Merge "languages: Use static array files for normalizer data"

[lhc/web/wiklou.git] / languages / Language.php

diff --git a/languages/Language.php b/languages/Language.php
index fc8ef87..1320a57 100644 (file)
--- a/languages/Language.php
+++ b/languages/Language.php
@@ -273,7 +273,7 @@ class Language {
         * language, script or variant codes actually exist in the repositories.
         *
         * Based on regexes by Mark Davis of the Unicode Consortium:
-        * http://unicode.org/repos/cldr/trunk/tools/java/org/unicode/cldr/util/data/langtagRegex.txt
+        * https://www.unicode.org/repos/cldr/trunk/tools/java/org/unicode/cldr/util/data/langtagRegex.txt
         *
         * @param string $code
         * @param bool $lenient Whether to allow '_' as separator. The default is only '-'.
@@ -792,12 +792,12 @@ class Language {
        /**
         * Get an array of language names, indexed by code.
         * @param null|string $inLanguage Code of language in which to return the names
-        *              Use null for autonyms (native names)
+        *              Use null for autonyms (native names)
         * @param string $include One of:
-        *              'all' all available languages
-        *              'mw' only if the language is defined in MediaWiki or wgExtraLanguageNames (default)
-        *              'mwfile' only if the language is in 'mw' *and* has a message file
-        * @return array Language code => language name
+        *              'all' all available languages
+        *              'mw' only if the language is defined in MediaWiki or wgExtraLanguageNames (default)
+        *              'mwfile' only if the language is in 'mw' *and* has a message file
+        * @return array Language code => language name (sorted by key)
         * @since 1.20
         */
        public static function fetchLanguageNames( $inLanguage = null, $include = 'mw' ) {
@@ -823,7 +823,7 @@ class Language {
         *              'all' all available languages
         *              'mw' only if the language is defined in MediaWiki or wgExtraLanguageNames (default)
         *              'mwfile' only if the language is in 'mw' *and* has a message file
-        * @return array Language code => language name
+        * @return array Language code => language name (sorted by key)
         */
        private static function fetchLanguageNamesUncached( $inLanguage = null, $include = 'mw' ) {
                global $wgExtraLanguageNames, $wgUsePigLatinVariant;
@@ -1951,8 +1951,8 @@ class Language {
         * Gets directionality of the first strongly directional codepoint, for embedBidi()
         *
         * This is the rule the BIDI algorithm uses to determine the directionality of
-        * paragraphs ( http://unicode.org/reports/tr9/#The_Paragraph_Level ) and
-        * FSI isolates ( http://unicode.org/reports/tr9/#Explicit_Directional_Isolates ).
+        * paragraphs ( https://www.unicode.org/reports/tr9/#The_Paragraph_Level ) and
+        * FSI isolates ( https://www.unicode.org/reports/tr9/#Explicit_Directional_Isolates ).
         *
         * TODO: Does not handle BIDI control characters inside the text.
         * TODO: Does not handle unallocated characters.
@@ -2990,8 +2990,8 @@ class Language {
                global $wgAllUnicodeFixes;
                $s = UtfNormal\Validator::cleanUp( $s );
                if ( $wgAllUnicodeFixes ) {
-                       $s = $this->transformUsingPairFile( 'normalize-ar.ser', $s );
-                       $s = $this->transformUsingPairFile( 'normalize-ml.ser', $s );
+                       $s = $this->transformUsingPairFile( 'normalize-ar.php', $s );
+                       $s = $this->transformUsingPairFile( 'normalize-ml.php', $s );
                }
 
                return $s;
@@ -3011,12 +3011,10 @@ class Language {
         * @throws MWException
         * @return string
         */
-       function transformUsingPairFile( $file, $string ) {
+       protected function transformUsingPairFile( $file, $string ) {
                if ( !isset( $this->transformData[$file] ) ) {
-                       $data = wfGetPrecompiledData( $file );
-                       if ( $data === false ) {
-                               throw new MWException( __METHOD__ . ": The transformation file $file is missing" );
-                       }
+                       global $IP;
+                       $data = require "$IP/languages/data/{$file}";
                        $this->transformData[$file] = new ReplacementArray( $data );
                }
                return $this->transformData[$file]->replace( $string );
@@ -3485,7 +3483,7 @@ class Language {
         * @param int $length Maximum length (including ellipsis)
         * @param string $ellipsis String to append to the truncated text
         * @param bool $adjustLength Subtract length of ellipsis from $length.
-        *      $adjustLength was introduced in 1.18, before that behaved as if false.
+        *      $adjustLength was introduced in 1.18, before that behaved as if false.
         * @return string
         */
        function truncate( $string, $length, $ellipsis = '...', $adjustLength = true ) {
@@ -4148,6 +4146,18 @@ class Language {
                return $this->mConverter;
        }
 
+       /**
+        * convert text to a variant
+        *
+        * @param string $text text to convert
+        * @param string|bool $variant variant to convert to, or false to use the user's preferred
+        *      variant (if logged in), or the project default variant
+        * @return string the converted string
+        */
+       public function autoConvert( $text, $variant = false ) {
+               return $this->mConverter->autoConvert( $text, $variant );
+       }
+
        /**
         * convert text to all supported variants
         *
@@ -4181,11 +4191,13 @@ class Language {
        /**
         * Convert a namespace index to a string in the preferred variant
         *
-        * @param int $ns
-        * @return string
+        * @param int $ns namespace index (https://www.mediawiki.org/wiki/Manual:Namespace)
+        * @param string|null $variant variant to convert to, or null to use the user's preferred
+        *      variant (if logged in), or the project default variant
+        * @return string a string representation of the namespace
         */
-       public function convertNamespace( $ns ) {
-               return $this->mConverter->convertNamespace( $ns );
+       public function convertNamespace( $ns, $variant = null ) {
+               return $this->mConverter->convertNamespace( $ns, $variant );
        }
 
        /**
@@ -4822,7 +4834,7 @@ class Language {
         * @param string $details HTML safe text between brackets
         * @param bool $oppositedm Add the direction mark opposite to your
         *   language, to display text properly
-        * @return HTML escaped string
+        * @return string HTML escaped
         */
        function specialList( $page, $details, $oppositedm = true ) {
                if ( !$details ) {