Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 54 |
|
0.00% |
0 / 5 |
CRAP | |
0.00% |
0 / 1 |
| Hooks | |
0.00% |
0 / 54 |
|
0.00% |
0 / 5 |
552 | |
0.00% |
0 / 1 |
| onParserFirstCallInit | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
42 | |||
| checkOverrideParam | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
6 | |||
| checkJosaonlyParam | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
6 | |||
| getJosa | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
42 | |||
| convertToJohabCode | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
56 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Extension:Josa |
| 4 | * Author: Bae Junehyeon <devunt@gmail.com> |
| 5 | * Original implementation by Park Shinjo <peremen@gmail.com> |
| 6 | * License: MIT License |
| 7 | */ |
| 8 | |
| 9 | namespace MediaWiki\Extension\Josa; |
| 10 | |
| 11 | use MediaWiki\Hook\ParserFirstCallInitHook; |
| 12 | use Parser; |
| 13 | |
| 14 | class Hooks implements ParserFirstCallInitHook { |
| 15 | /** @var array */ |
| 16 | private static $josaMap = [ |
| 17 | 'Eul/Ruel' => [ '을', '를', '을(를)' ], // 곶감을 / 사과를 |
| 18 | 'Eun/Neun' => [ '은', '는', '은(는)' ], // 곶감은 / 사과는 |
| 19 | 'E/Ga' => [ '이', '가', '이(가)' ], // 곶감이 / 사과가 |
| 20 | 'Gwa/Wa' => [ '과', '와', '과(와)' ], // 곶감과 / 사과와 |
| 21 | 'A/Ya' => [ '아', '야', '아(야)' ], // 태준아 / 철수야 |
| 22 | 'Euro/Ro' => [ '으로', '로', '(으)로' ], // 집으로 / 학교로 |
| 23 | 'E/' => [ '이', '', '(이)' ], // 태준이가 / 철수가 |
| 24 | ]; |
| 25 | |
| 26 | /** |
| 27 | * This dictionary matches each numbers and alphabets to the corresponding hangul pronounces. |
| 28 | * |
| 29 | * @var array |
| 30 | */ |
| 31 | private static $pronounceMap = [ |
| 32 | '0' => '영', '1' => '일', '2' => '이', |
| 33 | '3' => '삼', '4' => '사', '5' => '오', |
| 34 | '6' => '육', '7' => '칠', '8' => '팔', |
| 35 | '9' => '구', |
| 36 | |
| 37 | 'A' => '이', 'B' => '비', 'C' => '씨', |
| 38 | 'D' => '디', 'E' => '이', 'F' => '프', |
| 39 | 'G' => '지', 'H' => '치', 'I' => '이', |
| 40 | 'J' => '이', 'K' => '이', 'L' => '엘', |
| 41 | 'M' => '엠', 'N' => '엔', 'O' => '오', |
| 42 | 'P' => '피', 'Q' => '큐', 'R' => '알', |
| 43 | 'S' => '스', 'T' => '티', 'U' => '유', |
| 44 | 'V' => '이', 'W' => '유', 'X' => '스', |
| 45 | 'Y' => '이', 'Z' => '지', |
| 46 | ]; |
| 47 | |
| 48 | /** |
| 49 | * @param Parser $parser |
| 50 | */ |
| 51 | public function onParserFirstCallInit( $parser ) { |
| 52 | foreach ( self::$josaMap as $key => $value ) { |
| 53 | $parser->setFunctionHook( $key, static function ( |
| 54 | $parser, $str, $param1 = null, $param2 = null |
| 55 | ) use ( $key ) { |
| 56 | $josa = self::getJosa( $key, $str ); |
| 57 | |
| 58 | foreach ( [ $param1, $param2 ] as $param ) { |
| 59 | if ( $param === null ) { |
| 60 | break; |
| 61 | } |
| 62 | |
| 63 | $override = self::checkOverrideParam( $param ); |
| 64 | if ( $override !== false ) { |
| 65 | $josa = $override; |
| 66 | continue; |
| 67 | } |
| 68 | |
| 69 | $josaonly = self::checkJosaonlyParam( $param ); |
| 70 | if ( $josaonly === true ) { |
| 71 | $str = ''; |
| 72 | continue; |
| 73 | } |
| 74 | } |
| 75 | |
| 76 | return $str . $josa; |
| 77 | } ); |
| 78 | } |
| 79 | } |
| 80 | |
| 81 | /** |
| 82 | * @param string $param String to check |
| 83 | * @return bool|string Value of override param or false if it isn't presented |
| 84 | */ |
| 85 | public static function checkOverrideParam( $param ) { |
| 86 | if ( preg_match( '/^(override|덮어쓰기|오버라이드)=(.*)$/', $param, $matches ) === 0 ) { |
| 87 | return false; |
| 88 | } |
| 89 | return $matches[2]; |
| 90 | } |
| 91 | |
| 92 | /** |
| 93 | * @param string $param String to check |
| 94 | * @return bool Existence of josaonly param |
| 95 | */ |
| 96 | public static function checkJosaonlyParam( $param ) { |
| 97 | return ( $param === 'josaonly' || $param === '조사만' ); |
| 98 | } |
| 99 | |
| 100 | /** |
| 101 | * @param string $type Type of the last letter in the word (see JosaHooks::$josaMap's keys) |
| 102 | * @param string $str Word to determine the josa |
| 103 | * @return string Josa |
| 104 | */ |
| 105 | public static function getJosa( $type, $str ) { |
| 106 | $str = preg_replace( |
| 107 | '/[\x{0000}-\x{0027}\x{002A}-\x{002F}\x{003A}-\x{0040}\x{005B}-\x{0060}\x{007B}-\x{00FF}]/u', |
| 108 | '', |
| 109 | $str |
| 110 | ); |
| 111 | $chr = mb_substr( $str, -1, 1, 'utf-8' ); |
| 112 | if ( array_key_exists( $chr, self::$pronounceMap ) ) { |
| 113 | $chr = self::$pronounceMap[$chr]; |
| 114 | } |
| 115 | $code = self::convertToJohabCode( $chr ); |
| 116 | if ( !$code ) { |
| 117 | // Not hangul |
| 118 | $idx = 2; |
| 119 | } elseif ( ( $code - 0xAC00 ) % 28 === 0 ) { |
| 120 | // No trailing consonant |
| 121 | $idx = 1; |
| 122 | } elseif ( ( $type === 'Euro/Ro' ) && ( ( $code - 0xAC00 ) % 28 === 8 ) ) { |
| 123 | // $type is Euro/Ro and trailing consonant is rieul(ㄹ). |
| 124 | // This is an exception on Korean postposition rules. |
| 125 | $idx = 1; |
| 126 | } else { |
| 127 | // Trailing consonant exists |
| 128 | $idx = 0; |
| 129 | } |
| 130 | return self::$josaMap[$type][$idx]; |
| 131 | } |
| 132 | |
| 133 | /** |
| 134 | * @see https://ko.wikipedia.org/wiki/%ED%95%9C%EA%B8%80_%EC%83%81%EC%9A%A9_%EC%A1%B0%ED%95%A9%ED%98%95_%EC%9D%B8%EC%BD%94%EB%94%A9 |
| 135 | * |
| 136 | * @param string $str String to convert |
| 137 | * @return int|bool Converted Johab code |
| 138 | */ |
| 139 | private static function convertToJohabCode( $str ) { |
| 140 | $values = []; |
| 141 | $lookingFor = 1; |
| 142 | for ( $i = 0, $len = strlen( $str ); $i < $len; $i++ ) { |
| 143 | $thisValue = ord( $str[ $i ] ); |
| 144 | if ( $thisValue < 128 ) { |
| 145 | return false; |
| 146 | } else { |
| 147 | if ( count( $values ) === 0 ) { |
| 148 | $lookingFor = ( $thisValue < 224 ) ? 2 : 3; |
| 149 | } |
| 150 | $values[] = $thisValue; |
| 151 | if ( count( $values ) === $lookingFor ) { |
| 152 | $number = ( $lookingFor === 3 ) ? |
| 153 | ( ( $values[0] % 16 ) * 4096 ) + ( ( $values[1] % 64 ) * 64 ) + ( $values[2] % 64 ) : |
| 154 | ( ( $values[0] % 32 ) * 64 ) + ( $values[1] % 64 ); |
| 155 | return $number; |
| 156 | } |
| 157 | } |
| 158 | } |
| 159 | return false; |
| 160 | } |
| 161 | } |