require_once( dirname(__FILE__).'/../LanguageConverter.php' );
require_once( dirname(__FILE__).'/LanguageKk_cyrl.php' );
+define( 'KK_C_UC', 'АӘБВГҒДЕЁЖЗИЙКҚЛМНҢОӨПРСТУҰҮФХҺЦЧШЩЪЫІЬЭЮЯ' ); # Kazakh Cyrillic uppercase
+define( 'KK_C_LC', 'аәбвгғдеёжзийкқлмнңоөпрстуұүфхһцчшщъыіьэюя' ); # Kazakh Cyrillic lowercase
+define( 'KK_L_UC', 'AÄBCÇDEÉFGĞHIİÏJKLMNÑOÖPQRSŞTUÜVWXYÝZ' ); # Kazakh Latin uppercase
+define( 'KK_L_LC', 'aäbcçdeéfgğhıiïjklmnñoöpqrsştuüvwxyýz' ); # Kazakh Latin lowercase
+define( 'KK_A', 'اٵبۆگعدەجزيكقلمنڭوٶپرستۋۇٷفحھچشىٸ' ); # Kazakh Arabic
+
class KkConverter extends LanguageConverter {
- var $mLatinToCyrillic = array(
- 'YA' => 'Я', 'Ya' => 'Я', 'ya' => 'я', 'YE' => 'Е', 'Ye' => 'У', 'ye' => 'е',
- 'YO' => 'Ё', 'Yo' => 'Ё', 'yo' => 'ё', 'YU' => 'Ю', 'Yu' => 'Ю', 'yu' => 'ю',
- 'YW' => 'Ю', 'Yw' => 'Ю', 'yw' => 'ю',
-
- 'bʺ' => 'бъ', 'dʺ' => 'дъ', 'fʺ' => 'фъ', 'gʺ' => 'гъ', 'kʺ' => 'къ', 'lʺ' => 'лъ',
- 'mʺ' => 'мъ', 'nʺ' => 'нъ', 'pʺ' => 'пъ', 'rʺ' => 'ръ', 'sʺ' => 'съ', 'tʺ' => 'тъ',
- 'vʺ' => 'въ', 'zʺ' => 'зъ',
- /* 'jʺ' => 'жъ', 'cʺ' => 'цъ', 'çʺ' => 'чъ', 'şʺ' => 'шъ', */
-
- 'ŞÇʹ'=> 'ЩЬ', 'Şçʹ'=> 'Щь', 'Bʺ' => 'БЪ', 'Dʺ' => 'ДЪ', 'Fʺ' => 'ФЪ', 'Gʺ' => 'ГЪ', 'Kʺ' => 'КЪ', 'Lʺ' => 'ЛЪ',
- 'Mʺ' => 'МЪ', 'Nʺ' => 'НЪ', 'Pʺ' => 'ПЪ', 'Rʺ' => 'РЪ', 'Sʺ' => 'СЪ', 'Tʺ' => 'ТЪ',
- 'Vʺ' => 'ВЪ', 'Zʺ' => 'ЗЪ',
- /* 'Jʺ' => 'ЖЪ', 'Cʺ' => 'ЦЪ', 'Çʺ' => 'ЧЪ', 'Şʺ' => 'ШЪ', */
-
- 'şçʹ'=> 'щь', 'bʹ' => 'бь', 'dʹ' => 'дь', 'fʹ' => 'фь', 'gʹ' => 'гь', 'kʹ' => 'кь', 'lʹ' => 'ль',
- 'mʹ' => 'мь', 'nʹ' => 'нь', 'pʹ' => 'пь', 'rʹ' => 'рь', 'sʹ' => 'сь', 'tʹ' => 'ть',
- 'vʹ' => 'вь', 'zʹ' => 'зь', 'jʹ' => 'жь', 'cʹ' => 'ць', 'çʹ' => 'чь', 'şʹ' => 'шь',
-
- 'Bʹ' => 'БЬ', 'Dʹ' => 'ДЬ', 'Fʹ' => 'ФЬ', 'Gʹ' => 'ГЬ', 'Kʹ' => 'КЬ', 'Lʹ' => 'ЛЬ',
- 'Mʹ' => 'МЬ', 'Nʹ' => 'НЬ', 'Pʹ' => 'ПЬ', 'Rʹ' => 'РЬ', 'Sʹ' => 'СЬ', 'Tʹ' => 'ТЬ',
- 'Vʹ' => 'ВЬ', 'Zʹ' => 'ЗЬ', 'Jʹ' => 'ЖЬ', 'Cʹ' => 'ЦЬ', 'Çʹ' => 'ЧЬ', 'Şʹ' => 'ШЬ',
-
- 'ŞÇ' => 'Щ', 'Şç' => 'Щ', 'şç' => 'щ',
-
- 'a' => 'а', 'ä' => 'ә', 'b' => 'б', 'c' => 'ц', 'ç' => 'ч', 'd' => 'д', 'e' => 'е',
- 'é' => 'э', 'f' => 'ф', 'g' => 'г', 'ğ' => 'ғ', 'h' => 'һ', 'i' => 'і', 'ı' => 'ы',
- 'ï' => 'и', 'j' => 'ж', 'k' => 'к', 'l' => 'л', 'm' => 'м', 'n' => 'н', 'ñ' => 'ң',
- 'o' => 'о', 'ö' => 'ө', 'p' => 'п', 'q' => 'қ', 'r' => 'р', 's' => 'с', 'ş' => 'ш',
- 't' => 'т', 'u' => 'ұ', 'ü' => 'ү', 'v' => 'в', 'w' => 'у', 'x' => 'х', 'ý' => 'й',
- 'z' => 'з',
-
- 'A' => 'А', 'Ä' => 'Ә', 'B' => 'Б', 'C' => 'Ц', 'Ç' => 'Ч', 'D' => 'Д', 'E' => 'Е',
- 'É' => 'Э', 'F' => 'Ф', 'G' => 'Г', 'Ğ' => 'Ғ', 'H' => 'Һ', 'İ' => 'І', 'I' => 'Ы',
- 'Ï' => 'И', 'J' => 'Ж', 'K' => 'К', 'L' => 'Л', 'M' => 'М', 'N' => 'Н', 'Ñ' => 'Ң',
- 'O' => 'О', 'Ö' => 'Ө', 'P' => 'П', 'Q' => 'Қ', 'R' => 'Р', 'S' => 'С', 'Ş' => 'Ш',
- 'T' => 'Т', 'U' => 'Ұ', 'Ü' => 'Ү', 'V' => 'В', 'W' => 'У', 'Ý' => 'Й', 'X' => 'Х',
- 'Z' => 'З'
- );
-
- var $mCyrillicToLatin = array(
- 'а' => 'a', 'ә' => 'ä', 'б' => 'b', 'в' => 'v', 'г' => 'g', 'ғ' => 'ğ',
- 'д' => 'd', 'е' => 'e', 'ё' => 'yo', 'ж' => 'j', 'з' => 'z', 'и' => 'ï',
- 'й' => 'ý', 'к' => 'k', 'қ' => 'q', 'л' => 'l', 'м' => 'm', 'н' => 'n',
- 'ң' => 'ñ', 'о' => 'o', 'ө' => 'ö', 'п' => 'p', 'р' => 'r', 'с' => 's',
- 'т' => 't', 'у' => 'w', 'ұ' => 'u', 'ү' => 'ü', 'ф' => 'f', 'х' => 'x',
- 'һ' => 'h', 'ц' => 'c', 'ч' => 'ç', 'ш' => 'ş', 'щ' => 'şç', 'ъ' => 'ʺ',
- 'ы' => 'ı', 'ь' => 'ʹ', 'і' => 'i', 'э' => 'é', 'ю' => 'yw', 'я' => 'ya',
-
- 'А' => 'A', 'Ә' => 'Ä', 'Б' => 'B', 'В' => 'V', 'Г' => 'G', 'Ғ' => 'Ğ',
- 'Д' => 'D', 'Е' => 'E', 'Ё' => 'Yo', 'Ж' => 'J', 'З' => 'Z', 'И' => 'Ï',
- 'Й' => 'Ý', 'К' => 'K', 'Қ' => 'Q', 'Л' => 'L', 'М' => 'M', 'Н' => 'N',
- 'Ң' => 'Ñ', 'О' => 'O', 'Ө' => 'Ö', 'П' => 'P', 'Р' => 'R', 'С' => 'S',
- 'Т' => 'T', 'У' => 'W', 'Ұ' => 'U', 'Ү' => 'Ü', 'Ф' => 'F', 'Х' => 'X',
- 'Һ' => 'H', 'Ц' => 'C', 'Ч' => 'Ç', 'Ш' => 'Ş', 'Щ' => 'Şç', 'Ъ' => 'ʺ',
- 'Ы' => 'I', 'Ь' => 'ʹ', 'І' => 'İ', 'Э' => 'É', 'Ю' => 'Yw', 'Я' => 'Ya'
- );
-
- var $mCyrillicToArabic = array(
- 'ла' => 'لا', 'лА' => 'لا', 'ЛА' => 'لا', 'Ла' => 'لا',
-
- 'а' => 'ا', 'ә' => 'ٵ', 'б' => 'ب', 'в' => 'ۆ', 'г' => 'گ', 'ғ' => 'ع',
- 'д' => 'د', 'е' => 'ە', 'ё' => 'يو', 'ж' => 'ج', 'з' => 'ز', 'и' => 'ي',
- 'й' => 'ي', 'к' => 'ك', 'қ' => 'ق', 'л' => 'ل', 'м' => 'م', 'н' => 'ن',
- 'ң' => 'ڭ', 'о' => 'و', 'ө' => 'ٶ', 'п' => 'پ', 'р' => 'ر', 'с' => 'س',
- 'т' => 'ت', 'у' => 'ۋ', 'ұ' => 'ۇ', 'ү' => 'ٷ', 'ф' => 'ف', 'х' => 'ح',
- 'һ' => 'ھ', 'ц' => 'تس', 'ч' => 'چ', 'ш' => 'ش', 'щ' => 'شش', 'ъ' => 'ي',
- 'ы' => 'ى', 'ь' => 'ي', 'і' => 'ٸ', 'э' => 'ە', 'ю' => 'يۋ', 'я' => 'يا',
-
- 'А' => 'ا', 'Ә' => 'ٵ', 'Б' => 'ب', 'В' => 'ۆ', 'Г' => 'گ', 'Ғ' => 'ع',
- 'Д' => 'د', 'Е' => 'ە', 'Ё' => 'يو', 'Ж' => 'ج', 'З' => 'ز', 'И' => 'ي',
- 'Й' => 'ي', 'К' => 'ك', 'Қ' => 'ق', 'Л' => 'ل', 'М' => 'م', 'Н' => 'ن',
- 'Ң' => 'ڭ', 'О' => 'و', 'Ө' => 'ٶ', 'П' => 'پ', 'Р' => 'ر', 'С' => 'س',
- 'Т' => 'ت', 'У' => 'ۋ', 'Ұ' => 'ۇ', 'Ү' => 'ٷ', 'Ф' => 'ف', 'Х' => 'ح',
- 'Һ' => 'ھ', 'Ц' => 'تس', 'Ч' => 'چ', 'Ш' => 'ش', 'Щ' => 'شش', 'Ъ' => 'ي',
- 'Ы' => 'ى', 'Ь' => 'ي', 'І' => 'ٸ', 'Э' => 'ە', 'Ю' => 'يۋ', 'Я' => 'يا',
-
- '?' => '؟',
- '%' => '٪',
- ',' => '،',
- ';' => '؛'
- );
-
- var $mLatinToArabic = array(
- 'la' => 'لا', 'lA' => 'لا', 'LA' => 'لا', 'La' => 'لا',
-
- 'a' => 'ا', 'ä' => 'ٵ', 'b' => 'ب', 'v' => 'ۆ', 'g' => 'گ', 'ğ' => 'ع',
- 'd' => 'د', 'e' => 'ە', 'yo' => 'يو', 'j' => 'ج', 'z' => 'ز', 'ï' => 'ي',
- 'ý' => 'ي', 'k' => 'ك', 'q' => 'ق', 'l' => 'ل', 'm' => 'م', 'n' => 'ن',
- 'ñ' => 'ڭ', 'o' => 'و', 'ö' => 'ٶ', 'p' => 'پ', 'r' => 'ر', 's' => 'س',
- 't' => 'ت', 'w' => 'ۋ', 'u' => 'ۇ', 'ü' => 'ٷ', 'f' => 'ف', 'x' => 'ح',
- 'h' => 'ھ', 'c' => 'تس', 'ç' => 'چ', 'ş' => 'ش', 'şş' => 'شش', '″' => 'ي',
- 'ı' => 'ى', '′' => 'ي', 'i' => 'ٸ', 'é' => 'ە',
-
- 'A' => 'ا', 'Ä' => 'ٵ', 'B' => 'ب', 'V' => 'ۆ', 'G' => 'گ', 'Ğ' => 'ع',
- 'D' => 'د', 'E' => 'ە', 'YO' => 'يو', 'J' => 'ج', 'Z' => 'ز', 'Ï' => 'ي',
- 'Ý' => 'ي', 'K' => 'ك', 'Q' => 'ق', 'L' => 'ل', 'M' => 'م', 'N' => 'ن',
- 'Ñ' => 'ڭ', 'O' => 'و', 'Ö' => 'ٶ', 'P' => 'پ', 'R' => 'ر', 'S' => 'س',
- 'T' => 'ت', 'W' => 'ۋ', 'U' => 'ۇ', 'Ü' => 'ٷ', 'F' => 'ف', 'X' => 'ح',
- 'H' => 'ھ', 'C' => 'تس', 'Ç' => 'چ', 'Ş' => 'ش', 'ŞŞ' => 'شش', '″' => 'ي',
- 'I' => 'ى', '′' => 'ي', 'İ' => 'ٸ', 'É' => 'ە',
-
- '?' => '؟',
- '%' => '٪',
- ',' => '،',
- ';' => '؛'
- );
function loadDefaultTables() {
+ // require( dirname(__FILE__)."/../../includes/KkConversion.php" );
+ // Placeholder for future implementing. Remove variables declarations
+ // after generating KkConversion.php
+ $kk2Cyrl = array();
+ $kk2Latn = array();
+ $kk2Arab = array();
+ $kk2KZ = array();
+ $kk2TR = array();
+ $kk2CN = array();
+
$this->mTables = array(
- 'kk-cyrl' => new ReplacementArray( $this->mLatinToCyrillic ),
- 'kk-latn' => new ReplacementArray( $this->mCyrillicToLatin ),
- 'kk-arab' => new ReplacementArray( array_merge($this->mCyrillicToArabic, $this->mLatinToArabic) ),
- 'kk-kz' => new ReplacementArray( $this->mLatinToCyrillic ),
- 'kk-tr' => new ReplacementArray( $this->mCyrillicToLatin ),
- 'kk-cn' => new ReplacementArray( array_merge($this->mCyrillicToArabic, $this->mLatinToArabic) ),
- 'kk' => new ReplacementArray()
+ 'kk-cyrl' => new ReplacementArray( $kk2Cyrl ),
+ 'kk-latn' => new ReplacementArray( $kk2Latn ),
+ 'kk-arab' => new ReplacementArray( $kk2Arab ),
+ 'kk-kz' => new ReplacementArray( array_merge($kk2Cyrl, $kk2KZ) ),
+ 'kk-tr' => new ReplacementArray( array_merge($kk2Latn, $kk2TR) ),
+ 'kk-cn' => new ReplacementArray( array_merge($kk2Arab, $kk2CN) ),
+ 'kk' => new ReplacementArray()
);
+
+ self::loadRegs();
}
function postLoadTables() {
$this->mTables['kk-cn']->merge( $this->mTables['kk-arab'] );
}
+ function loadRegs() {
+
+ $this->mCyrl2Latn = array(
+ '/№/u' => 'No.',
+ ## Е after vowels
+ '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])Е/u' => '$1YE',
+ '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])е/ui' => '$1ye',
+ ## leading ЁЮЯЩ
+ '/^Ё(['.KK_C_UC.']|$)/u' => 'YO$1', '/^Ё(['.KK_C_LC.']|$)/u' => 'Yo$1',
+ '/^Ю(['.KK_C_UC.']|$)/u' => 'YU$1', '/^Ю(['.KK_C_LC.']|$)/u' => 'Yu$1',
+ '/^Я(['.KK_C_UC.']|$)/u' => 'YA$1', '/^Я(['.KK_C_LC.']|$)/u' => 'Ya$1',
+ '/^Щ(['.KK_C_UC.']|$)/u' => 'ŞÇ$1', '/^Щ(['.KK_C_LC.']|$)/u' => 'Şç$1',
+ ## other ЁЮЯ
+ '/Ё/u' => 'YO', '/ё/u' => 'yo',
+ '/Ю/u' => 'YU', '/ю/u' => 'yu',
+ '/Я/u' => 'YA', '/я/u' => 'ya',
+ '/Щ/u' => 'ŞÇ', '/щ/u' => 'şç',
+ ## soft and hard signs
+ '/[ъЪ]/u' => 'ʺ', '/[ьЬ]/u' => 'ʹ',
+ ## other characters
+ '/А/u' => 'A', '/а/u' => 'a', '/Ә/u' => 'Ä', '/ә/u' => 'ä',
+ '/Б/u' => 'B', '/б/u' => 'b', '/В/u' => 'V', '/в/u' => 'v',
+ '/Г/u' => 'G', '/г/u' => 'g', '/Ғ/u' => 'Ğ', '/ғ/u' => 'ğ',
+ '/Д/u' => 'D', '/д/u' => 'd', '/Е/u' => 'E', '/е/u' => 'e',
+ '/Ж/u' => 'J', '/ж/u' => 'j', '/З/u' => 'Z', '/з/u' => 'z',
+ '/И/u' => 'Ï', '/и/u' => 'ï', '/Й/u' => 'Ý', '/й/u' => 'ý',
+ '/К/u' => 'K', '/к/u' => 'k', '/Қ/u' => 'Q', '/қ/u' => 'q',
+ '/Л/u' => 'L', '/л/u' => 'l', '/М/u' => 'M', '/м/u' => 'm',
+ '/Н/u' => 'N', '/н/u' => 'n', '/Ң/u' => 'Ñ', '/ң/u' => 'ñ',
+ '/О/u' => 'O', '/о/u' => 'o', '/Ө/u' => 'Ö', '/ө/u' => 'ö',
+ '/П/u' => 'P', '/п/u' => 'p', '/Р/u' => 'R', '/р/u' => 'r',
+ '/С/u' => 'S', '/с/u' => 's', '/Т/u' => 'T', '/т/u' => 't',
+ '/У/u' => 'W', '/у/u' => 'w', '/Ұ/u' => 'U', '/ұ/u' => 'u',
+ '/Ү/u' => 'Ü', '/ү/u' => 'ü', '/Ф/u' => 'F', '/ф/u' => 'f',
+ '/Х/u' => 'X', '/х/u' => 'x', '/Һ/u' => 'H', '/һ/u' => 'h',
+ '/Ц/u' => 'C', '/ц/u' => 'c', '/Ч/u' => 'Ç', '/ч/u' => 'ç',
+ '/Ш/u' => 'Ş', '/ш/u' => 'ş', '/Ы/u' => 'I', '/ы/u' => 'ı',
+ '/І/u' => 'İ', '/і/u' => 'i', '/Э/u' => 'É', '/э/u' => 'é',
+ );
+
+ $this->mLatn2Cyrl = array(
+ '/No\./u' => '№',
+ ## Şç
+ '/ŞÇʹ/u'=> 'ЩЬ', '/Şçʹ/u'=> 'Щь', '/Şçʹ/u'=> 'Щь',
+ '/Ş[Çç]/u' => 'Щ', '/şç/u' => 'щ',
+ ## soft and hard signs
+ '/(['.KK_L_UC.'])ʺ(['.KK_L_UC.'])/u' => '$1Ъ$2',
+ '/ʺ(['.KK_L_LC.'])/u' => 'ъ$1',
+ '/(['.KK_L_UC.'])ʹ(['.KK_L_UC.'])/u' => '$1Ь$2',
+ '/ʹ(['.KK_L_LC.'])/u' => 'ь$1',
+ '/ʺ/u' => 'ъ',
+ '/ʹ/u' => 'ь',
+ ## Ye Yo Yu Ya.
+ '/Y[Ee]/u' => 'Е', '/ye/u' => 'е',
+ '/Y[Oo]/u' => 'Ё', '/yo/u' => 'ё',
+ '/Y[UWuw]/u' => 'Ю', '/y[uw]/u' => 'ю',
+ '/Y[Aa]/u' => 'Я', '/ya/u' => 'я',
+ ## other characters
+ '/A/u' => 'А', '/a/u' => 'а', '/Ä/u' => 'Ә', '/ä/u' => 'ә',
+ '/B/u' => 'Б', '/b/u' => 'б', '/C/u' => 'Ц', '/c/u' => 'ц',
+ '/Ç/u' => 'Ч', '/ç/u' => 'ч', '/D/u' => 'Д', '/d/u' => 'д',
+ '/E/u' => 'Е', '/e/u' => 'е', '/É/u' => 'Э', '/é/u' => 'э',
+ '/F/u' => 'Ф', '/f/u' => 'ф', '/G/u' => 'Г', '/g/u' => 'г',
+ '/Ğ/u' => 'Ғ', '/ğ/u' => 'ғ', '/H/u' => 'Һ', '/h/u' => 'һ',
+ '/I/u' => 'Ы', '/ı/u' => 'ы', '/İ/u' => 'І', '/i/u' => 'і',
+ '/Ï/u' => 'И', '/ï/u' => 'и', '/J/u' => 'Ж', '/j/u' => 'ж',
+ '/K/u' => 'К', '/k/u' => 'к', '/L/u' => 'Л', '/l/u' => 'л',
+ '/M/u' => 'М', '/m/u' => 'м', '/N/u' => 'Н', '/n/u' => 'н',
+ '/Ñ/u' => 'Ң', '/ñ/u' => 'ң', '/O/u' => 'О', '/o/u' => 'о',
+ '/Ö/u' => 'Ө', '/ö/u' => 'ө', '/P/u' => 'П', '/p/u' => 'п',
+ '/Q/u' => 'Қ', '/q/u' => 'қ', '/R/u' => 'Р', '/r/u' => 'р',
+ '/S/u' => 'С', '/s/u' => 'с', '/Ş/u' => 'Ш', '/ş/u' => 'ш',
+ '/T/u' => 'Т', '/t/u' => 'т', '/U/u' => 'Ұ', '/u/u' => 'ұ',
+ '/Ü/u' => 'Ү', '/ü/u' => 'ү', '/V/u' => 'В', '/v/u' => 'в',
+ '/W/u' => 'У', '/w/u' => 'у', '/Ý/u' => 'Й', '/ý/u' => 'й',
+ '/X/u' => 'Х', '/x/u' => 'х', '/Z/u' => 'З', '/z/u' => 'з',
+ );
+
+ $this->mCyLa2Arab = array(
+ ## Cyrillic -> Arabic
+ '/\№/u' => 'نٶ.',
+ '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])е/ui' => '$1يە',
+ '/а/ui' => 'ا', '/ә/ui' => 'ٵ', '/б/ui' => 'ب', '/в/ui' => 'ۆ',
+ '/г/ui' => 'گ', '/ғ/ui' => 'ع', '/д/ui' => 'د', '/[еэ]/ui' => 'ە',
+ '/ё/ui' => 'يو', '/ж/ui' => 'ج', '/з/ui' => 'ز', '/и/ui' => 'ي',
+ '/й/ui' => 'ي', '/к/ui' => 'ك', '/қ/ui' => 'ق', '/л/ui' => 'ل',
+ '/м/ui' => 'م', '/н/ui' => 'ن', '/ң/ui' => 'ڭ', '/о/ui' => 'و',
+ '/ө/ui' => 'ٶ', '/п/ui' => 'پ', '/р/ui' => 'ر', '/с/ui' => 'س',
+ '/т/ui' => 'ت', '/у/ui' => 'ۋ', '/ұ/ui' => 'ۇ', '/ү/ui' => 'ٷ',
+ '/ф/ui' => 'ف', '/х/ui' => 'ح', '/һ/ui' => 'ھ', '/ц/ui' => 'تس',
+ '/ч/ui' => 'چ', '/ш/ui' => 'ش', '/щ/ui' => 'شش', '/[ъь]/ui' => '',
+ '/ы/ui' => 'ى', '/і/ui' => 'ٸ', '/ю/ui' => 'يۋ', '/я/ui' => 'يا',
+
+ ## Latin -> Arabic // commented for now...
+ /*'/No\./u' => 'نٶ.',
+ '/a/ui' => 'ا', '/ä/ui' => 'ٵ', '/b/ui' => 'ب', '/c/ui' => 'تس',
+ '/ç/ui' => 'چ', '/d/ui' => 'د', '/[eé]/ui' => 'ە', '/f/ui' => 'ف',
+ '/g/ui' => 'گ', '/ğ/ui' => 'ع', '/h/ui' => 'ھ', '/[ıI]/u' => 'ى',
+ '/[iİ]/u' => 'ٸ', '/ï/ui' => 'ي', '/j/ui' => 'ج', '/k/ui' => 'ك',
+ '/l/ui' => 'ل', '/m/ui' => 'م', '/n/ui' => 'ن', '/ñ/ui' => 'ڭ',
+ '/o/ui' => 'و', '/ö/ui' => 'ٶ', '/p/ui' => 'پ', '/q/ui' => 'ق',
+ '/r/ui' => 'ر', '/s/ui' => 'س', '/ş/ui' => 'ش', '/t/ui' => 'ت',
+ '/u/ui' => 'ۇ', '/ü/ui' => 'ٷ', '/v/ui' => 'ۆ', '/w/ui' => 'ۋ',
+ '/x/ui' => 'ح', '/[yý]/ui' => 'ي', '/z/ui' => 'ز', '/[ʺʹ]/ui' => '',*/
+
+ ## Punctuation -> Arabic
+ '/\?/' => '؟', # ؟
+ '/\,/' => '،', # ،
+ '/\;/' => '؛' , # ؛
+ '/\%/' => '٪', # ٪
+ ## Digits -> Arabic
+ '/0/' => '۰', # ۰
+ '/1/' => '۱', # ۱
+ '/2/' => '۲', # ۲
+ '/3/' => '۳', # ۳
+ '/4/' => '۴', # ۴
+ '/5/' => '۵', # ۵
+ '/6/' => '۶', # ۶
+ '/7/' => '۷', # ۷
+ '/8/' => '۸', # ۸
+ '/9/' => '۹', # ۹
+ );
+
+ }
+
/* rules should be defined as -{ekavian | iyekavian-} -or-
-{code:text | code:text | ...}-
update: delete all rule parsing because it's not used
- currently, and just produces a couple of bugs
+ currently, and just produces a couple of bugs
*/
function parseManualRule($rule, $flags=array()) {
if(in_array('T',$flags)){
/*
* A function wrapper:
- * - if there is no selected variant, leave the link
- * names as they were
- * - do not try to find variants for usernames
+ * - if there is no selected variant, leave the link
+ * names as they were
+ * - do not try to find variants for usernames
*/
function findVariantLink( &$link, &$nt ) {
// check for user namespace
$link=$oldlink;
}
- /*
- * We want our external link captions to be converted in variants,
- * so we return the original text instead -{$text}-, except for URLs
- */
- function markNoConversion($text, $noParse=false) {
- if($noParse || preg_match("/^https?:\/\/|ftp:\/\/|irc:\/\//",$text))
- return parent::markNoConversion($text);
- return $text;
- }
-
/*
* An ugly function wrapper for parsing Image titles
* (to prevent image name conversion)
}
/**
- * It translates text into variant, specials:
- * - ommiting roman numbers
+ * It translates text into variant
*/
- function translate($text, $toVariant){
- $breaks = '[^\w\x80-\xff]';
+ function translate( $text, $toVariant ){
+ global $wgContLanguageCode;
+ $text = parent::translate( $text, $toVariant );
- // regexp for roman numbers
- $roman = 'M{0,4}(CM|CD|D?C{0,3})(XC|XL|L?X{0,3})(IX|IV|V?I{0,3})';
-
- $reg = '/^'.$roman.'$|^'.$roman.$breaks.'|'.$breaks.$roman.'$|'.$breaks.$roman.$breaks.'/';
-
- $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
-
- $m = array_shift($matches);
- if( !isset( $this->mTables[$toVariant] ) ) {
- throw new MWException( "Broken variant table: " . implode( ',', array_keys( $this->mTables ) ) );
+ $letters = '';
+ switch( $toVariant ) {
+ case 'kk-cyrl':
+ case 'kk-kz':
+ $letters = KK_L_UC . KK_L_LC . 'ʺʹ0123456789';
+ $wgContLanguageCode = 'kk';
+ break;
+ case 'kk-latn':
+ case 'kk-tr':
+ $letters = KK_C_UC . KK_C_LC . '№0123456789';
+ $wgContLanguageCode = 'kk-Latn';
+ break;
+ case 'kk-arab':
+ case 'kk-cn':
+ // $letters = KK_C_UC.KK_C_LC.KK_L_UC.KK_L_LC.'ʺʹ%№0123456789?,;';
+ $letters = KK_C_UC . KK_C_LC . '%№0123456789?,;';
+ $wgContLanguageCode = 'kk-Arab';
+ break;
+ default:
+ $wgContLanguageCode = 'kk';
+ return $text;
}
- $ret = $this->mTables[$toVariant]->replace( $m[0] );
- $mstart = $m[1]+strlen($m[0]);
- foreach($matches as $m) {
- $ret .= substr($text, $mstart, $m[1]-$mstart);
- $ret .= parent::translate($m[0], $toVariant);
+ // disable conversion variables like $1, $2...
+ $varsfix = '\$[0-9]';
+
+ $matches = preg_split( '/' . $varsfix . '[^' . $letters . ']+/u', $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
+ $mstart = 0;
+ foreach( $matches as $m ) {
+ $ret .= substr( $text, $mstart, $m[1]-$mstart );
+ $ret .= $this->regsConverter( $m[0], $toVariant );
$mstart = $m[1] + strlen($m[0]);
}
-
return $ret;
}
+ function regsConverter( $text, $toVariant ) {
+ if ($text == '') return $text;
+
+ $pat = array();
+ $rep = array();
+ switch( $toVariant ) {
+ case 'kk-arab':
+ case 'kk-cn':
+ foreach( $this->mCyLa2Arab as $pat => $rep ) {
+ $text = preg_replace( $pat, $rep, $text );
+ }
+ return $text;
+ break;
+ case 'kk-latn':
+ case 'kk-tr':
+ foreach( $this->mCyrl2Latn as $pat => $rep ) {
+ $text = preg_replace( $pat, $rep, $text );
+ }
+ return $text;
+ break;
+ case 'kk-cyrl':
+ case 'kk-cn':
+ foreach( $this->mLatn2Cyrl as $pat => $rep ) {
+ $text = preg_replace( $pat, $rep, $text );
+ }
+ return $text;
+ break;
+ default:
+ return $text;
+ }
+ }
+
+ /*
+ * We want our external link captions to be converted in variants,
+ * so we return the original text instead -{$text}-, except for URLs
+ */
+ function markNoConversion( $text, $noParse=false ) {
+ if( $noParse || preg_match( "/^https?:\/\/|ftp:\/\/|irc:\/\//", $text ) )
+ return parent::markNoConversion( $text );
+ return $text;
+ }
+
+ function convertCategoryKey( $key ) {
+ return $this->autoConvert( $key, 'kk' );
+ }
+
}
/* class that handles Cyrillic, Latin and Arabic scripts for Kazakh
- right now it only distinguish kk_cyrl, kk_latn, kk_arab, kk_kz, kk_tr and kk_cn.
+ right now it only distinguish kk_cyrl, kk_latn, kk_arab and kk_kz, kk_tr, kk_cn.
*/
class LanguageKk extends LanguageKk_cyrl {
$wgHooks['ArticleSaveComplete'][] = $this->mConverter;
}
- function convertGrammar( $word, $case ) {
- $fname="LanguageKk::convertGrammar";
- wfProfileIn( $fname );
-
- switch ( $this->getPreferredVariant() ) {
- case 'kk-cn':
- case 'kk-arab':
- $word = parent::convertGrammar( $word, $case, $variant='kk-arab' );
- break;
- case 'kk-tr':
- case 'kk-latn':
- $word = parent::convertGrammar( $word, $case, $variant='kk-latn' );
- break;
- case 'kk-kz':
- case 'kk-cyrl':
- case 'kk':
- $word = parent::convertGrammar( $word, $case, $variant='kk-cyrl' );
- break;
- default:
- $word = parent::convertGrammar( $word, $case, $variant='kk-cyrl' );
+ /**
+ * Work around for right-to-left direction support in kk-arab and kk-cn
+ *
+ * @return bool
+ */
+ function isRTL() {
+ $variant = $this->getPreferredVariant();
+ if ( $variant == 'kk-arab' || $variant == 'kk-cn' ) {
+ return true;
+ } else {
+ return parent::isRTL();
}
-
- wfProfileOut( $fname );
- return $word;
}
/*
- * It fixes issue ucfirst with transforming 'i' to 'İ'
+ * It fixes issue with ucfirst for transforming 'i' to 'İ'
*
*/
function ucfirst ( $string ) {
- if ( ($this->getPreferredVariant() == 'kk-tr' || $this->getPreferredVariant() == 'kk-latn') && $string[0] == 'i' ) {
+ $variant = $this->getPreferredVariant();
+ if ( ($variant == 'kk-latn' || $variant == 'kk-tr') && $string[0] == 'i' ) {
$string = 'İ' . substr( $string, 1 );
} else {
$string = parent::ucfirst( $string );
}
/*
- * It fixes issue for lcfirst with transforming 'I' to 'ı'
+ * It fixes issue with lcfirst for transforming 'I' to 'ı'
*
*/
function lcfirst ( $string ) {
- if ( ($this->getPreferredVariant() == 'kk-tr' || $this->getPreferredVariant() == 'kk-latn') && $string[0] == 'I' ) {
+ $variant = $this->getPreferredVariant();
+ if ( ($variant == 'kk-latn' || $variant == 'kk-tr') && $string[0] == 'I' ) {
$string = 'ı' . substr( $string, 1 );
} else {
$string = parent::lcfirst( $string );
return $string;
}
-}
+ function convertGrammar( $word, $case ) {
+ $fname="LanguageKk::convertGrammar";
+ wfProfileIn( $fname );
+
+ $variant = $this->getPreferredVariant();
+ switch ( $variant ) {
+ case 'kk-arab':
+ case 'kk-cn':
+ $word = parent::convertGrammar( $word, $case, $variant = 'kk-arab' );
+ break;
+ case 'kk-latn':
+ case 'kk-tr':
+ $word = parent::convertGrammar( $word, $case, $variant = 'kk-latn' );
+ break;
+ case 'kk-cyrl':
+ case 'kk-kz':
+ case 'kk':
+ default:
+ $word = parent::convertGrammar( $word, $case, $variant = 'kk-cyrl' );
+ }
+ wfProfileOut( $fname );
+ return $word;
+ }
+}
dépôts / lhc / web / wiklou.git / commitdiff