Convert unicode range to actual character range (Kotlin & Swift & TypeScript)

convertUnicodeRangeToCharRange.kt

import android.icu.text.BreakIterator

fun convertUnicodeRangeToCharRange(str: String, unicodeRange: IntRange): IntRange? {
    val charIterator = BreakIterator.getCharacterInstance()
    charIterator.setText(str)

    var charStart = -1
    var charEnd = -1
    var currentUnicodeIndex = 0
    var charIndex = 0

    while (charIterator.next() != BreakIterator.DONE) {
        if (currentUnicodeIndex == unicodeRange.first) {
            charStart = charIndex
        }
        if (currentUnicodeIndex == unicodeRange.last + 1) {
            charEnd = charIndex
            break
        }

        currentUnicodeIndex += charIterator.current() - charIterator.previous()
        charIterator.next()
        charIndex++
    }

    if (charEnd == -1) {
        charEnd = charIndex
    }

    if (charStart == -1) {
        return null
    }

    return IntRange(charStart, charEnd - 1)
}

fun main() {
    val str = "🙂‍↕️ hello world"
    val unicodeRange = 0..6

    val charRange = convertUnicodeRangeToCharRange(str, unicodeRange)
    println("Converted character range: $charRange")
    if (charRange != null) {
        println("Substring: ${str.substring(charRange)}")
        // Output: 0..2
    }
}

convertUnicodeRangeToCharRange.swift

import Foundation

func convertUnicodeRangeToCharRange(_ str: String, unicodeRange: NSRange) -> Range<String.Index>? {
  guard
    let start16 = str.utf16.index(
      str.utf16.startIndex, offsetBy: unicodeRange.location, limitedBy: str.utf16.endIndex),
    let end16 = str.utf16.index(
      start16, offsetBy: unicodeRange.length, limitedBy: str.utf16.endIndex),
    let start = String.Index(start16, within: str),
    let end = String.Index(end16, within: str)
  else {
    return nil
  }
  return start..<end
}

let str = "🙂‍↕️ hello world"
let unicodeRange = NSRange(location: 0, length: 6)

if let charRange = convertUnicodeRangeToCharRange(str, unicodeRange: unicodeRange) {
  print("Character range: \"\(charRange)\"")
  // Output: Character range: "🙂‍↕️ "
} else {
  print("Invalid range")
}

convertUnicodeRangeToCharRange.ts

function convertUnicodeRangeToCharRange(str: string, unicodeRange: { start: number; end: number }) {
  const segmenter = new Intl.Segmenter();
  const segments = Array.from(segmenter.segment(str));

  let charStart = 0;
  let charEnd = 0;
  let currentUnicodeIndex = 0;

  for (const [index, segment] of segments.entries()) {
    if (currentUnicodeIndex === unicodeRange.start) {
      charStart = index;
    }
    if (currentUnicodeIndex === unicodeRange.end) {
      charEnd = index;
      break;
    }
    currentUnicodeIndex += segment.segment.length;
  }

  if (charEnd === 0 && currentUnicodeIndex === unicodeRange.end) {
    charEnd = segments.length;
  }

  return { start: charStart, end: charEnd };
}

const str1 = '🙂‍↕️ hello world';
const unicodeRange = { start: 0, end: 6 };

const charRange = convertUnicodeRangeToCharRange(str1, unicodeRange);
console.log(charRange);
// Output: { start: 0, end: 2 }