finance-duck/internal/banking/csv.go

package banking

import (
	"bufio"
	"encoding/csv"
	"fmt"
	"io"
	"strings"
	"time"
	"unicode"

	"finance-duck/internal/domain"
)

// ParseCSV accepts N26 English and German account-activity exports, including
// their older Date/Datum and newer Booking Date/Buchungsdatum schemas. Supported
// columns: Date/Datum/Booking Date/Buchungsdatum, Value Date/Wertstellung/
// Wertstellungsdatum, Payee/Partner Name/Zahlungsempfänger/Empfänger/Auftraggeber,
// Account number/Kontonummer/IBAN, Payment reference/Verwendungszweck,
// Payment type/Transaktionstyp, Amount (EUR)/Betrag (EUR), and optional
// Currency/Währung and Transaction ID/Transaktions-ID. Foreign-original-amount,
// exchange-rate and category columns are deliberately not used for account money.
// Comma and semicolon delimiters, UTF-8 BOM, CRLF, RFC4180 quoted multiline
// descriptions, ISO and German dates, decimal comma and decimal point are accepted.
// Missing required booking-date or account-amount columns fail the entire import.
func ParseCSV(input io.Reader, account domain.Account) ([]domain.Facts, error) {
	if account.ID == "" {
		return nil, fmt.Errorf("CSV requires a selected account")
	}
	reader := bufio.NewReader(input)
	first, err := reader.ReadString('\n')
	if err != nil && err != io.EOF {
		return nil, fmt.Errorf("read CSV header: %w", err)
	}
	first = strings.TrimPrefix(first, "\ufeff")
	delimiter := ','
	// Count separators outside quotes; descriptions may contain either delimiter.
	quoted := false
	commas, semicolons := 0, 0
	for _, r := range first {
		if r == '"' {
			quoted = !quoted
		}
		if !quoted {
			if r == ',' {
				commas++
			}
			if r == ';' {
				semicolons++
			}
		}
	}
	if semicolons > commas {
		delimiter = ';'
	}
	parser := csv.NewReader(io.MultiReader(strings.NewReader(first), reader))
	parser.Comma = delimiter
	headers, err := parser.Read()
	if err != nil {
		return nil, fmt.Errorf("invalid N26 CSV header")
	}
	columns := make(map[string]int)
	amountCurrency := ""
	for i, h := range headers {
		name := headerName(h)
		key := ""
		switch name {
		case "date", "datum", "booking date", "buchungsdatum":
			key = "date"
		case "value date", "wertstellung", "wertstellungsdatum", "valutadatum":
			key = "value"
		case "payee", "partner name", "zahlungsempfänger", "zahlungsempfänger name", "empfänger", "empfänger/auftraggeber", "partnername", "name zahlungspartner":
			key = "party"
		case "account number", "partner iban", "kontonummer", "iban", "konto":
			key = "iban"
		case "payment reference", "verwendungszweck", "reference", "beschreibung":
			key = "description"
		case "payment type", "transaktionstyp", "zahlungstyp", "type", "typ":
			key = "type"
		case "currency", "währung":
			key = "currency"
		case "transaction id", "transaktions-id", "transaktions id":
			key = "external"
		case "amount", "betrag":
			key = "amount"
		default:
			for _, prefix := range []string{"amount (", "betrag ("} {
				if strings.HasPrefix(name, prefix) && strings.HasSuffix(name, ")") {
					candidate := strings.ToUpper(strings.TrimSuffix(strings.TrimPrefix(name, prefix), ")"))
					if validCurrency(candidate) {
						key = "amount"
						amountCurrency = candidate
					}
				}
			}
		}
		if key != "" {
			if _, exists := columns[key]; exists {
				return nil, fmt.Errorf("duplicate N26 CSV column %s", key)
			}
			columns[key] = i
		}
	}
	if _, ok := columns["date"]; !ok {
		return nil, fmt.Errorf("N26 CSV requires Date/Datum or Booking Date/Buchungsdatum")
	}
	if _, ok := columns["amount"]; !ok {
		return nil, fmt.Errorf("N26 CSV requires Amount (currency)/Betrag (currency)")
	}
	get := func(row []string, key string) string {
		if i, ok := columns[key]; ok {
			return strings.TrimSpace(row[i])
		}
		return ""
	}
	result := make([]domain.Facts, 0)
	for rowNumber := 2; ; rowNumber++ {
		row, err := parser.Read()
		if err == io.EOF {
			break
		}
		if err != nil {
			return nil, fmt.Errorf("invalid N26 CSV record %d", rowNumber)
		}
		date, err := parseDate(get(row, "date"))
		if err != nil {
			return nil, fmt.Errorf("invalid booking date in CSV record %d", rowNumber)
		}
		value := get(row, "value")
		if value != "" {
			value, err = parseDate(value)
			if err != nil {
				return nil, fmt.Errorf("invalid value date in CSV record %d", rowNumber)
			}
		}
		amount, err := parseCSVAmount(get(row, "amount"))
		if err != nil {
			return nil, fmt.Errorf("invalid account amount in CSV record %d", rowNumber)
		}
		currency := strings.ToUpper(get(row, "currency"))
		if currency == "" {
			currency = amountCurrency
		}
		if currency == "" {
			currency = strings.ToUpper(account.Currency)
		}
		if !validCurrency(currency) || (amountCurrency != "" && currency != amountCurrency) || (account.Currency != "" && currency != strings.ToUpper(account.Currency)) {
			return nil, fmt.Errorf("invalid or conflicting account currency in CSV record %d", rowNumber)
		}
		description := get(row, "description")
		if description == "" {
			description = get(row, "type")
		}
		result = append(result, domain.Facts{Source: "n26_csv", AccountID: account.ID, BookingDate: date, ValueDate: value, Amount: amount, Currency: currency, RawDescription: description, ExternalID: get(row, "external"), Counterparty: get(row, "party"), CounterpartyIBAN: normalizeIBAN(get(row, "iban"))})
	}
	return result, nil
}

func headerName(s string) string {
	return strings.ToLower(strings.Join(strings.Fields(strings.TrimPrefix(s, "\ufeff")), " "))
}
func normalizeIBAN(s string) string {
	return strings.ToUpper(strings.Map(func(r rune) rune {
		if unicode.IsSpace(r) {
			return -1
		}
		return r
	}, s))
}
func validCurrency(s string) bool {
	if len(s) != 3 {
		return false
	}
	for _, c := range s {
		if c < 'A' || c > 'Z' {
			return false
		}
	}
	return true
}
func parseDate(s string) (string, error) {
	for _, layout := range []string{"2006-01-02", "02.01.2006", "2.1.2006"} {
		if d, e := time.Parse(layout, s); e == nil {
			return d.Format("2006-01-02"), nil
		}
	}
	return "", fmt.Errorf("invalid date")
}
func parseCSVAmount(s string) (domain.Money, error) {
	s = strings.TrimPrefix(strings.TrimSpace(s), "+")
	// German grouping is only accepted when every group is exactly three digits.
	if strings.Contains(s, ",") {
		if strings.Count(s, ",") != 1 {
			return "", fmt.Errorf("invalid decimal separator")
		}
		pair := strings.SplitN(s, ",", 2)
		if strings.Contains(pair[0], ".") {
			groups := strings.Split(strings.TrimLeft(pair[0], "+-"), ".")
			if len(groups[0]) < 1 || len(groups[0]) > 3 {
				return "", fmt.Errorf("invalid grouping")
			}
			for _, g := range groups[1:] {
				if len(g) != 3 {
					return "", fmt.Errorf("invalid grouping")
				}
			}
			pair[0] = strings.ReplaceAll(pair[0], ".", "")
		}
		s = pair[0] + "." + pair[1]
	}
	return domain.ParseMoney(s)
}