WIP

src/inference/test_field_extractor.py

@@ -266,6 +266,38 @@ class TestNormalizePaymentLine:
         assert is_valid is True
         # Bankgiro should be normalized despite spaces
 
+    def test_payment_line_with_spaces_in_check_digits(self, extractor):
+        """Test payment line with spaces around check digits: #41 # instead of #41#."""
+        text = "# 6026726908 # 736 00 9 > 5692041 #41 #"
+        result, is_valid, error = extractor._normalize_payment_line(text)
+
+        assert result is not None
+        assert is_valid is True
+        assert "6026726908" in result
+        assert "736 00" in result
+        assert "5692041#41#" in result
+
+    def test_payment_line_with_ocr_spaces_in_amount(self, extractor):
+        """Test payment line with OCR-induced spaces in amount: '12 0 0 00' -> '1200 00'."""
+        text = "# 11000770600242 # 12 0 0 00 5 3082963#41#"
+        result, is_valid, error = extractor._normalize_payment_line(text)
+
+        assert result is not None
+        assert is_valid is True
+        assert "11000770600242" in result
+        assert "1200 00" in result
+        assert "3082963#41#" in result
+
+    def test_payment_line_without_greater_symbol(self, extractor):
+        """Test payment line with missing > symbol (low-DPI OCR issue)."""
+        text = "# 11000770600242 # 1200 00 5 3082963#41#"
+        result, is_valid, error = extractor._normalize_payment_line(text)
+
+        assert result is not None
+        assert is_valid is True
+        assert "11000770600242" in result
+        assert "1200 00" in result
+
 
 class TestNormalizeCustomerNumber:
     """Tests for customer number normalization."""
@@ -284,6 +316,33 @@ class TestNormalizeCustomerNumber:
         result, is_valid, error = extractor._normalize_customer_number("JTY5763")
         assert result is not None
 
+    def test_format_without_dash(self, extractor):
+        """Test customer number format without dash: Dwq 211X -> DWQ 211-X."""
+        text = "Dwq 211X Billo SE 106 43 Stockholm"
+        result, is_valid, error = extractor._normalize_customer_number(text)
+
+        assert result is not None
+        assert is_valid is True
+        assert result == "DWQ 211-X"
+
+    def test_swedish_postal_code_exclusion(self, extractor):
+        """Test that Swedish postal codes are excluded: SE 106 43 should not be extracted."""
+        text = "SE 106 43 Stockholm"
+        result, is_valid, error = extractor._normalize_customer_number(text)
+
+        # Should not extract postal code
+        assert result is None or "SE 106" not in result
+
+    def test_customer_number_with_postal_code_in_text(self, extractor):
+        """Test extracting customer number when postal code is also present."""
+        text = "Customer: ABC 123X, Address: SE 106 43 Stockholm"
+        result, is_valid, error = extractor._normalize_customer_number(text)
+
+        assert result is not None
+        assert "ABC" in result
+        # Should not extract postal code
+        assert "SE 106" not in result if result else True
+
 
 class TestNormalizeSupplierOrgNumber:
     """Tests for supplier organization number normalization."""