index.js

const {join} = require('path')
const {readFileSync} = require('fs')
const parse = require('csv-parse/lib/sync')
const {deburr} = require('lodash')

const rules = parse(readFileSync(join(__dirname, 'normadresse.csv')), {
  columns: true
}).reduce((rules, rule) => {
  const step = parseFloat(rule.etape, 10)

  rules[step] = rules[step] || []

  rules[step].push({
    long: rule.long,
    short: rule.court.replace(/\\g<(\d+)>/g, '$$$1')
  })

  return rules
}, {})

function selectShortWords(input, output, maxLength) {
  const long = input.split(' ')
  const short = output.split(' ')

  for (let i = short.length - 1; i >= 0; i--) {
    if (short[i] === '@') {
      long[i - 1] = short[i - 1]
      delete long[i]
      delete short[i]
    }
  }

  let next
  for (let i = 1; i < short.length; ++i) {
    next = [
      ...short.slice(0, i).filter(w => w),
      ...long.slice(i).filter(w => w)
    ].join(' ')

    if (next.length <= maxLength) {
      break
    }
  }

  return next
}

function normalize(input, maxLength) {
  input = deburr(input)
    .toUpperCase()
    .replace(/[^A-Z0-9s]/g, ' ')
    .replace(/ {2}/g, ' ')

  let output = input
  if (output.length <= maxLength) {
    return output
  }

  // 1 - abréviation du type de voie
  for (const rule of rules[1]) {
    output = output.replace(new RegExp(rule.long), rule.short)
  }

  output = selectShortWords(input, output, maxLength)
  if (output.length <= maxLength) {
    return output
  }

  input = output

  // 2 - abréviation des titres militaires, religieux et civils
  for (let step = 0; step < 2; step++) {
    for (const rule of rules[2]) {
      output = output.replace(new RegExp(` ${rule.long} `), ` ${rule.short} `)
    }
  }

  output = selectShortWords(input, output, maxLength)
  if (output.length <= maxLength) {
    return output
  }

  input = output

  // 4 - abréviations générales
  for (let step = 0; step < 3; step++) {
    for (const rule of rules[4]) {
      output = output.replace(new RegExp(`(^| )${rule.long} `), ` ${rule.short.toLowerCase()} `).trim()
    }
  }

  output = selectShortWords(input, output, maxLength)
  if (output.length <= maxLength) {
    return output
  }

  input = output

  // 5 - abréviation type de voies
  for (let step = 0; step < 2; step++) {
    for (const rule of rules[5]) {
      output = output.replace(new RegExp(` ${rule.long.trim()} `), ` ${rule.short.trim().toLowerCase()} `)
    }

    for (const rule of rules[1]) {
      output = output.replace(new RegExp(` ${rule.long.trim()} `), ` ${rule.short.trim().toLowerCase()} `)
    }
  }

  output = selectShortWords(input, output, maxLength)
  if (output.length <= maxLength) {
    return output
  }

  input = output

  // 3 - abréviations prénoms sauf pour ST prénoms
  let words = output.split(' ')
  for (let i = 1; i < words.length - 1; i++) {
    const word = words[i]

    if (words[i - 1].slice(0, 5) !== 'SAINT') {
      for (const rule of rules[3]) {
        if (new RegExp(`${rule.long}`).test(word)) {
          words[i] = rule.short.toLowerCase()
        }
      }
    }

    output = words.join(' ')
  }

  output = selectShortWords(input, output, maxLength)
  if (output.length <= maxLength) {
    return output
  }

  input = output

  // 6 - abréviation saint/sainte et prolonge(e)/inférieur(e)
  for (let step = 0; step < 2; step++) {
    for (const rule of rules[6]) {
      output = output.replace(new RegExp(rule.long), rule.short.toLowerCase())
    }
  }

  output = selectShortWords(input, output, maxLength)
  if (output.length <= maxLength) {
    return output
  }

  input = output

  // 5bis - type de voie en début...
  for (const rule of rules[5]) {
    output = output.replace(new RegExp(`^${rule.long.trim()} `), `${rule.short.trim().toLowerCase()} `)
  }

  output = selectShortWords(input, output, maxLength)
  if (output.length <= maxLength) {
    return output
  }

  input = output

  // 9 - remplacement des particules des noms propres pour ne pas les supprimer
  for (const rule of rules[9]) {
    output = output.replace(new RegExp(rule.long), rule.short)
  }

  // 10 - élimination des articles
  for (let step = 0; step < 6; step++) {
    output = output.replace(/ (LE|LA|LES|AU|AUX|DE|DU|DES|D|ET|A|L|SUR|EN) /, ' ')

    if (output.length <= maxLength) {
      return output
    }
  }

  // 11 - abréviations résiduelle
  words = output.split(' ')
  for (let i = 1; i < words.length - 1; ++i) {
    const word = words[i]

    if (word === word.toUpperCase() && word.length > 1 && word.charCodeAt(0) >= 'A'.charCodeAt(0)) {
      words[i] = word[0]

      output = words.join(' ')
      if (output.length <= maxLength) {
        return output
      }
    }
  }

  // 12 - élimination des articles
  for (let step = 0; step < 4; step++) {
    output = output.replace(/ (le|la|les|au|aux|de|du|des|d|et|a|l|sur) /, ' ')

    if (output.length <= maxLength) {
      return output
    }
  }

  return output
}

module.exports = function (input, maxLength = 32) {
  return normalize(input, maxLength).toUpperCase()
}