netgescon-day0/app/Services/Contabilita/UnicreditSemicolonCsvParser.php

<?php

namespace App\Services\Contabilita;

use Carbon\Carbon;

class UnicreditSemicolonCsvParser
{
    /**
     * Parser per export CSV (separatore ;) con righe introduttive e header tipo:
     * Data contabile;Data valuta;Descrizione;Accrediti;Addebiti;...
     *
     * @return array{
     *   rows: array<int, array{data: Carbon, valuta: Carbon|null, descrizione: string, importo: float, causale: string|null, raw_line: string}>,
     *   meta: array{header_row:int, delimiter:string, parsed_rows:int, skipped_rows:int}
     * }
     */
    public function parse(string $content): array
    {
        $content = str_replace(["\r\n", "\r"], "\n", $content);
        $lines = explode("\n", $content);

        $delimiter = ';';
        $headerIndex = null;

        $scanMax = min(count($lines), 300);
        for ($i = 0; $i < $scanMax; $i++) {
            $line = trim((string) $lines[$i]);
            if ($line === '') {
                continue;
            }

            if (
                stripos($line, 'Data contabile') !== false
                && stripos($line, 'Accrediti') !== false
                && stripos($line, 'Addebiti') !== false
            ) {
                $headerIndex = $i;
                break;
            }
        }

        if ($headerIndex === null) {
            throw new \RuntimeException('Formato CSV non riconosciuto: header "Data contabile" non trovato');
        }

        $header = str_getcsv((string) $lines[$headerIndex], $delimiter);
        $header = array_map(fn($h) => $this->normHeader((string) $h), $header);

        $hmap = [];
        foreach ($header as $idx => $h) {
            if ($h === '') {
                continue;
            }
            if (!array_key_exists($h, $hmap)) {
                $hmap[$h] = (int) $idx;
            }
        }

        $idxData = $hmap['datacontabile'] ?? null;
        $idxValuta = $hmap['datavaluta'] ?? null;
        $idxDescr = $hmap['descrizione'] ?? null;
        $idxAccrediti = $hmap['accrediti'] ?? null;
        $idxAddebiti = $hmap['addebiti'] ?? null;
        $idxDescrEstesa = $hmap['descrizioneestesa'] ?? null;

        if ($idxData === null || $idxDescr === null || ($idxAccrediti === null && $idxAddebiti === null)) {
            throw new \RuntimeException('Formato CSV non riconosciuto: colonne richieste mancanti');
        }

        $rows = [];
        $skipped = 0;

        for ($i = $headerIndex + 1; $i < count($lines); $i++) {
            $line = (string) $lines[$i];
            if (trim($line) === '') {
                continue;
            }

            $cols = str_getcsv($line, $delimiter);
            if (count($cols) < count($header)) {
                $cols = array_pad($cols, count($header), null);
            }

            $rawData = $cols[$idxData] ?? null;
            $data = $this->parseDate($rawData);
            if (! $data) {
                $skipped++;
                continue;
            }

            $valuta = null;
            if ($idxValuta !== null) {
                $valuta = $this->parseDate($cols[$idxValuta] ?? null);
            }

            $descrizione = trim((string) ($cols[$idxDescr] ?? ''));
            $descrEstesa = $idxDescrEstesa !== null ? trim((string) ($cols[$idxDescrEstesa] ?? '')) : '';
            if ($descrizione === '' && $descrEstesa !== '') {
                $descrizione = $descrEstesa;
            }
            if ($descrizione === '') {
                $skipped++;
                continue;
            }

            $accrediti = $idxAccrediti !== null ? $this->parseMoney($cols[$idxAccrediti] ?? null) : null;
            $addebiti = $idxAddebiti !== null ? $this->parseMoney($cols[$idxAddebiti] ?? null) : null;

            $importo = null;
            if ($accrediti !== null && abs($accrediti) > 0.0) {
                $importo = abs($accrediti);
            } elseif ($addebiti !== null && abs($addebiti) > 0.0) {
                $importo = -abs($addebiti);
            }

            if ($importo === null) {
                $skipped++;
                continue;
            }

            $rows[] = [
                'data' => $data,
                'valuta' => $valuta,
                'descrizione' => $descrizione,
                'importo' => (float) $importo,
                'causale' => null,
                'raw_line' => 'csv:' . ($i + 1),
            ];
        }

        return [
            'rows' => $rows,
            'meta' => [
                'header_row' => (int) ($headerIndex + 1),
                'delimiter' => $delimiter,
                'parsed_rows' => (int) count($rows),
                'skipped_rows' => (int) $skipped,
            ],
        ];
    }

    private function parseDate(mixed $value): ?Carbon
    {
        $s = trim((string) ($value ?? ''));
        if ($s === '') {
            return null;
        }

        foreach (['d/m/y', 'd/m/Y', 'd.m.Y', 'd.m.y', 'Y-m-d'] as $fmt) {
            try {
                return Carbon::createFromFormat($fmt, $s)->startOfDay();
            } catch (\Throwable) {
                // keep trying
            }
        }

        return null;
    }

    private function parseMoney(mixed $value): ?float
    {
        if ($value === null) {
            return null;
        }

        $s = trim((string) $value);
        if ($s === '') {
            return null;
        }

        $s = str_replace(["\u{00A0}", "\t", ' '], '', $s);
        $s = str_ireplace(['EUR', 'EURO', '€'], '', $s);
        $s = preg_replace('/[^0-9,\.\-\+]/', '', $s) ?? $s;

        // formato italiano: 1.059,26
        $s = str_replace('.', '', $s);
        $s = str_replace(',', '.', $s);

        if (! is_numeric($s)) {
            return null;
        }

        return (float) $s;
    }

    private function normHeader(string $value): string
    {
        $value = trim(mb_strtolower($value));
        $value = str_replace([" ", "\t", "\u{00A0}"], '', $value);
        $value = str_replace(['à', 'á', 'â', 'ä'], 'a', $value);
        $value = str_replace(['è', 'é', 'ê', 'ë'], 'e', $value);
        $value = str_replace(['ì', 'í', 'î', 'ï'], 'i', $value);
        $value = str_replace(['ò', 'ó', 'ô', 'ö'], 'o', $value);
        $value = str_replace(['ù', 'ú', 'û', 'ü'], 'u', $value);
        $value = preg_replace('/[^a-z0-9]/', '', $value) ?? $value;
        return $value;
    }
}