
  I: 'Ī',
  j: 'ĵ',
  J: 'Ĵ',
  k: 'ķ',
  K: 'Ķ',
  l: 'ŀ',
  L: 'Ŀ',
  m: 'ḿ',
  M: 'Ḿ',
  n: 'ƞ',
  N: 'Ƞ',
  o: 'ǿ',
  O: 'Ǿ',
  p: 'ƥ',
  P: 'Ƥ',
  q: 'ɋ',
  Q: 'Ɋ',
  r: 'ř',
  R: 'Ř',
  s: 'ş',
  S: 'Ş',
  t: 'ŧ',
  T: 'Ŧ',
  v: 'ṽ',
  V: 'Ṽ',
  u: 'ŭ',
  U: 'Ŭ',
  w: 'ẇ',
  W: 'Ẇ',
  x: 'ẋ',
  X: 'Ẋ',
  y: 'ẏ',
  Y: 'Ẏ',
  z: 'ẑ',
  Z: 'Ẑ',
}

// Inspired by pseudo-localization from Netflix (https://netflixtechblog.com/pseudo-localization-netflix-12fff76fbcbe)
export const pseudoLocalizeString = (string: string): string => {
  let pseudoLocalizedText = string
    .split('')
    .map(char => ACCENTED_MAP[char as keyof typeof ACCENTED_MAP] || char)
    .join('')

  // Emulate 35% longer text by duplicating vowels
  const targetLength = Math.ceil(string.length * 1.35)

  const vowelMatches = [...string.toLowerCase().matchAll(/[aeiou­]/gi)]

  const distribution = distribute(
    vowelMatches.length,
    targetLength - string.length
  )

  let offset = 0
  for (let i = 0; i < vowelMatches.length; i++) {
    const match = vowelMatches[i]
    const vowelInstances = distribution[i]
    const pos = match.index + offset

    const vowel = pseudoLocalizedText[pos]

    // Insert additional vowel instances
    pseudoLocalizedText =
      pseudoLocalizedText.slice(0, pos) +
      vowel.repeat(vowelInstances + 1) +
      pseudoLocalizedText.slice(pos + 1)

    offset += vowelInstances
  }

  return pseudoLocalizedText
}

// Create a front-loaded distribution. E.g, length of 5 and spend of 12 = [3, 3, 2, 2, 2]
function distribute(length: number, spend: number): number[] {
  const base = Math.floor(spend / length)
  const remainder = spend % length

  return Array(length)
    .fill(base)
    .map((n, i) => (i < remainder ? n + 1 : n))
}