languages/LanguageZh.php

   1 <?php
   2 require_once( "LanguageZh_cn.php");
   3 require_once( "LanguageZh_tw.php");
   4 require_once( "LanguageZh_sg.php");
   5 require_once( "LanguageZh_hk.php");
   6
   7 /*
   8    hook to refresh the cache of conversion tables when
   9    MediaWiki:zhconversiontable* is updated
  10 */
  11 function zhOnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) {
  12         $titleobj = $article->getTitle();
  13         if($titleobj->getNamespace() == NS_MEDIAWIKI) {
  14                 global $wgContLang; // should be an LanguageZh.
  15                 if(get_class($wgContLang) != 'languagezh')
  16                         return true;
  17
  18                 $title = $titleobj->getDBkey();
  19                 $t = explode('/', $title, 2);
  20                 if( $t[0] == 'Zhconversiontable' ) {
  21                         if(!in_array($t[1], array('zh-cn', 'zh-tw', 'zh-sg', 'zh-hk')))
  22                                 return true;
  23                         $wgContLang->reloadTables();
  24                 }
  25         }
  26 }
  27
  28 $wgHooks['ArticleSaveComplete'][] = 'zhOnArticleSaveComplete';
  29
  30 /* class that handles both Traditional and Simplified Chinese
  31    right now it only distinguish zh_cn and zh_tw (actuall, zh_cn and
  32    non-zh_cn), will add support for zh_sg, zh_hk, etc, later.
  33 */
  34 class LanguageZh extends LanguageZh_cn {
  35
  36         var $mZhLanguageCode=false;
  37         var $mTables=false; //the mapping tables
  38         var $mTablesLoaded = false;
  39         var $mCacheKey;
  40         var $mDoTitleConvert = true;
  41         function LanguageZh() {
  42                 global $wgDBname;
  43                 $this->mCacheKey = $wgDBname . ":zhtables";
  44         }
  45
  46         function reloadTables() {
  47                 global $wgMemc;
  48                 $wgMemc->delete($this->mCacheKey);
  49                 $this->mTablesLoaded=false;
  50                 $this->loadTables();
  51         }
  52
  53         // load conversion tables either from the cache or the disk
  54         function loadTables() {
  55                 global $wgMemc;
  56                 if( $this->mTablesLoaded )
  57                         return;
  58                 $this->mTablesLoaded = true;
  59                 $this->mTables = $wgMemc->get( $this->mCacheKey );
  60                 if( empty( $this->mTables ) ) {
  61                         global $wgMessageCache;
  62                         require( "includes/ZhConversion.php" );
  63                         $this->mTables = array();
  64                         $this->mTables['zh-cn'] = $zh2CN;
  65                         $this->mTables['zh-tw'] = $zh2TW;
  66                         $this->mTables['zh-sg'] = $zh2SG;
  67                         $this->mTables['zh-hk'] = $zh2HK;
  68                         if( is_object( $wgMessageCache ) ){
  69                                 $cached = $this->parseCachedTable( $wgMessageCache->get( 'zhconversiontable/zh-cn', true, true, true ) );
  70                                 $this->mTables['zh-cn'] = array_merge($this->mTables['zh-cn'], $cached);
  71
  72                                 $cached = $this->parseCachedTable( $wgMessageCache->get( 'zhconversiontable/zh-tw', true, true, true ) );
  73                                 $this->mTables['zh-tw'] = array_merge($this->mTables['zh-tw'], $cached);
  74
  75                                 $cached = $this->parseCachedTable( $wgMessageCache->get( 'zhconversiontable/zh-sg', true, true, true ) );
  76                                 $this->mTables['zh-sg'] = array_merge($this->mTables['zh-sg'], $cached);
  77
  78                                 $cached = $this->parseCachedTable( $wgMessageCache->get( 'zhconversiontable/zh-hk', true, true, true ) );
  79                                 $this->mTables['zh-hk'] = array_merge($this->mTables['zh-hk'], $cached);
  80
  81                         }
  82                         $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
  83                 }
  84         }
  85
  86         /*
  87                 parse the conversion table stored in the cache
  88
  89                 the table should be in the following format:
  90
  91                         -{
  92                                 word => word ;
  93                                 word => word ;
  94                                 ...
  95                         -}
  96         */
  97         function parseCachedTable($txt) {
  98                 /* $txt should be enclosed by -{ and }- */
  99                 $a = explode( '-{', $txt);
 100                 if( count($a) < 2)
 101                         return array();
 102                 array_shift($a);
 103                 $b = explode( '}-', $a[0]);
 104
 105                 $stripped = str_replace(array('*','#'), '', $b[0]);
 106                 $table = explode( ';', $stripped );
 107                 $ret = array();
 108                 foreach( $table as $t ) {
 109                         $m = explode( '=>', $t );
 110                         if( count( $m ) != 2)
 111                                 continue;
 112                         $ret[trim($m[0])] = trim($m[1]);
 113                 }
 114                 return $ret;
 115         }
 116
 117         /*
 118                 get preferred language variants.
 119         */
 120         function getPreferredVariant() {
 121                 global $wgUser, $wgRequest;
 122
 123                 if($this->mZhLanguageCode)
 124                         return $this->mZhLanguageCode;
 125
 126                 // see if the preference is set in the request
 127                 $zhreq = $wgRequest->getText( 'variant' );
 128                 if( in_array( $zhreq, $this->getVariants() ) ) {
 129                         $this->mZhLanguageCode = $zhreq;
 130                         return $zhreq;
 131                 }
 132
 133                 // get language variant preference from logged in users
 134                 if($wgUser->getID()!=0) {
 135                         $this->mZhLanguageCode = $wgUser->getOption('variant');
 136                 }
 137
 138                 if( !$this->mZhLanguageCode ) {
 139                         // see if some zh- variant is set in the http header,
 140                         $this->mZhLanguageCode="zh";
 141                         $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 142                         $zh = strstr($header, 'zh-');
 143                         if($zh) {
 144                                 $this->mZhLanguageCode = substr($zh,0,5);
 145                         }
 146                 }
 147                 return $this->mZhLanguageCode;
 148         }
 149
 150         # this should give much better diff info
 151         function segmentForDiff( $text ) {
 152                 return preg_replace(
 153                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 154                         "' ' .\"$1\"", $text);
 155         }
 156
 157         function unsegmentForDiff( $text ) {
 158                 return preg_replace(
 159                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 160                         "\"$1\"", $text);
 161         }
 162
 163         function autoConvert($text, $toVariant=false) {
 164                 $fname="LanguageZh::autoConvert";
 165                 wfProfileIn( $fname );
 166
 167                 if(!$this->mTablesLoaded)
 168                         $this->loadTables();
 169
 170                 if(!$toVariant)
 171                         $toVariant = $this->getPreferredVariant();
 172                 $ret = '';
 173                 switch( $toVariant ) {
 174                         case 'zh-cn': $ret = strtr($text, $this->mTables['zh-cn']);break;
 175                         case 'zh-tw': $ret = strtr($text, $this->mTables['zh-tw']);break;
 176                         case 'zh-sg': $ret = strtr(strtr($text, $this->mTables['zh-cn']), $this->mTables['zh-sg']);break;
 177                         case 'zh-hk': $ret = strtr(strtr($text, $this->mTables['zh-tw']), $this->mTables['zh-hk']);break;
 178                         default: $ret = $text;
 179                 }
 180                 wfProfileOut( $fname );
 181                 return $ret;
 182         }
 183
 184         function autoConvertToAllVariants($text) {
 185                 $fname="LanguageZh::autoConvertToAllVariants";
 186                 wfProfileIn( $fname );
 187                 if( !$this->mTablesLoaded )
 188                         $this->loadTables();
 189
 190                 $ret = array();
 191                 $ret['zh-cn'] = strtr($text, $this->mTables['zh-cn']);
 192                 $ret['zh-tw'] = strtr($text, $this->mTables['zh-tw']);
 193                 $ret['zh-sg'] = strtr(strtr($text, $this->mTables['zh-cn']), $this->mTables['zh-sg']);
 194                 $ret['zh-hk'] = strtr(strtr($text, $this->mTables['zh-tw']), $this->mTables['zh-hk']);
 195                 wfProfileOut( $fname );
 196                 return $ret;
 197         }
 198
 199         # convert text to different variants of a language. the automatic
 200         # conversion is done in autoConvert(). here we parse the text
 201         # marked with -{}-, which specifies special conversions of the
 202         # text that can not be accomplished in autoConvert()
 203         #
 204         # syntax of the markup:
 205         # -{code1:text1;code2:text2;...}-  or
 206         # -{text}- in which case no conversion should take place for text
 207         function convert( $text , $isTitle=false) {
 208                 global $wgDisableLangConversion;
 209                 if($wgDisableLangConversion)
 210                         return $text;
 211
 212                 $mw =& MagicWord::get( MAG_NOTITLECONVERT );
 213                 if( $mw->matchAndRemove( $text ) )
 214                         $this->mDoTitleConvert = false;
 215
 216                 // no conversion if redirecting
 217                 $mw =& MagicWord::get( MAG_REDIRECT );
 218                 if( $mw->matchStart( $text ))
 219                         return $text;
 220
 221                 if( $isTitle ) {
 222                         if( !$this->mDoTitleConvert )
 223                                 return $text;
 224
 225                         global $wgRequest;
 226                         $isredir = $wgRequest->getText( 'redirect', 'yes' );
 227                         $action = $wgRequest->getText( 'action' );
 228                         if ( $isredir == 'no' || $action == 'edit' ) {
 229                                 return $text;
 230                         }
 231                         else {
 232                                 return $this->autoConvert($text);
 233                         }
 234                 }
 235
 236                 $plang = $this->getPreferredVariant();
 237                 $fallback = $this->getVariantFallback($plang);
 238
 239                 $tarray = explode("-{", $text);
 240                 $tfirst = array_shift($tarray);
 241                 $text = $this->autoConvert($tfirst);
 242                 foreach($tarray as $txt) {
 243                         $marked = explode("}-", $txt);
 244
 245                         $choice = explode(";", $marked{0});
 246                         if(!array_key_exists(1, $choice)) {
 247                                 /* a single choice */
 248                                 $text .= $choice{0};
 249                         } else {
 250                                 $choice1=false;
 251                                 $choice2=false;
 252                                 foreach($choice as $c) {
 253                                         $v = explode(":", $c);
 254                                         if(!array_key_exists(1, $v)) {
 255                                                 //syntax error in the markup, give up
 256                                                 break;
 257                                         }
 258                                         $code = trim($v{0});
 259                                         $content = trim($v{1});
 260                                         if($code == $plang) {
 261                                                 $choice1 = $content;
 262                                                 break;
 263                                         }
 264                                         if($code == $fallback)
 265                                                 $choice2 = $content;
 266                                 }
 267                                 if ( $choice1 )
 268                                         $text .= $choice1;
 269                                 elseif ( $choice2 )
 270                                         $text .= $choice2;
 271                                 else
 272                                         $text .= $marked{0};
 273                         }
 274                         if(array_key_exists(1, $marked))
 275                                 $text .= $this->autoConvert($marked{1});
 276                 }
 277
 278                 return $text;
 279         }
 280
 281
 282         function getVariants() {
 283                 return array("zh", "zh-cn", "zh-tw", "zh-sg", "zh-hk");
 284         }
 285
 286         function getVariantFallback($v) {
 287                 switch ($v) {
 288                 case 'zh': return 'zh-cn'; break;
 289                 case 'zh-cn': return 'zh-sg'; break;
 290                 case 'zh-sg': return 'zh-cn'; break;
 291                 case 'zh-tw': return 'zh-hk'; break;
 292                 case 'zh-hk': return 'zh-tw'; break;
 293                 }
 294                 return false;
 295         }
 296
 297         // word segmentation
 298         function stripForSearch( $string ) {
 299                 $fname="LanguageZh::stripForSearch";
 300                 wfProfileIn( $fname );
 301
 302                 // eventually this should be a word segmentation
 303                 // for now just treat each character as a word
 304                 $t = preg_replace(
 305                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 306                                 "' ' .\"$1\"", $string);
 307
 308         //always convert to zh-cn before indexing. it should be
 309                 //better to use zh-cn for search, since conversion from
 310                 //Traditional to Simplified is less ambiguous than the
 311                 //other way around
 312
 313                 $t = $this->autoConvert($t, 'zh-cn');
 314                 $t = LanguageUtf8::stripForSearch( $t );
 315                 wfProfileOut( $fname );
 316                 return $t;
 317
 318         }
 319
 320         function convertForSearchResult( $termsArray ) {
 321                 $terms = implode( '|', $termsArray );
 322                 $terms = implode( '|', $this->autoConvertToAllVariants( $terms ) );
 323                 $ret = array_unique( explode('|', $terms) );
 324                 return $ret;
 325         }
 326
 327         function findVariantLink( &$link, &$nt ) {
 328                 static $count=0; //used to limit this operation
 329                 static $cache=array();
 330                 global $wgDisableLangConversion;
 331                 $pref = $this->getPreferredVariant();
 332                 if( $wgDisableLangConversion || $pref == 'zh' || $count > 50)
 333                         return;
 334                 $count++;
 335                 $variants = $this->autoConvertToAllVariants($link);
 336                 if($variants == false) //give up
 337                         return;
 338                 foreach( $variants as $v ) {
 339                         if(isset($cache[$v]))
 340                                 continue;
 341                         $cache[$v] = 1;
 342                         $varnt = Title::newFromText( $v );
 343                         if( $varnt && $varnt->getArticleID() > 0 ) {
 344                                 $nt = $varnt;
 345                                 $link = $v;
 346                                 break;
 347                         }
 348                 }
 349         }
 350
 351         function getExtraHashOptions() {
 352                 global $wgUser;
 353                 $variant = $this->getPreferredVariant();
 354                 return '!' . $variant ;
 355         }
 356 }
 357 ?>