Fix for bug 9413 and the related Malayalam issue reported on wikitech-l.

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

diff --git a/languages/Language.php b/languages/Language.php
index 0478246..690d592 100644 (file)
--- a/languages/Language.php
+++ b/languages/Language.php
@@ -62,6 +62,11 @@ class Language {
        var $minSearchLength;
        var $mExtendedSpecialPageAliases;
 
+       /**
+        * ReplacementArray object caches
+        */
+       var $transformData = array();
+
        static public $dataCache;
        static public $mLangObjCache = array();
 
@@ -1865,6 +1870,36 @@ class Language {
                }
        }
 
+       /**
+        * Convert a UTF-8 string to normal form C. In Malayalam and Arabic, this
+        * also cleans up certain backwards-compatible sequences, converting them 
+        * to the modern Unicode equivalent.
+        *
+        * This is language-specific for performance reasons only.
+        */
+       function normalize( $s ) {
+               return UtfNormal::cleanUp( $s );
+       }
+
+       /**
+        * Transform a string using serialized data stored in the given file (which
+        * must be in the serialized subdirectory of $IP). The file contains pairs
+        * mapping source characters to destination characters. 
+        *
+        * The data is cached in process memory. This will go faster if you have the 
+        * FastStringSearch extension.
+        */
+       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" );
+                       }
+                       $this->transformData[$file] = new ReplacementArray( $data );
+               }
+               return $this->transformData[$file]->replace( $string );
+       }
+
        /**
         * For right-to-left language support
         *