LanguageNameUtils.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Languages;
 
use HashBagOStuff;
use Hooks;
use MediaWiki\Config\ServiceOptions;
use MediaWikiTitleCodec;
use MWException;
use Wikimedia\Assert\Assert;
 
class LanguageNameUtils {
    const AUTONYMS = null;
 
    const ALL = 'all';
 
    const DEFINED = 'mw';
 
    const SUPPORTED = 'mwfile';
 
    private $options;
 
    private $languageNameCache;
 
    private $validCodeCache = [];
 
    public const CONSTRUCTOR_OPTIONS = [
        'ExtraLanguageNames',
        'UsePigLatinVariant',
    ];
 
    public function __construct( ServiceOptions $options ) {
        $options->assertRequiredOptions( self::CONSTRUCTOR_OPTIONS );
        $this->options = $options;
    }
 
    public function isSupportedLanguage( $code ) {
        if ( !$this->isValidBuiltInCode( $code ) ) {
            return false;
        }
 
        if ( $code === 'qqq' ) {
            // Special code for internal use, not supported even though there is a qqq.json
            return false;
        }
 
        return is_readable( $this->getMessagesFileName( $code ) ) ||
            is_readable( $this->getJsonMessagesFileName( $code ) );
    }
 
    public function isValidCode( $code ) {
        Assert::parameterType( 'string', $code, '$code' );
        if ( !isset( $this->validCodeCache[$code] ) ) {
            // People think language codes are HTML-safe, so enforce it.  Ideally we should only
            // allow a-zA-Z0-9- but .+ and other chars are often used for {{int:}} hacks.  See bugs
            // T39564, T39587, T38938.
            $this->validCodeCache[$code] =
                // Protect against path traversal
                strcspn( $code, ":/\\\000&<>'\"" ) === strlen( $code ) &&
                !preg_match( MediaWikiTitleCodec::getTitleInvalidRegex(), $code );
        }
        return $this->validCodeCache[$code];
    }
 
    public function isValidBuiltInCode( $code ) {
        Assert::parameterType( 'string', $code, '$code' );
 
        return (bool)preg_match( '/^[a-z0-9-]{2,}$/', $code );
    }
 
    public function isKnownLanguageTag( $tag ) {
        // Quick escape for invalid input to avoid exceptions down the line when code tries to
        // process tags which are not valid at all.
        if ( !$this->isValidBuiltInCode( $tag ) ) {
            return false;
        }
 
        if ( isset( Data\Names::$names[$tag] ) || $this->getLanguageName( $tag, $tag ) !== '' ) {
            return true;
        }
 
        return false;
    }
 
    public function getLanguageNames( $inLanguage = self::AUTONYMS, $include = self::DEFINED ) {
        $cacheKey = $inLanguage === self::AUTONYMS ? 'null' : $inLanguage;
        $cacheKey .= ":$include";
        if ( !$this->languageNameCache ) {
            $this->languageNameCache = new HashBagOStuff( [ 'maxKeys' => 20 ] );
        }
 
        $ret = $this->languageNameCache->get( $cacheKey );
        if ( !$ret ) {
            $ret = $this->getLanguageNamesUncached( $inLanguage, $include );
            $this->languageNameCache->set( $cacheKey, $ret );
        }
        return $ret;
    }
 
    private function getLanguageNamesUncached( $inLanguage, $include ) {
        // If passed an invalid language code to use, fallback to en
        if ( $inLanguage !== self::AUTONYMS && !$this->isValidCode( $inLanguage ) ) {
            $inLanguage = 'en';
        }
 
        $names = [];
 
        if ( $inLanguage !== self::AUTONYMS ) {
            # TODO: also include for self::AUTONYMS, when this code is more efficient
            Hooks::run( 'LanguageGetTranslatedLanguageNames', [ &$names, $inLanguage ] );
        }
 
        $mwNames = $this->options->get( 'ExtraLanguageNames' ) + Data\Names::$names;
        if ( $this->options->get( 'UsePigLatinVariant' ) ) {
            // Pig Latin (for variant development)
            $mwNames['en-x-piglatin'] = 'Igpay Atinlay';
        }
 
        foreach ( $mwNames as $mwCode => $mwName ) {
            # - Prefer own MediaWiki native name when not using the hook
            # - For other names just add if not added through the hook
            if ( $mwCode === $inLanguage || !isset( $names[$mwCode] ) ) {
                $names[$mwCode] = $mwName;
            }
        }
 
        if ( $include === self::ALL ) {
            ksort( $names );
            return $names;
        }
 
        $returnMw = [];
        $coreCodes = array_keys( $mwNames );
        foreach ( $coreCodes as $coreCode ) {
            $returnMw[$coreCode] = $names[$coreCode];
        }
 
        if ( $include === self::SUPPORTED ) {
            $namesMwFile = [];
            # We do this using a foreach over the codes instead of a directory loop so that messages
            # files in extensions will work correctly.
            foreach ( $returnMw as $code => $value ) {
                if ( is_readable( $this->getMessagesFileName( $code ) ) ||
                    is_readable( $this->getJsonMessagesFileName( $code ) )
                ) {
                    $namesMwFile[$code] = $names[$code];
                }
            }
 
            ksort( $namesMwFile );
            return $namesMwFile;
        }
 
        ksort( $returnMw );
        # self::DEFINED option; default if it's not one of the other two options
        # (self::ALL/self::SUPPORTED)
        return $returnMw;
    }
 
    public function getLanguageName( $code, $inLanguage = self::AUTONYMS, $include = self::ALL ) {
        $code = strtolower( $code );
        $array = $this->getLanguageNames( $inLanguage, $include );
        return $array[$code] ?? '';
    }
 
    public function getFileName( $prefix, $code, $suffix = '.php' ) {
        if ( !$this->isValidBuiltInCode( $code ) ) {
            throw new MWException( "Invalid language code \"$code\"" );
        }
 
        return $prefix . str_replace( '-', '_', ucfirst( $code ) ) . $suffix;
    }
 
    public function getMessagesFileName( $code ) {
        global $IP;
        $file = $this->getFileName( "$IP/languages/messages/Messages", $code, '.php' );
        Hooks::run( 'Language::getMessagesFileName', [ $code, &$file ] );
        return $file;
    }
 
    public function getJsonMessagesFileName( $code ) {
        global $IP;
 
        if ( !$this->isValidBuiltInCode( $code ) ) {
            throw new MWException( "Invalid language code \"$code\"" );
        }
 
        return "$IP/languages/i18n/$code.json";
    }
}