fix: 코드 효율 개선 (#124)

* fix: 모든 글자의 자소를 나누지 않고 마지막 글자만 계산합니다.

* fix: /\s/g to /\s+/g 일괄치환 효율 증가 bench: https://jsfiddle.net/crucify/g74b5d1h/

* fix: NFD 로 변환하여 자소계산 과정 생략하고 초성 ㄱ-ㅎ 만 남기게 수정

* fix: 자소계산을 생략하고 받침이 있는지만 검사합니다.

* fix: 유니코드 상수화 가독성 향상

* fix: 기존함수 받침문제 테스트 추가 (이 브랜치에서 해결됨)

* Create sixty-clouds-repeat.md

---------

Co-authored-by: 박찬혁 <pgg6713@gmail.com>

.changeset/sixty-clouds-repeat.md

@@ -0,0 +1,5 @@
+---
+"es-hangul": patch
+---
+
+fix: 코드 효율 개선

src/chosungIncludes.ts

@@ -2,13 +2,13 @@ import { disassembleHangulToGroups } from './disassemble';
 import { canBeChosung, getChosung } from './utils';
 
 export function chosungIncludes(x: string, y: string) {
-  const trimmedY = y.replace(/\s/g, '');
+  const trimmedY = y.replace(/\s+/g, '');
 
   if (!isOnlyChosung(trimmedY)) {
     return false;
   }
 
-  const chosungX = getChosung(x).replace(/\s/g, '');
+  const chosungX = getChosung(x).replace(/\s+/g, '');
   const chosungY = trimmedY;
 
   return chosungX.includes(chosungY);

src/constants.ts

@@ -3,6 +3,16 @@ export const COMPLETE_HANGUL_END_CHARCODE = '힣'.charCodeAt(0);
 export const NUMBER_OF_JONGSUNG = 28;
 export const NUMBER_OF_JUNGSUNG = 21;
 
+const _JASO_HANGUL_NFD = [...'각힣'.normalize('NFD')].map(char => char.charCodeAt(0)); // NFC 에 정의되지 않은 문자는 포함하지 않음
+export const JASO_HANGUL_NFD = {
+  START_CHOSEONG: _JASO_HANGUL_NFD[0], // ㄱ
+  START_JUNGSEONG: _JASO_HANGUL_NFD[1], // ㅏ
+  START_JONGSEONG: _JASO_HANGUL_NFD[2], // ㄱ
+  END_CHOSEONG: _JASO_HANGUL_NFD[3], // ㅎ
+  END_JUNGSEONG: _JASO_HANGUL_NFD[4], // ㅣ
+  END_JONGSEONG: _JASO_HANGUL_NFD[5], // ㅎ
+}
+
 /**
  * ㄱ -> 'ㄱ'
  * ㄳ -> 'ㄱㅅ' 으로 나눈다.

src/removeLastHangulCharacter.spec.ts

@@ -3,6 +3,7 @@ import { removeLastHangulCharacter } from './removeLastHangulCharacter';
 describe('removeLastHangulCharacter', () => {
   it('마지막 문자가 겹받침인 경우 홑받침으로 바꾼다.', () => {
     expect(removeLastHangulCharacter('안녕하세요 값')).toBe('안녕하세요 갑');
+    expect(removeLastHangulCharacter('안녕하세요 값이')).toBe('안녕하세요 값ㅇ');
   });
   it('마지막 문자가 초성과 중성의 조합으로 끝날 경우 초성만 남긴다.', () => {
     expect(removeLastHangulCharacter('프론트엔드')).toBe('프론트엔ㄷ');

src/removeLastHangulCharacter.ts

@@ -20,25 +20,13 @@ import { excludeLastElement } from './_internal';
  * removeLastHangulCharacter('신세계') // 신세ㄱ
  */
 export function removeLastHangulCharacter(words: string) {
-  const disassembledGroups = disassembleHangulToGroups(words);
-  const lastCharacter = disassembledGroups[disassembledGroups.length - 1];
-
+  const lastCharacter = words[words.length - 1];
   if (lastCharacter == null) {
     return '';
   }
-
-  const withoutLastCharacter = disassembledGroups
-    .filter(v => v !== lastCharacter)
-    .map(([first, middle, last]) => {
-      if (middle != null) {
-        return combineHangulCharacter(first, middle, last);
-      }
-
-      return first;
-    });
-
-  const [[first, middle, last]] = excludeLastElement(lastCharacter);
+  const disassembleLastCharacter = disassembleHangulToGroups(lastCharacter);
+  const [[first, middle, last]] = excludeLastElement(disassembleLastCharacter[0]);
   const result = middle != null ? combineHangulCharacter(first, middle, last) : first;
 
-  return [...withoutLastCharacter, result].join('');
+  return [words.substring(0, words.length - 1), result].join('');
 }

src/utils.ts

@@ -1,10 +1,22 @@
 import {
+  COMPLETE_HANGUL_START_CHARCODE,
+  COMPLETE_HANGUL_END_CHARCODE,
   HANGUL_CHARACTERS_BY_FIRST_INDEX,
   HANGUL_CHARACTERS_BY_LAST_INDEX,
   HANGUL_CHARACTERS_BY_MIDDLE_INDEX,
+  NUMBER_OF_JONGSUNG,
+  JASO_HANGUL_NFD,
 } from './constants';
-import { disassembleHangul, disassembleHangulToGroups } from './disassemble';
-import { disassembleCompleteHangulCharacter } from './disassembleCompleteHangulCharacter';
+import { disassembleHangulToGroups } from './disassemble';
+
+const EXTRACT_CHOSEONG_REGEX = new RegExp(
+  `[^\\u${JASO_HANGUL_NFD.START_CHOSEONG.toString(16)}-\\u${JASO_HANGUL_NFD.END_CHOSEONG.toString(16)}ㄱ-ㅎ\\s]+`,
+  'ug'
+);
+const CHOOSE_NFD_CHOSEONG_REGEX = new RegExp(
+  `[\\u${JASO_HANGUL_NFD.START_CHOSEONG.toString(16)}-\\u${JASO_HANGUL_NFD.END_CHOSEONG.toString(16)}]`,
+  'g'
+);
 
 /**
  * @name hasBatchim
@@ -26,9 +38,14 @@ export function hasBatchim(str: string) {
   if (lastChar == null) {
     return false;
   }
+  const charCode = lastChar.charCodeAt(0);
+  const isCompleteHangul = COMPLETE_HANGUL_START_CHARCODE <= charCode && charCode <= COMPLETE_HANGUL_END_CHARCODE;
 
-  const disassembled = disassembleCompleteHangulCharacter(lastChar);
-  return disassembled != null && disassembled.last !== '';
+  if (!isCompleteHangul) {
+    return false;
+  }
+
+  return (charCode - COMPLETE_HANGUL_START_CHARCODE) % NUMBER_OF_JONGSUNG > 0;
 }
 
 /**
@@ -49,12 +66,18 @@ export function hasBatchim(str: string) {
 export function hasSingleBatchim(str: string) {
   const lastChar = str[str.length - 1];
 
-  if (lastChar == null || hasBatchim(lastChar) === false) {
+  if (lastChar == null) {
     return false;
   }
+  const charCode = lastChar.charCodeAt(0);
+  const isCompleteHangul = COMPLETE_HANGUL_START_CHARCODE <= charCode && charCode <= COMPLETE_HANGUL_END_CHARCODE;
 
-  const disassembled = disassembleHangul(lastChar);
-  return disassembled.length === 3;
+  if (!isCompleteHangul) {
+    return false;
+  }
+
+  const batchimCode = (charCode - COMPLETE_HANGUL_START_CHARCODE) % NUMBER_OF_JONGSUNG;
+  return HANGUL_CHARACTERS_BY_LAST_INDEX[batchimCode].length === 1;
 }
 
 /**
@@ -73,9 +96,9 @@ export function hasSingleBatchim(str: string) {
  * getChosung('띄어 쓰기') // 'ㄸㅇ ㅆㄱ'
  */
 export function getChosung(word: string) {
-  return disassembleHangulToGroups(word).reduce((chosung, [consonant]) => {
-    return `${chosung}${consonant}`;
-  }, '');
+  return word.normalize('NFD')
+    .replace(EXTRACT_CHOSEONG_REGEX, '') // NFD ㄱ-ㅎ, NFC ㄱ-ㅎ 외 문자 삭제
+    .replace(CHOOSE_NFD_CHOSEONG_REGEX, $0 => HANGUL_CHARACTERS_BY_FIRST_INDEX[$0.charCodeAt(0) - 0x1100]); // NFD to NFC
 }
 
 /**