유니코드 한글 자소 문자 처리

한글 음절 문자를 자소 단위로 분해하거나 또는 반대로 자소 문자들을 합쳐서 하나의 음절 문자로 만드는 자바 소스 코드이다. 이에 관한 간략한 배경 지식이 필요한 경우, 이 글을 참고하기 바란다.

package kr.pe.elex.hangeul;
/**
* Created by Elex on 2014-07-11.
*/public class HCharacter {
    // 한글 음절
           protected static final char HANGEUL_SYLLABLE_BEGIN = ' \uAC00' ;
    protected static final char HANGEUL_SYLLABLE_END = ' \uD7AF' ;

    // 한글 자모
           protected static final char HANGEUL_JAMO_BEGIN = ' \u1100 ';
    protected static final char HANGEUL_JAMO_END = ' \u11FF ';

    protected static final char HANGEUL_JAMO_CHOSEONG_BEGIN = ' \u1100' ;
    protected static final char HANGEUL_JAMO_CHOSEONG_END = ' \u1112' ;
    protected static final char HANGEUL_JAMO_CHOSEONG_OLD_BEGIN = ' \u1113' ;
    protected static final char HANGEUL_JAMO_CHOSEONG_OLD_END = ' \u115F' ;
    protected static final char HANGEUL_JAMO_JUNGSEONG_BEGIN = ' \u1161' ;
    protected static final char HANGEUL_JAMO_JUNGSEONG_END = ' \u1175' ;
    protected static final char HANGEUL_JAMO_JUNGSEONG_OLD_BEGIN = ' \u1176' ;
    protected static final char HANGEUL_JAMO_JUNGSEONG_OLD_END = ' \u11A7' ;
    protected static final char HANGEUL_JAMO_JONGSEONG_BEGIN = ' \u11A8' ;
    protected static final char HANGEUL_JAMO_JONGSEONG_END = ' \u11C2' ;
    protected static final char HANGEUL_JAMO_JONGSEONG_OLD_BEGIN = ' \u11C3' ;
    protected static final char HANGEUL_JAMO_JONGSEONG_OLD_END = ' \u11FF' ;

    // 한글 호환 자모
           protected static final char HANGEUL_COMPAT_JAMO_BEGIN = ' \u3131' ;
    protected static final char HANGEUL_COMPAT_JAMO_END = ' \u318E' ;

    protected static final char HANGEUL_COMPAT_JAMO_CONSONANT_BEGIN = ' \u3131' ;
    protected static final char HANGEUL_COMPAT_JAMO_CONSONANT_END = ' \u314E' ;
    protected static final char HANGEUL_COMPAT_JAMO_VOWEL_BEGIN = ' \u314F' ;
    protected static final char HANGEUL_COMPAT_JAMO_VOWEL_END = ' \u3163' ;
    protected static final char HANGEUL_COMPAT_JAMO_CONSONANT_OLD_BEGIN = ' \u3165' ;
    protected static final char HANGEUL_COMPAT_JAMO_CONSONANT_OLD_END = ' \u3186' ;
    protected static final char HANGEUL_COMPAT_JAMO_VOWEL_OLD_BEGIN = ' \u3187' ;
    protected static final char HANGEUL_COMPAT_JAMO_VOWEL_OLD_END = ' \u318E' ;

    //
    static final char HANGUL_CHOSEONG_KIYEOK = ' \u1100 ';
    static final char HANGUL_CHOSEONG_SSANGKIYEOK = ' \u1101 ';
    static final char HANGUL_CHOSEONG_NIEUN = ' \u1102 ';
    static final char HANGUL_CHOSEONG_TIKEUT = ' \u1103 ';
    static final char HANGUL_CHOSEONG_SSANGTIKEUT = ' \u1104 ';
    static final char HANGUL_CHOSEONG_RIEUL = ' \u1105 ';
    static final char HANGUL_CHOSEONG_MIEUM = ' \u1106 ';
    static final char HANGUL_CHOSEONG_PIEUP = ' \u1107 ';
    static final char HANGUL_CHOSEONG_SSANGPIEUP = ' \u1108 ';
    static final char HANGUL_CHOSEONG_SIOS = ' \u1109 ';
    static final char HANGUL_CHOSEONG_SSANGSIOS = ' \u110A ';
    static final char HANGUL_CHOSEONG_IEUNG = ' \u110B ';
    static final char HANGUL_CHOSEONG_CIEUC = ' \u110C ';
    static final char HANGUL_CHOSEONG_SSANGCIEUC = ' \u110D ';
    static final char HANGUL_CHOSEONG_CHIEUCH = ' \u110E ';
    static final char HANGUL_CHOSEONG_KHIEUKH = ' \u110F ';
    static final char HANGUL_CHOSEONG_THIEUTH = ' \u1110 ';
    static final char HANGUL_CHOSEONG_PHIEUPH = ' \u1111 ';
    static final char HANGUL_CHOSEONG_HIEUH = ' \u1112 ';

    static final char HANGUL_JUNGSEONG_A = ' \u1161' ;
    static final char HANGUL_JUNGSEONG_AE = ' \u1162 ';
    static final char HANGUL_JUNGSEONG_YA = ' \u1163 ';
    static final char HANGUL_JUNGSEONG_YAE = ' \u1164 ';
    static final char HANGUL_JUNGSEONG_EO = ' \u1165 ';
    static final char HANGUL_JUNGSEONG_E = ' \u1166' ;
    static final char HANGUL_JUNGSEONG_YEO = ' \u1167 ';
    static final char HANGUL_JUNGSEONG_YE = ' \u1168 ';
    static final char HANGUL_JUNGSEONG_O = ' \u1169' ;
    static final char HANGUL_JUNGSEONG_WA = ' \u116A ';
    static final char HANGUL_JUNGSEONG_WAE = ' \u116B ';
    static final char HANGUL_JUNGSEONG_OE = ' \u116C ';
    static final char HANGUL_JUNGSEONG_YO = ' \u116D ';
    static final char HANGUL_JUNGSEONG_U = ' \u116E' ;
    static final char HANGUL_JUNGSEONG_WEO = ' \u116F ';
    static final char HANGUL_JUNGSEONG_WE = ' \u1170 ';
    static final char HANGUL_JUNGSEONG_WI = ' \u1171 ';
    static final char HANGUL_JUNGSEONG_YU = ' \u1172 ';
    static final char HANGUL_JUNGSEONG_EU = ' \u1173 ';
    static final char HANGUL_JUNGSEONG_YI = ' \u1174 ';
    static final char HANGUL_JUNGSEONG_I = ' \u1175' ;

    static final char HANGUL_JONGSEONG_KIYEOK = ' \u11A8 ';
    static final char HANGUL_JONGSEONG_SSANGKIYEOK = ' \u11A9 ';
    static final char HANGUL_JONGSEONG_KIYEOK_SIOS = ' \u11AA ';
    static final char HANGUL_JONGSEONG_NIEUN = ' \u11AB ';
    static final char HANGUL_JONGSEONG_NIEUN_CIEUC = ' \u11AC ';
    static final char HANGUL_JONGSEONG_NIEUN_HIEUH = ' \u11AD ';
    static final char HANGUL_JONGSEONG_TIKEUT = ' \u11AE ';
    static final char HANGUL_JONGSEONG_RIEUL = ' \u11AF ';
    static final char HANGUL_JONGSEONG_RIEUL_KIYEOK = ' \u11B0' ;
    static final char HANGUL_JONGSEONG_RIEUL_MIEUM = ' \u11B1 ';
    static final char HANGUL_JONGSEONG_RIEUL_PIEUP = ' \u11B2 ';
    static final char HANGUL_JONGSEONG_RIEUL_SIOS = ' \u11B3 ';
    static final char HANGUL_JONGSEONG_RIEUL_THIEUTH = ' \u11B4' ;
    static final char HANGUL_JONGSEONG_RIEUL_PHIEUPH = ' \u11B5' ;
    static final char HANGUL_JONGSEONG_RIEUL_HIEUH = ' \u11B6 ';
    static final char HANGUL_JONGSEONG_MIEUM = ' \u11B7 ';
    static final char HANGUL_JONGSEONG_PIEUP = ' \u11B8 ';
    static final char HANGUL_JONGSEONG_PIEUP_SIOS = ' \u11B9 ';
    static final char HANGUL_JONGSEONG_SIOS = ' \u11BA ';
    static final char HANGUL_JONGSEONG_SSANGSIOS = ' \u11BB ';
    static final char HANGUL_JONGSEONG_IEUNG = ' \u11BC ';
    static final char HANGUL_JONGSEONG_CIEUC = ' \u11BD ';
    static final char HANGUL_JONGSEONG_CHIEUCH = ' \u11BE ';
    static final char HANGUL_JONGSEONG_KHIEUKH = ' \u11BF ';
    static final char HANGUL_JONGSEONG_THIEUTH = ' \u11C0 ';
    static final char HANGUL_JONGSEONG_PHIEUPH = ' \u11C1 ';
    static final char HANGUL_JONGSEONG_HIEUH = ' \u11C2 ';

    public static final char FILLER = 0 ;

    private char character ;
    private char initialConsonant , medialVowel , finalConsonant ;

    public HCharacter( char character) throws OutOfUnicodeRangeException {
          if (isSyllableLetter (character)) {
              this. character = character;
              this. finalConsonant = getJongSeong( this. character );
              this. medialVowel = getJungSeong( this. character );
              this. initialConsonant = getChoSeong( this. character );

         } else if ( isJamoLetter(character)) {
              this. character = character;
              this. finalConsonant = FILLER;
              this. medialVowel = FILLER;
              this. initialConsonant = FILLER;

         } else if ( isJamoCompatLetter(character)) {
              this. character = character;
              this. finalConsonant = FILLER;
              this. medialVowel = FILLER;
              this. initialConsonant = FILLER;

         } else {
              throw new OutOfUnicodeRangeException(character);
         }
    }

    public HCharacter( char initialConsonant, char medialVowel, char finalConsonant) throwsOutOfUnicodeRangeException {
          // 한글 호환 자모의 코드를 그냥 자모로 변환한다 .
          if (isJamoCompatLetter (initialConsonant)) {
              for ( char [] row : CONV_INITIAL_CONSONANTS ) {
                  if (row[ 1 ] == initialConsonant) {
                      initialConsonant = row[ 0 ];
                       break ;
                 }
             }
         }

          if (isJamoCompatLetter (medialVowel)) {
             medialVowel -= HANGEUL_COMPAT_JAMO_VOWEL_BEGIN ;
             medialVowel += HANGEUL_JAMO_JUNGSEONG_BEGIN ;
         }

          if (isJamoCompatLetter (finalConsonant)) {
              for ( char [] row : CONV_FINAL_CONSONANTS ) {
                  if (row[ 1 ] == finalConsonant) {
                      finalConsonant = row[ 0 ];
                       break ;
                 }
             }
         }

          // 옛한글은 완성형 문자 없음
                    if (!isJamoLetter(initialConsonant) || isOldLetter(initialConsonant))
              throw new OutOfUnicodeRangeException(initialConsonant);
          if (! isJamoLetter(medialVowel) || isOldLetter(medialVowel))
              throw new OutOfUnicodeRangeException(medialVowel);
          if (finalConsonant!= FILLER && (!isJamoLetter(finalConsonant) || isOldLetter(finalConsonant)))
              throw new OutOfUnicodeRangeException(finalConsonant);

          this. initialConsonant = initialConsonant;
          this. medialVowel = medialVowel;
          this. finalConsonant = finalConsonant;

          // 자모 결합
                    initialConsonant -= HANGEUL_JAMO_CHOSEONG_BEGIN ;
         medialVowel -= HANGEUL_JAMO_JUNGSEONG_BEGIN ;
          if (finalConsonant!= FILLER) {
             finalConsonant -= HANGEUL_JAMO_JONGSEONG_BEGIN ;
         }

          int val = 0;
         val += initialConsonant * 28 * 21 ;
         val += medialVowel * 28;
          if (finalConsonant!= FILLER) val += finalConsonant + 1 ;
         val += HANGEUL_SYLLABLE_BEGIN;

          this. character = ( char )val;
    }

    public HCharacter( char initialConsonant, char medialVowel) throws OutOfUnicodeRangeException {
          this(initialConsonant, medialVowel, FILLER );
    }

    public char getChar(){
          return character ;
    }

    public char getChoSeong() {
          return initialConsonant ;
    }

    public char getJungSeong() {
          return medialVowel ;
    }

    public char getJongSeong() {
          return finalConsonant ;
    }

    public char getChoSeongCompat() {
          return toCharCompat( initialConsonant );
    }

    public char getJungSeongCompat() {
          return toCharCompat( medialVowel );
    }

    public char getJongSeongCompat() {
          return toCharCompat( finalConsonant );
    }

    protected static char toCharCompat( char ch){
          if (! isJamoLetter(ch)) return ch;

          if (isVowelLetter (ch)){
              return ( char ) (ch - HANGEUL_JAMO_JUNGSEONG_BEGIN + HANGEUL_COMPAT_JAMO_VOWEL_BEGIN );
         } else {
              for ( char [] row : CONV_INITIAL_CONSONANTS ) {
                  if (row[ 0 ] == ch) {
                       return row[1 ];
                 }
             }
              for ( char [] row : CONV_FINAL_CONSONANTS ) {
                  if (row[ 0 ] == ch) {
                       return row[1 ];
                 }
             }
         }
          return ' \u3164 ';
    }

    protected static char getChoSeong( char character) {
          int val = ( int)character;
         val -= HANGEUL_SYLLABLE_BEGIN;
         val = val / 28 / 21 ;
         val += HANGEUL_JAMO_CHOSEONG_BEGIN;
          return ( char) val;
    }

    protected static char getJungSeong( char character) {
          int val = ( int)character;
         val -= HANGEUL_SYLLABLE_BEGIN;
         val = val / 28 % 21 ;
         val += HANGEUL_JAMO_JUNGSEONG_BEGIN;
          return ( char) val;
    }

    protected static char getJongSeong( char character) {
          int val = ( int)character;
         val -= HANGEUL_SYLLABLE_BEGIN;
         val = val % 28 - 1 ;
          if (val == - 1) return FILLER; //종성없음
                    val += HANGEUL_JAMO_JONGSEONG_BEGIN;
          return ( char) val;
    }

    public String toString(){
          return String. valueOf( character);
    }

    public boolean isJamoLetter(){
          return isJamoLetter( character );
    }

    protected static boolean isJamoLetter( char ch){
          return (ch>= HANGEUL_JAMO_BEGIN && ch<=HANGEUL_JAMO_END );
    }

    public boolean isJamoCompatLetter(){
          return isJamoCompatLetter( character );
    }

    protected static boolean isJamoCompatLetter( char ch){
          return (ch>= HANGEUL_COMPAT_JAMO_BEGIN && ch<=HANGEUL_COMPAT_JAMO_END );
    }

    public boolean isSyllableLetter(){
          return isSyllableLetter( character );
    }

    protected static boolean isSyllableLetter( char ch){
          return (ch>= HANGEUL_SYLLABLE_BEGIN && ch<=HANGEUL_SYLLABLE_END );
    }

    protected static boolean isVowelLetter( char ch) {
          return (ch>= HANGEUL_JAMO_JUNGSEONG_BEGIN && ch<=HANGEUL_JAMO_JUNGSEONG_END ) ||
                 (ch>= HANGEUL_COMPAT_JAMO_VOWEL_BEGIN && ch<=HANGEUL_COMPAT_JAMO_VOWEL_END ) ||
                 (ch>= HANGEUL_JAMO_JUNGSEONG_OLD_BEGIN && ch<=HANGEUL_JAMO_JUNGSEONG_OLD_END ) ||
                 (ch>= HANGEUL_COMPAT_JAMO_VOWEL_OLD_BEGIN && ch<=HANGEUL_COMPAT_JAMO_VOWEL_OLD_END );
    }

    protected static boolean isConsonantLetter( char ch) {
          if (isJamoLetter (ch) || isJamoCompatLetter(ch)) {
              return ! isVowelLetter(ch);
         }
          return false;
    }

    protected static boolean isOldLetter( char ch){
          return (ch>= HANGEUL_JAMO_CHOSEONG_OLD_BEGIN && ch<= HANGEUL_JAMO_CHOSEONG_OLD_END) ||
                 (ch>= HANGEUL_JAMO_JUNGSEONG_OLD_BEGIN && ch<=HANGEUL_JAMO_JUNGSEONG_OLD_END ) ||
                 (ch>= HANGEUL_JAMO_JONGSEONG_OLD_BEGIN && ch<=HANGEUL_JAMO_JONGSEONG_OLD_END ) ||
                 (ch>= HANGEUL_COMPAT_JAMO_CONSONANT_OLD_BEGIN && ch<=HANGEUL_COMPAT_JAMO_VOWEL_OLD_END );
    }

    private static final char [][] CONV_INITIAL_CONSONANTS = new char[][] {
             { HANGUL_CHOSEONG_KIYEOK, ' \u3131 '}, //ㄱ
                              { HANGUL_CHOSEONG_SSANGKIYEOK, ' \u3132 '}, //ㄲ
                              { HANGUL_CHOSEONG_NIEUN, ' \u3134 '}, //ㄴ
                              { HANGUL_CHOSEONG_TIKEUT, ' \u3137 '}, //ㄷ
                              { HANGUL_CHOSEONG_SSANGTIKEUT, ' \u3138 '}, //ㄸ
                              { HANGUL_CHOSEONG_RIEUL, ' \u3139 '}, //ㄹ
                              { HANGUL_CHOSEONG_MIEUM, ' \u3141 '}, //ㅁ
                              { HANGUL_CHOSEONG_PIEUP, ' \u3142 '}, //ㅂ
                              { HANGUL_CHOSEONG_SSANGPIEUP, ' \u3143 '}, //ㅃ
                              { HANGUL_CHOSEONG_SIOS, ' \u3145 '}, //ㅅ
                              { HANGUL_CHOSEONG_SSANGSIOS, ' \u3146 '}, //ㅆ
                              { HANGUL_CHOSEONG_IEUNG, ' \u3147 '}, //ㅇ
                              { HANGUL_CHOSEONG_CIEUC, ' \u3148 '}, //ㅈ
                              { HANGUL_CHOSEONG_SSANGCIEUC, ' \u3149 '}, //ㅉ
                              { HANGUL_CHOSEONG_CHIEUCH, ' \u314A '}, //ㅊ
                              { HANGUL_CHOSEONG_KHIEUKH, ' \u314B '}, //ㅋ
                              { HANGUL_CHOSEONG_THIEUTH, ' \u314C '}, //ㅌ
                              { HANGUL_CHOSEONG_PHIEUPH, ' \u314D '}, //ㅍ
                              { HANGUL_CHOSEONG_HIEUH, ' \u314E '} //ㅎ
          };

    private static final char [][] CONV_FINAL_CONSONANTS = new char[][] {
             { HANGUL_JONGSEONG_KIYEOK, ' \u3131 '}, //ㄱ
                              { HANGUL_JONGSEONG_SSANGKIYEOK, ' \u3132 '}, //ㄲ
                              { HANGUL_JONGSEONG_KIYEOK_SIOS, ' \u3133 '}, //ㄳ
                              { HANGUL_JONGSEONG_NIEUN, ' \u3134 '}, //ㄴ
                              { HANGUL_JONGSEONG_NIEUN_CIEUC, ' \u3135 '}, //ㄵ
                              { HANGUL_JONGSEONG_NIEUN_HIEUH, ' \u3136 '}, //ㄶ
                              { HANGUL_JONGSEONG_TIKEUT, ' \u3137 '}, //ㄷ
                              { HANGUL_JONGSEONG_RIEUL, ' \u3139 '}, //ㄹ
                              { HANGUL_JONGSEONG_RIEUL_KIYEOK, ' \u313A '}, //ㄺ
                              { HANGUL_JONGSEONG_RIEUL_MIEUM, ' \u313B '}, //ㄻ
                              { HANGUL_JONGSEONG_RIEUL_PIEUP, ' \u313C '}, //ㄼ
                              { HANGUL_JONGSEONG_RIEUL_SIOS, ' \u313D '}, //ㄽ
                              { HANGUL_JONGSEONG_RIEUL_THIEUTH , ' \u313E '}, //ㄾ
                              { HANGUL_JONGSEONG_RIEUL_PHIEUPH, ' \u313F '}, //ㄿ
                              { HANGUL_JONGSEONG_RIEUL_HIEUH, ' \u3140 '}, //ㅀ
                              { HANGUL_JONGSEONG_MIEUM, ' \u3141 '}, //ㅁ
                              { HANGUL_JONGSEONG_PIEUP, ' \u3142 '}, //ㅂ
                              { HANGUL_JONGSEONG_PIEUP_SIOS, ' \u3144 '}, //ㅄ
                              { HANGUL_JONGSEONG_SIOS, ' \u3145 '}, //ㅅ
                              { HANGUL_JONGSEONG_SSANGSIOS, ' \u3146 '}, //ㅆ
                              { HANGUL_JONGSEONG_IEUNG, ' \u3147 '}, //ㅇ
                              { HANGUL_JONGSEONG_CIEUC, ' \u3148 '}, //ㅈ
                              { HANGUL_JONGSEONG_CHIEUCH, ' \u314A '}, //ㅊ
                              { HANGUL_JONGSEONG_KHIEUKH, ' \u3148 '}, //ㅋ
                              { HANGUL_JONGSEONG_THIEUTH, ' \u314C '}, //ㅌ
                              { HANGUL_JONGSEONG_PHIEUPH, ' \u314D '}, //ㅍ
                              { HANGUL_JONGSEONG_HIEUH, ' \u314E '} //ㅎ
          };

    public class OutOfUnicodeRangeException extends Exception {
          private char character ;

          public OutOfUnicodeRangeException( char c) {
              this. character = c;
         }

          public String toString(){
              return Integer. toHexString(( int) character);
         }
    }

    public static void main(String[] args) throws OutOfUnicodeRangeException {
          //Usage 1.
         HCharacter character = new HCharacter( '호 ' );

        System. out.print(character.getChoSeongCompat());
        System. out.print(character.getJungSeongCompat());
        System. out.print(character.getJongSeongCompat());
         System. out.println();

          //Usage 2.
         System. out .println( new HCharacter( 'ㅇ ' , ' ㅏ '));
         System. out.println( new HCharacter( ' ㅌ ', 'ㅚ ' , ' ㄺ '));

    }
}