Bible Reference Tokenizer

readme.md

Links

• Deprecation notice: Removing anonymous gist creation
• A Compiler From Scratch

bible-reference-tokenizer.js

// const reference = 'Song of Solomon 7 NLT'
// const reference = 'Song of Solomon NLT'
// const reference = 'Song of Solomon'
// const reference = 'Proverbs 3:21-26 NLT'
// const reference = 'Proverbs 3 NLT'
// const reference = 'Proverbs 3: NLT'
const reference = 'Proverbs 3:1 NLT'

const LINE = '@'

const TOKEN_TYPES = [
  ['BOOK', /\b(song of solomon|song of songs|psalms|proverbs)\b/ui],
  ['PASSAGE', /(\b([\d]+[:]?[\d]?[-]?[\d]?)\b)/ui],
  ['TRANSLATION', /\b(NLT|NKJV|NIV)\b/ui],
]

const tokenizer = (source) => {
  let reference = source
  console.log({ reference })

  TOKEN_TYPES.forEach((tokenType, iteration) => {
    console.log(LINE.repeat(10), iteration, LINE.repeat(50))
    const [type, rule] = tokenType
    console.log({type, rule})

    const matched = reference.match(rule)
    console.log({ matched })

    if (matched) {
      let value = matched[1]
      reference = reference.slice(value.length).trim()
      let token = { type, value }
      console.log({ token, reference})
    }
  })
}

tokenizer(reference)