GoogleTranslateWebService.php

<?php
declare( strict_types = 1 );
 
namespace MediaWiki\Extension\Translate\WebService;
 
use FormatJson;
use MediaWiki\Http\HttpRequestFactory;
use Sanitizer;
 
class GoogleTranslateWebService extends TranslationWebService {
    private const PUBLIC_API = 'https://translation.googleapis.com/language/translate/v2';
    private $httpRequestFactory;
 
    public function __construct(
        HttpRequestFactory $httpRequestFactory,
        string $serviceName,
        array $config
    ) {
        parent::__construct( $serviceName, $config );
        $this->httpRequestFactory = $httpRequestFactory;
    }
 
    public function getType(): string {
        return 'mt';
    }
 
    protected function mapCode( string $code ): string {
        $map = [
            'be-tarask' => 'be',
            'zh-cn' => 'zh-CN',
            'zh-hans' => 'zh-CN',
            'zh-hant' => 'zh-TW',
            'zh-tw' => 'zh-TW',
        ];
 
        return $map[$code] ?? $code;
    }
 
    public function isSupportedLanguagePair( string $sourceLanguage, string $targetLanguage ): bool {
        $pairs = $this->getSupportedLanguagePairs();
        $from = $this->mapCode( $sourceLanguage );
        $to = $this->mapCode( $targetLanguage );
 
        // As long as the source & target language exist at Google it is fine
        return isset( $pairs[$from] ) && isset( $pairs[$to] ) && $from !== $to;
    }
 
    protected function doPairs(): array {
        if ( !isset( $this->config['key'] ) ) {
            throw new TranslationWebServiceConfigurationException( 'API key is not set' );
        }
 
        $api = $this->config['pairs'] ?? self::PUBLIC_API . '/languages';
        $params = [
            'key' => $this->config['key'],
        ];
 
        $json = $this->httpRequestFactory->get(
            wfAppendQuery( $api, wfArrayToCgi( $params ) ),
            [ 'timeout' => $this->config['timeout'] ],
            __METHOD__
        );
        $response = FormatJson::decode( $json );
 
        if ( !is_object( $response ) ) {
            $exception = 'Malformed reply from remote server: ' . (string)$json;
            throw new TranslationWebServiceException( $exception );
        }
 
        $pairs = [];
        foreach ( $response->data->languages as $language ) {
            // Google can translate from any language to any language
            $pairs[$language->language] = true;
        }
 
        return $pairs;
    }
 
    protected function getQuery( string $text, string $sourceLanguage, string $targetLanguage ): TranslationQuery {
        if ( !isset( $this->config['key'] ) ) {
            throw new TranslationWebServiceConfigurationException( 'API key is not set' );
        }
        # https://cloud.google.com/translate/docs/reference/translate
        if ( strlen( $text ) > 10000 ) {
            // There is no limitation but we don't want the translation service to be abused, don't we?
            throw new TranslationWebServiceInvalidInputException( 'Source text too long' );
        }
 
        $url = $this->config['url'] ?? self::PUBLIC_API;
        $text = trim( $text );
        $text = $this->wrapUntranslatable( $text );
 
        return TranslationQuery::factory( $url )
            ->timeout( intval( $this->config['timeout'] ?? 3 ) )
            ->postWithData( wfArrayToCgi( [
                'key' => $this->config['key'],
                'q' => $text,
                'target' => $targetLanguage,
                'source' => $sourceLanguage,
                'format' => 'html',
            ] ) );
    }
 
    protected function parseResponse( TranslationQueryResponse $reply ): string {
        $body = $reply->getBody();
        $response = FormatJson::decode( $body );
        if ( !is_object( $response ) ) {
            throw new TranslationWebServiceException( 'Invalid json: ' . serialize( $body ) );
        }
        $text = Sanitizer::decodeCharReferences( $response->data->translations[0]->translatedText );
        $text = $this->unwrapUntranslatable( $text );
 
        return trim( $text );
    }
}