- Generated on Wed Oct 1 2014 21:36:29 for MediaWiki by
1.7.6.1
|
MediaWiki
REL1_19
|
00001 <?php 00002 00003 require_once( dirname( __FILE__ ) . '/../LanguageConverter.php' ); 00004 require_once( dirname( __FILE__ ) . '/LanguageKk_cyrl.php' ); 00005 00006 define( 'KK_C_UC', 'АӘБВГҒДЕЁЖЗИЙКҚЛМНҢОӨПРСТУҰҮФХҺЦЧШЩЪЫІЬЭЮЯ' ); # Kazakh Cyrillic uppercase 00007 define( 'KK_C_LC', 'аәбвгғдеёжзийкқлмнңоөпрстуұүфхһцчшщъыіьэюя' ); # Kazakh Cyrillic lowercase 00008 define( 'KK_L_UC', 'AÄBCÇDEÉFGĞHIİÏJKLMNÑOÖPQRSŞTUÜVWXYÝZ' ); # Kazakh Latin uppercase 00009 define( 'KK_L_LC', 'aäbcçdeéfgğhıiïjklmnñoöpqrsştuüvwxyýz' ); # Kazakh Latin lowercase 00010 // define( 'KK_A', 'ٴابپتجحدرزسشعفقكلمنڭەوۇۋۆىيچھ' ); # Kazakh Arabic 00011 define( 'H_HAMZA', 'ٴ' ); # U+0674 ARABIC LETTER HIGH HAMZA 00012 // define( 'ZWNJ', '' ); # U+200C ZERO WIDTH NON-JOINER 00013 00019 class KkConverter extends LanguageConverter { 00020 00021 protected $mCyrl2Latn, $mLatn2Cyrl, $mCyLa2Arab; 00022 00030 function __construct( $langobj, $maincode, 00031 $variants = array(), 00032 $variantfallbacks = array(), 00033 $flags = array() ) { 00034 parent::__construct( $langobj, $maincode, 00035 $variants, $variantfallbacks, $flags ); 00036 00037 // No point delaying this since they're in code. 00038 // Waiting until loadDefaultTables() means they never get loaded 00039 // when the tables themselves are loaded from cache. 00040 $this->loadRegs(); 00041 } 00042 00043 function loadDefaultTables() { 00044 // require( dirname(__FILE__)."/../../includes/KkConversion.php" ); 00045 // Placeholder for future implementing. Remove variables declarations 00046 // after generating KkConversion.php 00047 $kk2Cyrl = array(); 00048 $kk2Latn = array(); 00049 $kk2Arab = array(); 00050 $kk2KZ = array(); 00051 $kk2TR = array(); 00052 $kk2CN = array(); 00053 00054 $this->mTables = array( 00055 'kk-cyrl' => new ReplacementArray( $kk2Cyrl ), 00056 'kk-latn' => new ReplacementArray( $kk2Latn ), 00057 'kk-arab' => new ReplacementArray( $kk2Arab ), 00058 'kk-kz' => new ReplacementArray( array_merge( $kk2Cyrl, $kk2KZ ) ), 00059 'kk-tr' => new ReplacementArray( array_merge( $kk2Latn, $kk2TR ) ), 00060 'kk-cn' => new ReplacementArray( array_merge( $kk2Arab, $kk2CN ) ), 00061 'kk' => new ReplacementArray() 00062 ); 00063 } 00064 00065 function postLoadTables() { 00066 $this->mTables['kk-kz']->merge( $this->mTables['kk-cyrl'] ); 00067 $this->mTables['kk-tr']->merge( $this->mTables['kk-latn'] ); 00068 $this->mTables['kk-cn']->merge( $this->mTables['kk-arab'] ); 00069 } 00070 00071 function loadRegs() { 00072 00073 $this->mCyrl2Latn = array( 00074 # # Punctuation 00075 '/№/u' => 'No.', 00076 # # Е after vowels 00077 '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])Е/u' => '$1YE', 00078 '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])е/ui' => '$1ye', 00079 # # leading ЁЮЯЩ 00080 '/^Ё([' . KK_C_UC . ']|$)/u' => 'YO$1', '/^Ё([' . KK_C_LC . ']|$)/u' => 'Yo$1', 00081 '/^Ю([' . KK_C_UC . ']|$)/u' => 'YU$1', '/^Ю([' . KK_C_LC . ']|$)/u' => 'Yu$1', 00082 '/^Я([' . KK_C_UC . ']|$)/u' => 'YA$1', '/^Я([' . KK_C_LC . ']|$)/u' => 'Ya$1', 00083 '/^Щ([' . KK_C_UC . ']|$)/u' => 'ŞÇ$1', '/^Щ([' . KK_C_LC . ']|$)/u' => 'Şç$1', 00084 # # other ЁЮЯ 00085 '/Ё/u' => 'YO', '/ё/u' => 'yo', 00086 '/Ю/u' => 'YU', '/ю/u' => 'yu', 00087 '/Я/u' => 'YA', '/я/u' => 'ya', 00088 '/Щ/u' => 'ŞÇ', '/щ/u' => 'şç', 00089 # # soft and hard signs 00090 '/[ъЪ]/u' => 'ʺ', '/[ьЬ]/u' => 'ʹ', 00091 # # other characters 00092 '/А/u' => 'A', '/а/u' => 'a', '/Ә/u' => 'Ä', '/ә/u' => 'ä', 00093 '/Б/u' => 'B', '/б/u' => 'b', '/В/u' => 'V', '/в/u' => 'v', 00094 '/Г/u' => 'G', '/г/u' => 'g', '/Ғ/u' => 'Ğ', '/ғ/u' => 'ğ', 00095 '/Д/u' => 'D', '/д/u' => 'd', '/Е/u' => 'E', '/е/u' => 'e', 00096 '/Ж/u' => 'J', '/ж/u' => 'j', '/З/u' => 'Z', '/з/u' => 'z', 00097 '/И/u' => 'Ï', '/и/u' => 'ï', '/Й/u' => 'Ý', '/й/u' => 'ý', 00098 '/К/u' => 'K', '/к/u' => 'k', '/Қ/u' => 'Q', '/қ/u' => 'q', 00099 '/Л/u' => 'L', '/л/u' => 'l', '/М/u' => 'M', '/м/u' => 'm', 00100 '/Н/u' => 'N', '/н/u' => 'n', '/Ң/u' => 'Ñ', '/ң/u' => 'ñ', 00101 '/О/u' => 'O', '/о/u' => 'o', '/Ө/u' => 'Ö', '/ө/u' => 'ö', 00102 '/П/u' => 'P', '/п/u' => 'p', '/Р/u' => 'R', '/р/u' => 'r', 00103 '/С/u' => 'S', '/с/u' => 's', '/Т/u' => 'T', '/т/u' => 't', 00104 '/У/u' => 'W', '/у/u' => 'w', '/Ұ/u' => 'U', '/ұ/u' => 'u', 00105 '/Ү/u' => 'Ü', '/ү/u' => 'ü', '/Ф/u' => 'F', '/ф/u' => 'f', 00106 '/Х/u' => 'X', '/х/u' => 'x', '/Һ/u' => 'H', '/һ/u' => 'h', 00107 '/Ц/u' => 'C', '/ц/u' => 'c', '/Ч/u' => 'Ç', '/ч/u' => 'ç', 00108 '/Ш/u' => 'Ş', '/ш/u' => 'ş', '/Ы/u' => 'I', '/ы/u' => 'ı', 00109 '/І/u' => 'İ', '/і/u' => 'i', '/Э/u' => 'É', '/э/u' => 'é', 00110 ); 00111 00112 $this->mLatn2Cyrl = array( 00113 # # Punctuation 00114 '/#|No\./' => '№', 00115 # # Şç 00116 '/ŞÇʹ/u' => 'ЩЬ', '/Şçʹ/u' => 'Щь', '/Şçʹ/u' => 'Щь', 00117 '/Ş[Çç]/u' => 'Щ', '/şç/u' => 'щ', 00118 # # soft and hard signs 00119 '/([' . KK_L_UC . '])ʺ([' . KK_L_UC . '])/u' => '$1Ъ$2', 00120 '/ʺ([' . KK_L_LC . '])/u' => 'ъ$1', 00121 '/([' . KK_L_UC . '])ʹ([' . KK_L_UC . '])/u' => '$1Ь$2', 00122 '/ʹ([' . KK_L_LC . '])/u' => 'ь$1', 00123 '/ʺ/u' => 'ъ', 00124 '/ʹ/u' => 'ь', 00125 # # Ye Yo Yu Ya. 00126 '/Y[Ee]/u' => 'Е', '/ye/u' => 'е', 00127 '/Y[Oo]/u' => 'Ё', '/yo/u' => 'ё', 00128 '/Y[UWuw]/u' => 'Ю', '/y[uw]/u' => 'ю', 00129 '/Y[Aa]/u' => 'Я', '/ya/u' => 'я', 00130 # # other characters 00131 '/A/u' => 'А', '/a/u' => 'а', '/Ä/u' => 'Ә', '/ä/u' => 'ә', 00132 '/B/u' => 'Б', '/b/u' => 'б', '/C/u' => 'Ц', '/c/u' => 'ц', 00133 '/Ç/u' => 'Ч', '/ç/u' => 'ч', '/D/u' => 'Д', '/d/u' => 'д', 00134 '/E/u' => 'Е', '/e/u' => 'е', '/É/u' => 'Э', '/é/u' => 'э', 00135 '/F/u' => 'Ф', '/f/u' => 'ф', '/G/u' => 'Г', '/g/u' => 'г', 00136 '/Ğ/u' => 'Ғ', '/ğ/u' => 'ғ', '/H/u' => 'Һ', '/h/u' => 'һ', 00137 '/I/u' => 'Ы', '/ı/u' => 'ы', '/İ/u' => 'І', '/i/u' => 'і', 00138 '/Ï/u' => 'И', '/ï/u' => 'и', '/J/u' => 'Ж', '/j/u' => 'ж', 00139 '/K/u' => 'К', '/k/u' => 'к', '/L/u' => 'Л', '/l/u' => 'л', 00140 '/M/u' => 'М', '/m/u' => 'м', '/N/u' => 'Н', '/n/u' => 'н', 00141 '/Ñ/u' => 'Ң', '/ñ/u' => 'ң', '/O/u' => 'О', '/o/u' => 'о', 00142 '/Ö/u' => 'Ө', '/ö/u' => 'ө', '/P/u' => 'П', '/p/u' => 'п', 00143 '/Q/u' => 'Қ', '/q/u' => 'қ', '/R/u' => 'Р', '/r/u' => 'р', 00144 '/S/u' => 'С', '/s/u' => 'с', '/Ş/u' => 'Ш', '/ş/u' => 'ш', 00145 '/T/u' => 'Т', '/t/u' => 'т', '/U/u' => 'Ұ', '/u/u' => 'ұ', 00146 '/Ü/u' => 'Ү', '/ü/u' => 'ү', '/V/u' => 'В', '/v/u' => 'в', 00147 '/W/u' => 'У', '/w/u' => 'у', '/Ý/u' => 'Й', '/ý/u' => 'й', 00148 '/X/u' => 'Х', '/x/u' => 'х', '/Z/u' => 'З', '/z/u' => 'з', 00149 ); 00150 00151 $this->mCyLa2Arab = array( 00152 # # Punctuation -> Arabic 00153 '/#|№|No\./u' => '', # ؀ 00154 '/\,/' => '،', # ، 00155 '/;/' => '؛', # ؛ 00156 '/\?/' => '؟', # ؟ 00157 '/%/' => '٪', # ٪ 00158 '/\*/' => '٭', # ٭ 00159 # # Digits -> Arabic 00160 '/0/' => '۰', # ۰ 00161 '/1/' => '۱', # ۱ 00162 '/2/' => '۲', # ۲ 00163 '/3/' => '۳', # ۳ 00164 '/4/' => '۴', # ۴ 00165 '/5/' => '۵', # ۵ 00166 '/6/' => '۶', # ۶ 00167 '/7/' => '۷', # ۷ 00168 '/8/' => '۸', # ۸ 00169 '/9/' => '۹', # ۹ 00170 # # Cyrillic -> Arabic 00171 '/Аллаһ/ui' => 'ﷲ', 00172 '/([АӘЕЁИОӨҰҮЭЮЯЪЬ])е/ui' => '$1يە', 00173 '/[еэ]/ui' => 'ە', '/[ъь]/ui' => '', 00174 '/[аә]/ui' => 'ا', '/[оө]/ui' => 'و', '/[ұү]/ui' => 'ۇ', '/[ыі]/ui' => 'ى', 00175 '/[и]/ui' => 'ىي', '/ё/ui' => 'يو', '/ю/ui' => 'يۋ', '/я/ui' => 'يا', '/[й]/ui' => 'ي', 00176 '/ц/ui' => 'تس', '/щ/ui' => 'شش', 00177 '/һ/ui' => 'ح', '/ч/ui' => 'تش', 00178 # '/һ/ui' => 'ھ', '/ч/ui' => 'چ', 00179 '/б/ui' => 'ب', '/в/ui' => 'ۆ', '/г/ui' => 'گ', '/ғ/ui' => 'ع', 00180 '/д/ui' => 'د', '/ж/ui' => 'ج', '/з/ui' => 'ز', '/к/ui' => 'ك', 00181 '/қ/ui' => 'ق', '/л/ui' => 'ل', '/м/ui' => 'م', '/н/ui' => 'ن', 00182 '/ң/ui' => 'ڭ', '/п/ui' => 'پ', '/р/ui' => 'ر', '/с/ui' => 'س', 00183 '/т/ui' => 'ت', '/у/ui' => 'ۋ', '/ф/ui' => 'ف', '/х/ui' => 'ح', 00184 '/ш/ui' => 'ش', 00185 # # Latin -> Arabic // commented for now... 00186 /*'/Allah/ui' => 'ﷲ', 00187 '/[eé]/ui' => 'ە', '/[yý]/ui' => 'ي', '/[ʺʹ]/ui' => '', 00188 '/[aä]/ui' => 'ا', '/[oö]/ui' => 'و', '/[uü]/ui' => 'ۇ', 00189 '/[ï]/ui' => 'ىي', '/[ıIiİ]/u' => 'ى', 00190 '/c/ui' => 'تس', 00191 '/ç/ui' => 'تش', '/h/ui' => 'ح', 00192 #'/ç/ui' => 'چ', '/h/ui' => 'ھ', 00193 '/b/ui' => 'ب','/d/ui' => 'د', 00194 '/f/ui' => 'ف', '/g/ui' => 'گ', '/ğ/ui' => 'ع', 00195 '/j/ui' => 'ج', '/k/ui' => 'ك', '/l/ui' => 'ل', '/m/ui' => 'م', 00196 '/n/ui' => 'ن', '/ñ/ui' => 'ڭ', '/p/ui' => 'پ', '/q/ui' => 'ق', 00197 '/r/ui' => 'ر', '/s/ui' => 'س', '/ş/ui' => 'ش', '/t/ui' => 'ت', 00198 '/v/ui' => 'ۆ', '/w/ui' => 'ۋ', '/x/ui' => 'ح', '/z/ui' => 'ز',*/ 00199 ); 00200 } 00201 00213 function parseManualRule( $rule, $flags = array() ) { 00214 if ( in_array( 'T', $flags ) ) { 00215 return parent::parseManualRule( $rule, $flags ); 00216 } 00217 00218 $carray = array(); 00219 // otherwise ignore all formatting 00220 foreach ( $this->mVariants as $v ) { 00221 $carray[$v] = $rule; 00222 } 00223 00224 return $carray; 00225 } 00226 00237 function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) { 00238 // check for user namespace 00239 if ( is_object( $nt ) ) { 00240 $ns = $nt->getNamespace(); 00241 if ( $ns == NS_USER || $ns == NS_USER_TALK ) 00242 return; 00243 } 00244 00245 $oldlink = $link; 00246 parent::findVariantLink( $link, $nt, $ignoreOtherCond ); 00247 if ( $this->getPreferredVariant() == $this->mMainLanguageCode ) 00248 $link = $oldlink; 00249 } 00250 00260 function autoConvert( $text, $toVariant = false ) { 00261 global $wgTitle; 00262 if ( is_object( $wgTitle ) && $wgTitle->getNameSpace() == NS_FILE ) { 00263 $imagename = $wgTitle->getNsText(); 00264 if ( preg_match( "/^$imagename:/", $text ) ) return $text; 00265 } 00266 return parent::autoConvert( $text, $toVariant ); 00267 } 00268 00277 function translate( $text, $toVariant ) { 00278 global $wgLanguageCode; 00279 $text = parent::translate( $text, $toVariant ); 00280 00281 switch( $toVariant ) { 00282 case 'kk-cyrl': 00283 case 'kk-kz': 00284 $letters = KK_L_UC . KK_L_LC . 'ʺʹ#0123456789'; 00285 $wgLanguageCode = 'kk'; 00286 break; 00287 case 'kk-latn': 00288 case 'kk-tr': 00289 $letters = KK_C_UC . KK_C_LC . '№0123456789'; 00290 $wgLanguageCode = 'kk-Latn'; 00291 break; 00292 case 'kk-arab': 00293 case 'kk-cn': 00294 $letters = KK_C_UC . KK_C_LC . /*KK_L_UC.KK_L_LC.'ʺʹ'.*/',;\?%\*№0123456789'; 00295 $wgLanguageCode = 'kk-Arab'; 00296 break; 00297 default: 00298 $wgLanguageCode = 'kk'; 00299 return $text; 00300 } 00301 // disable conversion variables like $1, $2... 00302 $varsfix = '\$[0-9]'; 00303 00304 $matches = preg_split( '/' . $varsfix . '[^' . $letters . ']+/u', $text, -1, PREG_SPLIT_OFFSET_CAPTURE ); 00305 $mstart = 0; 00306 $ret = ''; 00307 foreach ( $matches as $m ) { 00308 $ret .= substr( $text, $mstart, $m[1] -$mstart ); 00309 $ret .= $this->regsConverter( $m[0], $toVariant ); 00310 $mstart = $m[1] + strlen( $m[0] ); 00311 } 00312 return $ret; 00313 } 00314 00320 function regsConverter( $text, $toVariant ) { 00321 if ( $text == '' ) { 00322 return $text; 00323 } 00324 00325 switch( $toVariant ) { 00326 case 'kk-arab': 00327 case 'kk-cn': 00328 $letters = KK_C_LC . KK_C_UC/*.KK_L_LC.KK_L_UC*/; 00329 $front = 'әөүіӘӨҮІ'/*.'äöüiÄÖÜİ'*/; 00330 $excludes = 'еэгғкқЕЭГҒКҚ'/*.'eégğkqEÉGĞKQ'*/; 00331 // split text to words 00332 $matches = preg_split( '/[\b\s\-\.:]+/', $text, -1, PREG_SPLIT_OFFSET_CAPTURE ); 00333 $mstart = 0; 00334 $ret = ''; 00335 foreach ( $matches as $m ) { 00336 $ret .= substr( $text, $mstart, $m[1] - $mstart ); 00337 // is matched the word to front vowels? 00338 // exclude a words matched to е, э, г, к, к, қ, 00339 // them should be without hamza 00340 if ( preg_match( '/[' . $front . ']/u', $m[0] ) && !preg_match( '/[' . $excludes . ']/u', $m[0] ) ) { 00341 $ret .= preg_replace( '/[' . $letters . ']+/u', H_HAMZA . '$0', $m[0] ); 00342 } else { 00343 $ret .= $m[0]; 00344 } 00345 $mstart = $m[1] + strlen( $m[0] ); 00346 } 00347 $text =& $ret; 00348 foreach ( $this->mCyLa2Arab as $pat => $rep ) { 00349 $text = preg_replace( $pat, $rep, $text ); 00350 } 00351 return $text; 00352 break; 00353 case 'kk-latn': 00354 case 'kk-tr': 00355 foreach ( $this->mCyrl2Latn as $pat => $rep ) { 00356 $text = preg_replace( $pat, $rep, $text ); 00357 } 00358 return $text; 00359 break; 00360 case 'kk-cyrl': 00361 case 'kk-kz': 00362 foreach ( $this->mLatn2Cyrl as $pat => $rep ) { 00363 $text = preg_replace( $pat, $rep, $text ); 00364 } 00365 return $text; 00366 break; 00367 default: 00368 return $text; 00369 } 00370 } 00371 00381 function markNoConversion( $text, $noParse = false ) { 00382 if ( $noParse || preg_match( "/^https?:\/\/|ftp:\/\/|irc:\/\//", $text ) ) 00383 return parent::markNoConversion( $text ); 00384 return $text; 00385 } 00386 00391 function convertCategoryKey( $key ) { 00392 return $this->autoConvert( $key, 'kk' ); 00393 } 00394 00395 } 00396 00403 class LanguageKk extends LanguageKk_cyrl { 00404 00405 function __construct() { 00406 global $wgHooks; 00407 parent::__construct(); 00408 00409 $variants = array( 'kk', 'kk-cyrl', 'kk-latn', 'kk-arab', 'kk-kz', 'kk-tr', 'kk-cn' ); 00410 $variantfallbacks = array( 00411 'kk' => 'kk-cyrl', 00412 'kk-cyrl' => 'kk', 00413 'kk-latn' => 'kk', 00414 'kk-arab' => 'kk', 00415 'kk-kz' => 'kk-cyrl', 00416 'kk-tr' => 'kk-latn', 00417 'kk-cn' => 'kk-arab' 00418 ); 00419 00420 $this->mConverter = new KkConverter( $this, 'kk', $variants, $variantfallbacks ); 00421 00422 $wgHooks['ArticleSaveComplete'][] = $this->mConverter; 00423 } 00424 00430 function isRTL() { 00431 $variant = $this->getPreferredVariant(); 00432 if ( $variant == 'kk-arab' || $variant == 'kk-cn' ) { 00433 return true; 00434 } else { 00435 return parent::isRTL(); 00436 } 00437 } 00438 00446 function ucfirst ( $string ) { 00447 $variant = $this->getPreferredVariant(); 00448 if ( ( $variant == 'kk-latn' || $variant == 'kk-tr' ) && $string[0] == 'i' ) { 00449 $string = 'İ' . substr( $string, 1 ); 00450 } else { 00451 $string = parent::ucfirst( $string ); 00452 } 00453 return $string; 00454 } 00455 00463 function lcfirst ( $string ) { 00464 $variant = $this->getPreferredVariant(); 00465 if ( ( $variant == 'kk-latn' || $variant == 'kk-tr' ) && $string[0] == 'I' ) { 00466 $string = 'ı' . substr( $string, 1 ); 00467 } else { 00468 $string = parent::lcfirst( $string ); 00469 } 00470 return $string; 00471 } 00472 00478 function convertGrammar( $word, $case ) { 00479 wfProfileIn( __METHOD__ ); 00480 00481 $variant = $this->getPreferredVariant(); 00482 switch ( $variant ) { 00483 case 'kk-arab': 00484 case 'kk-cn': 00485 $word = parent::convertGrammarKk_arab( $word, $case ); 00486 break; 00487 case 'kk-latn': 00488 case 'kk-tr': 00489 $word = parent::convertGrammarKk_latn( $word, $case ); 00490 break; 00491 case 'kk-cyrl': 00492 case 'kk-kz': 00493 case 'kk': 00494 default: 00495 $word = parent::convertGrammarKk_cyrl( $word, $case ); 00496 } 00497 00498 wfProfileOut( __METHOD__ ); 00499 return $word; 00500 } 00501 }
1.7.6.1