Steuerkram/src/Services/BankImportParser.php

<?php

declare(strict_types=1);

namespace Steuerkram\Services;

class BankImportParser
{
    /**
     * Parst eine hochgeladene Bank-Datei (CSV/TXT/XLSX) anhand des Formats.
     * Gibt ein Array von Zeilen zurück, jede Zeile als assoc array.
     *
     * @throws \RuntimeException bei Parsefehlern
     */
    public function parse(string $filePath, string $extension, array $format): array
    {
        $ext = strtolower($extension);

        if ($ext === 'xlsx') {
            $lines = $this->readXlsx($filePath);
        } else {
            $lines = $this->readCsv($filePath, $format['delimiter'], $format['enclosure']);
        }

        // Übersprungene Zeilen und optionale Headerzeile entfernen
        $skipRows = (int)$format['skip_rows'];
        $lines    = array_slice($lines, $skipRows);
        if (!empty($format['has_header_row'])) {
            $lines = array_slice($lines, 1);
        }

        $colBuchungstag  = (int)$format['col_buchungstag'];
        $colWertstellung = (int)$format['col_wertstellung'];
        $colVorgang      = (int)$format['col_vorgang'];
        $colBuchungstext = (int)$format['col_buchungstext'];
        $colWert         = (int)$format['col_wert'];

        $result = [];
        foreach ($lines as $i => $cells) {
            // Leerzeilen überspringen
            $joined = implode('', $cells);
            if (trim($joined) === '') {
                continue;
            }

            $buchungstag  = $this->parseDate($cells[$colBuchungstag]  ?? '', $format['date_format']);
            $wertstellung = $this->parseDate($cells[$colWertstellung] ?? '', $format['date_format']);
            $wert         = $this->parseDecimal(
                $cells[$colWert] ?? '0',
                $format['decimal_separator'],
                $format['thousands_separator']
            );

            if ($buchungstag === null || $wertstellung === null) {
                continue;
            }

            $result[] = [
                'buchungstag'  => $buchungstag,
                'wertstellung' => $wertstellung,
                'vorgang'      => trim($cells[$colVorgang]      ?? ''),
                'buchungstext' => trim($cells[$colBuchungstext] ?? ''),
                'wert'         => $wert,
                'raw_line'     => implode($format['delimiter'], $cells),
            ];
        }

        return $result;
    }

    private function readCsv(string $filePath, string $delimiter, string $enclosure): array
    {
        $content = file_get_contents($filePath);
        if ($content === false) {
            throw new \RuntimeException('Datei konnte nicht gelesen werden.');
        }

        // BOM entfernen
        $content = preg_replace('/^\xEF\xBB\xBF/', '', $content);

        // Verschiedene Zeilenenden normalisieren
        $content = str_replace(["\r\n", "\r"], "\n", $content);
        $lines   = explode("\n", $content);

        $result = [];
        foreach ($lines as $line) {
            if (trim($line) === '') {
                $result[] = [];
                continue;
            }
            $result[] = $this->parseCsvLine($line, $delimiter, $enclosure);
        }
        return $result;
    }

    private function parseCsvLine(string $line, string $delimiter, string $enclosure): array
    {
        $cells = [];
        $len   = strlen($line);
        $i     = 0;

        while ($i < $len) {
            if ($line[$i] === $enclosure) {
                // Quoted cell
                $i++;
                $cell = '';
                while ($i < $len) {
                    if ($line[$i] === $enclosure) {
                        if ($i + 1 < $len && $line[$i + 1] === $enclosure) {
                            $cell .= $enclosure;
                            $i += 2;
                        } else {
                            $i++;
                            break;
                        }
                    } else {
                        $cell .= $line[$i++];
                    }
                }
                $cells[] = $cell;
                // Delimiter überspringen
                if ($i < $len && $line[$i] === $delimiter) {
                    $i++;
                }
            } else {
                // Unquoted cell
                $end  = strpos($line, $delimiter, $i);
                if ($end === false) {
                    $cells[] = substr($line, $i);
                    break;
                }
                $cells[] = substr($line, $i, $end - $i);
                $i       = $end + 1;
            }
        }

        return $cells;
    }

    private function readXlsx(string $filePath): array
    {
        if (!class_exists('ZipArchive')) {
            throw new \RuntimeException('ZipArchive nicht verfügbar – XLSX kann nicht gelesen werden.');
        }

        $zip = new \ZipArchive();
        if ($zip->open($filePath) !== true) {
            throw new \RuntimeException('XLSX-Datei konnte nicht geöffnet werden.');
        }

        // Shared Strings laden (optional)
        $sharedStrings = [];
        $ssXml = $zip->getFromName('xl/sharedStrings.xml');
        if ($ssXml !== false) {
            $ss = simplexml_load_string($ssXml);
            foreach ($ss->si ?? [] as $si) {
                $text = '';
                if (isset($si->t)) {
                    $text = (string)$si->t;
                } else {
                    foreach ($si->r ?? [] as $r) {
                        $text .= (string)($r->t ?? '');
                    }
                }
                $sharedStrings[] = $text;
            }
        }

        // Erstes Worksheet lesen
        $sheetXml = $zip->getFromName('xl/worksheets/sheet1.xml');
        $zip->close();

        if ($sheetXml === false) {
            throw new \RuntimeException('Kein Arbeitsblatt in der XLSX-Datei gefunden.');
        }

        $sheet   = simplexml_load_string($sheetXml);
        $rows    = [];
        $maxCol  = 0;

        foreach ($sheet->sheetData->row ?? [] as $row) {
            $rowArr = [];
            foreach ($row->c as $cell) {
                [$colIdx] = $this->xlsxCellRef((string)$cell['r']);
                if ($colIdx > $maxCol) $maxCol = $colIdx;

                $type  = (string)($cell['t'] ?? '');
                $value = (string)($cell->v ?? '');

                if ($type === 's') {
                    $value = $sharedStrings[(int)$value] ?? '';
                } elseif ($type === 'inlineStr') {
                    $value = (string)($cell->is->t ?? '');
                }

                $rowArr[$colIdx] = $value;
            }

            // Lücken auffüllen
            $filled = [];
            for ($c = 0; $c <= $maxCol; $c++) {
                $filled[] = $rowArr[$c] ?? '';
            }
            $rows[] = $filled;
        }

        return $rows;
    }

    private function xlsxCellRef(string $ref): array
    {
        preg_match('/^([A-Z]+)(\d+)$/', $ref, $m);
        $col = 0;
        foreach (str_split($m[1] ?? 'A') as $char) {
            $col = $col * 26 + (ord($char) - ord('A') + 1);
        }
        return [$col - 1, (int)($m[2] ?? 1) - 1];
    }

    private function parseDate(string $value, string $format): ?string
    {
        $value = trim($value);
        if ($value === '') return null;

        $d = \DateTime::createFromFormat($format, $value);
        if (!$d) return null;

        return $d->format('Y-m-d');
    }

    private function parseDecimal(string $value, string $decimal, string $thousands): float
    {
        $value = trim($value);
        // Tausendertrennzeichen entfernen, Dezimaltrennzeichen → Punkt
        $value = str_replace($thousands, '', $value);
        $value = str_replace($decimal, '.', $value);
        $value = preg_replace('/[^\d.\-]/', '', $value);
        return (float)$value;
    }
}