a parse for a language that has only parentheses, whitespace, and words

simoni.js

// this is a parser for a language described in this tweet: https://twitter.com/msimoni/status/1721647625294782972
const LPAR = 40
const RPAR = 41

const isWord = cp => 32 < cp && cp !== LPAR && cp !== RPAR

export const parse = inputString => {
  const warnings = []
  const topLevelArray = []
  let currentArray = topLevelArray
  const stack = [currentArray]

  const wordBuffer = []
  let i = 0

  for (const character of inputString) {
    const codePoint = character.codePointAt(0)
    if (isWord(codePoint)) wordBuffer.push(codePoint)
    // if the state was a word, but no more, emit the word
    if (wordBuffer.length && !isWord(codePoint)) {
      currentArray.push(String.fromCodePoint(...wordBuffer))
      wordBuffer.length = 0
    }
    switch (codePoint) {
      case LPAR: {
        const newArray = []
        currentArray.push(newArray)
        stack.push(newArray)
        currentArray = newArray
        break
      }
      case RPAR: {
        // only pop if we have something to pop, this allows for extra closing parens
        if (stack.length === 1)
          warnings.push({ type: 'extraClosingParen', index: i })
        else {
          stack.pop()
          currentArray = stack.at(-1)
        }
        break
      }
    }
    i++
  }
  if (wordBuffer.length) currentArray.push(String.fromCodePoint(...wordBuffer))

  if (stack.length !== 1)
    warnings.push({ type: 'unclosedParens', number: stack.length - 1 })
  return { forms: topLevelArray, warnings }
}