Re-structure the project.

tests/test_payment_line_parser.py

@@ -0,0 +1,282 @@
+"""
+Tests for payment line parser.
+"""
+
+import pytest
+import sys
+from pathlib import Path
+
+# Add project root to path
+project_root = Path(__file__).parent.parent
+sys.path.insert(0, str(project_root))
+
+from src.inference.payment_line_parser import PaymentLineParser, PaymentLineData
+
+
+class TestPaymentLineParser:
+    """Test PaymentLineParser class."""
+
+    @pytest.fixture
+    def parser(self):
+        """Create parser instance."""
+        return PaymentLineParser()
+
+    def test_parse_full_format_with_amount(self, parser):
+        """Test parsing full format with amount."""
+        text = "# 94228110015950070 # 15658 00 8 > 48666036#14#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.ocr_number == "94228110015950070"
+        assert data.amount == "15658.00"
+        assert data.account_number == "48666036"
+        assert data.record_type == "8"
+        assert data.check_digits == "14"
+        assert data.parse_method == "full"
+
+    def test_parse_with_spaces_in_amount(self, parser):
+        """Test parsing with OCR-induced spaces in amount."""
+        text = "# 11000770600242 # 12 0 0 00 5 > 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.ocr_number == "11000770600242"
+        assert data.amount == "1200.00"  # Spaces removed
+        assert data.account_number == "3082963"
+        assert data.record_type == "5"
+        assert data.check_digits == "41"
+
+    def test_parse_with_spaces_in_check_digits(self, parser):
+        """Test parsing with spaces around check digits: #41 # instead of #41#."""
+        text = "# 6026726908 # 736 00 9 > 5692041 #41 #"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.ocr_number == "6026726908"
+        assert data.amount == "736.00"
+        assert data.account_number == "5692041"
+        assert data.check_digits == "41"
+
+    def test_parse_without_greater_than_symbol(self, parser):
+        """Test parsing when > symbol is missing (OCR error)."""
+        text = "# 11000770600242 # 1200 00 5 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.ocr_number == "11000770600242"
+        assert data.amount == "1200.00"
+        assert data.account_number == "3082963"
+
+    def test_parse_format_without_amount(self, parser):
+        """Test parsing format without amount."""
+        text = "# 11000770600242 # > 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.ocr_number == "11000770600242"
+        assert data.amount is None
+        assert data.account_number == "3082963"
+        assert data.check_digits == "41"
+        assert data.parse_method == "no_amount"
+
+    def test_parse_account_only_format(self, parser):
+        """Test parsing account-only format."""
+        text = "> 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.ocr_number == ""
+        assert data.amount is None
+        assert data.account_number == "3082963"
+        assert data.check_digits == "41"
+        assert data.parse_method == "account_only"
+        assert "Partial" in data.error
+
+    def test_parse_invalid_format(self, parser):
+        """Test parsing invalid format."""
+        text = "This is not a payment line"
+        data = parser.parse(text)
+
+        assert not data.is_valid
+        assert data.error is not None
+        assert "No valid payment line format" in data.error
+
+    def test_parse_empty_text(self, parser):
+        """Test parsing empty text."""
+        data = parser.parse("")
+
+        assert not data.is_valid
+        assert data.error == "Empty payment line text"
+
+    def test_format_machine_readable_full(self, parser):
+        """Test formatting full data to machine-readable format."""
+        data = PaymentLineData(
+            ocr_number="94228110015950070",
+            amount="15658.00",
+            account_number="48666036",
+            record_type="8",
+            check_digits="14",
+            raw_text="original",
+            is_valid=True
+        )
+
+        formatted = parser.format_machine_readable(data)
+
+        assert "# 94228110015950070 #" in formatted
+        assert "15658 00 8" in formatted
+        assert "48666036#14#" in formatted
+
+    def test_format_machine_readable_no_amount(self, parser):
+        """Test formatting data without amount."""
+        data = PaymentLineData(
+            ocr_number="11000770600242",
+            amount=None,
+            account_number="3082963",
+            record_type=None,
+            check_digits="41",
+            raw_text="original",
+            is_valid=True
+        )
+
+        formatted = parser.format_machine_readable(data)
+
+        assert "# 11000770600242 #" in formatted
+        assert "3082963#41#" in formatted
+
+    def test_format_machine_readable_account_only(self, parser):
+        """Test formatting account-only data."""
+        data = PaymentLineData(
+            ocr_number="",
+            amount=None,
+            account_number="3082963",
+            record_type=None,
+            check_digits="41",
+            raw_text="original",
+            is_valid=True
+        )
+
+        formatted = parser.format_machine_readable(data)
+
+        assert "> 3082963#41#" in formatted
+
+    def test_format_for_field_extractor_valid(self, parser):
+        """Test formatting for FieldExtractor API (valid data)."""
+        text = "# 6026726908 # 736 00 9 > 5692041#41#"
+        data = parser.parse(text)
+
+        formatted, is_valid, error = parser.format_for_field_extractor(data)
+
+        assert is_valid
+        assert formatted is not None
+        assert "# 6026726908 #" in formatted
+        assert "736 00" in formatted
+
+    def test_format_for_field_extractor_invalid(self, parser):
+        """Test formatting for FieldExtractor API (invalid data)."""
+        text = "invalid payment line"
+        data = parser.parse(text)
+
+        formatted, is_valid, error = parser.format_for_field_extractor(data)
+
+        assert not is_valid
+        assert formatted is None
+        assert error is not None
+
+
+class TestRealWorldExamples:
+    """Test with real-world payment line examples from the codebase."""
+
+    @pytest.fixture
+    def parser(self):
+        """Create parser instance."""
+        return PaymentLineParser()
+
+    def test_billo310_payment_line(self, parser):
+        """Test Billo310 PDF payment line (from issue report)."""
+        # This is the payment line that had Amount extraction issue
+        text = "# 6026726908 # 736 00 9 > 5692041 #41 #"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.amount == "736.00"  # Correct amount
+        assert data.account_number == "5692041"
+
+    def test_billo363_payment_line(self, parser):
+        """Test Billo363 PDF payment line."""
+        text = "# 11000770600242 # 12 0 0 00 5 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.amount == "1200.00"
+        assert data.ocr_number == "11000770600242"
+
+    def test_payment_line_with_spaces_in_account(self, parser):
+        """Test payment line with spaces in account number."""
+        text = "# 94228110015950070 # 15658 00 8 > 4 8 6 6 6 0 3 6#14#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.account_number == "48666036"  # Spaces removed
+
+    def test_multiple_spaces_in_amounts(self, parser):
+        """Test handling multiple spaces in amount."""
+        text = "# 11000770600242 #  1  2  0  0  00 5 > 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.amount == "1200.00"
+
+
+class TestEdgeCases:
+    """Test edge cases and error conditions."""
+
+    @pytest.fixture
+    def parser(self):
+        """Create parser instance."""
+        return PaymentLineParser()
+
+    def test_very_long_ocr_number(self, parser):
+        """Test with very long OCR number."""
+        text = "# 123456789012345678901234567890 # 1000 00 5 > 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.ocr_number == "123456789012345678901234567890"
+
+    def test_zero_amount(self, parser):
+        """Test with zero amount."""
+        text = "# 11000770600242 # 0 00 5 > 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.amount == "0.00"
+
+    def test_large_amount(self, parser):
+        """Test with large amount."""
+        text = "# 11000770600242 # 999999 99 5 > 3082963#41#"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.amount == "999999.99"
+
+    def test_text_with_extra_characters(self, parser):
+        """Test with extra characters around payment line."""
+        text = "Some text before # 6026726908 # 736 00 9 > 5692041#41# and after"
+        data = parser.parse(text)
+
+        assert data.is_valid
+        assert data.amount == "736.00"
+
+    def test_none_input(self, parser):
+        """Test with None input."""
+        data = parser.parse(None)
+
+        assert not data.is_valid
+        assert data.error is not None
+
+    def test_whitespace_only(self, parser):
+        """Test with whitespace only."""
+        data = parser.parse("   \t\n  ")
+
+        assert not data.is_valid
+        assert "Empty" in data.error