webmoney/backend/app/Services/Import/CsvParser.php

<?php

namespace App\Services\Import;

class CsvParser implements FileParserInterface
{
    protected static array $supportedExtensions = ['csv', 'txt'];

    /**
     * Parse the CSV file and return all data
     */
    public function parse(string $filePath, array $options = []): array
    {
        $headerRow = $options['header_row'] ?? 0;
        $dataStartRow = $options['data_start_row'] ?? 1;
        $delimiter = $options['delimiter'] ?? $this->detectDelimiter($filePath);
        $enclosure = $options['enclosure'] ?? '"';
        $encoding = $options['encoding'] ?? $this->detectEncoding($filePath);
        
        $data = [];
        $headers = [];
        $rowIndex = 0;
        
        $handle = fopen($filePath, 'r');
        
        if ($handle === false) {
            throw new \RuntimeException("Could not open file: $filePath");
        }
        
        while (($row = fgetcsv($handle, 0, $delimiter, $enclosure)) !== false) {
            // Converter encoding se necessário
            if ($encoding !== 'UTF-8') {
                $row = array_map(function ($value) use ($encoding) {
                    return mb_convert_encoding($value, 'UTF-8', $encoding);
                }, $row);
            }
            
            if ($rowIndex === $headerRow) {
                $headers = $row;
                $rowIndex++;
                continue;
            }
            
            if ($rowIndex < $dataStartRow) {
                $rowIndex++;
                continue;
            }
            
            // Verificar se não está vazia
            $nonEmpty = array_filter($row, fn($v) => $v !== null && $v !== '');
            if (!empty($nonEmpty)) {
                $data[] = $row;
            }
            
            $rowIndex++;
        }
        
        fclose($handle);
        
        return [
            'headers' => $headers,
            'data' => $data,
            'total_rows' => count($data),
            'detected_delimiter' => $delimiter,
            'detected_encoding' => $encoding,
        ];
    }

    /**
     * Get headers from CSV file
     */
    public function getHeaders(string $filePath, array $options = []): array
    {
        $headerRow = $options['header_row'] ?? 0;
        $delimiter = $options['delimiter'] ?? $this->detectDelimiter($filePath);
        $enclosure = $options['enclosure'] ?? '"';
        $encoding = $options['encoding'] ?? $this->detectEncoding($filePath);
        
        $handle = fopen($filePath, 'r');
        
        if ($handle === false) {
            throw new \RuntimeException("Could not open file: $filePath");
        }
        
        $rowIndex = 0;
        $headers = [];
        
        while (($row = fgetcsv($handle, 0, $delimiter, $enclosure)) !== false) {
            if ($rowIndex === $headerRow) {
                $headers = $row;
                break;
            }
            $rowIndex++;
        }
        
        fclose($handle);
        
        // Converter encoding se necessário
        if ($encoding !== 'UTF-8') {
            $headers = array_map(function ($value) use ($encoding) {
                return mb_convert_encoding($value, 'UTF-8', $encoding);
            }, $headers);
        }
        
        return $headers;
    }

    /**
     * Get preview data
     */
    public function getPreview(string $filePath, int $rows = 10, array $options = []): array
    {
        $delimiter = $options['delimiter'] ?? $this->detectDelimiter($filePath);
        $enclosure = $options['enclosure'] ?? '"';
        $encoding = $options['encoding'] ?? $this->detectEncoding($filePath);
        
        $preview = [];
        $rowCount = 0;
        $totalRows = 0;
        
        $handle = fopen($filePath, 'r');
        
        if ($handle === false) {
            throw new \RuntimeException("Could not open file: $filePath");
        }
        
        // Primeiro, contar todas as linhas
        while (fgets($handle) !== false) {
            $totalRows++;
        }
        
        // Voltar ao início
        rewind($handle);
        
        // Ler preview
        while (($row = fgetcsv($handle, 0, $delimiter, $enclosure)) !== false) {
            if ($rowCount >= $rows) {
                break;
            }
            
            // Converter encoding se necessário
            if ($encoding !== 'UTF-8') {
                $row = array_map(function ($value) use ($encoding) {
                    return mb_convert_encoding($value, 'UTF-8', $encoding);
                }, $row);
            }
            
            $preview[] = [
                'row_index' => $rowCount,
                'data' => $row,
            ];
            
            $rowCount++;
        }
        
        fclose($handle);
        
        return [
            'preview' => $preview,
            'total_rows' => $totalRows,
            'columns_count' => !empty($preview) ? count($preview[0]['data']) : 0,
            'detected_delimiter' => $delimiter,
            'detected_encoding' => $encoding,
        ];
    }

    /**
     * Detect CSV delimiter
     */
    protected function detectDelimiter(string $filePath): string
    {
        $delimiters = [',', ';', "\t", '|'];
        $counts = array_fill_keys($delimiters, 0);
        
        $handle = fopen($filePath, 'r');
        if ($handle === false) {
            return ',';
        }
        
        // Ler primeiras 5 linhas
        $lines = 0;
        while (($line = fgets($handle)) !== false && $lines < 5) {
            foreach ($delimiters as $d) {
                $counts[$d] += substr_count($line, $d);
            }
            $lines++;
        }
        
        fclose($handle);
        
        // Retornar o delimitador mais frequente
        arsort($counts);
        return array_key_first($counts);
    }

    /**
     * Detect file encoding
     */
    protected function detectEncoding(string $filePath): string
    {
        $content = file_get_contents($filePath, false, null, 0, 10000);
        
        // Verificar BOM UTF-8
        if (substr($content, 0, 3) === "\xEF\xBB\xBF") {
            return 'UTF-8';
        }
        
        // Tentar detectar encoding
        $encoding = mb_detect_encoding($content, ['UTF-8', 'ISO-8859-1', 'Windows-1252', 'ASCII'], true);
        
        return $encoding ?: 'UTF-8';
    }

    /**
     * Check if parser supports the extension
     */
    public static function supports(string $extension): bool
    {
        return in_array(strtolower($extension), self::$supportedExtensions);
    }
}