MediaWiki
REL1_22
|
00001 <?php 00024 require_once __DIR__ . '/../LanguageConverter.php'; 00025 require_once __DIR__ . '/LanguageKk_cyrl.php'; 00026 00027 define( 'KK_C_UC', 'АӘБВГҒДЕЁЖЗИЙКҚЛМНҢОӨПРСТУҰҮФХҺЦЧШЩЪЫІЬЭЮЯ' ); # Kazakh Cyrillic uppercase 00028 define( 'KK_C_LC', 'аәбвгғдеёжзийкқлмнңоөпрстуұүфхһцчшщъыіьэюя' ); # Kazakh Cyrillic lowercase 00029 define( 'KK_L_UC', 'AÄBCÇDEÉFGĞHIİÏJKLMNÑOÖPQRSŞTUÜVWXYÝZ' ); # Kazakh Latin uppercase 00030 define( 'KK_L_LC', 'aäbcçdeéfgğhıiïjklmnñoöpqrsştuüvwxyýz' ); # Kazakh Latin lowercase 00031 // define( 'KK_A', 'ٴابپتجحدرزسشعفقكلمنڭەوۇۋۆىيچھ' ); # Kazakh Arabic 00032 define( 'H_HAMZA', 'ٴ' ); # U+0674 ARABIC LETTER HIGH HAMZA 00033 // define( 'ZWNJ', '' ); # U+200C ZERO WIDTH NON-JOINER 00034 00040 class KkConverter extends LanguageConverter { 00041 00042 protected $mCyrl2Latn, $mLatn2Cyrl, $mCyLa2Arab; 00043 00051 function __construct( $langobj, $maincode, 00052 $variants = array(), 00053 $variantfallbacks = array(), 00054 $flags = array() ) { 00055 parent::__construct( $langobj, $maincode, 00056 $variants, $variantfallbacks, $flags ); 00057 00058 // No point delaying this since they're in code. 00059 // Waiting until loadDefaultTables() means they never get loaded 00060 // when the tables themselves are loaded from cache. 00061 $this->loadRegs(); 00062 } 00063 00064 function loadDefaultTables() { 00065 // require __DIR__."/../../includes/KkConversion.php"; 00066 // Placeholder for future implementing. Remove variables declarations 00067 // after generating KkConversion.php 00068 $kk2Cyrl = array(); 00069 $kk2Latn = array(); 00070 $kk2Arab = array(); 00071 $kk2KZ = array(); 00072 $kk2TR = array(); 00073 $kk2CN = array(); 00074 00075 $this->mTables = array( 00076 'kk-cyrl' => new ReplacementArray( $kk2Cyrl ), 00077 'kk-latn' => new ReplacementArray( $kk2Latn ), 00078 'kk-arab' => new ReplacementArray( $kk2Arab ), 00079 'kk-kz' => new ReplacementArray( array_merge( $kk2Cyrl, $kk2KZ ) ), 00080 'kk-tr' => new ReplacementArray( array_merge( $kk2Latn, $kk2TR ) ), 00081 'kk-cn' => new ReplacementArray( array_merge( $kk2Arab, $kk2CN ) ), 00082 'kk' => new ReplacementArray() 00083 ); 00084 } 00085 00086 function postLoadTables() { 00087 $this->mTables['kk-kz']->merge( $this->mTables['kk-cyrl'] ); 00088 $this->mTables['kk-tr']->merge( $this->mTables['kk-latn'] ); 00089 $this->mTables['kk-cn']->merge( $this->mTables['kk-arab'] ); 00090 } 00091 00092 function loadRegs() { 00093 00094 $this->mCyrl2Latn = array( 00095 # # Punctuation 00096 '/№/u' => 'No.', 00097 # # Е after vowels 00098 '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])Е/u' => '$1YE', 00099 '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])е/ui' => '$1ye', 00100 # # leading ЁЮЯЩ 00101 '/^Ё([' . KK_C_UC . ']|$)/u' => 'YO$1', '/^Ё([' . KK_C_LC . ']|$)/u' => 'Yo$1', 00102 '/^Ю([' . KK_C_UC . ']|$)/u' => 'YU$1', '/^Ю([' . KK_C_LC . ']|$)/u' => 'Yu$1', 00103 '/^Я([' . KK_C_UC . ']|$)/u' => 'YA$1', '/^Я([' . KK_C_LC . ']|$)/u' => 'Ya$1', 00104 '/^Щ([' . KK_C_UC . ']|$)/u' => 'ŞÇ$1', '/^Щ([' . KK_C_LC . ']|$)/u' => 'Şç$1', 00105 # # other ЁЮЯ 00106 '/Ё/u' => 'YO', '/ё/u' => 'yo', 00107 '/Ю/u' => 'YU', '/ю/u' => 'yu', 00108 '/Я/u' => 'YA', '/я/u' => 'ya', 00109 '/Щ/u' => 'ŞÇ', '/щ/u' => 'şç', 00110 # # soft and hard signs 00111 '/[ъЪ]/u' => 'ʺ', '/[ьЬ]/u' => 'ʹ', 00112 # # other characters 00113 '/А/u' => 'A', '/а/u' => 'a', '/Ә/u' => 'Ä', '/ә/u' => 'ä', 00114 '/Б/u' => 'B', '/б/u' => 'b', '/В/u' => 'V', '/в/u' => 'v', 00115 '/Г/u' => 'G', '/г/u' => 'g', '/Ғ/u' => 'Ğ', '/ғ/u' => 'ğ', 00116 '/Д/u' => 'D', '/д/u' => 'd', '/Е/u' => 'E', '/е/u' => 'e', 00117 '/Ж/u' => 'J', '/ж/u' => 'j', '/З/u' => 'Z', '/з/u' => 'z', 00118 '/И/u' => 'Ï', '/и/u' => 'ï', '/Й/u' => 'Ý', '/й/u' => 'ý', 00119 '/К/u' => 'K', '/к/u' => 'k', '/Қ/u' => 'Q', '/қ/u' => 'q', 00120 '/Л/u' => 'L', '/л/u' => 'l', '/М/u' => 'M', '/м/u' => 'm', 00121 '/Н/u' => 'N', '/н/u' => 'n', '/Ң/u' => 'Ñ', '/ң/u' => 'ñ', 00122 '/О/u' => 'O', '/о/u' => 'o', '/Ө/u' => 'Ö', '/ө/u' => 'ö', 00123 '/П/u' => 'P', '/п/u' => 'p', '/Р/u' => 'R', '/р/u' => 'r', 00124 '/С/u' => 'S', '/с/u' => 's', '/Т/u' => 'T', '/т/u' => 't', 00125 '/У/u' => 'W', '/у/u' => 'w', '/Ұ/u' => 'U', '/ұ/u' => 'u', 00126 '/Ү/u' => 'Ü', '/ү/u' => 'ü', '/Ф/u' => 'F', '/ф/u' => 'f', 00127 '/Х/u' => 'X', '/х/u' => 'x', '/Һ/u' => 'H', '/һ/u' => 'h', 00128 '/Ц/u' => 'C', '/ц/u' => 'c', '/Ч/u' => 'Ç', '/ч/u' => 'ç', 00129 '/Ш/u' => 'Ş', '/ш/u' => 'ş', '/Ы/u' => 'I', '/ы/u' => 'ı', 00130 '/І/u' => 'İ', '/і/u' => 'i', '/Э/u' => 'É', '/э/u' => 'é', 00131 ); 00132 00133 $this->mLatn2Cyrl = array( 00134 # # Punctuation 00135 '/#|No\./' => '№', 00136 # # Şç 00137 '/ŞÇʹ/u' => 'ЩЬ', '/Şçʹ/u' => 'Щь', 00138 '/Ş[Çç]/u' => 'Щ', '/şç/u' => 'щ', 00139 # # soft and hard signs 00140 '/([' . KK_L_UC . '])ʺ([' . KK_L_UC . '])/u' => '$1Ъ$2', 00141 '/ʺ([' . KK_L_LC . '])/u' => 'ъ$1', 00142 '/([' . KK_L_UC . '])ʹ([' . KK_L_UC . '])/u' => '$1Ь$2', 00143 '/ʹ([' . KK_L_LC . '])/u' => 'ь$1', 00144 '/ʺ/u' => 'ъ', 00145 '/ʹ/u' => 'ь', 00146 # # Ye Yo Yu Ya. 00147 '/Y[Ee]/u' => 'Е', '/ye/u' => 'е', 00148 '/Y[Oo]/u' => 'Ё', '/yo/u' => 'ё', 00149 '/Y[UWuw]/u' => 'Ю', '/y[uw]/u' => 'ю', 00150 '/Y[Aa]/u' => 'Я', '/ya/u' => 'я', 00151 # # other characters 00152 '/A/u' => 'А', '/a/u' => 'а', '/Ä/u' => 'Ә', '/ä/u' => 'ә', 00153 '/B/u' => 'Б', '/b/u' => 'б', '/C/u' => 'Ц', '/c/u' => 'ц', 00154 '/Ç/u' => 'Ч', '/ç/u' => 'ч', '/D/u' => 'Д', '/d/u' => 'д', 00155 '/E/u' => 'Е', '/e/u' => 'е', '/É/u' => 'Э', '/é/u' => 'э', 00156 '/F/u' => 'Ф', '/f/u' => 'ф', '/G/u' => 'Г', '/g/u' => 'г', 00157 '/Ğ/u' => 'Ғ', '/ğ/u' => 'ғ', '/H/u' => 'Һ', '/h/u' => 'һ', 00158 '/I/u' => 'Ы', '/ı/u' => 'ы', '/İ/u' => 'І', '/i/u' => 'і', 00159 '/Ï/u' => 'И', '/ï/u' => 'и', '/J/u' => 'Ж', '/j/u' => 'ж', 00160 '/K/u' => 'К', '/k/u' => 'к', '/L/u' => 'Л', '/l/u' => 'л', 00161 '/M/u' => 'М', '/m/u' => 'м', '/N/u' => 'Н', '/n/u' => 'н', 00162 '/Ñ/u' => 'Ң', '/ñ/u' => 'ң', '/O/u' => 'О', '/o/u' => 'о', 00163 '/Ö/u' => 'Ө', '/ö/u' => 'ө', '/P/u' => 'П', '/p/u' => 'п', 00164 '/Q/u' => 'Қ', '/q/u' => 'қ', '/R/u' => 'Р', '/r/u' => 'р', 00165 '/S/u' => 'С', '/s/u' => 'с', '/Ş/u' => 'Ш', '/ş/u' => 'ш', 00166 '/T/u' => 'Т', '/t/u' => 'т', '/U/u' => 'Ұ', '/u/u' => 'ұ', 00167 '/Ü/u' => 'Ү', '/ü/u' => 'ү', '/V/u' => 'В', '/v/u' => 'в', 00168 '/W/u' => 'У', '/w/u' => 'у', '/Ý/u' => 'Й', '/ý/u' => 'й', 00169 '/X/u' => 'Х', '/x/u' => 'х', '/Z/u' => 'З', '/z/u' => 'з', 00170 ); 00171 00172 $this->mCyLa2Arab = array( 00173 # # Punctuation -> Arabic 00174 '/#|№|No\./u' => '', # ؀ 00175 '/\,/' => '،', # ، 00176 '/;/' => '؛', # ؛ 00177 '/\?/' => '؟', # ؟ 00178 '/%/' => '٪', # ٪ 00179 '/\*/' => '٭', # ٭ 00180 # # Digits -> Arabic 00181 '/0/' => '۰', # ۰ 00182 '/1/' => '۱', # ۱ 00183 '/2/' => '۲', # ۲ 00184 '/3/' => '۳', # ۳ 00185 '/4/' => '۴', # ۴ 00186 '/5/' => '۵', # ۵ 00187 '/6/' => '۶', # ۶ 00188 '/7/' => '۷', # ۷ 00189 '/8/' => '۸', # ۸ 00190 '/9/' => '۹', # ۹ 00191 # # Cyrillic -> Arabic 00192 '/Аллаһ/ui' => 'ﷲ', 00193 '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])е/ui' => '$1يە', 00194 '/[еэ]/ui' => 'ە', '/[ъь]/ui' => '', 00195 '/[аә]/ui' => 'ا', '/[оө]/ui' => 'و', '/[ұү]/ui' => 'ۇ', '/[ыі]/ui' => 'ى', 00196 '/[и]/ui' => 'ىي', '/ё/ui' => 'يو', '/ю/ui' => 'يۋ', '/я/ui' => 'يا', '/[й]/ui' => 'ي', 00197 '/ц/ui' => 'تس', '/щ/ui' => 'شش', 00198 '/һ/ui' => 'ح', '/ч/ui' => 'تش', 00199 # '/һ/ui' => 'ھ', '/ч/ui' => 'چ', 00200 '/б/ui' => 'ب', '/в/ui' => 'ۆ', '/г/ui' => 'گ', '/ғ/ui' => 'ع', 00201 '/д/ui' => 'د', '/ж/ui' => 'ج', '/з/ui' => 'ز', '/к/ui' => 'ك', 00202 '/қ/ui' => 'ق', '/л/ui' => 'ل', '/м/ui' => 'م', '/н/ui' => 'ن', 00203 '/ң/ui' => 'ڭ', '/п/ui' => 'پ', '/р/ui' => 'ر', '/с/ui' => 'س', 00204 '/т/ui' => 'ت', '/у/ui' => 'ۋ', '/ф/ui' => 'ف', '/х/ui' => 'ح', 00205 '/ш/ui' => 'ش', 00206 # # Latin -> Arabic // commented for now... 00207 /*'/Allah/ui' => 'ﷲ', 00208 '/[eé]/ui' => 'ە', '/[yý]/ui' => 'ي', '/[ʺʹ]/ui' => '', 00209 '/[aä]/ui' => 'ا', '/[oö]/ui' => 'و', '/[uü]/ui' => 'ۇ', 00210 '/[ï]/ui' => 'ىي', '/[ıIiİ]/u' => 'ى', 00211 '/c/ui' => 'تس', 00212 '/ç/ui' => 'تش', '/h/ui' => 'ح', 00213 #'/ç/ui' => 'چ', '/h/ui' => 'ھ', 00214 '/b/ui' => 'ب','/d/ui' => 'د', 00215 '/f/ui' => 'ف', '/g/ui' => 'گ', '/ğ/ui' => 'ع', 00216 '/j/ui' => 'ج', '/k/ui' => 'ك', '/l/ui' => 'ل', '/m/ui' => 'م', 00217 '/n/ui' => 'ن', '/ñ/ui' => 'ڭ', '/p/ui' => 'پ', '/q/ui' => 'ق', 00218 '/r/ui' => 'ر', '/s/ui' => 'س', '/ş/ui' => 'ش', '/t/ui' => 'ت', 00219 '/v/ui' => 'ۆ', '/w/ui' => 'ۋ', '/x/ui' => 'ح', '/z/ui' => 'ز',*/ 00220 ); 00221 } 00222 00234 function parseManualRule( $rule, $flags = array() ) { 00235 if ( in_array( 'T', $flags ) ) { 00236 return parent::parseManualRule( $rule, $flags ); 00237 } 00238 00239 $carray = array(); 00240 // otherwise ignore all formatting 00241 foreach ( $this->mVariants as $v ) { 00242 $carray[$v] = $rule; 00243 } 00244 00245 return $carray; 00246 } 00247 00258 function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) { 00259 // check for user namespace 00260 if ( is_object( $nt ) ) { 00261 $ns = $nt->getNamespace(); 00262 if ( $ns == NS_USER || $ns == NS_USER_TALK ) { 00263 return; 00264 } 00265 } 00266 00267 $oldlink = $link; 00268 parent::findVariantLink( $link, $nt, $ignoreOtherCond ); 00269 if ( $this->getPreferredVariant() == $this->mMainLanguageCode ) { 00270 $link = $oldlink; 00271 } 00272 } 00273 00283 function autoConvert( $text, $toVariant = false ) { 00284 global $wgTitle; 00285 if ( is_object( $wgTitle ) && $wgTitle->getNamespace() == NS_FILE ) { 00286 $imagename = $wgTitle->getNsText(); 00287 if ( preg_match( "/^$imagename:/", $text ) ) { 00288 return $text; 00289 } 00290 } 00291 return parent::autoConvert( $text, $toVariant ); 00292 } 00293 00302 function translate( $text, $toVariant ) { 00303 global $wgLanguageCode; 00304 $text = parent::translate( $text, $toVariant ); 00305 00306 switch ( $toVariant ) { 00307 case 'kk-cyrl': 00308 case 'kk-kz': 00309 $letters = KK_L_UC . KK_L_LC . 'ʺʹ#0123456789'; 00310 $wgLanguageCode = 'kk'; 00311 break; 00312 case 'kk-latn': 00313 case 'kk-tr': 00314 $letters = KK_C_UC . KK_C_LC . '№0123456789'; 00315 $wgLanguageCode = 'kk-Latn'; 00316 break; 00317 case 'kk-arab': 00318 case 'kk-cn': 00319 $letters = KK_C_UC . KK_C_LC . /*KK_L_UC.KK_L_LC.'ʺʹ'.*/',;\?%\*№0123456789'; 00320 $wgLanguageCode = 'kk-Arab'; 00321 break; 00322 default: 00323 $wgLanguageCode = 'kk'; 00324 return $text; 00325 } 00326 // disable conversion variables like $1, $2... 00327 $varsfix = '\$[0-9]'; 00328 00329 $matches = preg_split( '/' . $varsfix . '[^' . $letters . ']+/u', $text, -1, PREG_SPLIT_OFFSET_CAPTURE ); 00330 $mstart = 0; 00331 $ret = ''; 00332 foreach ( $matches as $m ) { 00333 $ret .= substr( $text, $mstart, $m[1] -$mstart ); 00334 $ret .= $this->regsConverter( $m[0], $toVariant ); 00335 $mstart = $m[1] + strlen( $m[0] ); 00336 } 00337 return $ret; 00338 } 00339 00345 function regsConverter( $text, $toVariant ) { 00346 if ( $text == '' ) { 00347 return $text; 00348 } 00349 00350 switch ( $toVariant ) { 00351 case 'kk-arab': 00352 case 'kk-cn': 00353 $letters = KK_C_LC . KK_C_UC/*.KK_L_LC.KK_L_UC*/; 00354 $front = 'әөүіӘӨҮІ'/*.'äöüiÄÖÜİ'*/; 00355 $excludes = 'еэгғкқЕЭГҒКҚ'/*.'eégğkqEÉGĞKQ'*/; 00356 // split text to words 00357 $matches = preg_split( '/[\b\s\-\.:]+/', $text, -1, PREG_SPLIT_OFFSET_CAPTURE ); 00358 $mstart = 0; 00359 $ret = ''; 00360 foreach ( $matches as $m ) { 00361 $ret .= substr( $text, $mstart, $m[1] - $mstart ); 00362 // is matched the word to front vowels? 00363 // exclude a words matched to е, э, г, к, к, қ, 00364 // them should be without hamza 00365 if ( preg_match( '/[' . $front . ']/u', $m[0] ) && !preg_match( '/[' . $excludes . ']/u', $m[0] ) ) { 00366 $ret .= preg_replace( '/[' . $letters . ']+/u', H_HAMZA . '$0', $m[0] ); 00367 } else { 00368 $ret .= $m[0]; 00369 } 00370 $mstart = $m[1] + strlen( $m[0] ); 00371 } 00372 $text =& $ret; 00373 foreach ( $this->mCyLa2Arab as $pat => $rep ) { 00374 $text = preg_replace( $pat, $rep, $text ); 00375 } 00376 return $text; 00377 break; 00378 case 'kk-latn': 00379 case 'kk-tr': 00380 foreach ( $this->mCyrl2Latn as $pat => $rep ) { 00381 $text = preg_replace( $pat, $rep, $text ); 00382 } 00383 return $text; 00384 break; 00385 case 'kk-cyrl': 00386 case 'kk-kz': 00387 foreach ( $this->mLatn2Cyrl as $pat => $rep ) { 00388 $text = preg_replace( $pat, $rep, $text ); 00389 } 00390 return $text; 00391 break; 00392 default: 00393 return $text; 00394 } 00395 } 00396 00401 function convertCategoryKey( $key ) { 00402 return $this->autoConvert( $key, 'kk' ); 00403 } 00404 00405 } 00406 00413 class LanguageKk extends LanguageKk_cyrl { 00414 00415 function __construct() { 00416 global $wgHooks; 00417 parent::__construct(); 00418 00419 $variants = array( 'kk', 'kk-cyrl', 'kk-latn', 'kk-arab', 'kk-kz', 'kk-tr', 'kk-cn' ); 00420 $variantfallbacks = array( 00421 'kk' => 'kk-cyrl', 00422 'kk-cyrl' => 'kk', 00423 'kk-latn' => 'kk', 00424 'kk-arab' => 'kk', 00425 'kk-kz' => 'kk-cyrl', 00426 'kk-tr' => 'kk-latn', 00427 'kk-cn' => 'kk-arab' 00428 ); 00429 00430 $this->mConverter = new KkConverter( $this, 'kk', $variants, $variantfallbacks ); 00431 00432 $wgHooks['PageContentSaveComplete'][] = $this->mConverter; 00433 } 00434 00442 function ucfirst( $string ) { 00443 $variant = $this->getPreferredVariant(); 00444 if ( ( $variant == 'kk-latn' || $variant == 'kk-tr' ) && $string[0] == 'i' ) { 00445 $string = 'İ' . substr( $string, 1 ); 00446 } else { 00447 $string = parent::ucfirst( $string ); 00448 } 00449 return $string; 00450 } 00451 00459 function lcfirst( $string ) { 00460 $variant = $this->getPreferredVariant(); 00461 if ( ( $variant == 'kk-latn' || $variant == 'kk-tr' ) && $string[0] == 'I' ) { 00462 $string = 'ı' . substr( $string, 1 ); 00463 } else { 00464 $string = parent::lcfirst( $string ); 00465 } 00466 return $string; 00467 } 00468 00474 function convertGrammar( $word, $case ) { 00475 wfProfileIn( __METHOD__ ); 00476 00477 $variant = $this->getPreferredVariant(); 00478 switch ( $variant ) { 00479 case 'kk-arab': 00480 case 'kk-cn': 00481 $word = parent::convertGrammarKk_arab( $word, $case ); 00482 break; 00483 case 'kk-latn': 00484 case 'kk-tr': 00485 $word = parent::convertGrammarKk_latn( $word, $case ); 00486 break; 00487 case 'kk-cyrl': 00488 case 'kk-kz': 00489 case 'kk': 00490 default: 00491 $word = parent::convertGrammarKk_cyrl( $word, $case ); 00492 } 00493 00494 wfProfileOut( __METHOD__ ); 00495 return $word; 00496 } 00497 }