languages/ConverterRule.php

   1 <?php
   2 /**
   3  * This program is free software; you can redistribute it and/or modify
   4  * it under the terms of the GNU General Public License as published by
   5  * the Free Software Foundation; either version 2 of the License, or
   6  * (at your option) any later version.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  11  * GNU General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along
  14  * with this program; if not, write to the Free Software Foundation, Inc.,
  15  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  16  * http://www.gnu.org/copyleft/gpl.html
  17  *
  18  * @file
  19  * @ingroup Language
  20  */
  21
  22 /**
  23  * Parser for rules of language conversion , parse rules in -{ }- tag.
  24  * @ingroup Language
  25  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
  26  */
  27 class ConverterRule {
  28         public $mText; // original text in -{text}-
  29         public $mConverter; // LanguageConverter object
  30         public $mRuleDisplay = '';
  31         public $mRuleTitle = false;
  32         public $mRules = '';// string : the text of the rules
  33         public $mRulesAction = 'none';
  34         public $mFlags = [];
  35         public $mVariantFlags = [];
  36         public $mConvTable = [];
  37         public $mBidtable = [];// array of the translation in each variant
  38         public $mUnidtable = [];// array of the translation in each variant
  39
  40         /**
  41          * @param string $text The text between -{ and }-
  42          * @param LanguageConverter $converter
  43          */
  44         public function __construct( $text, $converter ) {
  45                 $this->mText = $text;
  46                 $this->mConverter = $converter;
  47         }
  48
  49         /**
  50          * Check if variants array in convert array.
  51          *
  52          * @param array|string $variants Variant language code
  53          * @return string Translated text
  54          */
  55         public function getTextInBidtable( $variants ) {
  56                 $variants = (array)$variants;
  57                 if ( !$variants ) {
  58                         return false;
  59                 }
  60                 foreach ( $variants as $variant ) {
  61                         if ( isset( $this->mBidtable[$variant] ) ) {
  62                                 return $this->mBidtable[$variant];
  63                         }
  64                 }
  65                 return false;
  66         }
  67
  68         /**
  69          * Parse flags with syntax -{FLAG| ... }-
  70          * @private
  71          */
  72         function parseFlags() {
  73                 $text = $this->mText;
  74                 $flags = [];
  75                 $variantFlags = [];
  76
  77                 $sepPos = strpos( $text, '|' );
  78                 if ( $sepPos !== false ) {
  79                         $validFlags = $this->mConverter->mFlags;
  80                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
  81                         foreach ( $f as $ff ) {
  82                                 $ff = trim( $ff );
  83                                 if ( isset( $validFlags[$ff] ) ) {
  84                                         $flags[$validFlags[$ff]] = true;
  85                                 }
  86                         }
  87                         $text = strval( substr( $text, $sepPos + 1 ) );
  88                 }
  89
  90                 if ( !$flags ) {
  91                         $flags['S'] = true;
  92                 } elseif ( isset( $flags['R'] ) ) {
  93                         $flags = [ 'R' => true ];// remove other flags
  94                 } elseif ( isset( $flags['N'] ) ) {
  95                         $flags = [ 'N' => true ];// remove other flags
  96                 } elseif ( isset( $flags['-'] ) ) {
  97                         $flags = [ '-' => true ];// remove other flags
  98                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
  99                         $flags['H'] = true;
 100                 } elseif ( isset( $flags['H'] ) ) {
 101                         // replace A flag, and remove other flags except T
 102                         $temp = [ '+' => true, 'H' => true ];
 103                         if ( isset( $flags['T'] ) ) {
 104                                 $temp['T'] = true;
 105                         }
 106                         if ( isset( $flags['D'] ) ) {
 107                                 $temp['D'] = true;
 108                         }
 109                         $flags = $temp;
 110                 } else {
 111                         if ( isset( $flags['A'] ) ) {
 112                                 $flags['+'] = true;
 113                                 $flags['S'] = true;
 114                         }
 115                         if ( isset( $flags['D'] ) ) {
 116                                 unset( $flags['S'] );
 117                         }
 118                         // try to find flags like "zh-hans", "zh-hant"
 119                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
 120                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
 121                         if ( $variantFlags ) {
 122                                 $variantFlags = array_flip( $variantFlags );
 123                                 $flags = [];
 124                         }
 125                 }
 126                 $this->mVariantFlags = $variantFlags;
 127                 $this->mRules = $text;
 128                 $this->mFlags = $flags;
 129         }
 130
 131         /**
 132          * Generate conversion table.
 133          * @private
 134          */
 135         function parseRules() {
 136                 $rules = $this->mRules;
 137                 $bidtable = [];
 138                 $unidtable = [];
 139                 $variants = $this->mConverter->mVariants;
 140                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
 141
 142                 // Split according to $varsep_pattern, but ignore semicolons from HTML entities
 143                 $rules = preg_replace( '/(&[#a-zA-Z0-9]+);/', "$1\x01", $rules );
 144                 $choice = preg_split( $varsep_pattern, $rules );
 145                 $choice = str_replace( "\x01", ';', $choice );
 146
 147                 foreach ( $choice as $c ) {
 148                         $v = explode( ':', $c, 2 );
 149                         if ( count( $v ) != 2 ) {
 150                                 // syntax error, skip
 151                                 continue;
 152                         }
 153                         $to = trim( $v[1] );
 154                         $v = trim( $v[0] );
 155                         $u = explode( '=>', $v, 2 );
 156                         // if $to is empty (which is also used as $from in bidtable),
 157                         // strtr() could return a wrong result.
 158                         if ( count( $u ) == 1 && $to !== '' && in_array( $v, $variants ) ) {
 159                                 $bidtable[$v] = $to;
 160                         } elseif ( count( $u ) == 2 ) {
 161                                 $from = trim( $u[0] );
 162                                 $v = trim( $u[1] );
 163                                 // if $from is empty, strtr() could return a wrong result.
 164                                 if ( array_key_exists( $v, $unidtable )
 165                                         && !is_array( $unidtable[$v] )
 166                                         && $from !== ''
 167                                         && in_array( $v, $variants ) ) {
 168                                         $unidtable[$v] = [ $from => $to ];
 169                                 } elseif ( $from !== '' && in_array( $v, $variants ) ) {
 170                                         $unidtable[$v][$from] = $to;
 171                                 }
 172                         }
 173                         // syntax error, pass
 174                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
 175                                 $bidtable = [];
 176                                 $unidtable = [];
 177                                 break;
 178                         }
 179                 }
 180                 $this->mBidtable = $bidtable;
 181                 $this->mUnidtable = $unidtable;
 182         }
 183
 184         /**
 185          * @private
 186          *
 187          * @return string
 188          */
 189         function getRulesDesc() {
 190                 $codesep = $this->mConverter->mDescCodeSep;
 191                 $varsep = $this->mConverter->mDescVarSep;
 192                 $text = '';
 193                 foreach ( $this->mBidtable as $k => $v ) {
 194                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
 195                 }
 196                 foreach ( $this->mUnidtable as $k => $a ) {
 197                         foreach ( $a as $from => $to ) {
 198                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
 199                                         "$codesep$to$varsep";
 200                         }
 201                 }
 202                 return $text;
 203         }
 204
 205         /**
 206          * Parse rules conversion.
 207          * @private
 208          *
 209          * @param string $variant
 210          *
 211          * @return string
 212          */
 213         function getRuleConvertedStr( $variant ) {
 214                 $bidtable = $this->mBidtable;
 215                 $unidtable = $this->mUnidtable;
 216
 217                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
 218                         return $this->mRules;
 219                 } else {
 220                         // display current variant in bidirectional array
 221                         $disp = $this->getTextInBidtable( $variant );
 222                         // or display current variant in fallbacks
 223                         if ( $disp === false ) {
 224                                 $disp = $this->getTextInBidtable(
 225                                         $this->mConverter->getVariantFallbacks( $variant ) );
 226                         }
 227                         // or display current variant in unidirectional array
 228                         if ( $disp === false && array_key_exists( $variant, $unidtable ) ) {
 229                                 $disp = array_values( $unidtable[$variant] )[0];
 230                         }
 231                         // or display first text under disable manual convert
 232                         if ( $disp === false && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
 233                                 if ( count( $bidtable ) > 0 ) {
 234                                         $disp = array_values( $bidtable )[0];
 235                                 } else {
 236                                         $disp = array_values( array_values( $unidtable )[0] )[0];
 237                                 }
 238                         }
 239                         return $disp;
 240                 }
 241         }
 242
 243         /**
 244          * Similar to getRuleConvertedStr(), but this prefers to use original
 245          * page title if $variant === $this->mConverter->mMainLanguageCode
 246          * and may return false in this case (so this title conversion rule
 247          * will be ignored and the original title is shown).
 248          *
 249          * @since 1.22
 250          * @param string $variant The variant code to display page title in
 251          * @return string|bool The converted title or false if just page name
 252          */
 253         function getRuleConvertedTitle( $variant ) {
 254                 if ( $variant === $this->mConverter->mMainLanguageCode ) {
 255                         // If a string targeting exactly this variant is set,
 256                         // use it. Otherwise, just return false, so the real
 257                         // page name can be shown (and because variant === main,
 258                         // there'll be no further automatic conversion).
 259                         $disp = $this->getTextInBidtable( $variant );
 260                         if ( $disp ) {
 261                                 return $disp;
 262                         }
 263                         if ( array_key_exists( $variant, $this->mUnidtable ) ) {
 264                                 $disp = array_values( $this->mUnidtable[$variant] )[0];
 265                         }
 266                         // Assigned above or still false.
 267                         return $disp;
 268                 } else {
 269                         return $this->getRuleConvertedStr( $variant );
 270                 }
 271         }
 272
 273         /**
 274          * Generate conversion table for all text.
 275          * @private
 276          */
 277         function generateConvTable() {
 278                 // Special case optimisation
 279                 if ( !$this->mBidtable && !$this->mUnidtable ) {
 280                         $this->mConvTable = [];
 281                         return;
 282                 }
 283
 284                 $bidtable = $this->mBidtable;
 285                 $unidtable = $this->mUnidtable;
 286                 $manLevel = $this->mConverter->mManualLevel;
 287
 288                 $vmarked = [];
 289                 foreach ( $this->mConverter->mVariants as $v ) {
 290                         /* for bidirectional array
 291                                 fill in the missing variants, if any,
 292                                 with fallbacks */
 293                         if ( !isset( $bidtable[$v] ) ) {
 294                                 $variantFallbacks =
 295                                         $this->mConverter->getVariantFallbacks( $v );
 296                                 $vf = $this->getTextInBidtable( $variantFallbacks );
 297                                 if ( $vf ) {
 298                                         $bidtable[$v] = $vf;
 299                                 }
 300                         }
 301
 302                         if ( isset( $bidtable[$v] ) ) {
 303                                 foreach ( $vmarked as $vo ) {
 304                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
 305                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
 306                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
 307                                         // to introduce a custom mapping between
 308                                         // words WordZh and WordTw in the whole text
 309                                         if ( $manLevel[$v] == 'bidirectional' ) {
 310                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
 311                                         }
 312                                         if ( $manLevel[$vo] == 'bidirectional' ) {
 313                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
 314                                         }
 315                                 }
 316                                 $vmarked[] = $v;
 317                         }
 318                         /* for unidirectional array fill to convert tables */
 319                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
 320                                 && isset( $unidtable[$v] )
 321                         ) {
 322                                 if ( isset( $this->mConvTable[$v] ) ) {
 323                                         $this->mConvTable[$v] = $unidtable[$v] + $this->mConvTable[$v];
 324                                 } else {
 325                                         $this->mConvTable[$v] = $unidtable[$v];
 326                                 }
 327                         }
 328                 }
 329         }
 330
 331         /**
 332          * Parse rules and flags.
 333          * @param string $variant Variant language code
 334          */
 335         public function parse( $variant = null ) {
 336                 if ( !$variant ) {
 337                         $variant = $this->mConverter->getPreferredVariant();
 338                 }
 339
 340                 $this->parseFlags();
 341                 $flags = $this->mFlags;
 342
 343                 // convert to specified variant
 344                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
 345                 if ( $this->mVariantFlags ) {
 346                         // check if current variant in flags
 347                         if ( isset( $this->mVariantFlags[$variant] ) ) {
 348                                 // then convert <text to convert> to current language
 349                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
 350                                         $variant );
 351                         } else {
 352                                 // if current variant no in flags,
 353                                 // then we check its fallback variants.
 354                                 $variantFallbacks =
 355                                         $this->mConverter->getVariantFallbacks( $variant );
 356                                 if ( is_array( $variantFallbacks ) ) {
 357                                         foreach ( $variantFallbacks as $variantFallback ) {
 358                                                 // if current variant's fallback exist in flags
 359                                                 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
 360                                                         // then convert <text to convert> to fallback language
 361                                                         $this->mRules =
 362                                                                 $this->mConverter->autoConvert( $this->mRules,
 363                                                                         $variantFallback );
 364                                                         break;
 365                                                 }
 366                                         }
 367                                 }
 368                         }
 369                         $this->mFlags = $flags = [ 'R' => true ];
 370                 }
 371
 372                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
 373                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
 374                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
 375                         $this->parseRules();
 376                 }
 377                 $rules = $this->mRules;
 378
 379                 if ( !$this->mBidtable && !$this->mUnidtable ) {
 380                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
 381                                 // fill all variants if text in -{A/H/-|text}- is non-empty but without rules
 382                                 if ( $rules !== '' ) {
 383                                         foreach ( $this->mConverter->mVariants as $v ) {
 384                                                 $this->mBidtable[$v] = $rules;
 385                                         }
 386                                 }
 387                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
 388                                 $this->mFlags = $flags = [ 'R' => true ];
 389                         }
 390                 }
 391
 392                 $this->mRuleDisplay = false;
 393                 foreach ( $flags as $flag => $unused ) {
 394                         switch ( $flag ) {
 395                                 case 'R':
 396                                         // if we don't do content convert, still strip the -{}- tags
 397                                         $this->mRuleDisplay = $rules;
 398                                         break;
 399                                 case 'N':
 400                                         // process N flag: output current variant name
 401                                         $ruleVar = trim( $rules );
 402                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
 403                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
 404                                         } else {
 405                                                 $this->mRuleDisplay = '';
 406                                         }
 407                                         break;
 408                                 case 'D':
 409                                         // process D flag: output rules description
 410                                         $this->mRuleDisplay = $this->getRulesDesc();
 411                                         break;
 412                                 case 'H':
 413                                         // process H,- flag or T only: output nothing
 414                                         $this->mRuleDisplay = '';
 415                                         break;
 416                                 case '-':
 417                                         $this->mRulesAction = 'remove';
 418                                         $this->mRuleDisplay = '';
 419                                         break;
 420                                 case '+':
 421                                         $this->mRulesAction = 'add';
 422                                         $this->mRuleDisplay = '';
 423                                         break;
 424                                 case 'S':
 425                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
 426                                         break;
 427                                 case 'T':
 428                                         $this->mRuleTitle = $this->getRuleConvertedTitle( $variant );
 429                                         $this->mRuleDisplay = '';
 430                                         break;
 431                                 default:
 432                                         // ignore unknown flags (but see error case below)
 433                         }
 434                 }
 435                 if ( $this->mRuleDisplay === false ) {
 436                         $this->mRuleDisplay = '<span class="error">'
 437                                 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
 438                                 . '</span>';
 439                 }
 440
 441                 $this->generateConvTable();
 442         }
 443
 444         /**
 445          * Checks if there are conversion rules.
 446          * @return bool
 447          */
 448         public function hasRules() {
 449                 return $this->mRules !== '';
 450         }
 451
 452         /**
 453          * Get display text on markup -{...}-
 454          * @return string
 455          */
 456         public function getDisplay() {
 457                 return $this->mRuleDisplay;
 458         }
 459
 460         /**
 461          * Get converted title.
 462          * @return string
 463          */
 464         public function getTitle() {
 465                 return $this->mRuleTitle;
 466         }
 467
 468         /**
 469          * Return how deal with conversion rules.
 470          * @return string
 471          */
 472         public function getRulesAction() {
 473                 return $this->mRulesAction;
 474         }
 475
 476         /**
 477          * Get conversion table. (bidirectional and unidirectional
 478          * conversion table)
 479          * @return array
 480          */
 481         public function getConvTable() {
 482                 return $this->mConvTable;
 483         }
 484
 485         /**
 486          * Get conversion rules string.
 487          * @return string
 488          */
 489         public function getRules() {
 490                 return $this->mRules;
 491         }
 492
 493         /**
 494          * Get conversion flags.
 495          * @return array
 496          */
 497         public function getFlags() {
 498                 return $this->mFlags;
 499         }
 500 }