3 * Contains the LanguageConverter class and ConverterRule class
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
25 * Base class for language conversion.
28 * @author Zhengzhu Feng <zhengzhu@gmail.com>
29 * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
31 class LanguageConverter
{
34 * languages supporting variants
38 static public $languagesWithVariants = array(
50 public $mMainLanguageCode;
51 public $mVariants, $mVariantFallbacks, $mVariantNames;
52 public $mTablesLoaded = false;
54 // 'bidirectional' 'unidirectional' 'disable' for each variant
58 * @var String: memcached key name
64 public $mDescCodeSep = ':', $mDescVarSep = ';';
65 public $mUcfirst = false;
66 public $mConvRuleTitle = false;
69 public $mHeaderVariant;
70 public $mMaxDepth = 10;
71 public $mVarSeparatorPattern;
73 const CACHE_VERSION_KEY
= 'VERSION 7';
78 * @param $langobj Language: the Language Object
79 * @param $maincode String: the main language code of this language
80 * @param $variants Array: the supported variants of this language
81 * @param $variantfallbacks Array: the fallback language of each variant
82 * @param $flags Array: defining the custom strings that maps to the flags
83 * @param $manualLevel Array: limit for supported variants
85 public function __construct( $langobj, $maincode, $variants = array(),
86 $variantfallbacks = array(), $flags = array(),
87 $manualLevel = array() ) {
88 global $wgDisabledVariants;
89 $this->mLangObj
= $langobj;
90 $this->mMainLanguageCode
= $maincode;
91 $this->mVariants
= array_diff( $variants, $wgDisabledVariants );
92 $this->mVariantFallbacks
= $variantfallbacks;
93 $this->mVariantNames
= Language
::fetchLanguageNames();
94 $this->mCacheKey
= wfMemcKey( 'conversiontables', $maincode );
95 $defaultflags = array(
96 // 'S' show converted text
97 // '+' add rules for alltext
98 // 'E' the gave flags is error
99 // these flags above are reserved for program
100 'A' => 'A', // add rule for convert code (all text convert)
101 'T' => 'T', // title convert
102 'R' => 'R', // raw content
103 'D' => 'D', // convert description (subclass implement)
104 '-' => '-', // remove convert (not implement)
105 'H' => 'H', // add rule for convert code (but no display in placed code)
106 'N' => 'N' // current variant name
108 $this->mFlags
= array_merge( $defaultflags, $flags );
109 foreach ( $this->mVariants
as $v ) {
110 if ( array_key_exists( $v, $manualLevel ) ) {
111 $this->mManualLevel
[$v] = $manualLevel[$v];
113 $this->mManualLevel
[$v] = 'bidirectional';
115 $this->mFlags
[$v] = $v;
120 * Get all valid variants.
121 * Call this instead of using $this->mVariants directly.
123 * @return Array: contains all valid variants
125 public function getVariants() {
126 return $this->mVariants
;
130 * In case some variant is not defined in the markup, we need
131 * to have some fallback. For example, in zh, normally people
132 * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
133 * when zh-sg is preferred but not defined, we will pick zh-hans
134 * in this case. Right now this is only used by zh.
136 * @param $variant String: the language code of the variant
137 * @return String|array: The code of the fallback language or the
138 * main code if there is no fallback
140 public function getVariantFallbacks( $variant ) {
141 if ( isset( $this->mVariantFallbacks
[$variant] ) ) {
142 return $this->mVariantFallbacks
[$variant];
144 return $this->mMainLanguageCode
;
148 * Get the title produced by the conversion rule.
149 * @return String: The converted title text
151 public function getConvRuleTitle() {
152 return $this->mConvRuleTitle
;
156 * Get preferred language variant.
157 * @return String: the preferred language code
159 public function getPreferredVariant() {
160 global $wgDefaultLanguageVariant, $wgUser;
162 $req = $this->getURLVariant();
164 if ( $wgUser->isLoggedIn() && !$req ) {
165 $req = $this->getUserVariant();
167 $req = $this->getHeaderVariant();
170 if ( $wgDefaultLanguageVariant && !$req ) {
171 $req = $this->validateVariant( $wgDefaultLanguageVariant );
174 // This function, unlike the other get*Variant functions, is
175 // not memoized (i.e. there return value is not cached) since
176 // new information might appear during processing after this
178 if ( $this->validateVariant( $req ) ) {
181 return $this->mMainLanguageCode
;
185 * Get default variant.
186 * This function would not be affected by user's settings
187 * @return String: the default variant code
189 public function getDefaultVariant() {
190 global $wgDefaultLanguageVariant;
192 $req = $this->getURLVariant();
195 $req = $this->getHeaderVariant();
198 if ( $wgDefaultLanguageVariant && !$req ) {
199 $req = $this->validateVariant( $wgDefaultLanguageVariant );
205 return $this->mMainLanguageCode
;
209 * Validate the variant
210 * @param $variant String: the variant to validate
211 * @return Mixed: returns the variant if it is valid, null otherwise
213 public function validateVariant( $variant = null ) {
214 if ( $variant !== null && in_array( $variant, $this->mVariants
) ) {
221 * Get the variant specified in the URL
223 * @return Mixed: variant if one found, false otherwise.
225 public function getURLVariant() {
228 if ( $this->mURLVariant
) {
229 return $this->mURLVariant
;
232 // see if the preference is set in the request
233 $ret = $wgRequest->getText( 'variant' );
236 $ret = $wgRequest->getVal( 'uselang' );
239 $this->mURLVariant
= $this->validateVariant( $ret );
240 return $this->mURLVariant
;
244 * Determine if the user has a variant set.
246 * @return Mixed: variant if one found, false otherwise.
248 protected function getUserVariant() {
249 global $wgUser, $wgContLang;
251 // memoizing this function wreaks havoc on parserTest.php
253 if ( $this->mUserVariant ) {
254 return $this->mUserVariant;
258 // Get language variant preference from logged in users
259 // Don't call this on stub objects because that causes infinite
260 // recursion during initialisation
261 if ( $wgUser->isLoggedIn() ) {
262 if ( $this->mMainLanguageCode
== $wgContLang->getCode() ) {
263 $ret = $wgUser->getOption( 'variant' );
265 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode
);
268 // figure out user lang without constructing wgLang to avoid
269 // infinite recursion
270 $ret = $wgUser->getOption( 'language' );
273 $this->mUserVariant
= $this->validateVariant( $ret );
274 return $this->mUserVariant
;
278 * Determine the language variant from the Accept-Language header.
280 * @return Mixed: variant if one found, false otherwise.
282 protected function getHeaderVariant() {
285 if ( $this->mHeaderVariant
) {
286 return $this->mHeaderVariant
;
289 // see if some supported language variant is set in the
291 $languages = array_keys( $wgRequest->getAcceptLang() );
292 if ( empty( $languages ) ) {
296 $fallbackLanguages = array();
297 foreach ( $languages as $language ) {
298 $this->mHeaderVariant
= $this->validateVariant( $language );
299 if ( $this->mHeaderVariant
) {
303 // To see if there are fallbacks of current language.
304 // We record these fallback variants, and process
306 $fallbacks = $this->getVariantFallbacks( $language );
307 if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode
) {
308 $fallbackLanguages[] = $fallbacks;
309 } elseif ( is_array( $fallbacks ) ) {
311 array_merge( $fallbackLanguages, $fallbacks );
315 if ( !$this->mHeaderVariant
) {
316 // process fallback languages now
317 $fallback_languages = array_unique( $fallbackLanguages );
318 foreach ( $fallback_languages as $language ) {
319 $this->mHeaderVariant
= $this->validateVariant( $language );
320 if ( $this->mHeaderVariant
) {
326 return $this->mHeaderVariant
;
330 * Dictionary-based conversion.
331 * This function would not parse the conversion rules.
332 * If you want to parse rules, try to use convert() or
335 * @param $text String the text to be converted
336 * @param $toVariant bool|string the target language code
337 * @return String the converted text
339 public function autoConvert( $text, $toVariant = false ) {
340 wfProfileIn( __METHOD__
);
345 $toVariant = $this->getPreferredVariant();
347 wfProfileOut( __METHOD__
);
352 if ( $this->guessVariant( $text, $toVariant ) ) {
353 wfProfileOut( __METHOD__
);
357 /* we convert everything except:
358 1. HTML markups (anything between < and >)
360 3. placeholders created by the parser
363 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
364 $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
369 // this one is needed when the text is inside an HTML markup
370 $htmlfix = '|<[^>]+$|^[^<>]*>';
372 // disable convert to variants between <code> tags
373 $codefix = '<code>.+?<\/code>|';
374 // disable conversion of <script> tags
375 $scriptfix = '<script.*?>.*?<\/script>|';
376 // disable conversion of <pre> tags
377 $prefix = '<pre.*?>.*?<\/pre>|';
379 $reg = '/' . $codefix . $scriptfix . $prefix .
380 '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
385 // Guard against delimiter nulls in the input
386 $text = str_replace( "\000", '', $text );
388 $markupMatches = null;
389 $elementMatches = null;
390 while ( $startPos < strlen( $text ) ) {
391 if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE
, $startPos ) ) {
392 $elementPos = $markupMatches[0][1];
393 $element = $markupMatches[0][0];
395 $elementPos = strlen( $text );
399 // Queue the part before the markup for translation in a batch
400 $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
402 // Advance to the next position
403 $startPos = $elementPos +
strlen( $element );
405 // Translate any alt or title attributes inside the matched element
407 && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element, $elementMatches )
409 $attrs = Sanitizer
::decodeTagAttributes( $elementMatches[2] );
411 foreach ( array( 'title', 'alt' ) as $attrName ) {
412 if ( !isset( $attrs[$attrName] ) ) {
415 $attr = $attrs[$attrName];
416 // Don't convert URLs
417 if ( !strpos( $attr, '://' ) ) {
418 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
421 // Remove HTML tags to avoid disrupting the layout
422 $attr = preg_replace( '/<[^>]+>/', '', $attr );
423 if ( $attr !== $attrs[$attrName] ) {
424 $attrs[$attrName] = $attr;
429 $element = $elementMatches[1] . Html
::expandAttributes( $attrs ) .
433 $literalBlob .= $element . "\000";
436 // Do the main translation batch
437 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
439 // Put the output back together
440 $translatedIter = StringUtils
::explode( "\000", $translatedBlob );
441 $literalIter = StringUtils
::explode( "\000", $literalBlob );
443 while ( $translatedIter->valid() && $literalIter->valid() ) {
444 $output .= $translatedIter->current();
445 $output .= $literalIter->current();
446 $translatedIter->next();
447 $literalIter->next();
450 wfProfileOut( __METHOD__
);
455 * Translate a string to a variant.
456 * Doesn't parse rules or do any of that other stuff, for that use
457 * convert() or convertTo().
459 * @param $text String: text to convert
460 * @param $variant String: variant language code
461 * @return String: translated text
463 public function translate( $text, $variant ) {
464 wfProfileIn( __METHOD__
);
465 // If $text is empty or only includes spaces, do nothing
466 // Otherwise translate it
467 if ( trim( $text ) ) {
469 $text = $this->mTables
[$variant]->replace( $text );
471 wfProfileOut( __METHOD__
);
476 * Call translate() to convert text to all valid variants.
478 * @param $text String: the text to be converted
479 * @return Array: variant => converted text
481 public function autoConvertToAllVariants( $text ) {
482 wfProfileIn( __METHOD__
);
486 foreach ( $this->mVariants
as $variant ) {
487 $ret[$variant] = $this->translate( $text, $variant );
490 wfProfileOut( __METHOD__
);
495 * Apply manual conversion rules.
497 * @param $convRule ConverterRule Object of ConverterRule
499 protected function applyManualConv( $convRule ) {
500 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
502 // Bug 24072: $mConvRuleTitle was overwritten by other manual
503 // rule(s) not for title, this breaks the title conversion.
504 $newConvRuleTitle = $convRule->getTitle();
505 if ( $newConvRuleTitle ) {
506 // So I add an empty check for getTitle()
507 $this->mConvRuleTitle
= $newConvRuleTitle;
510 // merge/remove manual conversion rules to/from global table
511 $convTable = $convRule->getConvTable();
512 $action = $convRule->getRulesAction();
513 foreach ( $convTable as $variant => $pair ) {
514 if ( !$this->validateVariant( $variant ) ) {
518 if ( $action == 'add' ) {
519 foreach ( $pair as $from => $to ) {
520 // to ensure that $from and $to not be left blank
521 // so $this->translate() could always return a string
522 if ( $from ||
$to ) {
523 // more efficient than array_merge(), about 2.5 times.
524 $this->mTables
[$variant]->setPair( $from, $to );
527 } elseif ( $action == 'remove' ) {
528 $this->mTables
[$variant]->removeArray( $pair );
534 * Auto convert a Title object to a readable string in the
537 * @param $title Title a object of Title
538 * @return String: converted title text
540 public function convertTitle( $title ) {
541 $variant = $this->getPreferredVariant();
542 $index = $title->getNamespace();
543 if ( $index !== NS_MAIN
) {
544 $text = $this->convertNamespace( $index, $variant ) . ':';
548 $text .= $this->translate( $title->getText(), $variant );
553 * Get the namespace display name in the preferred variant.
555 * @param $index int namespace id
556 * @param $variant string|null variant code or null for preferred variant
557 * @return String: namespace name for display
559 public function convertNamespace( $index, $variant = null ) {
560 if ( $variant === null ) {
561 $variant = $this->getPreferredVariant();
563 if ( $index === NS_MAIN
) {
566 // First check if a message gives a converted name in the target variant.
567 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
568 if ( $nsConvMsg->exists() ) {
569 return $nsConvMsg->plain();
571 // Then check if a message gives a converted name in content language
572 // which needs extra translation to the target variant.
573 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
574 if ( $nsConvMsg->exists() ) {
575 return $this->translate( $nsConvMsg->plain(), $variant );
577 // No message exists, retrieve it from the target variant's namespace names.
578 $langObj = $this->mLangObj
->factory( $variant );
579 return $langObj->getFormattedNsText( $index );
584 * Convert text to different variants of a language. The automatic
585 * conversion is done in autoConvert(). Here we parse the text
586 * marked with -{}-, which specifies special conversions of the
587 * text that can not be accomplished in autoConvert().
589 * Syntax of the markup:
590 * -{code1:text1;code2:text2;...}- or
591 * -{flags|code1:text1;code2:text2;...}- or
592 * -{text}- in which case no conversion should take place for text
594 * @param $text String: text to be converted
595 * @return String: converted text
597 public function convert( $text ) {
598 $variant = $this->getPreferredVariant();
599 return $this->convertTo( $text, $variant );
603 * Same as convert() except a extra parameter to custom variant.
605 * @param $text String: text to be converted
606 * @param $variant String: the target variant code
607 * @return String: converted text
609 public function convertTo( $text, $variant ) {
610 global $wgDisableLangConversion;
611 if ( $wgDisableLangConversion ) {
614 // Reset converter state for a new converter run.
615 $this->mConvRuleTitle
= false;
616 return $this->recursiveConvertTopLevel( $text, $variant );
620 * Recursively convert text on the outside. Allow to use nested
621 * markups to custom rules.
623 * @param $text String: text to be converted
624 * @param $variant String: the target variant code
625 * @param $depth Integer: depth of recursion
626 * @return String: converted text
628 protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
631 $length = strlen( $text );
632 $shouldConvert = !$this->guessVariant( $text, $variant );
634 while ( $startPos < $length ) {
635 $pos = strpos( $text, '-{', $startPos );
637 if ( $pos === false ) {
638 // No more markup, append final segment
639 $fragment = substr( $text, $startPos );
640 $out .= $shouldConvert ?
$this->autoConvert( $fragment, $variant ) : $fragment;
645 // Append initial segment
646 $fragment = substr( $text, $startPos, $pos - $startPos );
647 $out .= $shouldConvert ?
$this->autoConvert( $fragment, $variant ) : $fragment;
652 // Do recursive conversion
653 $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth +
1 );
660 * Recursively convert text on the inside.
662 * @param $text String: text to be converted
663 * @param $variant String: the target variant code
664 * @param $startPos int
665 * @param $depth Integer: depth of recursion
667 * @throws MWException
668 * @return String: converted text
670 protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
671 // Quick sanity check (no function calls)
672 if ( $text[$startPos] !== '-' ||
$text[$startPos +
1] !== '{' ) {
673 throw new MWException( __METHOD__
. ': invalid input string' );
678 $warningDone = false;
679 $length = strlen( $text );
681 while ( $startPos < $length ) {
683 preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE
, $startPos );
693 // Append initial segment
694 $inner .= substr( $text, $startPos, $pos - $startPos );
702 if ( $depth >= $this->mMaxDepth
) {
704 if ( !$warningDone ) {
705 $inner .= '<span class="error">' .
706 wfMessage( 'language-converter-depth-warning' )
707 ->numParams( $this->mMaxDepth
)->inContentLanguage()->text() .
714 // Recursively parse another rule
715 $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth +
1 );
720 $rule = new ConverterRule( $inner, $this );
721 $rule->parse( $variant );
722 $this->applyManualConv( $rule );
723 return $rule->getDisplay();
725 throw new MWException( __METHOD__
. ': invalid regex match' );
730 if ( $startPos < $length ) {
731 $inner .= substr( $text, $startPos );
734 return '-{' . $this->autoConvert( $inner, $variant );
738 * If a language supports multiple variants, it is possible that
739 * non-existing link in one variant actually exists in another variant.
740 * This function tries to find it. See e.g. LanguageZh.php
742 * @param $link String: the name of the link
743 * @param $nt Mixed: the title object of the link
744 * @param $ignoreOtherCond Boolean: to disable other conditions when
745 * we need to transclude a template or update a category's link
746 * @return Null, the input parameters may be modified upon return
748 public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
749 # If the article has already existed, there is no need to
750 # check it again, otherwise it may cause a fault.
751 if ( is_object( $nt ) && $nt->exists() ) {
755 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
757 $isredir = $wgRequest->getText( 'redirect', 'yes' );
758 $action = $wgRequest->getText( 'action' );
759 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
760 $disableLinkConversion = $wgDisableLangConversion
761 ||
$wgDisableTitleConversion;
762 $linkBatch = new LinkBatch();
766 if ( $disableLinkConversion ||
767 ( !$ignoreOtherCond &&
770 ||
$action == 'submit'
771 ||
$linkconvert == 'no'
772 ||
$wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
776 if ( is_object( $nt ) ) {
777 $ns = $nt->getNamespace();
780 $variants = $this->autoConvertToAllVariants( $link );
781 if ( !$variants ) { // give up
787 foreach ( $variants as $v ) {
789 $varnt = Title
::newFromText( $v, $ns );
790 if ( !is_null( $varnt ) ) {
791 $linkBatch->addObj( $varnt );
797 // fetch all variants in single query
798 $linkBatch->execute();
800 foreach ( $titles as $varnt ) {
801 if ( $varnt->getArticleID() > 0 ) {
803 $link = $varnt->getText();
810 * Returns language specific hash options.
814 public function getExtraHashOptions() {
815 $variant = $this->getPreferredVariant();
816 return '!' . $variant;
820 * Guess if a text is written in a variant. This should be implemented in subclasses.
822 * @param string $text the text to be checked
823 * @param string $variant language code of the variant to be checked for
824 * @return bool true if $text appears to be written in $variant, false if not
826 * @author Nikola Smolenski <smolensk@eunet.rs>
829 public function guessVariant( $text, $variant ) {
834 * Load default conversion tables.
835 * This method must be implemented in derived class.
838 * @throws MWException
840 function loadDefaultTables() {
841 $name = get_class( $this );
842 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
846 * Load conversion tables either from the cache or the disk.
848 * @param $fromCache Boolean: load from memcached? Defaults to true.
850 function loadTables( $fromCache = true ) {
851 global $wgLangConvMemc;
853 if ( $this->mTablesLoaded
) {
857 wfProfileIn( __METHOD__
);
858 $this->mTablesLoaded
= true;
859 $this->mTables
= false;
861 wfProfileIn( __METHOD__
. '-cache' );
862 $this->mTables
= $wgLangConvMemc->get( $this->mCacheKey
);
863 wfProfileOut( __METHOD__
. '-cache' );
865 if ( !$this->mTables ||
!array_key_exists( self
::CACHE_VERSION_KEY
, $this->mTables
) ) {
866 wfProfileIn( __METHOD__
. '-recache' );
867 // not in cache, or we need a fresh reload.
868 // We will first load the default tables
869 // then update them using things in MediaWiki:Conversiontable/*
870 $this->loadDefaultTables();
871 foreach ( $this->mVariants
as $var ) {
872 $cached = $this->parseCachedTable( $var );
873 $this->mTables
[$var]->mergeArray( $cached );
876 $this->postLoadTables();
877 $this->mTables
[self
::CACHE_VERSION_KEY
] = true;
879 $wgLangConvMemc->set( $this->mCacheKey
, $this->mTables
, 43200 );
880 wfProfileOut( __METHOD__
. '-recache' );
882 wfProfileOut( __METHOD__
);
886 * Hook for post processing after conversion tables are loaded.
888 function postLoadTables() { }
891 * Reload the conversion tables.
895 function reloadTables() {
896 if ( $this->mTables
) {
897 unset( $this->mTables
);
899 $this->mTablesLoaded
= false;
900 $this->loadTables( false );
904 * Parse the conversion table stored in the cache.
906 * The tables should be in blocks of the following form:
913 * To make the tables more manageable, subpages are allowed
914 * and will be parsed recursively if $recursive == true.
916 * @param $code String: language code
917 * @param $subpage String: subpage name
918 * @param $recursive Boolean: parse subpages recursively? Defaults to true.
922 function parseCachedTable( $code, $subpage = '', $recursive = true ) {
923 static $parsed = array();
925 $key = 'Conversiontable/' . $code;
927 $key .= '/' . $subpage;
929 if ( array_key_exists( $key, $parsed ) ) {
933 $parsed[$key] = true;
935 if ( $subpage === '' ) {
936 $txt = MessageCache
::singleton()->getMsgFromNamespace( $key, $code );
939 $title = Title
::makeTitleSafe( NS_MEDIAWIKI
, $key );
940 if ( $title && $title->exists() ) {
941 $revision = Revision
::newFromTitle( $title );
943 if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT
) {
944 $txt = $revision->getContent( Revision
::RAW
)->getNativeData();
947 // @todo in the future, use a specialized content model, perhaps based on json!
952 # Nothing to parse if there's no text
953 if ( $txt === false ||
$txt === null ||
$txt === '' ) {
957 // get all subpage links of the form
958 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
959 $linkhead = $this->mLangObj
->getNsText( NS_MEDIAWIKI
) .
961 $subs = StringUtils
::explode( '[[', $txt );
963 foreach ( $subs as $sub ) {
964 $link = explode( ']]', $sub, 2 );
965 if ( count( $link ) != 2 ) {
968 $b = explode( '|', $link[0], 2 );
969 $b = explode( '/', trim( $b[0] ), 3 );
970 if ( count( $b ) == 3 ) {
976 if ( $b[0] == $linkhead && $b[1] == $code ) {
977 $sublinks[] = $sublink;
981 // parse the mappings in this page
982 $blocks = StringUtils
::explode( '-{', $txt );
985 foreach ( $blocks as $block ) {
987 // Skip the part before the first -{
991 $mappings = explode( '}-', $block, 2 );
992 $stripped = str_replace( array( "'", '"', '*', '#' ), '', $mappings[0] );
993 $table = StringUtils
::explode( ';', $stripped );
994 foreach ( $table as $t ) {
995 $m = explode( '=>', $t, 3 );
996 if ( count( $m ) != 2 ) {
999 // trim any trailling comments starting with '//'
1000 $tt = explode( '//', $m[1], 2 );
1001 $ret[trim( $m[0] )] = trim( $tt[0] );
1005 // recursively parse the subpages
1007 foreach ( $sublinks as $link ) {
1008 $s = $this->parseCachedTable( $code, $link, $recursive );
1009 $ret = array_merge( $ret, $s );
1013 if ( $this->mUcfirst
) {
1014 foreach ( $ret as $k => $v ) {
1015 $ret[$this->mLangObj
->ucfirst( $k )] = $this->mLangObj
->ucfirst( $v );
1022 * Enclose a string with the "no conversion" tag. This is used by
1023 * various functions in the Parser.
1025 * @param $text String: text to be tagged for no conversion
1026 * @param $noParse Boolean: unused
1027 * @return String: the tagged text
1029 public function markNoConversion( $text, $noParse = false ) {
1030 # don't mark if already marked
1031 if ( strpos( $text, '-{' ) ||
strpos( $text, '}-' ) ) {
1035 $ret = "-{R|$text}-";
1040 * Convert the sorting key for category links. This should make different
1041 * keys that are variants of each other map to the same key.
1043 * @param $key string
1047 function convertCategoryKey( $key ) {
1052 * Hook to refresh the cache of conversion tables when
1053 * MediaWiki:Conversiontable* is updated.
1056 * @param $page WikiPage object
1057 * @param $user Object: User object for the current user
1058 * @param $content Content: new page content
1059 * @param $summary String: edit summary of the edit
1060 * @param $isMinor Boolean: was the edit marked as minor?
1061 * @param $isWatch Boolean: did the user watch this page or not?
1063 * @param $flags int Bitfield
1064 * @param $revision Object: new Revision object or null
1065 * @return Boolean: true
1067 function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1068 $isWatch, $section, $flags, $revision ) {
1069 $titleobj = $page->getTitle();
1070 if ( $titleobj->getNamespace() == NS_MEDIAWIKI
) {
1071 $title = $titleobj->getDBkey();
1072 $t = explode( '/', $title, 3 );
1074 if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1075 if ( $this->validateVariant( $t[1] ) ) {
1076 $this->reloadTables();
1084 * Armour rendered math against conversion.
1085 * Escape special chars in parsed math text. (in most cases are img elements)
1087 * @param $text String: text to armour against conversion
1088 * @return String: armoured text where { and } have been converted to
1090 * @deprecated since 1.22 is no longer used
1092 public function armourMath( $text ) {
1093 // convert '-{' and '}-' to '-{' and '}-' to prevent
1094 // any unwanted markup appearing in the math image tag.
1095 $text = strtr( $text, array( '-{' => '-{', '}-' => '}-' ) );
1100 * Get the cached separator pattern for ConverterRule::parseRules()
1102 function getVarSeparatorPattern() {
1103 if ( is_null( $this->mVarSeparatorPattern
) ) {
1104 // varsep_pattern for preg_split:
1105 // text should be splited by ";" only if a valid variant
1106 // name exist after the markup, for example:
1107 // -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1108 // <span style="font-size:120%;">yyy</span>;}-
1109 // we should split it as:
1111 // [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1112 // [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1116 foreach ( $this->mVariants
as $variant ) {
1117 // zh-hans:xxx;zh-hant:yyy
1118 $pat .= $variant . '\s*:|';
1119 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1120 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1123 $this->mVarSeparatorPattern
= $pat;
1125 return $this->mVarSeparatorPattern
;
1130 * Parser for rules of language conversion , parse rules in -{ }- tag.
1132 * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1134 class ConverterRule
{
1135 public $mText; // original text in -{text}-
1136 public $mConverter; // LanguageConverter object
1137 public $mRuleDisplay = '';
1138 public $mRuleTitle = false;
1139 public $mRules = '';// string : the text of the rules
1140 public $mRulesAction = 'none';
1141 public $mFlags = array();
1142 public $mVariantFlags = array();
1143 public $mConvTable = array();
1144 public $mBidtable = array();// array of the translation in each variant
1145 public $mUnidtable = array();// array of the translation in each variant
1150 * @param $text String: the text between -{ and }-
1151 * @param $converter LanguageConverter object
1153 public function __construct( $text, $converter ) {
1154 $this->mText
= $text;
1155 $this->mConverter
= $converter;
1159 * Check if variants array in convert array.
1161 * @param $variants Array or string: variant language code
1162 * @return String: translated text
1164 public function getTextInBidtable( $variants ) {
1165 $variants = (array)$variants;
1169 foreach ( $variants as $variant ) {
1170 if ( isset( $this->mBidtable
[$variant] ) ) {
1171 return $this->mBidtable
[$variant];
1178 * Parse flags with syntax -{FLAG| ... }-
1181 function parseFlags() {
1182 $text = $this->mText
;
1184 $variantFlags = array();
1186 $sepPos = strpos( $text, '|' );
1187 if ( $sepPos !== false ) {
1188 $validFlags = $this->mConverter
->mFlags
;
1189 $f = StringUtils
::explode( ';', substr( $text, 0, $sepPos ) );
1190 foreach ( $f as $ff ) {
1192 if ( isset( $validFlags[$ff] ) ) {
1193 $flags[$validFlags[$ff]] = true;
1196 $text = strval( substr( $text, $sepPos +
1 ) );
1201 } elseif ( isset( $flags['R'] ) ) {
1202 $flags = array( 'R' => true );// remove other flags
1203 } elseif ( isset( $flags['N'] ) ) {
1204 $flags = array( 'N' => true );// remove other flags
1205 } elseif ( isset( $flags['-'] ) ) {
1206 $flags = array( '-' => true );// remove other flags
1207 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1209 } elseif ( isset( $flags['H'] ) ) {
1210 // replace A flag, and remove other flags except T
1211 $temp = array( '+' => true, 'H' => true );
1212 if ( isset( $flags['T'] ) ) {
1215 if ( isset( $flags['D'] ) ) {
1220 if ( isset( $flags['A'] ) ) {
1224 if ( isset( $flags['D'] ) ) {
1225 unset( $flags['S'] );
1227 // try to find flags like "zh-hans", "zh-hant"
1228 // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1229 $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter
->mVariants
);
1230 if ( $variantFlags ) {
1231 $variantFlags = array_flip( $variantFlags );
1235 $this->mVariantFlags
= $variantFlags;
1236 $this->mRules
= $text;
1237 $this->mFlags
= $flags;
1241 * Generate conversion table.
1244 function parseRules() {
1245 $rules = $this->mRules
;
1246 $bidtable = array();
1247 $unidtable = array();
1248 $variants = $this->mConverter
->mVariants
;
1249 $varsep_pattern = $this->mConverter
->getVarSeparatorPattern();
1251 // Split according to $varsep_pattern, but ignore semicolons from HTML entities
1252 $rules = preg_replace( '/(&[#a-zA-Z0-9]+);/', "$1\x01", $rules );
1253 $choice = preg_split( $varsep_pattern, $rules );
1254 $choice = str_replace( "\x01", ';', $choice );
1256 foreach ( $choice as $c ) {
1257 $v = explode( ':', $c, 2 );
1258 if ( count( $v ) != 2 ) {
1259 // syntax error, skip
1262 $to = trim( $v[1] );
1264 $u = explode( '=>', $v, 2 );
1265 // if $to is empty, strtr() could return a wrong result
1266 if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1267 $bidtable[$v] = $to;
1268 } elseif ( count( $u ) == 2 ) {
1269 $from = trim( $u[0] );
1271 if ( array_key_exists( $v, $unidtable )
1272 && !is_array( $unidtable[$v] )
1274 && in_array( $v, $variants ) ) {
1275 $unidtable[$v] = array( $from => $to );
1276 } elseif ( $to && in_array( $v, $variants ) ) {
1277 $unidtable[$v][$from] = $to;
1280 // syntax error, pass
1281 if ( !isset( $this->mConverter
->mVariantNames
[$v] ) ) {
1282 $bidtable = array();
1283 $unidtable = array();
1287 $this->mBidtable
= $bidtable;
1288 $this->mUnidtable
= $unidtable;
1296 function getRulesDesc() {
1297 $codesep = $this->mConverter
->mDescCodeSep
;
1298 $varsep = $this->mConverter
->mDescVarSep
;
1300 foreach ( $this->mBidtable
as $k => $v ) {
1301 $text .= $this->mConverter
->mVariantNames
[$k] . "$codesep$v$varsep";
1303 foreach ( $this->mUnidtable
as $k => $a ) {
1304 foreach ( $a as $from => $to ) {
1305 $text .= $from . '⇒' . $this->mConverter
->mVariantNames
[$k] .
1306 "$codesep$to$varsep";
1313 * Parse rules conversion.
1320 function getRuleConvertedStr( $variant ) {
1321 $bidtable = $this->mBidtable
;
1322 $unidtable = $this->mUnidtable
;
1324 if ( count( $bidtable ) +
count( $unidtable ) == 0 ) {
1325 return $this->mRules
;
1327 // display current variant in bidirectional array
1328 $disp = $this->getTextInBidtable( $variant );
1329 // or display current variant in fallbacks
1331 $disp = $this->getTextInBidtable(
1332 $this->mConverter
->getVariantFallbacks( $variant ) );
1334 // or display current variant in unidirectional array
1335 if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1336 $disp = array_values( $unidtable[$variant] );
1339 // or display frist text under disable manual convert
1340 if ( !$disp && $this->mConverter
->mManualLevel
[$variant] == 'disable' ) {
1341 if ( count( $bidtable ) > 0 ) {
1342 $disp = array_values( $bidtable );
1345 $disp = array_values( $unidtable );
1346 $disp = array_values( $disp[0] );
1355 * Similar to getRuleConvertedStr(), but this prefers to use original
1356 * page title if $variant === $this->mConverter->mMainLanguageCode
1357 * and may return false in this case (so this title conversion rule
1358 * will be ignored and the original title is shown).
1361 * @param $variant The variant code to display page title in
1362 * @return String|false The converted title or false if just page name
1364 function getRuleConvertedTitle( $variant ) {
1365 if ( $variant === $this->mConverter
->mMainLanguageCode
) {
1366 // If a string targeting exactly this variant is set,
1367 // use it. Otherwise, just return false, so the real
1368 // page name can be shown (and because variant === main,
1369 // there'll be no further automatic conversion).
1370 $disp = $this->getTextInBidtable( $variant );
1374 if ( array_key_exists( $variant, $this->mUnidtable
) ) {
1375 $disp = array_values( $this->mUnidtable
[$variant] );
1378 // Assigned above or still false.
1381 return $this->getRuleConvertedStr( $variant );
1386 * Generate conversion table for all text.
1389 function generateConvTable() {
1390 // Special case optimisation
1391 if ( !$this->mBidtable
&& !$this->mUnidtable
) {
1392 $this->mConvTable
= array();
1396 $bidtable = $this->mBidtable
;
1397 $unidtable = $this->mUnidtable
;
1398 $manLevel = $this->mConverter
->mManualLevel
;
1401 foreach ( $this->mConverter
->mVariants
as $v ) {
1402 /* for bidirectional array
1403 fill in the missing variants, if any,
1405 if ( !isset( $bidtable[$v] ) ) {
1407 $this->mConverter
->getVariantFallbacks( $v );
1408 $vf = $this->getTextInBidtable( $variantFallbacks );
1410 $bidtable[$v] = $vf;
1414 if ( isset( $bidtable[$v] ) ) {
1415 foreach ( $vmarked as $vo ) {
1416 // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1417 // or -{H|zh:WordZh;zh-tw:WordTw}-
1418 // or -{-|zh:WordZh;zh-tw:WordTw}-
1419 // to introduce a custom mapping between
1420 // words WordZh and WordTw in the whole text
1421 if ( $manLevel[$v] == 'bidirectional' ) {
1422 $this->mConvTable
[$v][$bidtable[$vo]] = $bidtable[$v];
1424 if ( $manLevel[$vo] == 'bidirectional' ) {
1425 $this->mConvTable
[$vo][$bidtable[$v]] = $bidtable[$vo];
1430 /* for unidirectional array fill to convert tables */
1431 if ( ( $manLevel[$v] == 'bidirectional' ||
$manLevel[$v] == 'unidirectional' )
1432 && isset( $unidtable[$v] )
1434 if ( isset( $this->mConvTable
[$v] ) ) {
1435 $this->mConvTable
[$v] = array_merge( $this->mConvTable
[$v], $unidtable[$v] );
1437 $this->mConvTable
[$v] = $unidtable[$v];
1444 * Parse rules and flags.
1445 * @param $variant String: variant language code
1447 public function parse( $variant = null ) {
1449 $variant = $this->mConverter
->getPreferredVariant();
1452 $this->parseFlags();
1453 $flags = $this->mFlags
;
1455 // convert to specified variant
1456 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1457 if ( $this->mVariantFlags
) {
1458 // check if current variant in flags
1459 if ( isset( $this->mVariantFlags
[$variant] ) ) {
1460 // then convert <text to convert> to current language
1461 $this->mRules
= $this->mConverter
->autoConvert( $this->mRules
,
1464 // if current variant no in flags,
1465 // then we check its fallback variants.
1467 $this->mConverter
->getVariantFallbacks( $variant );
1468 if ( is_array( $variantFallbacks ) ) {
1469 foreach ( $variantFallbacks as $variantFallback ) {
1470 // if current variant's fallback exist in flags
1471 if ( isset( $this->mVariantFlags
[$variantFallback] ) ) {
1472 // then convert <text to convert> to fallback language
1474 $this->mConverter
->autoConvert( $this->mRules
,
1481 $this->mFlags
= $flags = array( 'R' => true );
1484 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1485 // decode => HTML entities modified by Sanitizer::removeHTMLtags
1486 $this->mRules
= str_replace( '=>', '=>', $this->mRules
);
1487 $this->parseRules();
1489 $rules = $this->mRules
;
1491 if ( !$this->mBidtable
&& !$this->mUnidtable
) {
1492 if ( isset( $flags['+'] ) ||
isset( $flags['-'] ) ) {
1493 // fill all variants if text in -{A/H/-|text} without rules
1494 foreach ( $this->mConverter
->mVariants
as $v ) {
1495 $this->mBidtable
[$v] = $rules;
1497 } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1498 $this->mFlags
= $flags = array( 'R' => true );
1502 $this->mRuleDisplay
= false;
1503 foreach ( $flags as $flag => $unused ) {
1506 // if we don't do content convert, still strip the -{}- tags
1507 $this->mRuleDisplay
= $rules;
1510 // process N flag: output current variant name
1511 $ruleVar = trim( $rules );
1512 if ( isset( $this->mConverter
->mVariantNames
[$ruleVar] ) ) {
1513 $this->mRuleDisplay
= $this->mConverter
->mVariantNames
[$ruleVar];
1515 $this->mRuleDisplay
= '';
1519 // process D flag: output rules description
1520 $this->mRuleDisplay
= $this->getRulesDesc();
1523 // process H,- flag or T only: output nothing
1524 $this->mRuleDisplay
= '';
1527 $this->mRulesAction
= 'remove';
1528 $this->mRuleDisplay
= '';
1531 $this->mRulesAction
= 'add';
1532 $this->mRuleDisplay
= '';
1535 $this->mRuleDisplay
= $this->getRuleConvertedStr( $variant );
1538 $this->mRuleTitle
= $this->getRuleConvertedTitle( $variant );
1539 $this->mRuleDisplay
= '';
1542 // ignore unknown flags (but see error case below)
1545 if ( $this->mRuleDisplay
=== false ) {
1546 $this->mRuleDisplay
= '<span class="error">'
1547 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
1551 $this->generateConvTable();
1555 * @todo FIXME: code this function :)
1557 public function hasRules() {
1562 * Get display text on markup -{...}-
1565 public function getDisplay() {
1566 return $this->mRuleDisplay
;
1570 * Get converted title.
1573 public function getTitle() {
1574 return $this->mRuleTitle
;
1578 * Return how deal with conversion rules.
1581 public function getRulesAction() {
1582 return $this->mRulesAction
;
1586 * Get conversion table. (bidirectional and unidirectional
1590 public function getConvTable() {
1591 return $this->mConvTable
;
1595 * Get conversion rules string.
1598 public function getRules() {
1599 return $this->mRules
;
1603 * Get conversion flags.
1606 public function getFlags() {
1607 return $this->mFlags
;