Update paddle, and support invoice line item

tests/validation/__init__.py

@@ -0,0 +1 @@
+"""Validation tests."""

tests/validation/test_vat_validator.py

@@ -0,0 +1,323 @@
+"""
+Tests for VAT Validator
+
+Tests cross-validation of VAT information from multiple sources.
+"""
+
+import pytest
+from backend.validation.vat_validator import (
+    VATValidationResult,
+    VATValidator,
+    MathCheckResult,
+)
+from backend.vat.vat_extractor import VATBreakdown, VATSummary
+from backend.table.line_items_extractor import LineItem, LineItemsResult
+
+
+class TestMathCheckResult:
+    """Tests for MathCheckResult dataclass."""
+
+    def test_create_math_check_result(self):
+        """Test creating a math check result."""
+        result = MathCheckResult(
+            rate=25.0,
+            base_amount=10000.0,
+            expected_vat=2500.0,
+            actual_vat=2500.0,
+            is_valid=True,
+            tolerance=0.01,
+        )
+        assert result.rate == 25.0
+        assert result.is_valid is True
+
+    def test_math_check_with_tolerance(self):
+        """Test math check within tolerance."""
+        result = MathCheckResult(
+            rate=25.0,
+            base_amount=10000.0,
+            expected_vat=2500.0,
+            actual_vat=2500.01,  # Within tolerance
+            is_valid=True,
+            tolerance=0.02,
+        )
+        assert result.is_valid is True
+
+
+class TestVATValidationResult:
+    """Tests for VATValidationResult dataclass."""
+
+    def test_create_validation_result(self):
+        """Test creating a validation result."""
+        result = VATValidationResult(
+            is_valid=True,
+            confidence_score=0.95,
+            math_checks=[],
+            total_check=True,
+            line_items_vs_summary=True,
+            amount_consistency=True,
+            needs_review=False,
+            review_reasons=[],
+        )
+        assert result.is_valid is True
+        assert result.confidence_score == 0.95
+        assert result.needs_review is False
+
+    def test_validation_result_with_review_reasons(self):
+        """Test validation result requiring review."""
+        result = VATValidationResult(
+            is_valid=False,
+            confidence_score=0.4,
+            math_checks=[],
+            total_check=False,
+            line_items_vs_summary=None,
+            amount_consistency=False,
+            needs_review=True,
+            review_reasons=["Math check failed", "Total mismatch"],
+        )
+        assert result.is_valid is False
+        assert result.needs_review is True
+        assert len(result.review_reasons) == 2
+
+
+class TestVATValidator:
+    """Tests for VATValidator."""
+
+    def test_validate_simple_vat(self):
+        """Test validating simple single-rate VAT."""
+        validator = VATValidator()
+
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="2 500,00", source="regex")
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 500,00",
+            total_incl_vat="12 500,00",
+            confidence=0.9,
+        )
+
+        result = validator.validate(vat_summary)
+
+        assert result.is_valid is True
+        assert result.confidence_score >= 0.9
+        assert result.total_check is True
+
+    def test_validate_multiple_vat_rates(self):
+        """Test validating multiple VAT rates."""
+        validator = VATValidator()
+
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="8 000,00", vat_amount="2 000,00", source="regex"),
+                VATBreakdown(rate=12.0, base_amount="2 000,00", vat_amount="240,00", source="regex"),
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 240,00",
+            total_incl_vat="12 240,00",
+            confidence=0.9,
+        )
+
+        result = validator.validate(vat_summary)
+
+        assert result.is_valid is True
+        assert len(result.math_checks) == 2
+
+    def test_validate_math_check_failure(self):
+        """Test detecting math check failure."""
+        validator = VATValidator()
+
+        # VAT amount doesn't match rate
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="3 000,00", source="regex")  # Should be 2500
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="3 000,00",
+            total_incl_vat="13 000,00",
+            confidence=0.9,
+        )
+
+        result = validator.validate(vat_summary)
+
+        assert result.is_valid is False
+        assert result.needs_review is True
+        assert any("Math" in reason or "math" in reason for reason in result.review_reasons)
+
+    def test_validate_total_mismatch(self):
+        """Test detecting total amount mismatch."""
+        validator = VATValidator()
+
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="2 500,00", source="regex")
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 500,00",
+            total_incl_vat="15 000,00",  # Wrong - should be 12500
+            confidence=0.9,
+        )
+
+        result = validator.validate(vat_summary)
+
+        assert result.total_check is False
+        assert result.needs_review is True
+
+    def test_validate_with_line_items(self):
+        """Test validation with line items comparison."""
+        validator = VATValidator()
+
+        line_items = LineItemsResult(
+            items=[
+                LineItem(row_index=0, description="Item 1", amount="5 000,00", vat_rate="25"),
+                LineItem(row_index=1, description="Item 2", amount="5 000,00", vat_rate="25"),
+            ],
+            header_row=["Description", "Amount"],
+            raw_html="<table>...</table>",
+        )
+
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="2 500,00", source="regex")
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 500,00",
+            total_incl_vat="12 500,00",
+            confidence=0.9,
+        )
+
+        result = validator.validate(vat_summary, line_items=line_items)
+
+        assert result.line_items_vs_summary is not None
+
+    def test_validate_amount_consistency(self):
+        """Test consistency check with extracted amount field."""
+        validator = VATValidator()
+
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="2 500,00", source="regex")
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 500,00",
+            total_incl_vat="12 500,00",
+            confidence=0.9,
+        )
+
+        # Existing amount field from YOLO extraction
+        existing_amount = "12 500,00"
+
+        result = validator.validate(vat_summary, existing_amount=existing_amount)
+
+        assert result.amount_consistency is True
+
+    def test_validate_amount_inconsistency(self):
+        """Test detecting amount field inconsistency."""
+        validator = VATValidator()
+
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="2 500,00", source="regex")
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 500,00",
+            total_incl_vat="12 500,00",
+            confidence=0.9,
+        )
+
+        # Different amount from YOLO extraction
+        existing_amount = "15 000,00"
+
+        result = validator.validate(vat_summary, existing_amount=existing_amount)
+
+        assert result.amount_consistency is False
+        assert result.needs_review is True
+
+    def test_validate_empty_summary(self):
+        """Test validating empty VAT summary."""
+        validator = VATValidator()
+
+        vat_summary = VATSummary(
+            breakdowns=[],
+            total_excl_vat=None,
+            total_vat=None,
+            total_incl_vat=None,
+            confidence=0.0,
+        )
+
+        result = validator.validate(vat_summary)
+
+        assert result.confidence_score == 0.0
+        assert result.is_valid is False
+
+    def test_validate_without_base_amounts(self):
+        """Test validation when base amounts are not available."""
+        validator = VATValidator()
+
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount=None, vat_amount="2 500,00", source="regex")
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 500,00",
+            total_incl_vat="12 500,00",
+            confidence=0.9,
+        )
+
+        result = validator.validate(vat_summary)
+
+        # Should still validate totals even without per-rate base amounts
+        assert result.total_check is True
+
+    def test_confidence_score_calculation(self):
+        """Test confidence score calculation."""
+        validator = VATValidator()
+
+        # All checks pass - high confidence
+        vat_summary_good = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="2 500,00", source="regex")
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 500,00",
+            total_incl_vat="12 500,00",
+            confidence=0.95,
+        )
+        result_good = validator.validate(vat_summary_good)
+
+        # Some checks fail - lower confidence
+        vat_summary_bad = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="3 000,00", source="regex")
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="3 000,00",
+            total_incl_vat="12 500,00",  # Doesn't match
+            confidence=0.5,
+        )
+        result_bad = validator.validate(vat_summary_bad)
+
+        assert result_good.confidence_score > result_bad.confidence_score
+
+    def test_tolerance_configuration(self):
+        """Test configurable tolerance for math checks."""
+        # Strict tolerance
+        validator_strict = VATValidator(tolerance=0.001)
+        # Lenient tolerance
+        validator_lenient = VATValidator(tolerance=1.0)
+
+        vat_summary = VATSummary(
+            breakdowns=[
+                VATBreakdown(rate=25.0, base_amount="10 000,00", vat_amount="2 500,50", source="regex")  # Off by 0.50
+            ],
+            total_excl_vat="10 000,00",
+            total_vat="2 500,50",
+            total_incl_vat="12 500,50",
+            confidence=0.9,
+        )
+
+        result_strict = validator_strict.validate(vat_summary)
+        result_lenient = validator_lenient.validate(vat_summary)
+
+        # Strict should fail, lenient should pass
+        assert result_strict.math_checks[0].is_valid is False
+        assert result_lenient.math_checks[0].is_valid is True