CustomUppercaseCollation.php

Go to the documentation of this file.

<?php
use MediaWiki\Languages\LanguageFactory;
 
class CustomUppercaseCollation extends NumericUppercaseCollation {
 
    private $alphabet;
 
    private $puaSubset;
 
    private $firstLetters;
 
    public function __construct(
        LanguageFactory $languageFactory,
        array $alphabet,
        $digitTransformLang
    ) {
        if ( count( $alphabet ) < 1 || count( $alphabet ) >= 4096 ) {
            throw new UnexpectedValueException( "Alphabet must be < 4096 items" );
        }
        $this->firstLetters = $alphabet;
        $digitTransformLang = $digitTransformLang instanceof Language
            ? $digitTransformLang
            : $languageFactory->getLanguage( $digitTransformLang );
        // For digraphs, only the first letter is capitalized in input
        $this->alphabet = array_map( [ $digitTransformLang, 'uc' ], $alphabet );
 
        $this->puaSubset = [];
        $len = count( $alphabet );
        for ( $i = 0; $i < $len; $i++ ) {
            $this->puaSubset[] = "\xF3\xB3" . chr( (int)floor( $i / 64 ) + 128 ) . chr( ( $i % 64 ) + 128 );
        }
 
        // Sort these arrays so that any trigraphs, digraphs etc. are first
        // (and they get replaced first in convertToPua()).
        $lengths = array_map( 'mb_strlen', $this->alphabet );
        array_multisort( $lengths, SORT_DESC, $this->firstLetters, $this->alphabet, $this->puaSubset );
 
        parent::__construct( $languageFactory, $digitTransformLang );
    }
 
    private function convertToPua( $string ) {
        return str_replace( $this->alphabet, $this->puaSubset, $string );
    }
 
    public function getSortKey( $string ) {
        return $this->convertToPua( parent::getSortKey( $string ) );
    }
 
    public function getFirstLetter( $string ) {
        $sortkey = $this->getSortKey( $string );
 
        // In case a title begins with a character from our alphabet, return the corresponding
        // first-letter. (This also happens if the title has a corresponding PUA code in it, to avoid
        // inconsistent behaviour. This class mostly assumes that people will not use PUA codes.)
        $index = array_search( substr( $sortkey, 0, 4 ), $this->puaSubset );
        if ( $index !== false ) {
            return $this->firstLetters[ $index ];
        }
 
        // String begins with a character outside of our alphabet, fall back
        return parent::getFirstLetter( $string );
    }
}