Make ArbitraryBase Unicode-aware (#1299)

* Make ArbitraryBase Unicode-aware

https://mathiasbynens.be/notes/javascript-unicode#counting-symbols

* Fix performance bug and add Unicode test

* Add BigInt version and push output chars to array

Author: lionel-rowe

Conversions/ArbitraryBase.js

@@ -1,20 +1,36 @@
 /**
-* Converts a string from one base to other
+* Divide two numbers and get the result of floor division and remainder
+* @param {number} dividend
+* @param {number} divisor
+* @returns {[result: number, remainder: number]}
+*/
+const floorDiv = (dividend, divisor) => {
+  const remainder = dividend % divisor
+  const result = Math.floor(dividend / divisor)
+
+  return [result, remainder]
+}
+
+/**
+* Converts a string from one base to other. Loses accuracy above the value of `Number.MAX_SAFE_INTEGER`.
 * @param {string} stringInBaseOne String in input base
 * @param {string} baseOneCharacters Character set for the input base
 * @param {string} baseTwoCharacters Character set for the output base
 * @returns {string}
 */
-const convertArbitraryBase = (stringInBaseOne, baseOneCharacters, baseTwoCharacters) => {
-  if ([stringInBaseOne, baseOneCharacters, baseTwoCharacters].map(arg => typeof arg).some(type => type !== 'string')) {
+const convertArbitraryBase = (stringInBaseOne, baseOneCharacterString, baseTwoCharacterString) => {
+  if ([stringInBaseOne, baseOneCharacterString, baseTwoCharacterString].map(arg => typeof arg).some(type => type !== 'string')) {
     throw new TypeError('Only string arguments are allowed')
   }
-  [baseOneCharacters, baseTwoCharacters].forEach(baseString => {
-    const charactersInBase = [...baseString]
+
+  const baseOneCharacters = [...baseOneCharacterString]
+  const baseTwoCharacters = [...baseTwoCharacterString]
+
+  for (const charactersInBase of [baseOneCharacters, baseTwoCharacters]) {
     if (charactersInBase.length !== new Set(charactersInBase).size) {
       throw new TypeError('Duplicate characters in character set are not allowed')
     }
-  })
+  }
   const reversedStringOneChars = [...stringInBaseOne].reverse()
   const stringOneBase = baseOneCharacters.length
   let value = 0
@@ -27,24 +43,57 @@ const convertArbitraryBase = (stringInBaseOne, baseOneCharacters, baseTwoCharact
     value += (digitNumber * placeValue)
     placeValue *= stringOneBase
   }
-  let stringInBaseTwo = ''
+  const outputChars = []
   const stringTwoBase = baseTwoCharacters.length
   while (value > 0) {
-    const remainder = value % stringTwoBase
-    stringInBaseTwo = baseTwoCharacters.charAt(remainder) + stringInBaseTwo
-    value /= stringTwoBase
+    const [divisionResult, remainder] = floorDiv(value, stringTwoBase)
+    outputChars.push(baseTwoCharacters[remainder])
+    value = divisionResult
   }
-  const baseTwoZero = baseTwoCharacters.charAt(0)
-  return stringInBaseTwo.replace(new RegExp(`^${baseTwoZero}+`), '')
+  return outputChars.reverse().join('') || baseTwoCharacters[0]
 }
 
-export { convertArbitraryBase }
+/**
+* Converts a arbitrary-length string from one base to other. Doesn't lose accuracy.
+* @param {string} stringInBaseOne String in input base
+* @param {string} baseOneCharacters Character set for the input base
+* @param {string} baseTwoCharacters Character set for the output base
+* @returns {string}
+*/
+const convertArbitraryBaseBigIntVersion = (stringInBaseOne, baseOneCharacterString, baseTwoCharacterString) => {
+  if ([stringInBaseOne, baseOneCharacterString, baseTwoCharacterString].map(arg => typeof arg).some(type => type !== 'string')) {
+    throw new TypeError('Only string arguments are allowed')
+  }
 
-// > convertArbitraryBase('98', '0123456789', '01234567')
-// '142'
+  const baseOneCharacters = [...baseOneCharacterString]
+  const baseTwoCharacters = [...baseTwoCharacterString]
 
-// > convertArbitraryBase('98', '0123456789', 'abcdefgh')
-// 'bec'
+  for (const charactersInBase of [baseOneCharacters, baseTwoCharacters]) {
+    if (charactersInBase.length !== new Set(charactersInBase).size) {
+      throw new TypeError('Duplicate characters in character set are not allowed')
+    }
+  }
+  const reversedStringOneChars = [...stringInBaseOne].reverse()
+  const stringOneBase = BigInt(baseOneCharacters.length)
+  let value = 0n
+  let placeValue = 1n
+  for (const digit of reversedStringOneChars) {
+    const digitNumber = BigInt(baseOneCharacters.indexOf(digit))
+    if (digitNumber === -1n) {
+      throw new TypeError(`Not a valid character: ${digit}`)
+    }
+    value += (digitNumber * placeValue)
+    placeValue *= stringOneBase
+  }
+  const outputChars = []
+  const stringTwoBase = BigInt(baseTwoCharacters.length)
+  while (value > 0n) {
+    const divisionResult = value / stringTwoBase
+    const remainder = value % stringTwoBase
+    outputChars.push(baseTwoCharacters[remainder])
+    value = divisionResult
+  }
+  return outputChars.reverse().join('') || baseTwoCharacters[0]
+}
 
-// > convertArbitraryBase('129', '0123456789', '01234567')
-// '201'
+export { convertArbitraryBase, convertArbitraryBaseBigIntVersion }

Conversions/test/ArbitraryBase.test.js

@@ -1,4 +1,4 @@
-import { convertArbitraryBase } from '../ArbitraryBase'
+import { convertArbitraryBase, convertArbitraryBaseBigIntVersion } from '../ArbitraryBase'
 
 test('Check the answer of convertArbitraryBase(98, 0123456789, 01234567) is 142', () => {
   const res = convertArbitraryBase('98', '0123456789', '01234567')
@@ -34,3 +34,23 @@ test('Check the answer of convertArbitraryBase(111, 0123456789, abcdefgh) is bfh
   const res = convertArbitraryBase('111', '0123456789', 'abcdefgh')
   expect(res).toBe('bfh')
 })
+
+test('Unicode awareness', () => {
+  const res = convertArbitraryBase('98', '0123456789', '💝🎸🦄')
+  expect(res).toBe('🎸💝🎸🦄🦄')
+})
+
+test('zero', () => {
+  const res = convertArbitraryBase('0', '0123456789', 'abc')
+  expect(res).toBe('a')
+})
+
+test('BigInt version with input string of arbitrary length', () => {
+  const resBigIntVersion = convertArbitraryBaseBigIntVersion(
+    String(10n ** 100n),
+    '0123456789',
+    '0123456789abcdefghijklmnopqrstuvwxyz'
+  )
+
+  expect(resBigIntVersion).toBe((10n ** 100n).toString(36))
+})