FendxPHP/fendx-framework/fendx-i18n/src/Tools/Extractor/TranslationKeyExtractor.php

<?php
declare(strict_types=1);

namespace Fendx\I18n\Tools\Extractor;

use Fendx\I18n\Tools\Extractor\Parser\PhpParser;
use Fendx\I18n\Tools\Extractor\Parser\JsParser;
use Fendx\I18n\Tools\Extractor\Parser\HtmlParser;
use Fendx\I18n\Tools\Extractor\Parser\TwigParser;
use Fendx\I18n\Tools\Extractor\Parser\VueParser;

class TranslationKeyExtractor
{
    protected PhpParser $phpParser;
    protected JsParser $jsParser;
    protected HtmlParser $htmlParser;
    protected TwigParser $twigParser;
    protected VueParser $vueParser;
    protected array $config = [];
    protected array $extractedKeys = [];
    protected array $fileScanned = [];

    public function __construct(array $config = [])
    {
        $this->config = array_merge($this->getDefaultConfig(), $config);
        $this->phpParser = new PhpParser($this->config);
        $this->jsParser = new JsParser($this->config);
        $this->htmlParser = new HtmlParser($this->config);
        $this->twigParser = new TwigParser($this->config);
        $this->vueParser = new VueParser($this->config);
    }

    /**
     * Extract translation keys from directory.
     */
    public function extractFromDirectory(string $directory, array $options = []): array
    {
        $this->extractedKeys = [];
        $this->fileScanned = [];
        
        $recursive = $options['recursive'] ?? true;
        $patterns = $options['patterns'] ?? $this->config['file_patterns'];
        $excludePatterns = $options['exclude_patterns'] ?? $this->config['exclude_patterns'];
        
        $this->scanDirectory($directory, $recursive, $patterns, $excludePatterns);
        
        return $this->getExtractedKeys();
    }

    /**
     * Extract translation keys from file.
     */
    public function extractFromFile(string $filepath): array
    {
        $this->extractedKeys = [];
        $this->fileScanned = [];
        
        if (!file_exists($filepath)) {
            throw new \InvalidArgumentException("File not found: {$filepath}");
        }
        
        $this->scanFile($filepath);
        
        return $this->getExtractedKeys();
    }

    /**
     * Extract translation keys from string content.
     */
    public function extractFromString(string $content, string $filename = 'unknown'): array
    {
        $this->extractedKeys = [];
        $this->fileScanned = [];
        
        $this->parseContent($content, $filename);
        
        return $this->getExtractedKeys();
    }

    /**
     * Scan directory for files.
     */
    protected function scanDirectory(string $directory, bool $recursive, array $patterns, array $excludePatterns): void
    {
        $iterator = new \RecursiveIteratorIterator(
            $recursive ? new \RecursiveDirectoryIterator($directory) : new \DirectoryIterator($directory),
            \RecursiveIteratorIterator::SELF_FIRST
        );

        foreach ($iterator as $file) {
            if ($file->isDir()) {
                continue;
            }

            $filepath = $file->getPathname();
            
            // Check exclude patterns
            if ($this->matchesExcludePatterns($filepath, $excludePatterns)) {
                continue;
            }
            
            // Check include patterns
            if ($this->matchesPatterns($filepath, $patterns)) {
                $this->scanFile($filepath);
            }
        }
    }

    /**
     * Scan single file.
     */
    protected function scanFile(string $filepath): void
    {
        if (in_array($filepath, $this->fileScanned)) {
            return;
        }

        $content = file_get_contents($filepath);
        if ($content === false) {
            return;
        }

        $this->fileScanned[] = $filepath;
        $this->parseContent($content, $filepath);
    }

    /**
     * Parse content based on file type.
     */
    protected function parseContent(string $content, string $filename): void
    {
        $extension = strtolower(pathinfo($filename, PATHINFO_EXTENSION));
        
        switch ($extension) {
            case 'php':
                $this->extractFromPhp($content, $filename);
                break;
            case 'js':
            case 'jsx':
            case 'ts':
            case 'tsx':
                $this->extractFromJs($content, $filename);
                break;
            case 'html':
            case 'htm':
                $this->extractFromHtml($content, $filename);
                break;
            case 'twig':
                $this->extractFromTwig($content, $filename);
                break;
            case 'vue':
                $this->extractFromVue($content, $filename);
                break;
            default:
                // Try to detect content type
                if ($this->isPhpContent($content)) {
                    $this->extractFromPhp($content, $filename);
                } elseif ($this->isJsContent($content)) {
                    $this->extractFromJs($content, $filename);
                } elseif ($this->isHtmlContent($content)) {
                    $this->extractFromHtml($content, $filename);
                } elseif ($this->isTwigContent($content)) {
                    $this->extractFromTwig($content, $filename);
                }
                break;
        }
    }

    /**
     * Extract from PHP content.
     */
    protected function extractFromPhp(string $content, string $filename): void
    {
        $keys = $this->phpParser->extract($content);
        $this->addExtractedKeys($keys, $filename, 'php');
    }

    /**
     * Extract from JavaScript content.
     */
    protected function extractFromJs(string $content, string $filename): void
    {
        $keys = $this->jsParser->extract($content);
        $this->addExtractedKeys($keys, $filename, 'js');
    }

    /**
     * Extract from HTML content.
     */
    protected function extractFromHtml(string $content, string $filename): void
    {
        $keys = $this->htmlParser->extract($content);
        $this->addExtractedKeys($keys, $filename, 'html');
    }

    /**
     * Extract from Twig content.
     */
    protected function extractFromTwig(string $content, string $filename): void
    {
        $keys = $this->twigParser->extract($content);
        $this->addExtractedKeys($keys, $filename, 'twig');
    }

    /**
     * Extract from Vue content.
     */
    protected function extractFromVue(string $content, string $filename): void
    {
        $keys = $this->vueParser->extract($content);
        $this->addExtractedKeys($keys, $filename, 'vue');
    }

    /**
     * Add extracted keys.
     */
    protected function addExtractedKeys(array $keys, string $filename, string $type): void
    {
        foreach ($keys as $key => $info) {
            if (!isset($this->extractedKeys[$key])) {
                $this->extractedKeys[$key] = [
                    'key' => $key,
                    'files' => [],
                    'contexts' => [],
                    'parameters' => [],
                    'types' => [],
                    'line_numbers' => []
                ];
            }

            $this->extractedKeys[$key]['files'][] = $filename;
            $this->extractedKeys[$key]['types'][] = $type;
            
            if (isset($info['context'])) {
                $this->extractedKeys[$key]['contexts'][] = $info['context'];
            }
            
            if (isset($info['parameters'])) {
                $this->extractedKeys[$key]['parameters'] = array_merge(
                    $this->extractedKeys[$key]['parameters'],
                    $info['parameters']
                );
            }
            
            if (isset($info['line'])) {
                $this->extractedKeys[$key]['line_numbers'][] = $filename . ':' . $info['line'];
            }
        }
    }

    /**
     * Get extracted keys.
     */
    public function getExtractedKeys(): array
    {
        // Remove duplicates and sort
        $result = [];
        foreach ($this->extractedKeys as $key => $info) {
            $result[$key] = [
                'key' => $key,
                'files' => array_unique($info['files']),
                'contexts' => array_unique($info['contexts']),
                'parameters' => array_unique($info['parameters']),
                'types' => array_unique($info['types']),
                'line_numbers' => array_unique($info['line_numbers']),
                'usage_count' => count($info['files'])
            ];
        }

        ksort($result);
        return $result;
    }

    /**
     * Get keys by file.
     */
    public function getKeysByFile(): array
    {
        $keysByFile = [];
        
        foreach ($this->extractedKeys as $key => $info) {
            foreach ($info['files'] as $file) {
                if (!isset($keysByFile[$file])) {
                    $keysByFile[$file] = [];
                }
                $keysByFile[$file][] = $key;
            }
        }
        
        return $keysByFile;
    }

    /**
     * Get keys by type.
     */
    public function getKeysByType(): array
    {
        $keysByType = [];
        
        foreach ($this->extractedKeys as $key => $info) {
            foreach ($info['types'] as $type) {
                if (!isset($keysByType[$type])) {
                    $keysByType[$type] = [];
                }
                $keysByType[$type][] = $key;
            }
        }
        
        return $keysByType;
    }

    /**
     * Get unused keys (keys not found in any file).
     */
    public function getUnusedKeys(array $existingKeys): array
    {
        $extractedKeySet = array_keys($this->extractedKeys);
        $unusedKeys = array_diff($existingKeys, $extractedKeySet);
        
        return array_values($unusedKeys);
    }

    /**
     * Get missing keys (keys found but not in existing translations).
     */
    public function getMissingKeys(array $existingKeys): array
    {
        $extractedKeySet = array_keys($this->extractedKeys);
        $missingKeys = array_diff($extractedKeySet, $existingKeys);
        
        return array_values($missingKeys);
    }

    /**
     * Get keys with parameters.
     */
    public function getKeysWithParameters(): array
    {
        $keysWithParams = [];
        
        foreach ($this->extractedKeys as $key => $info) {
            if (!empty($info['parameters'])) {
                $keysWithParams[$key] = array_unique($info['parameters']);
            }
        }
        
        return $keysWithParams;
    }

    /**
     * Get duplicate keys (keys found in multiple contexts).
     */
    public function getDuplicateKeys(): array
    {
        $duplicates = [];
        
        foreach ($this->extractedKeys as $key => $info) {
            if (count($info['contexts']) > 1) {
                $duplicates[$key] = [
                    'contexts' => array_unique($info['contexts']),
                    'files' => array_unique($info['files'])
                ];
            }
        }
        
        return $duplicates;
    }

    /**
     * Generate translation template.
     */
    public function generateTemplate(array $existingKeys = null, string $language = 'en'): array
    {
        $template = [];
        $extractedKeys = array_keys($this->extractedKeys);
        
        if ($existingKeys) {
            // Merge with existing keys
            $allKeys = array_unique(array_merge($extractedKeys, $existingKeys));
        } else {
            $allKeys = $extractedKeys;
        }
        
        foreach ($allKeys as $key) {
            $template[$key] = $this->generatePlaceholder($key, $language);
        }
        
        return $template;
    }

    /**
     * Generate placeholder for translation key.
     */
    protected function generatePlaceholder(string $key, string $language): string
    {
        $parts = explode('.', $key);
        $lastPart = end($parts);
        
        // Convert to human readable format
        $placeholder = str_replace(['_', '-'], ' ', $lastPart);
        $placeholder = ucwords($placeholder);
        
        // Add language-specific prefix/suffix if needed
        switch ($language) {
            case 'zh-CN':
            case 'zh-TW':
                return $placeholder; // Chinese doesn't need articles
            case 'ja':
                return $placeholder; // Japanese doesn't need articles
            case 'ko':
                return $placeholder; // Korean doesn't need articles
            case 'fr':
                return "La {$placeholder}"; // French article
            case 'de':
                return "Der {$placeholder}"; // German article
            case 'es':
                return "El {$placeholder}"; // Spanish article
            case 'it':
                return "Il {$placeholder}"; // Italian article
            case 'ru':
                return "{$placeholder} (русский)"; // Russian suffix
            case 'ar':
                return "{$placeholder} (عربي)"; // Arabic suffix
            default:
                return "The {$placeholder}"; // English article
        }
    }

    /**
     * Export extracted keys.
     */
    public function export(string $format = 'json'): string
    {
        $data = [
            'extracted_keys' => $this->getExtractedKeys(),
            'statistics' => $this->getStatistics(),
            'generated_at' => date('Y-m-d H:i:s'),
            'config' => $this->config
        ];
        
        return match ($format) {
            'json' => json_encode($data, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE),
            'csv' => $this->exportToCsv(),
            'xlsx' => $this->exportToXlsx(),
            'php' => '<?php return ' . var_export($this->getExtractedKeys(), true) . ';',
            default => throw new \InvalidArgumentException("Unsupported export format: {$format}")
        };
    }

    /**
     * Export to CSV format.
     */
    protected function exportToCsv(): string
    {
        $csv = "Key,Files,Types,Contexts,Parameters,Usage Count\n";
        
        foreach ($this->getExtractedKeys() as $key => $info) {
            $csv .= '"' . $key . '",';
            $csv .= '"' . implode('; ', $info['files']) . '",';
            $csv .= '"' . implode('; ', $info['types']) . '",';
            $csv .= '"' . implode('; ', $info['contexts']) . '",';
            $csv .= '"' . implode('; ', $info['parameters']) . '",';
            $csv .= $info['usage_count'] . "\n";
        }
        
        return $csv;
    }

    /**
     * Export to XLSX format (basic implementation).
     */
    protected function exportToXlsx(): string
    {
        // This would require a proper XLSX library
        // For now, return CSV as placeholder
        return $this->exportToCsv();
    }

    /**
     * Get extraction statistics.
     */
    public function getStatistics(): array
    {
        $keysByType = $this->getKeysByType();
        $keysByFile = $this->getKeysByFile();
        
        return [
            'total_keys' => count($this->extractedKeys),
            'total_files_scanned' => count($this->fileScanned),
            'keys_by_type' => array_map('count', $keysByType),
            'keys_by_file' => array_map('count', $keysByFile),
            'keys_with_parameters' => count($this->getKeysWithParameters()),
            'duplicate_keys' => count($this->getDuplicateKeys()),
            'most_used_keys' => $this->getMostUsedKeys(10),
            'file_types' => array_unique(array_merge(...array_values($keysByType)))
        ];
    }

    /**
     * Get most used keys.
     */
    protected function getMostUsedKeys(int $limit = 10): array
    {
        $keys = $this->getExtractedKeys();
        uasort($keys, fn($a, $b) => $b['usage_count'] - $a['usage_count']);
        
        return array_slice($keys, 0, $limit, true);
    }

    /**
     * Check if file matches patterns.
     */
    protected function matchesPatterns(string $filepath, array $patterns): bool
    {
        foreach ($patterns as $pattern) {
            if (fnmatch($pattern, basename($filepath)) || fnmatch($pattern, $filepath)) {
                return true;
            }
        }
        
        return false;
    }

    /**
     * Check if file matches exclude patterns.
     */
    protected function matchesExcludePatterns(string $filepath, array $patterns): bool
    {
        foreach ($patterns as $pattern) {
            if (fnmatch($pattern, basename($filepath)) || fnmatch($pattern, $filepath)) {
                return true;
            }
        }
        
        return false;
    }

    /**
     * Check if content is PHP.
     */
    protected function isPhpContent(string $content): bool
    {
        return str_contains($content, '<?php') || 
               preg_match('/\b(?:function|class|interface|trait|namespace|use|require|include)\b/', $content);
    }

    /**
     * Check if content is JavaScript.
     */
    protected function isJsContent(string $content): bool
    {
        return preg_match('/\b(?:function|const|let|var|import|export|class|extends)\b/', $content) ||
               str_contains($content, '=>') ||
               str_contains($content, 'React');
    }

    /**
     * Check if content is HTML.
     */
    protected function isHtmlContent(string $content): bool
    {
        return preg_match('/<[^>]+>/', $content) && 
               (str_contains($content, '<html') || str_contains($content, '<div') || str_contains($content, '<span'));
    }

    /**
     * Check if content is Twig.
     */
    protected function isTwigContent(string $content): bool
    {
        return preg_match('/\{\{.*?\}\}|\{%.*?%\}/', $content);
    }

    /**
     * Add custom parser.
     */
    public function addParser(string $extension, callable $parser): void
    {
        $this->customParsers[$extension] = $parser;
    }

    /**
     * Set custom patterns.
     */
    public function setPatterns(array $patterns): void
    {
        $this->config['file_patterns'] = $patterns;
    }

    /**
     * Set exclude patterns.
     */
    public function setExcludePatterns(array $patterns): void
    {
        $this->config['exclude_patterns'] = $patterns;
    }

    /**
     * Reset extractor state.
     */
    public function reset(): void
    {
        $this->extractedKeys = [];
        $this->fileScanned = [];
    }

    /**
     * Get default configuration.
     */
    protected function getDefaultConfig(): array
    {
        return [
            'file_patterns' => [
                '*.php',
                '*.js',
                '*.jsx',
                '*.ts',
                '*.tsx',
                '*.html',
                '*.htm',
                '*.twig',
                '*.vue'
            ],
            'exclude_patterns' => [
                'vendor/*',
                'node_modules/*',
                '.git/*',
                'storage/*',
                'bootstrap/*',
                'config/*',
                'routes/*',
                'tests/*',
                '*.min.js',
                '*.min.css',
                '*.cache'
            ],
            'php_patterns' => [
                '/trans\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/__\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/t\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/translate\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/Lang::get\s*\(\s*[\'"]([^\'"]+)[\'"]/'
            ],
            'js_patterns' => [
                '/trans\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/t\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/\$t\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/translate\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/i18n\.t\s*\(\s*[\'"]([^\'"]+)[\'"]/'
            ],
            'html_patterns' => [
                '/data-i18n\s*=\s*[\'"]([^\'"]+)[\'"]/',
                '/data-translate\s*=\s*[\'"]([^\'"]+)[\'"]/',
                '/data-trans\s*=\s*[\'"]([^\'"]+)[\'"]/'
            ],
            'twig_patterns' => [
                '/\{\{\s*trans\s*\([\'"]([^\'"]+)[\'"]/',
                '/\{\{\s*t\s*\([\'"]([^\'"]+)[\'"]/',
                '/\{\{\s*__\s*\([\'"]([^\'"]+)[\'"]/'
            ],
            'vue_patterns' => [
                '/\$t\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/t\s*\(\s*[\'"]([^\'"]+)[\'"]/',
                '/v-t\s*=\s*[\'"]([^\'"]+)[\'"]/'
            ]
        ];
    }

    /**
     * Get configuration.
     */
    public function getConfig(): array
    {
        return $this->config;
    }

    /**
     * Set configuration.
     */
    public function setConfig(array $config): void
    {
        $this->config = array_merge($this->config, $config);
    }

    /**
     * Create extractor instance.
     */
    public static function create(array $config = []): self
    {
        return new self($config);
    }

    /**
     * Create for PHP project.
     */
    public static function forPhp(): self
    {
        return new self([
            'file_patterns' => ['*.php', '*.twig'],
            'exclude_patterns' => ['vendor/*', 'node_modules/*', 'storage/*']
        ]);
    }

    /**
     * Create for JavaScript project.
     */
    public static function forJs(): self
    {
        return new self([
            'file_patterns' => ['*.js', '*.jsx', '*.ts', '*.tsx', '*.vue', '*.html'],
            'exclude_patterns' => ['node_modules/*', 'dist/*', 'build/*']
        ]);
    }

    /**
     * Create for full-stack project.
     */
    public static function forFullStack(): self
    {
        return new self([
            'file_patterns' => ['*.php', '*.js', '*.jsx', '*.ts', '*.tsx', '*.vue', '*.html', '*.twig'],
            'exclude_patterns' => ['vendor/*', 'node_modules/*', 'storage/*', 'dist/*', 'build/*']
        ]);
    }
}