[PHP] 한글 자음모음 분리 함수

  1 function utfCharToNumber($char) {
  2     $i = 0; 
  3     $number = ''; 
  4     $convmap = array(0x80, 0xffff, 0, 0xffff);
  5      
  6     $number = mb_encode_numericentity($char, $convmap, 'UTF-8'); 
  7  
  8     return $number; 
  9 } 
 10  
 11 function strToArray($str){ 
 12     $result = array(); 
 13     $stop = mb_strlen($str, 'UTF-8'); 
 14     for( $idx = 0; $idx < $stop; $idx++)  
 15     {  
 16         $result[] = mb_substr($str, $idx, 1, 'UTF-8');  
 17     }  
 18     return $result; 
 19 } 
 20  
 21 function parseInt($string) {  
 22     if(preg_match('/(\d+)/', $string, $array)) {  
 23         return $array[1]; 
 24     } else {  
 25         return 0; 
 26     }  
 27 }  
 28  
 29 function mb_chr($ord) 
 30 { 
 31     if ($ord < 128) return \chr($ord); // 1-byte 
 32     for ($i = 1; $i < 6 && $ord >= (1 << 5*$i+6); $i++); // units 
 33     $chr = \chr(($ord >> $i*6) + 256-(1 << 6-$i+1)); // start byte 
 34     for ($i -= 1; $i >= 0; $i--) // multi-bytes 
 35     { 
 36         $chr .= \chr((63 & ($ord >> $i*6)) + 128); 
 37     } 
 38     return $chr; 
 39 } 
 40  
 41 function make_jamo($str){ 
 42     //초성(19자) ㄱ ㄲ ㄴ ㄷ ㄸ ㄹ ㅁ ㅂ ㅃ ㅅ ㅆ ㅇ ㅈ ㅉ ㅊ ㅋ ㅌ ㅍ ㅎ 
 43     $ChoSeong = array (0x3131, 0x3132, 0x3134, 0x3137, 0x3138,
 44                           0x3139, 0x3141, 0x3142, 0x3143, 0x3145, 0x3146, 0x3147, 0x3148,
 45                         0x3149, 0x314a, 0x314b, 0x314c, 0x314d, 0x314e );
 46  
 47     //중성(21자) ㅏ ㅐ ㅑ ㅒ ㅓ ㅔ ㅕ ㅖ ㅗ ㅘ(9) ㅙ(10) ㅚ(11) ㅛ ㅜ ㅝ(14) ㅞ(15) ㅟ(16) ㅠ ㅡ ㅢ(19) ㅣ 
 48     $JungSeong = array (0x314f, 0x3150, 0x3151, 0x3152, 0x3153,
 49                         0x3154, 0x3155, 0x3156, 0x3157, 0x3158, 0x3159, 0x315a, 0x315b,
 50                         0x315c, 0x315d, 0x315e, 0x315f, 0x3160, 0x3161, 0x3162, 0x3163 );
 51  
 52     //종성(28자) <없음> ㄱ ㄲ ㄳ(3) ㄴ ㄵ(5) ㄶ(6) ㄷ ㄹ ㄺ(9) ㄻ(10) ㄼ(11) ㄽ(12) ㄾ(13) ㄿ(14) ㅀ(15) ㅁ ㅂ ㅄ(18) ㅅ ㅆ ㅇ ㅈ ㅊ ㅋ ㅌ ㅍ ㅎ 
 53     $JongSeong = array (0x0000, 0x3131, 0x3132, 0x3133, 0x3134,
 54                         0x3135, 0x3136, 0x3137, 0x3139, 0x313a, 0x313b, 0x313c, 0x313d,
 55                         0x313e, 0x313f, 0x3140, 0x3141, 0x3142, 0x3144, 0x3145, 0x3146,
 56                         0x3147, 0x3148, 0x314a, 0x314b, 0x314c, 0x314d, 0x314e );
 57  
 58     $chars = array(); 
 59     $result = array(); 
 60     $array_str = array(); 
 61     $array_str = strToArray($str); 
 62  
 63     for($i = 0 ; $i < mb_strlen($str, 'UTF-8'); $i++){ 
 64         $one_char_num = utfCharToNumber($array_str[$i]); 
 65         $one_char_num = substr($one_char_num, 2, mb_strlen($one_char_num, 'UTF-8')-3); 
 66  
 67         // "AC00:가" ~ "D7A3:힣"에 속한 글자만 분해.(한글 모두) 
 68         if($one_char_num >= 0xAC00 && $one_char_num <= 0xD7A3){ 
 69             $i1 = 0; 
 70             $i2 = 0; 
 71             $i3 = 0; 
 72  
 73             $i3 = $one_char_num - 0xAC00;
 74             $i1 = $i3 / (21 * 28);
 75             $i3 = $i3 % (21 * 28);
 76  
 77             $i2 = $i3 / 28; 
 78             $i3 = $i3 % 28; 
 79  
 80             $result[] = mb_chr($ChoSeong[parseInt($i1)]); 
 81  
 82             switch(parseInt($i2)){ 
 83                 case 9 : $result[] = 'ㅗㅏ'; break; 
 84                 case 10 : $result[] = 'ㅗㅐ'; break; 
 85                 case 11 : $result[] = 'ㅗㅣ'; break; 
 86                 case 14 : $result[] = 'ㅜㅓ'; break; 
 87                 case 15 : $result[] = 'ㅜㅔ'; break; 
 88                 case 16 : $result[] = 'ㅜㅣ'; break; 
 89                 case 19 : $result[] = 'ㅡㅣ'; break; 
 90  
 91                 default : $result[] = mb_chr($JungSeong[parseInt($i2)]); break;
 92             } 
 93  
 94             if ($i3 != 0x0000) // c가 0이 아니면, 즉 받침이 있으면 
 95             {                         
 96                 //복자음 분리 
 97                 switch(parseInt($i3)){ 
 98                     case 3 : $result[] = 'ㄱㅅ'; break; 
 99                     case 5 : $result[] = 'ㄴㅈ'; break;  
100                     case 6 : $result[] = 'ㄴㅎ'; break; 
101                     case 9 : $result[] = 'ㄹㄱ'; break; 
102                     case 10 : $result[] = 'ㄹㅁ'; break; 
103                     case 11 : $result[] = 'ㄹㅂ'; break; 
104                     case 12 : $result[] = 'ㄹㅅ'; break; 
105                     case 13 : $result[] = 'ㄹㅌ'; break; 
106                     case 14 : $result[] = 'ㄹㅍ'; break; 
107                     case 15 : $result[] = 'ㄹㅎ'; break; 
108                     case 18 : $result[] = 'ㅂㅅ'; break; 
109                     default : $result[] = mb_chr($JongSeong[parseInt($i3)]); break;
110                 } 
111             } 
112         }else{ 
113             $result[] = $array_str[$i]; 
114         } 
115     } 
116  
117     return $result; 
118 }