cmd/pdfamount/main.go

@@ -18,7 +18,36 @@ import (
 	onlyoffice "github.com/eslider/go-onlyoffice"
 )
 
-var amountRe = regexp.MustCompile(`(?i)(zu zahlender betrag|rechnungsbetrag)\s*[:\s]*([0-9][0-9.]*,[0-9]{2})`)
+// amountLabels are the payable-amount labels in priority order: the first
+// label present in a document wins. Within one label the last amount is taken,
+// because totals usually come last.
+//
+// "gesamtsumme" is not in the original list but is the real label on Diashop
+// invoices: "Gesamtsumme (inkl. Steuern)".
+var amountLabels = []string{
+	"zu zahlender betrag",
+	"rechnungsbetrag",
+	"rechnungsendbetrag",
+	"endbetrag",
+	"zahlbetrag",
+	"bruttobetrag",
+	"gesamtbetrag",
+	"gesamtsumme",
+	"betrag",
+	"total",
+	"summe",
+}
+
+// amountRes matches "<label> [optional (comment)] [: -] <number>" for every
+// label, in the same priority order as amountLabels.
+var amountRes = func() []*regexp.Regexp {
+	res := make([]*regexp.Regexp, 0, len(amountLabels))
+	for _, label := range amountLabels {
+		res = append(res, regexp.MustCompile(
+			`(?i)\b`+regexp.QuoteMeta(label)+`\b\s*(?:\([^)]*\))?\s*[:\-]?\s*([0-9]+(?:[.,][0-9]+)*)`))
+	}
+	return res
+}()
 
 func main() {
 	if len(os.Args) < 2 {
@@ -105,20 +134,63 @@ func pdfAmount(ctx context.Context, c *onlyoffice.Client, id string) (string, er
 	if err := cmd.Run(); err != nil {
 		return "", err
 	}
-	m := amountRe.FindStringSubmatch(buf.String())
+	return extractAmount(buf.String()), nil
-	if m == nil {
-		return "", nil
-	}
-	return parseDe(m[2]), nil
 }
 
-// parseDe turns "1.234,56" into 1234.56.
+// extractAmount returns the normalised ("1234.56") payable amount found in
-func parseDe(s string) string {
+// text, or "" if no known label matches.
-	s = strings.ReplaceAll(s, ".", "")
+func extractAmount(text string) string {
-	s = strings.ReplaceAll(s, ",", ".")
+	for _, re := range amountRes {
-	v, err := strconv.ParseFloat(s, 64)
+		ms := re.FindAllStringSubmatch(text, -1)
-	if err != nil {
+		if len(ms) == 0 {
-		return s
+			continue
+		}
+		if v, ok := normalizeAmount(ms[len(ms)-1][1]); ok {
+			return v
+		}
 	}
-	return strconv.FormatFloat(v, 'f', 2, 64)
+	return ""
+}
+
+// normalizeAmount turns "1.234,56" (DE), "1,234.56" (EN) or "1234.56" into
+// "1234.56". The rightmost separator is decimal only when followed by one or
+// two digits; otherwise every separator is a thousands separator.
+func normalizeAmount(s string) (string, bool) {
+	last := -1
+	for i := 0; i < len(s); i++ {
+		if s[i] == '.' || s[i] == ',' {
+			last = i
+		}
+	}
+	var dec byte
+	if last >= 0 {
+		digits := 0
+		for i := last + 1; i < len(s); i++ {
+			if s[i] < '0' || s[i] > '9' {
+				return "", false
+			}
+			digits++
+		}
+		if digits == 1 || digits == 2 {
+			dec = s[last]
+		}
+	}
+	var b strings.Builder
+	for i := 0; i < len(s); i++ {
+		switch c := s[i]; {
+		case c >= '0' && c <= '9':
+			b.WriteByte(c)
+		case (c == '.' || c == ',') && c == dec:
+			b.WriteByte('.')
+		case c == '.' || c == ',':
+			// thousands separator
+		default:
+			return "", false
+		}
+	}
+	v, err := strconv.ParseFloat(b.String(), 64)
+	if err != nil {
+		return "", false
+	}
+	return strconv.FormatFloat(v, 'f', 2, 64), true
 }

cmd/pdfamount/main_test.go

@@ -0,0 +1,130 @@
+package main
+
+import "testing"
+
+func TestExtractAmount(t *testing.T) {
+	tests := []struct {
+		name, text, want string
+	}{
+		{
+			name: "rechnungsbetrag de format",
+			text: "Rechnungsbetrag: 1.234,56 €",
+			want: "1234.56",
+		},
+		{
+			name: "rechnungsbetrag en thousands and dot",
+			text: "Rechnungsbetrag: 1,234.56",
+			want: "1234.56",
+		},
+		{
+			name: "rechnungsbetrag plain dot",
+			text: "Rechnungsbetrag: 1234.56",
+			want: "1234.56",
+		},
+		{
+			name: "rechnungsbetrag de comma only",
+			text: "Rechnungsbetrag: 1234,56",
+			want: "1234.56",
+		},
+		{
+			name: "currency suffix eur",
+			text: "Rechnungsbetrag: 1.234,56 EUR",
+			want: "1234.56",
+		},
+		{
+			name: "zu zahlender betrag wins over rechnungsbetrag",
+			text: "Zu zahlender Betrag: 10,00\nRechnungsbetrag: 99,00",
+			want: "10.00",
+		},
+		{
+			name: "rechnungsbetrag wins over endbetrag",
+			text: "Endbetrag: 20,00\nRechnungsbetrag: 30,00",
+			want: "30.00",
+		},
+		{
+			name: "bruttobetrag wins over bare betrag",
+			text: "Bruttobetrag: 50,00\nBetrag: 10,00",
+			want: "50.00",
+		},
+		{
+			name: "gesamtbetrag wins over bare betrag",
+			text: "Gesamtbetrag: 80,00\nBetrag: 10,00",
+			want: "80.00",
+		},
+		{
+			name: "last occurrence of same label wins",
+			text: "Rechnungsbetrag: 10,00\nRechnungsbetrag: 20,00",
+			want: "20.00",
+		},
+		{
+			name: "endbetrag fallback",
+			text: "Endbetrag: 42,00",
+			want: "42.00",
+		},
+		{
+			name: "zahlbetrag fallback without colon",
+			text: "Zahlbetrag 7,50 €",
+			want: "7.50",
+		},
+		{
+			name: "rechnungsendbetrag beats endbetrag",
+			text: "Rechnungsendbetrag: 12,00\nEndbetrag: 13,00",
+			want: "12.00",
+		},
+		{
+			name: "dkv style total line",
+			text: "Kundenbezogene Daten\n» TOTAL: 123,45 100,00 23,45 123,45\n",
+			want: "123.45",
+		},
+		{
+			name: "diashop style gesamtsumme with comment",
+			text: "Zwischensumme\n12,34 €\nZwischensumme\n12,34 €\nVersand & Bearbeitung\n4,95 €\nGesamtsumme (inkl. Steuern)\n17,29 €\n",
+			want: "17.29",
+		},
+		{
+			name: "diashop picks inclusive total last",
+			text: "Gesamtsumme (exkl. Steuern)\n12,34 €\nGesamtsumme (inkl. Steuern)\n17,29 €",
+			want: "17.29",
+		},
+		{
+			name: "no label",
+			text: "some text without any amount label 12,34",
+			want: "",
+		},
+	}
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			if got := extractAmount(tc.text); got != tc.want {
+				t.Fatalf("extractAmount()=%q want %q", got, tc.want)
+			}
+		})
+	}
+}
+
+func TestNormalizeAmount(t *testing.T) {
+	tests := []struct {
+		in   string
+		want string
+		ok   bool
+	}{
+		{"1.234,56", "1234.56", true},
+		{"1,234.56", "1234.56", true},
+		{"1234.56", "1234.56", true},
+		{"1234,56", "1234.56", true},
+		{"1.234.567,89", "1234567.89", true},
+		{"1,234,567.89", "1234567.89", true},
+		{"1.234", "1234.00", true},
+		{"12,5", "12.50", true},
+		{"12", "12.00", true},
+		{"", "0.00", false},
+	}
+	for _, tc := range tests {
+		got, ok := normalizeAmount(tc.in)
+		if ok != tc.ok {
+			t.Fatalf("normalizeAmount(%q) ok=%v want %v", tc.in, ok, tc.ok)
+		}
+		if ok && got != tc.want {
+			t.Fatalf("normalizeAmount(%q)=%q want %q", tc.in, got, tc.want)
+		}
+	}
+}