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 * Convert link text to all valid variants.
496 * In the first, this function only convert text outside the
497 * "-{" "}-" markups. Since the "{" and "}" are not allowed in
498 * titles, the text will get all converted always.
499 * So I removed this feature and deprecated the function.
501 * @param $text String: the text to be converted
502 * @return Array: variant => converted text
503 * @deprecated since 1.17 Use autoConvertToAllVariants() instead
505 public function convertLinkToAllVariants( $text ) {
506 return $this->autoConvertToAllVariants( $text );
510 * Apply manual conversion rules.
512 * @param $convRule ConverterRule Object of ConverterRule
514 protected function applyManualConv( $convRule ) {
515 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
517 // Bug 24072: $mConvRuleTitle was overwritten by other manual
518 // rule(s) not for title, this breaks the title conversion.
519 $newConvRuleTitle = $convRule->getTitle();
520 if ( $newConvRuleTitle ) {
521 // So I add an empty check for getTitle()
522 $this->mConvRuleTitle
= $newConvRuleTitle;
525 // merge/remove manual conversion rules to/from global table
526 $convTable = $convRule->getConvTable();
527 $action = $convRule->getRulesAction();
528 foreach ( $convTable as $variant => $pair ) {
529 if ( !$this->validateVariant( $variant ) ) {
533 if ( $action == 'add' ) {
534 foreach ( $pair as $from => $to ) {
535 // to ensure that $from and $to not be left blank
536 // so $this->translate() could always return a string
537 if ( $from ||
$to ) {
538 // more efficient than array_merge(), about 2.5 times.
539 $this->mTables
[$variant]->setPair( $from, $to );
542 } elseif ( $action == 'remove' ) {
543 $this->mTables
[$variant]->removeArray( $pair );
549 * Auto convert a Title object to a readable string in the
552 * @param $title Title a object of Title
553 * @return String: converted title text
555 public function convertTitle( $title ) {
556 $variant = $this->getPreferredVariant();
557 $index = $title->getNamespace();
558 if ( $index !== NS_MAIN
) {
559 $text = $this->convertNamespace( $index, $variant ) . ':';
563 $text .= $this->translate( $title->getText(), $variant );
568 * Get the namespace display name in the preferred variant.
570 * @param $index int namespace id
571 * @param $variant string|null variant code or null for preferred variant
572 * @return String: namespace name for display
574 public function convertNamespace( $index, $variant = null ) {
575 if ( $variant === null ) {
576 $variant = $this->getPreferredVariant();
578 if ( $index === NS_MAIN
) {
581 // First check if a message gives a converted name in the target variant.
582 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
583 if ( $nsConvMsg->exists() ) {
584 return $nsConvMsg->plain();
586 // Then check if a message gives a converted name in content language
587 // which needs extra translation to the target variant.
588 $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
589 if ( $nsConvMsg->exists() ) {
590 return $this->translate( $nsConvMsg->plain(), $variant );
592 // No message exists, retrieve it from the target variant's namespace names.
593 $langObj = $this->mLangObj
->factory( $variant );
594 return $langObj->getFormattedNsText( $index );
599 * Convert text to different variants of a language. The automatic
600 * conversion is done in autoConvert(). Here we parse the text
601 * marked with -{}-, which specifies special conversions of the
602 * text that can not be accomplished in autoConvert().
604 * Syntax of the markup:
605 * -{code1:text1;code2:text2;...}- or
606 * -{flags|code1:text1;code2:text2;...}- or
607 * -{text}- in which case no conversion should take place for text
609 * @param $text String: text to be converted
610 * @return String: converted text
612 public function convert( $text ) {
613 $variant = $this->getPreferredVariant();
614 return $this->convertTo( $text, $variant );
618 * Same as convert() except a extra parameter to custom variant.
620 * @param $text String: text to be converted
621 * @param $variant String: the target variant code
622 * @return String: converted text
624 public function convertTo( $text, $variant ) {
625 global $wgDisableLangConversion;
626 if ( $wgDisableLangConversion ) {
629 // Reset converter state for a new converter run.
630 $this->mConvRuleTitle
= false;
631 return $this->recursiveConvertTopLevel( $text, $variant );
635 * Recursively convert text on the outside. Allow to use nested
636 * markups to custom rules.
638 * @param $text String: text to be converted
639 * @param $variant String: the target variant code
640 * @param $depth Integer: depth of recursion
641 * @return String: converted text
643 protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
646 $length = strlen( $text );
647 $shouldConvert = !$this->guessVariant( $text, $variant );
649 while ( $startPos < $length ) {
650 $pos = strpos( $text, '-{', $startPos );
652 if ( $pos === false ) {
653 // No more markup, append final segment
654 $fragment = substr( $text, $startPos );
655 $out .= $shouldConvert ?
$this->autoConvert( $fragment, $variant ) : $fragment;
660 // Append initial segment
661 $fragment = substr( $text, $startPos, $pos - $startPos );
662 $out .= $shouldConvert ?
$this->autoConvert( $fragment, $variant ) : $fragment;
667 // Do recursive conversion
668 $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth +
1 );
675 * Recursively convert text on the inside.
677 * @param $text String: text to be converted
678 * @param $variant String: the target variant code
679 * @param $startPos int
680 * @param $depth Integer: depth of recursion
682 * @throws MWException
683 * @return String: converted text
685 protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
686 // Quick sanity check (no function calls)
687 if ( $text[$startPos] !== '-' ||
$text[$startPos +
1] !== '{' ) {
688 throw new MWException( __METHOD__
. ': invalid input string' );
693 $warningDone = false;
694 $length = strlen( $text );
696 while ( $startPos < $length ) {
698 preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE
, $startPos );
708 // Append initial segment
709 $inner .= substr( $text, $startPos, $pos - $startPos );
717 if ( $depth >= $this->mMaxDepth
) {
719 if ( !$warningDone ) {
720 $inner .= '<span class="error">' .
721 wfMessage( 'language-converter-depth-warning' )
722 ->numParams( $this->mMaxDepth
)->inContentLanguage()->text() .
729 // Recursively parse another rule
730 $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth +
1 );
735 $rule = new ConverterRule( $inner, $this );
736 $rule->parse( $variant );
737 $this->applyManualConv( $rule );
738 return $rule->getDisplay();
740 throw new MWException( __METHOD__
. ': invalid regex match' );
745 if ( $startPos < $length ) {
746 $inner .= substr( $text, $startPos );
749 return '-{' . $this->autoConvert( $inner, $variant );
753 * If a language supports multiple variants, it is possible that
754 * non-existing link in one variant actually exists in another variant.
755 * This function tries to find it. See e.g. LanguageZh.php
757 * @param $link String: the name of the link
758 * @param $nt Mixed: the title object of the link
759 * @param $ignoreOtherCond Boolean: to disable other conditions when
760 * we need to transclude a template or update a category's link
761 * @return Null, the input parameters may be modified upon return
763 public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
764 # If the article has already existed, there is no need to
765 # check it again, otherwise it may cause a fault.
766 if ( is_object( $nt ) && $nt->exists() ) {
770 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
772 $isredir = $wgRequest->getText( 'redirect', 'yes' );
773 $action = $wgRequest->getText( 'action' );
774 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
775 $disableLinkConversion = $wgDisableLangConversion
776 ||
$wgDisableTitleConversion;
777 $linkBatch = new LinkBatch();
781 if ( $disableLinkConversion ||
782 ( !$ignoreOtherCond &&
785 ||
$action == 'submit'
786 ||
$linkconvert == 'no'
787 ||
$wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
791 if ( is_object( $nt ) ) {
792 $ns = $nt->getNamespace();
795 $variants = $this->autoConvertToAllVariants( $link );
796 if ( !$variants ) { // give up
802 foreach ( $variants as $v ) {
804 $varnt = Title
::newFromText( $v, $ns );
805 if ( !is_null( $varnt ) ) {
806 $linkBatch->addObj( $varnt );
812 // fetch all variants in single query
813 $linkBatch->execute();
815 foreach ( $titles as $varnt ) {
816 if ( $varnt->getArticleID() > 0 ) {
818 $link = $varnt->getText();
825 * Returns language specific hash options.
829 public function getExtraHashOptions() {
830 $variant = $this->getPreferredVariant();
831 return '!' . $variant;
835 * Guess if a text is written in a variant. This should be implemented in subclasses.
837 * @param string $text the text to be checked
838 * @param string $variant language code of the variant to be checked for
839 * @return bool true if $text appears to be written in $variant, false if not
841 * @author Nikola Smolenski <smolensk@eunet.rs>
844 public function guessVariant( $text, $variant ) {
849 * Load default conversion tables.
850 * This method must be implemented in derived class.
853 * @throws MWException
855 function loadDefaultTables() {
856 $name = get_class( $this );
857 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
861 * Load conversion tables either from the cache or the disk.
863 * @param $fromCache Boolean: load from memcached? Defaults to true.
865 function loadTables( $fromCache = true ) {
866 global $wgLangConvMemc;
868 if ( $this->mTablesLoaded
) {
872 wfProfileIn( __METHOD__
);
873 $this->mTablesLoaded
= true;
874 $this->mTables
= false;
876 wfProfileIn( __METHOD__
. '-cache' );
877 $this->mTables
= $wgLangConvMemc->get( $this->mCacheKey
);
878 wfProfileOut( __METHOD__
. '-cache' );
880 if ( !$this->mTables ||
!array_key_exists( self
::CACHE_VERSION_KEY
, $this->mTables
) ) {
881 wfProfileIn( __METHOD__
. '-recache' );
882 // not in cache, or we need a fresh reload.
883 // We will first load the default tables
884 // then update them using things in MediaWiki:Conversiontable/*
885 $this->loadDefaultTables();
886 foreach ( $this->mVariants
as $var ) {
887 $cached = $this->parseCachedTable( $var );
888 $this->mTables
[$var]->mergeArray( $cached );
891 $this->postLoadTables();
892 $this->mTables
[self
::CACHE_VERSION_KEY
] = true;
894 $wgLangConvMemc->set( $this->mCacheKey
, $this->mTables
, 43200 );
895 wfProfileOut( __METHOD__
. '-recache' );
897 wfProfileOut( __METHOD__
);
901 * Hook for post processing after conversion tables are loaded.
903 function postLoadTables() { }
906 * Reload the conversion tables.
910 function reloadTables() {
911 if ( $this->mTables
) {
912 unset( $this->mTables
);
914 $this->mTablesLoaded
= false;
915 $this->loadTables( false );
919 * Parse the conversion table stored in the cache.
921 * The tables should be in blocks of the following form:
928 * To make the tables more manageable, subpages are allowed
929 * and will be parsed recursively if $recursive == true.
931 * @param $code String: language code
932 * @param $subpage String: subpage name
933 * @param $recursive Boolean: parse subpages recursively? Defaults to true.
937 function parseCachedTable( $code, $subpage = '', $recursive = true ) {
938 static $parsed = array();
940 $key = 'Conversiontable/' . $code;
942 $key .= '/' . $subpage;
944 if ( array_key_exists( $key, $parsed ) ) {
948 $parsed[$key] = true;
950 if ( $subpage === '' ) {
951 $txt = MessageCache
::singleton()->getMsgFromNamespace( $key, $code );
954 $title = Title
::makeTitleSafe( NS_MEDIAWIKI
, $key );
955 if ( $title && $title->exists() ) {
956 $revision = Revision
::newFromTitle( $title );
958 if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT
) {
959 $txt = $revision->getContent( Revision
::RAW
)->getNativeData();
962 // @todo in the future, use a specialized content model, perhaps based on json!
967 # Nothing to parse if there's no text
968 if ( $txt === false ||
$txt === null ||
$txt === '' ) {
972 // get all subpage links of the form
973 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
974 $linkhead = $this->mLangObj
->getNsText( NS_MEDIAWIKI
) .
976 $subs = StringUtils
::explode( '[[', $txt );
978 foreach ( $subs as $sub ) {
979 $link = explode( ']]', $sub, 2 );
980 if ( count( $link ) != 2 ) {
983 $b = explode( '|', $link[0], 2 );
984 $b = explode( '/', trim( $b[0] ), 3 );
985 if ( count( $b ) == 3 ) {
991 if ( $b[0] == $linkhead && $b[1] == $code ) {
992 $sublinks[] = $sublink;
996 // parse the mappings in this page
997 $blocks = StringUtils
::explode( '-{', $txt );
1000 foreach ( $blocks as $block ) {
1002 // Skip the part before the first -{
1006 $mappings = explode( '}-', $block, 2 );
1007 $stripped = str_replace( array( "'", '"', '*', '#' ), '', $mappings[0] );
1008 $table = StringUtils
::explode( ';', $stripped );
1009 foreach ( $table as $t ) {
1010 $m = explode( '=>', $t, 3 );
1011 if ( count( $m ) != 2 ) {
1014 // trim any trailling comments starting with '//'
1015 $tt = explode( '//', $m[1], 2 );
1016 $ret[trim( $m[0] )] = trim( $tt[0] );
1020 // recursively parse the subpages
1022 foreach ( $sublinks as $link ) {
1023 $s = $this->parseCachedTable( $code, $link, $recursive );
1024 $ret = array_merge( $ret, $s );
1028 if ( $this->mUcfirst
) {
1029 foreach ( $ret as $k => $v ) {
1030 $ret[$this->mLangObj
->ucfirst( $k )] = $this->mLangObj
->ucfirst( $v );
1037 * Enclose a string with the "no conversion" tag. This is used by
1038 * various functions in the Parser.
1040 * @param $text String: text to be tagged for no conversion
1041 * @param $noParse Boolean: unused
1042 * @return String: the tagged text
1044 public function markNoConversion( $text, $noParse = false ) {
1045 # don't mark if already marked
1046 if ( strpos( $text, '-{' ) ||
strpos( $text, '}-' ) ) {
1050 $ret = "-{R|$text}-";
1055 * Convert the sorting key for category links. This should make different
1056 * keys that are variants of each other map to the same key.
1058 * @param $key string
1062 function convertCategoryKey( $key ) {
1067 * Hook to refresh the cache of conversion tables when
1068 * MediaWiki:Conversiontable* is updated.
1071 * @param $page WikiPage object
1072 * @param $user Object: User object for the current user
1073 * @param $content Content: new page content
1074 * @param $summary String: edit summary of the edit
1075 * @param $isMinor Boolean: was the edit marked as minor?
1076 * @param $isWatch Boolean: did the user watch this page or not?
1078 * @param $flags int Bitfield
1079 * @param $revision Object: new Revision object or null
1080 * @return Boolean: true
1082 function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1083 $isWatch, $section, $flags, $revision ) {
1084 $titleobj = $page->getTitle();
1085 if ( $titleobj->getNamespace() == NS_MEDIAWIKI
) {
1086 $title = $titleobj->getDBkey();
1087 $t = explode( '/', $title, 3 );
1089 if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1090 if ( $this->validateVariant( $t[1] ) ) {
1091 $this->reloadTables();
1099 * Armour rendered math against conversion.
1100 * Escape special chars in parsed math text. (in most cases are img elements)
1102 * @param $text String: text to armour against conversion
1103 * @return String: armoured text where { and } have been converted to
1105 * @deprecated since 1.22 is no longer used
1107 public function armourMath( $text ) {
1108 // convert '-{' and '}-' to '-{' and '}-' to prevent
1109 // any unwanted markup appearing in the math image tag.
1110 $text = strtr( $text, array( '-{' => '-{', '}-' => '}-' ) );
1115 * Get the cached separator pattern for ConverterRule::parseRules()
1117 function getVarSeparatorPattern() {
1118 if ( is_null( $this->mVarSeparatorPattern
) ) {
1119 // varsep_pattern for preg_split:
1120 // text should be splited by ";" only if a valid variant
1121 // name exist after the markup, for example:
1122 // -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1123 // <span style="font-size:120%;">yyy</span>;}-
1124 // we should split it as:
1126 // [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1127 // [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1131 foreach ( $this->mVariants
as $variant ) {
1132 // zh-hans:xxx;zh-hant:yyy
1133 $pat .= $variant . '\s*:|';
1134 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1135 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1138 $this->mVarSeparatorPattern
= $pat;
1140 return $this->mVarSeparatorPattern
;
1145 * Parser for rules of language conversion , parse rules in -{ }- tag.
1147 * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1149 class ConverterRule
{
1150 public $mText; // original text in -{text}-
1151 public $mConverter; // LanguageConverter object
1152 public $mRuleDisplay = '';
1153 public $mRuleTitle = false;
1154 public $mRules = '';// string : the text of the rules
1155 public $mRulesAction = 'none';
1156 public $mFlags = array();
1157 public $mVariantFlags = array();
1158 public $mConvTable = array();
1159 public $mBidtable = array();// array of the translation in each variant
1160 public $mUnidtable = array();// array of the translation in each variant
1165 * @param $text String: the text between -{ and }-
1166 * @param $converter LanguageConverter object
1168 public function __construct( $text, $converter ) {
1169 $this->mText
= $text;
1170 $this->mConverter
= $converter;
1174 * Check if variants array in convert array.
1176 * @param $variants Array or string: variant language code
1177 * @return String: translated text
1179 public function getTextInBidtable( $variants ) {
1180 $variants = (array)$variants;
1184 foreach ( $variants as $variant ) {
1185 if ( isset( $this->mBidtable
[$variant] ) ) {
1186 return $this->mBidtable
[$variant];
1193 * Parse flags with syntax -{FLAG| ... }-
1196 function parseFlags() {
1197 $text = $this->mText
;
1199 $variantFlags = array();
1201 $sepPos = strpos( $text, '|' );
1202 if ( $sepPos !== false ) {
1203 $validFlags = $this->mConverter
->mFlags
;
1204 $f = StringUtils
::explode( ';', substr( $text, 0, $sepPos ) );
1205 foreach ( $f as $ff ) {
1207 if ( isset( $validFlags[$ff] ) ) {
1208 $flags[$validFlags[$ff]] = true;
1211 $text = strval( substr( $text, $sepPos +
1 ) );
1216 } elseif ( isset( $flags['R'] ) ) {
1217 $flags = array( 'R' => true );// remove other flags
1218 } elseif ( isset( $flags['N'] ) ) {
1219 $flags = array( 'N' => true );// remove other flags
1220 } elseif ( isset( $flags['-'] ) ) {
1221 $flags = array( '-' => true );// remove other flags
1222 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1224 } elseif ( isset( $flags['H'] ) ) {
1225 // replace A flag, and remove other flags except T
1226 $temp = array( '+' => true, 'H' => true );
1227 if ( isset( $flags['T'] ) ) {
1230 if ( isset( $flags['D'] ) ) {
1235 if ( isset( $flags['A'] ) ) {
1239 if ( isset( $flags['D'] ) ) {
1240 unset( $flags['S'] );
1242 // try to find flags like "zh-hans", "zh-hant"
1243 // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1244 $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter
->mVariants
);
1245 if ( $variantFlags ) {
1246 $variantFlags = array_flip( $variantFlags );
1250 $this->mVariantFlags
= $variantFlags;
1251 $this->mRules
= $text;
1252 $this->mFlags
= $flags;
1256 * Generate conversion table.
1259 function parseRules() {
1260 $rules = $this->mRules
;
1261 $bidtable = array();
1262 $unidtable = array();
1263 $variants = $this->mConverter
->mVariants
;
1264 $varsep_pattern = $this->mConverter
->getVarSeparatorPattern();
1266 // Split according to $varsep_pattern, but ignore semicolons from HTML entities
1267 $rules = preg_replace( '/(&[#a-zA-Z0-9]+);/', "$1\x01", $rules );
1268 $choice = preg_split( $varsep_pattern, $rules );
1269 $choice = str_replace( "\x01", ';', $choice );
1271 foreach ( $choice as $c ) {
1272 $v = explode( ':', $c, 2 );
1273 if ( count( $v ) != 2 ) {
1274 // syntax error, skip
1277 $to = trim( $v[1] );
1279 $u = explode( '=>', $v, 2 );
1280 // if $to is empty, strtr() could return a wrong result
1281 if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1282 $bidtable[$v] = $to;
1283 } elseif ( count( $u ) == 2 ) {
1284 $from = trim( $u[0] );
1286 if ( array_key_exists( $v, $unidtable )
1287 && !is_array( $unidtable[$v] )
1289 && in_array( $v, $variants ) ) {
1290 $unidtable[$v] = array( $from => $to );
1291 } elseif ( $to && in_array( $v, $variants ) ) {
1292 $unidtable[$v][$from] = $to;
1295 // syntax error, pass
1296 if ( !isset( $this->mConverter
->mVariantNames
[$v] ) ) {
1297 $bidtable = array();
1298 $unidtable = array();
1302 $this->mBidtable
= $bidtable;
1303 $this->mUnidtable
= $unidtable;
1311 function getRulesDesc() {
1312 $codesep = $this->mConverter
->mDescCodeSep
;
1313 $varsep = $this->mConverter
->mDescVarSep
;
1315 foreach ( $this->mBidtable
as $k => $v ) {
1316 $text .= $this->mConverter
->mVariantNames
[$k] . "$codesep$v$varsep";
1318 foreach ( $this->mUnidtable
as $k => $a ) {
1319 foreach ( $a as $from => $to ) {
1320 $text .= $from . '⇒' . $this->mConverter
->mVariantNames
[$k] .
1321 "$codesep$to$varsep";
1328 * Parse rules conversion.
1335 function getRuleConvertedStr( $variant ) {
1336 $bidtable = $this->mBidtable
;
1337 $unidtable = $this->mUnidtable
;
1339 if ( count( $bidtable ) +
count( $unidtable ) == 0 ) {
1340 return $this->mRules
;
1342 // display current variant in bidirectional array
1343 $disp = $this->getTextInBidtable( $variant );
1344 // or display current variant in fallbacks
1346 $disp = $this->getTextInBidtable(
1347 $this->mConverter
->getVariantFallbacks( $variant ) );
1349 // or display current variant in unidirectional array
1350 if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1351 $disp = array_values( $unidtable[$variant] );
1354 // or display frist text under disable manual convert
1355 if ( !$disp && $this->mConverter
->mManualLevel
[$variant] == 'disable' ) {
1356 if ( count( $bidtable ) > 0 ) {
1357 $disp = array_values( $bidtable );
1360 $disp = array_values( $unidtable );
1361 $disp = array_values( $disp[0] );
1370 * Similar to getRuleConvertedStr(), but this prefers to use original
1371 * page title if $variant === $this->mConverter->mMainLanguageCode
1372 * and may return false in this case (so this title conversion rule
1373 * will be ignored and the original title is shown).
1376 * @param $variant The variant code to display page title in
1377 * @return String|false The converted title or false if just page name
1379 function getRuleConvertedTitle( $variant ) {
1380 if ( $variant === $this->mConverter
->mMainLanguageCode
) {
1381 // If a string targeting exactly this variant is set,
1382 // use it. Otherwise, just return false, so the real
1383 // page name can be shown (and because variant === main,
1384 // there'll be no further automatic conversion).
1385 $disp = $this->getTextInBidtable( $variant );
1389 if ( array_key_exists( $variant, $this->mUnidtable
) ) {
1390 $disp = array_values( $this->mUnidtable
[$variant] );
1393 // Assigned above or still false.
1396 return $this->getRuleConvertedStr( $variant );
1401 * Generate conversion table for all text.
1404 function generateConvTable() {
1405 // Special case optimisation
1406 if ( !$this->mBidtable
&& !$this->mUnidtable
) {
1407 $this->mConvTable
= array();
1411 $bidtable = $this->mBidtable
;
1412 $unidtable = $this->mUnidtable
;
1413 $manLevel = $this->mConverter
->mManualLevel
;
1416 foreach ( $this->mConverter
->mVariants
as $v ) {
1417 /* for bidirectional array
1418 fill in the missing variants, if any,
1420 if ( !isset( $bidtable[$v] ) ) {
1422 $this->mConverter
->getVariantFallbacks( $v );
1423 $vf = $this->getTextInBidtable( $variantFallbacks );
1425 $bidtable[$v] = $vf;
1429 if ( isset( $bidtable[$v] ) ) {
1430 foreach ( $vmarked as $vo ) {
1431 // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1432 // or -{H|zh:WordZh;zh-tw:WordTw}-
1433 // or -{-|zh:WordZh;zh-tw:WordTw}-
1434 // to introduce a custom mapping between
1435 // words WordZh and WordTw in the whole text
1436 if ( $manLevel[$v] == 'bidirectional' ) {
1437 $this->mConvTable
[$v][$bidtable[$vo]] = $bidtable[$v];
1439 if ( $manLevel[$vo] == 'bidirectional' ) {
1440 $this->mConvTable
[$vo][$bidtable[$v]] = $bidtable[$vo];
1445 /* for unidirectional array fill to convert tables */
1446 if ( ( $manLevel[$v] == 'bidirectional' ||
$manLevel[$v] == 'unidirectional' )
1447 && isset( $unidtable[$v] )
1449 if ( isset( $this->mConvTable
[$v] ) ) {
1450 $this->mConvTable
[$v] = array_merge( $this->mConvTable
[$v], $unidtable[$v] );
1452 $this->mConvTable
[$v] = $unidtable[$v];
1459 * Parse rules and flags.
1460 * @param $variant String: variant language code
1462 public function parse( $variant = null ) {
1464 $variant = $this->mConverter
->getPreferredVariant();
1467 $this->parseFlags();
1468 $flags = $this->mFlags
;
1470 // convert to specified variant
1471 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1472 if ( $this->mVariantFlags
) {
1473 // check if current variant in flags
1474 if ( isset( $this->mVariantFlags
[$variant] ) ) {
1475 // then convert <text to convert> to current language
1476 $this->mRules
= $this->mConverter
->autoConvert( $this->mRules
,
1479 // if current variant no in flags,
1480 // then we check its fallback variants.
1482 $this->mConverter
->getVariantFallbacks( $variant );
1483 if ( is_array( $variantFallbacks ) ) {
1484 foreach ( $variantFallbacks as $variantFallback ) {
1485 // if current variant's fallback exist in flags
1486 if ( isset( $this->mVariantFlags
[$variantFallback] ) ) {
1487 // then convert <text to convert> to fallback language
1489 $this->mConverter
->autoConvert( $this->mRules
,
1496 $this->mFlags
= $flags = array( 'R' => true );
1499 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1500 // decode => HTML entities modified by Sanitizer::removeHTMLtags
1501 $this->mRules
= str_replace( '=>', '=>', $this->mRules
);
1502 $this->parseRules();
1504 $rules = $this->mRules
;
1506 if ( !$this->mBidtable
&& !$this->mUnidtable
) {
1507 if ( isset( $flags['+'] ) ||
isset( $flags['-'] ) ) {
1508 // fill all variants if text in -{A/H/-|text} without rules
1509 foreach ( $this->mConverter
->mVariants
as $v ) {
1510 $this->mBidtable
[$v] = $rules;
1512 } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1513 $this->mFlags
= $flags = array( 'R' => true );
1517 $this->mRuleDisplay
= false;
1518 foreach ( $flags as $flag => $unused ) {
1521 // if we don't do content convert, still strip the -{}- tags
1522 $this->mRuleDisplay
= $rules;
1525 // process N flag: output current variant name
1526 $ruleVar = trim( $rules );
1527 if ( isset( $this->mConverter
->mVariantNames
[$ruleVar] ) ) {
1528 $this->mRuleDisplay
= $this->mConverter
->mVariantNames
[$ruleVar];
1530 $this->mRuleDisplay
= '';
1534 // process D flag: output rules description
1535 $this->mRuleDisplay
= $this->getRulesDesc();
1538 // process H,- flag or T only: output nothing
1539 $this->mRuleDisplay
= '';
1542 $this->mRulesAction
= 'remove';
1543 $this->mRuleDisplay
= '';
1546 $this->mRulesAction
= 'add';
1547 $this->mRuleDisplay
= '';
1550 $this->mRuleDisplay
= $this->getRuleConvertedStr( $variant );
1553 $this->mRuleTitle
= $this->getRuleConvertedTitle( $variant );
1554 $this->mRuleDisplay
= '';
1557 // ignore unknown flags (but see error case below)
1560 if ( $this->mRuleDisplay
=== false ) {
1561 $this->mRuleDisplay
= '<span class="error">'
1562 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
1566 $this->generateConvTable();
1570 * @todo FIXME: code this function :)
1572 public function hasRules() {
1577 * Get display text on markup -{...}-
1580 public function getDisplay() {
1581 return $this->mRuleDisplay
;
1585 * Get converted title.
1588 public function getTitle() {
1589 return $this->mRuleTitle
;
1593 * Return how deal with conversion rules.
1596 public function getRulesAction() {
1597 return $this->mRulesAction
;
1601 * Get conversion table. (bidirectional and unidirectional
1605 public function getConvTable() {
1606 return $this->mConvTable
;
1610 * Get conversion rules string.
1613 public function getRules() {
1614 return $this->mRules
;
1618 * Get conversion flags.
1621 public function getFlags() {
1622 return $this->mFlags
;