LanguageJa.php

Go to the documentation of this file.

<?php
use MediaWiki\Language\Language;
 
class LanguageJa extends Language {
 
    private const WORD_SEG_HIRAGANA = '(?:\xe3(?:\x81[\x80-\xbf]|\x82[\x80-\x9f]))';
 
    private const WORD_SEG_KATAKANA = '(?:\xe3(?:\x82[\xa0-\xbf]|\x83[\x80-\xbf]))';
 
    private const WORD_SEG_KANJI =
        '(?:\xe3[\x88-\xbf][\x80-\xbf]|[\xe4-\xe8][\x80-\xbf]{2}|\xe9[\x80-\xa5][\x80-\xbf]|\xe9\xa6[\x80-\x99])';
 
    private const WORD_SEGMENTATION_REGEX =
        '/(' . self::WORD_SEG_HIRAGANA . '+|' . self::WORD_SEG_KATAKANA . '+|' . self::WORD_SEG_KANJI . '+)/';
 
    public function segmentByWord( $string ) {
        return self::insertSpace( $string, self::WORD_SEGMENTATION_REGEX );
    }
 
    public function emphasize( $text ) {
        return $text;
    }
}