X-Git-Url: http://git.cyclocoop.org/?a=blobdiff_plain;f=languages%2FLanguageConverter.php;h=137fe87d1e8cf69c36afbd862e9ac78e51819265;hb=f7bb180fef88665866ef3922e596edbf273ce78d;hp=6ab6e8f85252f887db8358d85cefefd197fdd6f2;hpb=6c2a31bbf5b2cf8cfcbd4f9df382217d69b5e18f;p=lhc%2Fweb%2Fwiklou.git diff --git a/languages/LanguageConverter.php b/languages/LanguageConverter.php index 6ab6e8f852..137fe87d1e 100644 --- a/languages/LanguageConverter.php +++ b/languages/LanguageConverter.php @@ -85,7 +85,7 @@ class LanguageConverter { * @param array $flags Defining the custom strings that maps to the flags * @param array $manualLevel Limit for supported variants */ - public function __construct( $langobj, $maincode, $variants = [], + public function __construct( Language $langobj, $maincode, $variants = [], $variantfallbacks = [], $flags = [], $manualLevel = [] ) { global $wgDisabledVariants; @@ -175,11 +175,13 @@ class LanguageConverter { $req = $this->validateVariant( $wgDefaultLanguageVariant ); } + $req = $this->validateVariant( $req ); + // This function, unlike the other get*Variant functions, is // not memoized (i.e. there return value is not cached) since // new information might appear during processing after this // is first called. - if ( $this->validateVariant( $req ) ) { + if ( $req ) { return $req; } return $this->mMainLanguageCode; @@ -215,9 +217,25 @@ class LanguageConverter { * @return mixed Returns the variant if it is valid, null otherwise */ public function validateVariant( $variant = null ) { - if ( $variant !== null && in_array( $variant, $this->mVariants ) ) { + if ( $variant === null ) { + return null; + } + // Our internal variants are always lower-case; the variant we + // are validating may have mixed case. + $variant = LanguageCode::replaceDeprecatedCodes( strtolower( $variant ) ); + if ( in_array( $variant, $this->mVariants ) ) { return $variant; } + // Browsers are supposed to use BCP 47 standard in the + // Accept-Language header, but not all of our internal + // mediawiki variant codes are BCP 47. Map BCP 47 code + // to our internal code. + foreach ( $this->mVariants as $v ) { + // Case-insensitive match (BCP 47 is mixed case) + if ( strtolower( LanguageCode::bcp47( $v ) ) === $variant ) { + return $v; + } + } return null; } @@ -296,7 +314,7 @@ class LanguageConverter { return $this->mHeaderVariant; } - // see if some supported language variant is set in the + // See if some supported language variant is set in the // HTTP header. $languages = array_keys( $wgRequest->getAcceptLang() ); if ( empty( $languages ) ) { @@ -548,17 +566,18 @@ class LanguageConverter { $convTable = $convRule->getConvTable(); $action = $convRule->getRulesAction(); foreach ( $convTable as $variant => $pair ) { - if ( !$this->validateVariant( $variant ) ) { + $v = $this->validateVariant( $variant ); + if ( !$v ) { continue; } if ( $action == 'add' ) { // More efficient than array_merge(), about 2.5 times. foreach ( $pair as $from => $to ) { - $this->mTables[$variant]->setPair( $from, $to ); + $this->mTables[$v]->setPair( $from, $to ); } } elseif ( $action == 'remove' ) { - $this->mTables[$variant]->removeArray( $pair ); + $this->mTables[$v]->removeArray( $pair ); } } } @@ -642,8 +661,12 @@ class LanguageConverter { * -{flags|code1:text1;code2:text2;...}- or * -{text}- in which case no conversion should take place for text * - * @param string $text Text to be converted - * @return string Converted text + * @warning Glossary state is maintained between calls. Never feed this + * method input that hasn't properly been escaped as it may result in + * an XSS in subsequent calls, even if those subsequent calls properly + * escape things. + * @param string $text Text to be converted, already html escaped. + * @return string Converted text (html) */ public function convert( $text ) { $variant = $this->getPreferredVariant(); @@ -653,9 +676,11 @@ class LanguageConverter { /** * Same as convert() except a extra parameter to custom variant. * - * @param string $text Text to be converted + * @param string $text Text to be converted, already html escaped + * @param-taint $text exec_html * @param string $variant The target variant code * @return string Converted text + * @return-taint escaped */ public function convertTo( $text, $variant ) { global $wgDisableLangConversion; @@ -773,7 +798,7 @@ class LanguageConverter { $warningDone = true; } $startPos += 2; - continue; + break; } // Recursively parse another rule $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 ); @@ -1150,8 +1175,21 @@ class LanguageConverter { // [1] => 'zh-hant:yyy' // [2] => '' // ] - $pat = '/;\s*(?='; + $expandedVariants = []; foreach ( $this->mVariants as $variant ) { + $expandedVariants[ $variant ] = 1; + // Accept standard BCP 47 names for variants as well. + $expandedVariants[ LanguageCode::bcp47( $variant ) ] = 1; + } + // Accept old deprecated names for variants + foreach ( LanguageCode::getDeprecatedCodeMapping() as $old => $new ) { + if ( isset( $expandedVariants[ $new ] ) ) { + $expandedVariants[ $old ] = 1; + } + } + + $pat = '/;\s*(?='; + foreach ( $expandedVariants as $variant => $ignore ) { // zh-hans:xxx;zh-hant:yyy $pat .= $variant . '\s*:|'; // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz