Fixed some doxygen warnings
[lhc/web/wiklou.git] / languages / LanguageConverter.php
1 <?php
2
3 /**
4 * Contains the LanguageConverter class and ConverterRule class
5 * @ingroup Language
6 *
7 * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
8 * @file
9 */
10
11 /**
12 * Base class for language conversion.
13 * @ingroup Language
14 *
15 * @author Zhengzhu Feng <zhengzhu@gmail.com>
16 * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
17 */
18 class LanguageConverter {
19 var $mMainLanguageCode;
20 var $mVariants, $mVariantFallbacks, $mVariantNames;
21 var $mTablesLoaded = false;
22 var $mTables;
23 var $mNamespaceTables;
24 // 'bidirectional' 'unidirectional' 'disable' for each variant
25 var $mManualLevel;
26 var $mCacheKey;
27 var $mLangObj;
28 var $mFlags;
29 var $mDescCodeSep = ':', $mDescVarSep = ';';
30 var $mUcfirst = false;
31 var $mConvRuleTitle = false;
32 var $mURLVariant;
33 var $mUserVariant;
34 var $mHeaderVariant;
35 var $mMaxDepth = 10;
36 var $mVarSeparatorPattern;
37
38 const CACHE_VERSION_KEY = 'VERSION 6';
39
40 /**
41 * Constructor
42 *
43 * @param $langobj The Language Object
44 * @param $maincode String: the main language code of this language
45 * @param $variants Array: the supported variants of this language
46 * @param $variantfallbacks Array: the fallback language of each variant
47 * @param $flags Array: defining the custom strings that maps to the flags
48 * @param $manualLevel Array: limit for supported variants
49 */
50 public function __construct( $langobj, $maincode,
51 $variants = array(),
52 $variantfallbacks = array(),
53 $flags = array(),
54 $manualLevel = array() ) {
55 $this->mLangObj = $langobj;
56 $this->mMainLanguageCode = $maincode;
57
58 global $wgDisabledVariants;
59 $this->mVariants = array();
60 foreach ( $variants as $variant ) {
61 if ( !in_array( $variant, $wgDisabledVariants ) ) {
62 $this->mVariants[] = $variant;
63 }
64 }
65 $this->mVariantFallbacks = $variantfallbacks;
66 global $wgLanguageNames;
67 $this->mVariantNames = $wgLanguageNames;
68 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
69 $f = array(
70 // 'S' show converted text
71 // '+' add rules for alltext
72 // 'E' the gave flags is error
73 // these flags above are reserved for program
74 'A' => 'A', // add rule for convert code (all text convert)
75 'T' => 'T', // title convert
76 'R' => 'R', // raw content
77 'D' => 'D', // convert description (subclass implement)
78 '-' => '-', // remove convert (not implement)
79 'H' => 'H', // add rule for convert code
80 // (but no display in placed code )
81 'N' => 'N' // current variant name
82 );
83 $this->mFlags = array_merge( $f, $flags );
84 foreach ( $this->mVariants as $v ) {
85 if ( array_key_exists( $v, $manualLevel ) ) {
86 $this->mManualLevel[$v] = $manualLevel[$v];
87 } else {
88 $this->mManualLevel[$v] = 'bidirectional';
89 }
90 $this->mNamespaceTables[$v] = array();
91 $this->mFlags[$v] = $v;
92 }
93 }
94
95 /**
96 * @public
97 */
98 function getVariants() {
99 return $this->mVariants;
100 }
101
102 /**
103 * In case some variant is not defined in the markup, we need
104 * to have some fallback. For example, in zh, normally people
105 * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
106 * when zh-sg is preferred but not defined, we will pick zh-hans
107 * in this case. Right now this is only used by zh.
108 *
109 * @param $v String: the language code of the variant
110 * @return string array The code of the fallback language or false if there
111 * is no fallback
112 */
113 public function getVariantFallbacks( $v ) {
114 if ( isset( $this->mVariantFallbacks[$v] ) ) {
115 return $this->mVariantFallbacks[$v];
116 }
117 return $this->mMainLanguageCode;
118 }
119
120 /**
121 * Get the title produced by the conversion rule.
122 * @returns String
123 */
124 function getConvRuleTitle() {
125 return $this->mConvRuleTitle;
126 }
127
128 /**
129 * Get preferred language variants.
130 * @param $fromUser Boolean: get it from $wgUser's preferences
131 * @param $fromHeader Boolean: get it from Accept-Language
132 * @return String: the preferred language code
133 */
134 public function getPreferredVariant( $fromUser = true, $fromHeader = false ) {
135 global $wgDefaultLanguageVariant;
136
137 $req = $this->getURLVariant();
138
139 if ( $fromUser && !$req ) {
140 $req = $this->getUserVariant();
141 }
142
143 if ( $fromHeader && !$req ) {
144 $req = $this->getHeaderVariant();
145 }
146
147 if ( $wgDefaultLanguageVariant && !$req ) {
148 $req = $this->validateVariant( $wgDefaultLanguageVariant );
149 }
150
151 // This function, unlike the other get*Variant functions, is
152 // not memoized (i.e. there return value is not cached) since
153 // new information might appear during processing after this
154 // is first called.
155 if ( $req ) {
156 return $req;
157 }
158 return $this->mMainLanguageCode;
159 }
160
161 /**
162 * Validate the variant
163 * @param $v String: the variant to validate
164 * @return Mixed: returns the variant if it is valid, null otherwise
165 */
166 function validateVariant( $v = null ) {
167 if ( $v !== null && in_array( $v, $this->mVariants ) ) {
168 return $v;
169 }
170 return null;
171 }
172
173 /**
174 * Get the variant specified in the URL
175 *
176 * @return Mixed: variant if one found, false otherwise.
177 */
178 function getURLVariant() {
179 global $wgRequest;
180 $ret = null;
181
182 if ( $this->mURLVariant ) {
183 return $this->mURLVariant;
184 }
185
186 // see if the preference is set in the request
187 $ret = $wgRequest->getText( 'variant' );
188
189 if ( !$ret ) {
190 $ret = $wgRequest->getVal( 'uselang' );
191 }
192
193 return $this->mURLVariant = $this->validateVariant( $ret );
194 }
195
196 /**
197 * Determine if the user has a variant set.
198 *
199 * @return Mixed: variant if one found, false otherwise.
200 */
201 function getUserVariant() {
202 global $wgUser;
203 $ret = null;
204
205 // memoizing this function wreaks havoc on parserTest.php
206 /* if ( $this->mUserVariant ) { */
207 /* return $this->mUserVariant; */
208 /* } */
209
210 // get language variant preference from logged in users
211 // Don't call this on stub objects because that causes infinite
212 // recursion during initialisation
213 if ( $wgUser->isLoggedIn() ) {
214 $ret = $wgUser->getOption( 'variant' );
215 }
216 else {
217 // figure out user lang without constructing wgLang to avoid
218 // infinite recursion
219 $ret = $wgUser->getOption( 'language' );
220 }
221
222 return $this->mUserVariant = $this->validateVariant( $ret );
223 }
224
225
226 /**
227 * Determine the language variant from the Accept-Language header.
228 *
229 * @return Mixed: variant if one found, false otherwise.
230 */
231 function getHeaderVariant() {
232 global $wgRequest;
233 $ret = null;
234
235 if ( $this->mHeaderVariant ) {
236 return $this->mHeaderVariant;
237 }
238
239 // see if some supported language variant is set in the
240 // http header.
241
242 $acceptLanguage = $wgRequest->getHeader( 'Accept-Language' );
243 if ( !$acceptLanguage ) {
244 return null;
245 }
246
247 // explode by comma
248 $result = StringUtils::explode( ',', strtolower( $acceptLanguage ) );
249 $languages = array();
250
251 foreach ( $result as $elem ) {
252 // if $elem likes 'zh-cn;q=0.9'
253 if ( ( $posi = strpos( $elem, ';' ) ) !== false ) {
254 // get the real language code likes 'zh-cn'
255 $languages[] = substr( $elem, 0, $posi );
256 } else {
257 $languages[] = $elem;
258 }
259 }
260
261 $fallback_languages = array();
262 foreach ( $languages as $language ) {
263 // strip whitespace
264 $language = trim( $language );
265 $this->mHeaderVariant = $this->validateVariant( $language );
266 if ( $this->mHeaderVariant ) {
267 break;
268 }
269
270 // To see if there are fallbacks of current language.
271 // We record these fallback variants, and process
272 // them later.
273 $fallbacks = $this->getVariantFallbacks( $language );
274 if ( is_string( $fallbacks ) ) {
275 $fallback_languages[] = $fallbacks;
276 } elseif ( is_array( $fallbacks ) ) {
277 $fallback_languages =
278 array_merge( $fallback_languages,
279 $fallbacks );
280 }
281 }
282
283 if ( !$this->mHeaderVariant ) {
284 // process fallback languages now
285 $fallback_languages = array_unique( $fallback_languages );
286 foreach ( $fallback_languages as $language ) {
287 $this->mHeaderVariant = $this->validateVariant( $language );
288 if ( $this->mHeaderVariant ) {
289 break;
290 }
291 }
292 }
293
294 return $this->mHeaderVariant;
295 }
296
297 /**
298 * Caption convert, base on preg_replace_callback.
299 *
300 * To convert text in "title" or "alt", like '<img alt="text" ... '
301 * or '<span title="text" ... '
302 *
303 * @return String like ' alt="yyyy"' or ' title="yyyy"'
304 * @private
305 */
306 function captionConvert( $matches ) {
307 $toVariant = $this->getPreferredVariant();
308 $title = $matches[1];
309 $text = $matches[2];
310 // we convert captions except URL
311 if ( !strpos( $text, '://' ) ) {
312 $text = $this->translate( $text, $toVariant );
313 }
314 return " $title=\"$text\"";
315 }
316
317 /**
318 * Dictionary-based conversion.
319 *
320 * @param $text String: the text to be converted
321 * @param $toVariant String: the target language code
322 * @return String: the converted text
323 * @private
324 */
325 function autoConvert( $text, $toVariant = false ) {
326 $fname = 'LanguageConverter::autoConvert';
327
328 wfProfileIn( $fname );
329
330 if ( !$this->mTablesLoaded ) {
331 $this->loadTables();
332 }
333
334 if ( !$toVariant ) {
335 $toVariant = $this->getPreferredVariant();
336 if ( !$toVariant ) {
337 return $text;
338 }
339 }
340
341 /* we convert everything except:
342 1. html markups (anything between < and >)
343 2. html entities
344 3. place holders created by the parser
345 */
346 global $wgParser;
347 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
348 $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
349 } else {
350 $marker = '';
351 }
352
353 // this one is needed when the text is inside an html markup
354 $htmlfix = '|<[^>]+$|^[^<>]*>';
355
356 // disable convert to variants between <code></code> tags
357 $codefix = '<code>.+?<\/code>|';
358 // disable convertsion of <script type="text/javascript"> ... </script>
359 $scriptfix = '<script.*?>.*?<\/script>|';
360 // disable conversion of <pre xxxx> ... </pre>
361 $prefix = '<pre.*?>.*?<\/pre>|';
362
363 $reg = '/' . $codefix . $scriptfix . $prefix .
364 '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
365
366 $matches = preg_split( $reg, $text, - 1, PREG_SPLIT_OFFSET_CAPTURE );
367
368 $m = array_shift( $matches );
369
370 $ret = $this->translate( $m[0], $toVariant );
371 $mstart = $m[1] + strlen( $m[0] );
372
373 // enable convertsion of '<img alt="xxxx" ... '
374 // or '<span title="xxxx" ... '
375 $captionpattern = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
376
377 $trtext = '';
378 $trtextmark = "\0";
379 $notrtext = array();
380 foreach ( $matches as $m ) {
381 $mark = substr( $text, $mstart, $m[1] - $mstart );
382 $mark = preg_replace_callback( $captionpattern,
383 array( &$this, 'captionConvert' ),
384 $mark );
385 // Let's convert the trtext only once,
386 // it would give us more performance improvement
387 $notrtext[] = $mark;
388 $trtext .= $m[0] . $trtextmark;
389 $mstart = $m[1] + strlen( $m[0] );
390 }
391 $notrtext[] = '';
392 $trtext = $this->translate( $trtext, $toVariant );
393 $trtext = StringUtils::explode( $trtextmark, $trtext );
394 foreach ( $trtext as $t ) {
395 $ret .= array_shift( $notrtext );
396 $ret .= $t;
397 }
398 wfProfileOut( $fname );
399 return $ret;
400 }
401
402 /**
403 * Translate a string to a variant.
404 * Doesn't process markup or do any of that other stuff, for that use
405 * convert().
406 *
407 * @param $text String: text to convert
408 * @param $variant String: variant language code
409 * @return String: translated text
410 * @private
411 */
412 function translate( $text, $variant ) {
413 wfProfileIn( __METHOD__ );
414 // If $text is empty or only includes spaces, do nothing
415 // Otherwise translate it
416 if ( trim( $text ) ) {
417 if ( !$this->mTablesLoaded ) {
418 $this->loadTables();
419 }
420 $text = $this->mTables[$variant]->replace( $text );
421 }
422 wfProfileOut( __METHOD__ );
423 return $text;
424 }
425
426 /**
427 * Convert text to all supported variants.
428 *
429 * @param $text String: the text to be converted
430 * @return Array of string
431 */
432 public function autoConvertToAllVariants( $text ) {
433 $fname = 'LanguageConverter::autoConvertToAllVariants';
434 wfProfileIn( $fname );
435 if ( !$this->mTablesLoaded ) {
436 $this->loadTables();
437 }
438
439 $ret = array();
440 foreach ( $this->mVariants as $variant ) {
441 $ret[$variant] = $this->translate( $text, $variant );
442 }
443
444 wfProfileOut( $fname );
445 return $ret;
446 }
447
448 /**
449 * Convert link text to all supported variants.
450 *
451 * @param $text String: the text to be converted
452 * @return Array of string
453 */
454 public function convertLinkToAllVariants( $text ) {
455 if ( !$this->mTablesLoaded ) {
456 $this->loadTables();
457 }
458
459 $ret = array();
460 $tarray = StringUtils::explode( '-{', $text );
461 $first = true;
462
463 foreach ( $tarray as $txt ) {
464 if ( $first ) {
465 $first = false;
466 foreach ( $this->mVariants as $variant ) {
467 $ret[$variant] = $this->translate( $txt, $variant );
468 }
469 continue;
470 }
471
472 $marked = explode( '}-', $txt, 2 );
473
474 foreach ( $this->mVariants as $variant ) {
475 $ret[$variant] .= '-{' . $marked[0] . '}-';
476 if ( array_key_exists( 1, $marked ) ) {
477 $ret[$variant] .= $this->translate( $marked[1], $variant );
478 }
479 }
480
481 }
482
483 return $ret;
484 }
485
486 /**
487 * Prepare manual conversion table.
488 * @private
489 */
490 function applyManualConv( $convRule ) {
491 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom
492 // conversion in title
493 $this->mConvRuleTitle = $convRule->getTitle();
494
495 // apply manual conversion table to global table
496 $convTable = $convRule->getConvTable();
497 $action = $convRule->getRulesAction();
498 foreach ( $convTable as $variant => $pair ) {
499 if ( !$this->validateVariant( $variant ) ) {
500 continue;
501 }
502
503 if ( $action == 'add' ) {
504 foreach ( $pair as $from => $to ) {
505 // to ensure that $from and $to not be left blank
506 // so $this->translate() could always return a string
507 if ( $from || $to ) {
508 // more efficient than array_merge(), about 2.5 times.
509 $this->mTables[$variant]->setPair( $from, $to );
510 }
511 }
512 } elseif ( $action == 'remove' ) {
513 $this->mTables[$variant]->removeArray( $pair );
514 }
515 }
516 }
517
518 /**
519 * Convert namespace.
520 * @param $title String: the title included namespace
521 * @return Array of string
522 * @private
523 */
524 function convertNamespace( $title, $variant ) {
525 $splittitle = explode( ':', $title, 2 );
526 if ( count( $splittitle ) < 2 ) {
527 return $title;
528 }
529 if ( isset( $this->mNamespaceTables[$variant][$splittitle[0]] ) ) {
530 $splittitle[0] = $this->mNamespaceTables[$variant][$splittitle[0]];
531 }
532 $ret = implode( ':', $splittitle );
533 return $ret;
534 }
535
536 /**
537 * Convert text to different variants of a language. The automatic
538 * conversion is done in autoConvert(). Here we parse the text
539 * marked with -{}-, which specifies special conversions of the
540 * text that can not be accomplished in autoConvert().
541 *
542 * Syntax of the markup:
543 * -{code1:text1;code2:text2;...}- or
544 * -{flags|code1:text1;code2:text2;...}- or
545 * -{text}- in which case no conversion should take place for text
546 *
547 * @param $text String: text to be converted
548 * @return String: converted text
549 */
550 public function convert( $text ) {
551 global $wgDisableLangConversion;
552 if ( $wgDisableLangConversion ) return $text;
553
554 $variant = $this->getPreferredVariant();
555
556 return $this->recursiveConvertTopLevel( $text, $variant );
557 }
558
559 protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
560 $startPos = 0;
561 $out = '';
562 $length = strlen( $text );
563 while ( $startPos < $length ) {
564 $m = false;
565 $pos = strpos( $text, '-{', $startPos );
566
567 if ( $pos === false ) {
568 // No more markup, append final segment
569 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
570 $startPos = $length;
571 return $out;
572 }
573
574 // Markup found
575 // Append initial segment
576 $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
577
578 // Advance position
579 $startPos = $pos;
580
581 // Do recursive conversion
582 $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
583 }
584
585 return $out;
586 }
587
588 protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
589 // Quick sanity check (no function calls)
590 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
591 throw new MWException( __METHOD__.': invalid input string' );
592 }
593
594 $startPos += 2;
595 $inner = '';
596 $warningDone = false;
597 $length = strlen( $text );
598
599 while ( $startPos < $length ) {
600 $m = false;
601 preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
602 if ( !$m ) {
603 // Unclosed rule
604 break;
605 }
606
607 $token = $m[0][0];
608 $pos = $m[0][1];
609
610 // Markup found
611 // Append initial segment
612 $inner .= substr( $text, $startPos, $pos - $startPos );
613
614 // Advance position
615 $startPos = $pos;
616
617 switch ( $token ) {
618 case '-{':
619 // Check max depth
620 if ( $depth >= $this->mMaxDepth ) {
621 $inner .= '-{';
622 if ( !$warningDone ) {
623 $inner .= '<span class="error">' .
624 wfMsgForContent( 'language-converter-depth-warning',
625 $this->mMaxDepth ) .
626 '</span>';
627 $warningDone = true;
628 }
629 $startPos += 2;
630 continue;
631 }
632 // Recursively parse another rule
633 $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
634 break;
635 case '}-':
636 // Apply the rule
637 $startPos += 2;
638 $rule = new ConverterRule( $inner, $this );
639 $rule->parse( $variant );
640 $this->applyManualConv( $rule );
641 return $rule->getDisplay();
642 default:
643 throw new MWException( __METHOD__.': invalid regex match' );
644 }
645 }
646
647 // Unclosed rule
648 if ( $startPos < $length ) {
649 $inner .= substr( $text, $startPos );
650 }
651 $startPos = $length;
652 return '-{' . $this->autoConvert( $inner, $variant );
653 }
654
655 /**
656 * If a language supports multiple variants, it is
657 * possible that non-existing link in one variant
658 * actually exists in another variant. This function
659 * tries to find it. See e.g. LanguageZh.php
660 *
661 * @param $link String: the name of the link
662 * @param $nt Mixed: the title object of the link
663 * @param $ignoreOtherCond Boolean: to disable other conditions when
664 * we need to transclude a template or update a category's link
665 * @return Null, the input parameters may be modified upon return
666 */
667 public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
668 # If the article has already existed, there is no need to
669 # check it again, otherwise it may cause a fault.
670 if ( is_object( $nt ) && $nt->exists() ) {
671 return;
672 }
673
674 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
675 $wgUser;
676 $isredir = $wgRequest->getText( 'redirect', 'yes' );
677 $action = $wgRequest->getText( 'action' );
678 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
679 $disableLinkConversion = $wgDisableLangConversion
680 || $wgDisableTitleConversion;
681 $linkBatch = new LinkBatch();
682
683 $ns = NS_MAIN;
684
685 if ( $disableLinkConversion ||
686 ( !$ignoreOtherCond &&
687 ( $isredir == 'no'
688 || $action == 'edit'
689 || $action == 'submit'
690 || $linkconvert == 'no'
691 || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
692 return;
693 }
694
695 if ( is_object( $nt ) ) {
696 $ns = $nt->getNamespace();
697 }
698
699 $variants = $this->autoConvertToAllVariants( $link );
700 if ( $variants == false ) { // give up
701 return;
702 }
703
704 $titles = array();
705
706 foreach ( $variants as $v ) {
707 if ( $v != $link ) {
708 $varnt = Title::newFromText( $v, $ns );
709 if ( !is_null( $varnt ) ) {
710 $linkBatch->addObj( $varnt );
711 $titles[] = $varnt;
712 }
713 }
714 }
715
716 // fetch all variants in single query
717 $linkBatch->execute();
718
719 foreach ( $titles as $varnt ) {
720 if ( $varnt->getArticleID() > 0 ) {
721 $nt = $varnt;
722 $link = $varnt->getText();
723 break;
724 }
725 }
726 }
727
728 /**
729 * Returns language specific hash options.
730 */
731 public function getExtraHashOptions() {
732 $variant = $this->getPreferredVariant();
733 return '!' . $variant ;
734 }
735
736 /**
737 * Load default conversion tables.
738 * This method must be implemented in derived class.
739 *
740 * @private
741 */
742 function loadDefaultTables() {
743 $name = get_class( $this );
744 wfDie( "Must implement loadDefaultTables() method in class $name" );
745 }
746
747 /**
748 * Load conversion tables either from the cache or the disk.
749 * @private
750 */
751 function loadTables( $fromcache = true ) {
752 global $wgMemc;
753 if ( $this->mTablesLoaded ) {
754 return;
755 }
756 wfProfileIn( __METHOD__ );
757 $this->mTablesLoaded = true;
758 $this->mTables = false;
759 if ( $fromcache ) {
760 wfProfileIn( __METHOD__ . '-cache' );
761 $this->mTables = $wgMemc->get( $this->mCacheKey );
762 wfProfileOut( __METHOD__ . '-cache' );
763 }
764 if ( !$this->mTables
765 || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
766 wfProfileIn( __METHOD__ . '-recache' );
767 // not in cache, or we need a fresh reload.
768 // we will first load the default tables
769 // then update them using things in MediaWiki:Zhconversiontable/*
770 $this->loadDefaultTables();
771 foreach ( $this->mVariants as $var ) {
772 $cached = $this->parseCachedTable( $var );
773 $this->mTables[$var]->mergeArray( $cached );
774 }
775
776 $this->postLoadTables();
777 $this->mTables[self::CACHE_VERSION_KEY] = true;
778
779 $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
780 wfProfileOut( __METHOD__ . '-recache' );
781 }
782 wfProfileOut( __METHOD__ );
783 }
784
785 /**
786 * Hook for post processig after conversion tables are loaded.
787 *
788 */
789 function postLoadTables() { }
790
791 /**
792 * Reload the conversion tables.
793 *
794 * @private
795 */
796 function reloadTables() {
797 if ( $this->mTables ) {
798 unset( $this->mTables );
799 }
800 $this->mTablesLoaded = false;
801 $this->loadTables( false );
802 }
803
804
805 /**
806 * Parse the conversion table stored in the cache.
807 *
808 * The tables should be in blocks of the following form:
809 * -{
810 * word => word ;
811 * word => word ;
812 * ...
813 * }-
814 *
815 * To make the tables more manageable, subpages are allowed
816 * and will be parsed recursively if $recursive == true.
817 *
818 */
819 function parseCachedTable( $code, $subpage = '', $recursive = true ) {
820 global $wgMessageCache;
821 static $parsed = array();
822
823 if ( !is_object( $wgMessageCache ) ) {
824 return array();
825 }
826
827 $key = 'Conversiontable/' . $code;
828 if ( $subpage ) {
829 $key .= '/' . $subpage;
830 }
831 if ( array_key_exists( $key, $parsed ) ) {
832 return array();
833 }
834
835 if ( strpos( $code, '/' ) === false ) {
836 $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
837 } else {
838 $title = Title::makeTitleSafe( NS_MEDIAWIKI,
839 "Conversiontable/$code" );
840 if ( $title && $title->exists() ) {
841 $article = new Article( $title );
842 $txt = $article->getContents();
843 } else {
844 $txt = '';
845 }
846 }
847
848 // get all subpage links of the form
849 // [[MediaWiki:conversiontable/zh-xx/...|...]]
850 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
851 ':Conversiontable';
852 $subs = StringUtils::explode( '[[', $txt );
853 $sublinks = array();
854 foreach ( $subs as $sub ) {
855 $link = explode( ']]', $sub, 2 );
856 if ( count( $link ) != 2 ) {
857 continue;
858 }
859 $b = explode( '|', $link[0], 2 );
860 $b = explode( '/', trim( $b[0] ), 3 );
861 if ( count( $b ) == 3 ) {
862 $sublink = $b[2];
863 } else {
864 $sublink = '';
865 }
866
867 if ( $b[0] == $linkhead && $b[1] == $code ) {
868 $sublinks[] = $sublink;
869 }
870 }
871
872
873 // parse the mappings in this page
874 $blocks = StringUtils::explode( '-{', $txt );
875 $ret = array();
876 $first = true;
877 foreach ( $blocks as $block ) {
878 if ( $first ) {
879 // Skip the part before the first -{
880 $first = false;
881 continue;
882 }
883 $mappings = explode( '}-', $block, 2 );
884 $stripped = str_replace( array( "'", '"', '*', '#' ), '',
885 $mappings[0] );
886 $table = StringUtils::explode( ';', $stripped );
887 foreach ( $table as $t ) {
888 $m = explode( '=>', $t, 3 );
889 if ( count( $m ) != 2 )
890 continue;
891 // trim any trailling comments starting with '//'
892 $tt = explode( '//', $m[1], 2 );
893 $ret[trim( $m[0] )] = trim( $tt[0] );
894 }
895 }
896 $parsed[$key] = true;
897
898
899 // recursively parse the subpages
900 if ( $recursive ) {
901 foreach ( $sublinks as $link ) {
902 $s = $this->parseCachedTable( $code, $link, $recursive );
903 $ret = array_merge( $ret, $s );
904 }
905 }
906
907 if ( $this->mUcfirst ) {
908 foreach ( $ret as $k => $v ) {
909 $ret[Language::ucfirst( $k )] = Language::ucfirst( $v );
910 }
911 }
912 return $ret;
913 }
914
915 /**
916 * Enclose a string with the "no conversion" tag. This is used by
917 * various functions in the Parser.
918 *
919 * @param $text String: text to be tagged for no conversion
920 * @param $noParse Unused (?)
921 * @return String: the tagged text
922 */
923 public function markNoConversion( $text, $noParse = false ) {
924 # don't mark if already marked
925 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
926 return $text;
927 }
928
929 $ret = "-{R|$text}-";
930 return $ret;
931 }
932
933 /**
934 * Convert the sorting key for category links. This should make different
935 * keys that are variants of each other map to the same key.
936 */
937 function convertCategoryKey( $key ) {
938 return $key;
939 }
940
941 /**
942 * Hook to refresh the cache of conversion tables when
943 * MediaWiki:conversiontable* is updated.
944 * @private
945 */
946 function OnArticleSaveComplete( $article, $user, $text, $summary, $isminor,
947 $iswatch, $section, $flags, $revision ) {
948 $titleobj = $article->getTitle();
949 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
950 $title = $titleobj->getDBkey();
951 $t = explode( '/', $title, 3 );
952 $c = count( $t );
953 if ( $c > 1 && $t[0] == 'Conversiontable' ) {
954 if ( $this->validateVariant( $t[1] ) ) {
955 $this->reloadTables();
956 }
957 }
958 }
959 return true;
960 }
961
962 /**
963 * Armour rendered math against conversion.
964 * Wrap math into rawoutput -{R| math }- syntax.
965 */
966 public function armourMath( $text ) {
967 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
968 // to avoid a unwanted '}-' appeared after the math-image.
969 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
970 $ret = "-{R|$text}-";
971 return $ret;
972 }
973
974 /**
975 * Get the cached separator pattern for ConverterRule::parseRules()
976 */
977 function getVarSeparatorPattern() {
978 if ( is_null( $this->mVarSeparatorPattern ) ) {
979 // varsep_pattern for preg_split:
980 // text should be splited by ";" only if a valid variant
981 // name exist after the markup, for example:
982 // -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
983 // <span style="font-size:120%;">yyy</span>;}-
984 // we should split it as:
985 // array(
986 // [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
987 // [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
988 // [2] => ''
989 // )
990 $pat = '/;\s*(?=';
991 foreach ( $this->mVariants as $variant ) {
992 // zh-hans:xxx;zh-hant:yyy
993 $pat .= $variant . '\s*:|';
994 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
995 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
996 }
997 $pat .= '\s*$)/';
998 $this->mVarSeparatorPattern = $pat;
999 }
1000 return $this->mVarSeparatorPattern;
1001 }
1002 }
1003
1004 /**
1005 * Parser for rules of language conversion , parse rules in -{ }- tag.
1006 * @ingroup Language
1007 * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1008 */
1009 class ConverterRule {
1010 var $mText; // original text in -{text}-
1011 var $mConverter; // LanguageConverter object
1012 var $mManualCodeError = '<strong class="error">code error!</strong>';
1013 var $mRuleDisplay = '';
1014 var $mRuleTitle = false;
1015 var $mRules = '';// string : the text of the rules
1016 var $mRulesAction = 'none';
1017 var $mFlags = array();
1018 var $mVariantFlags = array();
1019 var $mConvTable = array();
1020 var $mBidtable = array();// array of the translation in each variant
1021 var $mUnidtable = array();// array of the translation in each variant
1022
1023 /**
1024 * Constructor
1025 *
1026 * @param $text String: the text between -{ and }-
1027 * @param $converter LanguageConverter object
1028 */
1029 public function __construct( $text, $converter ) {
1030 $this->mText = $text;
1031 $this->mConverter = $converter;
1032 }
1033
1034 /**
1035 * Check if variants array in convert array.
1036 *
1037 * @param $variants Array or string: variant language code
1038 * @return String: translated text
1039 */
1040 public function getTextInBidtable( $variants ) {
1041 $variants = (array)$variants;
1042 if ( !$variants ) {
1043 return false;
1044 }
1045 foreach ( $variants as $variant ) {
1046 if ( isset( $this->mBidtable[$variant] ) ) {
1047 return $this->mBidtable[$variant];
1048 }
1049 }
1050 return false;
1051 }
1052
1053 /**
1054 * Parse flags with syntax -{FLAG| ... }-
1055 * @private
1056 */
1057 function parseFlags() {
1058 $text = $this->mText;
1059 $flags = array();
1060 $variantFlags = array();
1061
1062 $sepPos = strpos( $text, '|' );
1063 if ( $sepPos !== false ) {
1064 $validFlags = $this->mConverter->mFlags;
1065 $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1066 foreach ( $f as $ff ) {
1067 $ff = trim( $ff );
1068 if ( isset( $validFlags[$ff] ) ) {
1069 $flags[$validFlags[$ff]] = true;
1070 }
1071 }
1072 $text = strval( substr( $text, $sepPos + 1 ) );
1073 }
1074
1075 if ( !$flags ) {
1076 $flags['S'] = true;
1077 } elseif ( isset( $flags['R'] ) ) {
1078 $flags = array( 'R' => true );// remove other flags
1079 } elseif ( isset( $flags['N'] ) ) {
1080 $flags = array( 'N' => true );// remove other flags
1081 } elseif ( isset( $flags['-'] ) ) {
1082 $flags = array( '-' => true );// remove other flags
1083 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1084 $flags['H'] = true;
1085 } elseif ( isset( $flags['H'] ) ) {
1086 // replace A flag, and remove other flags except T
1087 $temp = array( '+' => true, 'H' => true );
1088 if ( isset( $flags['T'] ) ) {
1089 $temp['T'] = true;
1090 }
1091 if ( isset( $flags['D'] ) ) {
1092 $temp['D'] = true;
1093 }
1094 $flags = $temp;
1095 } else {
1096 if ( isset( $flags['A'] ) ) {
1097 $flags['+'] = true;
1098 $flags['S'] = true;
1099 }
1100 if ( isset( $flags['D'] ) ) {
1101 unset( $flags['S'] );
1102 }
1103 // try to find flags like "zh-hans", "zh-hant"
1104 // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1105 $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1106 if ( $variantFlags ) {
1107 $variantFlags = array_flip( $variantFlags );
1108 $flags = array();
1109 }
1110 }
1111 $this->mVariantFlags = $variantFlags;
1112 $this->mRules = $text;
1113 $this->mFlags = $flags;
1114 }
1115
1116 /**
1117 * Generate conversion table.
1118 * @private
1119 */
1120 function parseRules() {
1121 $rules = $this->mRules;
1122 $flags = $this->mFlags;
1123 $bidtable = array();
1124 $unidtable = array();
1125 $variants = $this->mConverter->mVariants;
1126 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1127
1128 $choice = preg_split( $varsep_pattern, $rules );
1129
1130 foreach ( $choice as $c ) {
1131 $v = explode( ':', $c, 2 );
1132 if ( count( $v ) != 2 ) {
1133 // syntax error, skip
1134 continue;
1135 }
1136 $to = trim( $v[1] );
1137 $v = trim( $v[0] );
1138 $u = explode( '=>', $v, 2 );
1139 // if $to is empty, strtr() could return a wrong result
1140 if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1141 $bidtable[$v] = $to;
1142 } elseif ( count( $u ) == 2 ) {
1143 $from = trim( $u[0] );
1144 $v = trim( $u[1] );
1145 if ( array_key_exists( $v, $unidtable )
1146 && !is_array( $unidtable[$v] )
1147 && $to
1148 && in_array( $v, $variants ) ) {
1149 $unidtable[$v] = array( $from => $to );
1150 } elseif ( $to && in_array( $v, $variants ) ) {
1151 $unidtable[$v][$from] = $to;
1152 }
1153 }
1154 // syntax error, pass
1155 if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1156 $bidtable = array();
1157 $unidtable = array();
1158 break;
1159 }
1160 }
1161 $this->mBidtable = $bidtable;
1162 $this->mUnidtable = $unidtable;
1163 }
1164
1165 /**
1166 * @private
1167 */
1168 function getRulesDesc() {
1169 $codesep = $this->mConverter->mDescCodeSep;
1170 $varsep = $this->mConverter->mDescVarSep;
1171 $text = '';
1172 foreach ( $this->mBidtable as $k => $v ) {
1173 $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1174 }
1175 foreach ( $this->mUnidtable as $k => $a ) {
1176 foreach ( $a as $from => $to ) {
1177 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1178 "$codesep$to$varsep";
1179 }
1180 }
1181 return $text;
1182 }
1183
1184 /**
1185 * Parse rules conversion.
1186 * @private
1187 */
1188 function getRuleConvertedStr( $variant ) {
1189 $bidtable = $this->mBidtable;
1190 $unidtable = $this->mUnidtable;
1191
1192 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1193 return $this->mRules;
1194 } else {
1195 // display current variant in bidirectional array
1196 $disp = $this->getTextInBidtable( $variant );
1197 // or display current variant in fallbacks
1198 if ( !$disp ) {
1199 $disp = $this->getTextInBidtable(
1200 $this->mConverter->getVariantFallbacks( $variant ) );
1201 }
1202 // or display current variant in unidirectional array
1203 if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1204 $disp = array_values( $unidtable[$variant] );
1205 $disp = $disp[0];
1206 }
1207 // or display frist text under disable manual convert
1208 if ( !$disp
1209 && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1210 if ( count( $bidtable ) > 0 ) {
1211 $disp = array_values( $bidtable );
1212 $disp = $disp[0];
1213 } else {
1214 $disp = array_values( $unidtable );
1215 $disp = array_values( $disp[0] );
1216 $disp = $disp[0];
1217 }
1218 }
1219 return $disp;
1220 }
1221 }
1222
1223 /**
1224 * Generate conversion table for all text.
1225 * @private
1226 */
1227 function generateConvTable() {
1228 // Special case optimisation
1229 if ( !$this->mBidtable && !$this->mUnidtable ) {
1230 $this->mConvTable = array();
1231 return;
1232 }
1233
1234 $bidtable = $this->mBidtable;
1235 $unidtable = $this->mUnidtable;
1236 $manLevel = $this->mConverter->mManualLevel;
1237
1238 $vmarked = array();
1239 foreach ( $this->mConverter->mVariants as $v ) {
1240 /* for bidirectional array
1241 fill in the missing variants, if any,
1242 with fallbacks */
1243 if ( !isset( $bidtable[$v] ) ) {
1244 $variantFallbacks =
1245 $this->mConverter->getVariantFallbacks( $v );
1246 $vf = $this->getTextInBidtable( $variantFallbacks );
1247 if ( $vf ) {
1248 $bidtable[$v] = $vf;
1249 }
1250 }
1251
1252 if ( isset( $bidtable[$v] ) ) {
1253 foreach ( $vmarked as $vo ) {
1254 // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1255 // or -{H|zh:WordZh;zh-tw:WordTw}-
1256 // or -{-|zh:WordZh;zh-tw:WordTw}-
1257 // to introduce a custom mapping between
1258 // words WordZh and WordTw in the whole text
1259 if ( $manLevel[$v] == 'bidirectional' ) {
1260 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1261 }
1262 if ( $manLevel[$vo] == 'bidirectional' ) {
1263 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1264 }
1265 }
1266 $vmarked[] = $v;
1267 }
1268 /*for unidirectional array fill to convert tables */
1269 if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1270 && isset( $unidtable[$v] ) )
1271 {
1272 if ( isset( $this->mConvTable[$v] ) ) {
1273 $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1274 } else {
1275 $this->mConvTable[$v] = $unidtable[$v];
1276 }
1277 }
1278 }
1279 }
1280
1281 /**
1282 * Parse rules and flags.
1283 * @public
1284 */
1285 function parse( $variant = NULL ) {
1286 if ( !$variant ) {
1287 $variant = $this->mConverter->getPreferredVariant();
1288 }
1289
1290 $variants = $this->mConverter->mVariants;
1291 $this->parseFlags();
1292 $flags = $this->mFlags;
1293
1294 // convert to specified variant
1295 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1296 if ( $this->mVariantFlags ) {
1297 // check if current variant in flags
1298 if ( isset( $this->mVariantFlags[$variant] ) ) {
1299 // then convert <text to convert> to current language
1300 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1301 $variant );
1302 } else { // if current variant no in flags,
1303 // then we check its fallback variants.
1304 $variantFallbacks =
1305 $this->mConverter->getVariantFallbacks( $variant );
1306 foreach ( $variantFallbacks as $variantFallback ) {
1307 // if current variant's fallback exist in flags
1308 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1309 // then convert <text to convert> to fallback language
1310 $this->mRules =
1311 $this->mConverter->autoConvert( $this->mRules,
1312 $variantFallback );
1313 break;
1314 }
1315 }
1316 }
1317 $this->mFlags = $flags = array( 'R' => true );
1318 }
1319
1320 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1321 // decode => HTML entities modified by Sanitizer::removeHTMLtags
1322 $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1323 $this->parseRules();
1324 }
1325 $rules = $this->mRules;
1326
1327 if ( !$this->mBidtable && !$this->mUnidtable ) {
1328 if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1329 // fill all variants if text in -{A/H/-|text} without rules
1330 foreach ( $this->mConverter->mVariants as $v ) {
1331 $this->mBidtable[$v] = $rules;
1332 }
1333 } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1334 $this->mFlags = $flags = array( 'R' => true );
1335 }
1336 }
1337
1338 $this->mRuleDisplay = false;
1339 foreach ( $flags as $flag => $unused ) {
1340 switch ( $flag ) {
1341 case 'R':
1342 // if we don't do content convert, still strip the -{}- tags
1343 $this->mRuleDisplay = $rules;
1344 break;
1345 case 'N':
1346 // process N flag: output current variant name
1347 $ruleVar = trim( $rules );
1348 if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1349 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1350 } else {
1351 $this->mRuleDisplay = '';
1352 }
1353 break;
1354 case 'D':
1355 // process D flag: output rules description
1356 $this->mRuleDisplay = $this->getRulesDesc();
1357 break;
1358 case 'H':
1359 // process H,- flag or T only: output nothing
1360 $this->mRuleDisplay = '';
1361 break;
1362 case '-':
1363 $this->mRulesAction = 'remove';
1364 $this->mRuleDisplay = '';
1365 break;
1366 case '+':
1367 $this->mRulesAction = 'add';
1368 $this->mRuleDisplay = '';
1369 break;
1370 case 'S':
1371 $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1372 break;
1373 case 'T':
1374 $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1375 $this->mRuleDisplay = '';
1376 break;
1377 default:
1378 // ignore unknown flags (but see error case below)
1379 }
1380 }
1381 if ( $this->mRuleDisplay === false ) {
1382 $this->mRuleDisplay = $this->mManualCodeError;
1383 }
1384
1385 $this->generateConvTable();
1386 }
1387
1388 /**
1389 * @public
1390 */
1391 function hasRules() {
1392 // TODO:
1393 }
1394
1395 /**
1396 * Get display text on markup -{...}-
1397 * @public
1398 */
1399 function getDisplay() {
1400 return $this->mRuleDisplay;
1401 }
1402
1403 /**
1404 * Get converted title.
1405 * @public
1406 */
1407 function getTitle() {
1408 return $this->mRuleTitle;
1409 }
1410
1411 /**
1412 * Return how deal with conversion rules.
1413 * @public
1414 */
1415 function getRulesAction() {
1416 return $this->mRulesAction;
1417 }
1418
1419 /**
1420 * Get conversion table. ( bidirectional and unidirectional
1421 * conversion table )
1422 * @public
1423 */
1424 function getConvTable() {
1425 return $this->mConvTable;
1426 }
1427
1428 /**
1429 * Get conversion rules string.
1430 * @public
1431 */
1432 function getRules() {
1433 return $this->mRules;
1434 }
1435
1436 /**
1437 * Get conversion flags.
1438 * @public
1439 */
1440 function getFlags() {
1441 return $this->mFlags;
1442 }
1443 }