kupshop/bundles/KupShop/I18nBundle/Util/TranslationEngine.php

<?php

declare(strict_types=1);

namespace KupShop\I18nBundle\Util;

use DeepL\DeepLException;
use DeepL\Translator;

class TranslationEngine
{
    private const GOOGLE_CHUNK = 3900;
    private static string $googleApiKey = 'AIzaSyCERmGVBqiuzN74C9l-CEMA08KwIgSQY_Y';
    private static string $deepLAuthKey = '491e37c3-4dc9-ed7b-7188-536b0598f034';

    public function getTranslation(string $text, string $sourceLanguage, string $language, bool $html = false): array
    {
        if (isFunctionalTests()) {
            $responseData['translatedText'] = $sourceLanguage.'->'.$language.': '.$text;

            return $responseData;
        }

        $deepl = (\Settings::getDefault()->use_deepl ?? 'N');

        getLogger()->notice('TranslationEngine:translate', ['char_count' => mb_strlen($text), 'from' => $sourceLanguage, 'to' => $language, 'deepl' => $deepl]);

        if ($deepl == 'Y') {
            return $this->getTranslationWithDeepL($text, $sourceLanguage, $language, $html);
        }

        return $this->getTranslationWithGoogle($text, $sourceLanguage, $language);
    }

    public function getTranslationMulti(array $textMulti, string $sourceLanguage, string $language): array
    {
        foreach ($textMulti as $key => $text) {
            if (empty($text)) {
                continue;
            }
            $translation = $this->getTranslation($text, $sourceLanguage, $language);
            if (empty($translation['error'])) {
                $textMulti[$key] = $translation['translatedText'];
            } else {
                unset($textMulti[$key]);
            }
        }

        return $textMulti;
    }

    private function getTranslationWithGoogle(string $text, string $sourceLanguage, string $language): array
    {
        // Google Translate nezná de-AT nebo de-CH, zná jen de-DE
        // potřebuju to teda, aby se použila němčina - jinak to nefunguje vubec
        $langAlias = [
            'at' => 'de',
            'ch' => 'de',
        ];

        foreach ([&$sourceLanguage, &$language] as &$lang) {
            if (isset($langAlias[$lang])) {
                $lang = $langAlias[$lang];
            }
        }

        $baseUrl = 'https://www.googleapis.com/language/translate/v2?key='.self::$googleApiKey; // .'&q='.rawurlencode($text).'&source='.$sourceLanguage.'&target='.$language

        $textChunks = $this->splitTextIntoChunks($text);
        $translatedChunks = [];

        foreach ($textChunks as $chunk) {
            $url = $baseUrl.'&q='.rawurlencode($chunk).'&source='.$sourceLanguage.'&target='.$language;

            $handle = curl_init($url);
            curl_setopt($handle, CURLOPT_RETURNTRANSFER, true);
            $response = curl_exec($handle);
            $responseData = json_decode($response ?: '{}', true);

            if (isset($responseData['error']) || curl_error($handle)) {
                curl_close($handle);

                return [
                    'error' => curl_error($handle) ?: ($responseData['error'] ?? 'Unknown error'),
                    'translatedText' => '',
                ];
            }

            $translatedChunks[] = html_entity_decode($responseData['data']['translations'][0]['translatedText'] ?? '', ENT_QUOTES);
            curl_close($handle);
        }

        // Combine all translated chunks into a single result
        $translatedText = implode(' ', $translatedChunks);

        // Replace malformed sequences
        $translatedText = str_replace(
            ['> , ', '> .'],
            ['>, ', '>.'],
            $translatedText
        );

        return [
            'error' => '',
            'translatedText' => $translatedText,
        ];
    }

    private function getTranslationWithDeepL(string $text, string $sourceLanguage, string $language, bool $html = false): array
    {
        $responseData = ['translatedText' => ''];

        try {
            $translator = new Translator(self::$deepLAuthKey);

            $language = $this->deeplLanguageMapping($language);
            $sourceLanguage = $this->deeplLanguageMapping($sourceLanguage, true);

            $options = [];

            if ($html) {
                $options['tag_handling'] = 'html';
                $options['non_splitting_tags'] = ['strong', 'em', 'b', 'i'];
            }

            $result = $translator->translateText($text, $sourceLanguage, $language, $options);
            if (!empty($result->text)) {
                $responseData['translatedText'] = $result->text;
            }
        } catch (DeepLException $exception) {
            $responseData['error'] = $exception->getMessage();
        }

        return $responseData;
    }

    protected function deeplLanguageMapping(string $language, bool $source = false): string
    {
        switch ($language) {
            case 'en':
                // if source language is 'en' => keep it 'en' otherwise use 'en-US'
                return $source ? 'en' : 'en-US';
            case 'at':
                return 'de';
        }

        return $language;
    }

    protected function splitTextIntoChunks(string $text): array
    {
        $chunks = [];
        $currentChunk = '';

        // Split text into words to preserve boundaries
        $words = preg_split('/(\s+)/u', $text, -1, PREG_SPLIT_DELIM_CAPTURE);

        foreach ($words as $word) {
            // Add word to the current chunk if it fits
            if (mb_strlen($currentChunk.$word) <= self::GOOGLE_CHUNK) {
                $currentChunk .= $word;
            } else {
                // Save the current chunk and start a new one
                $chunks[] = trim($currentChunk);
                $currentChunk = $word;
            }
        }

        // Add the remaining chunk
        if (!empty($currentChunk)) {
            $chunks[] = trim($currentChunk);
        }

        return $chunks;
    }
}