NumericUppercaseCollation.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Collation;
 
use MediaWiki\Language\Language;
use MediaWiki\Languages\LanguageFactory;
 
class NumericUppercaseCollation extends UppercaseCollation {
 
    private $digitTransformLang;
 
    public function __construct(
        LanguageFactory $languageFactory,
        $digitTransformLang
    ) {
        $this->digitTransformLang = $digitTransformLang instanceof Language
            ? $digitTransformLang
            : $languageFactory->getLanguage( $digitTransformLang );
        parent::__construct( $languageFactory );
    }
 
    public function getSortKey( $string ) {
        $sortkey = parent::getSortKey( $string );
        $sortkey = $this->convertDigits( $sortkey );
        // For each sequence of digits, insert the digit '0' and then the length of the sequence
        // (encoded in two bytes) before it. That's all folks, it sorts correctly now! The '0' ensures
        // correct position (where digits would normally sort), then the length will be compared putting
        // shorter numbers before longer ones; if identical, then the characters will be compared, which
        // generates the correct results for numbers of equal length.
        $sortkey = preg_replace_callback( '/\d+/', static function ( $matches ) {
            // Strip any leading zeros
            $number = ltrim( $matches[0], '0' );
            $len = strlen( $number );
            // This allows sequences of up to 65536 numeric characters to be handled correctly. One byte
            // would allow only for 256, which doesn't feel future-proof.
            $prefix = chr( (int)floor( $len / 256 ) ) . chr( $len % 256 );
            return '0' . $prefix . $number;
        }, $sortkey );
 
        return $sortkey;
    }
 
    private function convertDigits( $string ) {
        $table = $this->digitTransformLang->digitTransformTable();
        if ( $table ) {
            $table = array_filter( $table );
            $flipped = array_flip( $table );
            // Some languages seem to also have commas in this table.
            $flipped = array_filter( $flipped, 'is_numeric' );
            $string = strtr( $string, $flipped );
        }
        return $string;
    }
 
    public function getFirstLetter( $string ) {
        $convertedString = $this->convertDigits( $string );
 
        if ( preg_match( '/^\d/', $convertedString ) ) {
            return wfMessage( 'category-header-numerals' )
                ->numParams( 0, 9 )
                ->text();
        } else {
            return parent::getFirstLetter( $string );
        }
    }
}
 
class_alias( NumericUppercaseCollation::class, 'NumericUppercaseCollation' );