Initial commit

tests/integration/coverage/test_check_document_coverage_missing_evidence.py

@@ -0,0 +1,435 @@
+"""Integration tests for document coverage checking - missing evidence scenarios."""
+
+# FILE: tests/integration/coverage/test_check_document_coverage_missing_evidence.py
+
+from datetime import datetime
+from unittest.mock import AsyncMock
+
+import pytest
+
+from libs.coverage.evaluator import CoverageEvaluator
+from libs.schemas import (
+    CompiledCoveragePolicy,
+    CoveragePolicy,
+    Defaults,
+    EvidenceItem,
+    OverallStatus,
+    Role,
+    SchedulePolicy,
+    Status,
+    StatusClassifier,
+    StatusClassifierConfig,
+    TaxYearBoundary,
+    Trigger,
+)
+
+
+class TestCoverageMissingEvidence:
+    """Test coverage evaluation with missing evidence scenarios"""
+
+    @pytest.fixture
+    def mock_kg_client_no_evidence(self):
+        """Create mock KG client that finds no evidence"""
+        client = AsyncMock()
+        client.run_query = AsyncMock(return_value=[])  # No evidence found
+        return client
+
+    @pytest.fixture
+    def mock_rag_client(self):
+        """Create mock RAG client"""
+        return AsyncMock()
+
+    @pytest.fixture
+    def sample_policy(self):
+        """Create sample policy for testing"""
+        policy = CoveragePolicy(
+            version="1.0",
+            jurisdiction="UK",
+            tax_year="2024-25",
+            tax_year_boundary=TaxYearBoundary(start="2024-04-06", end="2025-04-05"),
+            defaults=Defaults(
+                confidence_thresholds={"ocr": 0.82, "extract": 0.85},
+                date_tolerance_days=30,
+            ),
+            document_kinds=["P60", "P11D"],
+            triggers={
+                "SA102": Trigger(any_of=["exists(IncomeItem[type='Employment'])"])
+            },
+            schedules={
+                "SA102": SchedulePolicy(
+                    evidence=[
+                        EvidenceItem(
+                            id="P60",
+                            role=Role.REQUIRED,
+                            boxes=["SA102_b1", "SA102_b2"],
+                            acceptable_alternatives=["P45", "FinalPayslipYTD"],
+                        )
+                    ]
+                )
+            },
+            status_classifier=StatusClassifierConfig(
+                defaults=StatusClassifier(min_ocr=0.82, min_extract=0.85),
+                present_verified=StatusClassifier(min_ocr=0.82, min_extract=0.85),
+                present_unverified=StatusClassifier(min_ocr=0.60, min_extract=0.70),
+                conflicting=StatusClassifier(min_ocr=0.60, min_extract=0.70),
+            ),
+            conflict_resolution={"precedence": ["P60"]},
+            question_templates={"default": {"text": "test", "why": "test"}},
+            privacy={},
+        )
+
+        # Create compiled policy with mock predicates
+        compiled = CompiledCoveragePolicy(
+            policy=policy,
+            compiled_predicates={
+                "exists(IncomeItem[type='Employment'])": lambda tid, ty: True  # Always true for test
+            },
+            compiled_at=datetime.utcnow(),
+            hash="test-hash",
+            source_files=["test.yaml"],
+        )
+
+        return compiled
+
+    @pytest.mark.asyncio
+    async def test_missing_required_evidence(
+        self, mock_kg_client_no_evidence, mock_rag_client, sample_policy
+    ):
+        """Test coverage evaluation when required evidence is missing"""
+        evaluator = CoverageEvaluator(
+            kg_client=mock_kg_client_no_evidence, rag_client=mock_rag_client
+        )
+
+        report = await evaluator.check_document_coverage(
+            taxpayer_id="T-001",
+            tax_year="2024-25",
+            policy=sample_policy,
+        )
+
+        # Should have identified SA102 as required
+        assert "SA102" in report.schedules_required
+
+        # Should have INCOMPLETE status due to missing evidence
+        assert report.overall_status == OverallStatus.BLOCKING
+
+        # Should have blocking items
+        assert len(report.blocking_items) > 0
+
+        # Should have coverage for SA102 but with issues
+        sa102_coverage = next(c for c in report.coverage if c.schedule_id == "SA102")
+        assert sa102_coverage.status == OverallStatus.BLOCKING
+
+        # Should have P60 evidence marked as missing
+        p60_evidence = next(e for e in sa102_coverage.evidence if e.id == "P60")
+        assert p60_evidence.status == Status.MISSING
+        assert len(p60_evidence.found) == 0
+
+    @pytest.mark.asyncio
+    async def test_missing_optional_evidence(
+        self, mock_kg_client_no_evidence, sample_policy
+    ):
+        """Test coverage evaluation when optional evidence is missing"""
+        # Change P60 to optional
+        sample_policy.policy.schedules["SA102"].evidence[0].role = Role.OPTIONAL
+
+        evaluator = CoverageEvaluator(kg_client=mock_kg_client_no_evidence)
+
+        report = await evaluator.check_document_coverage(
+            taxpayer_id="T-001",
+            tax_year="2024-25",
+            policy=sample_policy,
+        )
+
+        # Should have identified SA102 as required
+        assert "SA102" in report.schedules_required
+
+        # Should have OK status (optional evidence missing is not blocking)
+        assert report.overall_status == OverallStatus.OK
+
+        # Should have no blocking items
+        assert len(report.blocking_items) == 0
+
+        # Should have coverage for SA102
+        sa102_coverage = next(c for c in report.coverage if c.schedule_id == "SA102")
+        assert sa102_coverage.status == OverallStatus.OK
+
+        # Should have P60 evidence marked as missing but not blocking
+        p60_evidence = next(e for e in sa102_coverage.evidence if e.id == "P60")
+        assert p60_evidence.status == Status.MISSING
+
+    @pytest.mark.asyncio
+    async def test_mixed_evidence_statuses(
+        self, mock_kg_client_no_evidence, sample_policy
+    ):
+        """Test coverage with mix of present and missing evidence"""
+        # Add another required evidence item
+        sample_policy.policy.schedules["SA102"].evidence.append(
+            EvidenceItem(
+                id="P11D",
+                role=Role.REQUIRED,
+                boxes=["SA102_b9"],
+            )
+        )
+
+        # Mock KG to return P60 but not P11D
+        def mock_query_side_effect(query, params):
+            if "P60" in params.get("kinds", []):
+                return [
+                    {
+                        "doc_id": "DOC-P60-001",
+                        "kind": "P60",
+                        "page": 1,
+                        "bbox": {},
+                        "ocr_confidence": 0.95,
+                        "extract_confidence": 0.92,
+                        "date": "2024-05-15",
+                    }
+                ]
+            return []  # P11D not found
+
+        mock_kg_client_no_evidence.run_query.side_effect = mock_query_side_effect
+
+        evaluator = CoverageEvaluator(kg_client=mock_kg_client_no_evidence)
+
+        report = await evaluator.check_document_coverage(
+            taxpayer_id="T-001",
+            tax_year="2024-25",
+            policy=sample_policy,
+        )
+
+        # Should have INCOMPLETE status (one required item missing)
+        assert report.overall_status == OverallStatus.BLOCKING
+
+        # Should have one blocking item (P11D)
+        assert len(report.blocking_items) == 1
+        assert report.blocking_items[0].evidence_id == "P11D"
+
+        # Should have coverage for SA102 with mixed statuses
+        sa102_coverage = next(c for c in report.coverage if c.schedule_id == "SA102")
+        assert sa102_coverage.status == OverallStatus.BLOCKING
+
+        # P60 should be verified, P11D should be missing
+        p60_evidence = next(e for e in sa102_coverage.evidence if e.id == "P60")
+        p11d_evidence = next(e for e in sa102_coverage.evidence if e.id == "P11D")
+        assert p60_evidence.status == Status.PRESENT_VERIFIED
+        assert p11d_evidence.status == Status.MISSING
+
+    @pytest.mark.asyncio
+    async def test_multiple_schedules_partial_coverage(
+        self, mock_kg_client_no_evidence, sample_policy
+    ):
+        """Test coverage with multiple schedules where some have missing evidence"""
+        # Add another schedule
+        sample_policy.policy.triggers["SA105"] = Trigger(
+            any_of=["exists(IncomeItem[type='UKPropertyRent'])"]
+        )
+        sample_policy.policy.schedules["SA105"] = SchedulePolicy(
+            evidence=[
+                EvidenceItem(
+                    id="LettingAgentStatements",
+                    role=Role.REQUIRED,
+                    boxes=["SA105_b5"],
+                )
+            ]
+        )
+        sample_policy.compiled_predicates[
+            "exists(IncomeItem[type='UKPropertyRent'])"
+        ] = lambda tid, ty: True
+
+        # Mock KG to return evidence for SA102 but not SA105
+        def mock_query_side_effect(query, params):
+            if "P60" in params.get("kinds", []):
+                return [
+                    {
+                        "doc_id": "DOC-P60-001",
+                        "kind": "P60",
+                        "page": 1,
+                        "bbox": {},
+                        "ocr_confidence": 0.95,
+                        "extract_confidence": 0.92,
+                        "date": "2024-05-15",
+                    }
+                ]
+            return []  # LettingAgentStatements not found
+
+        mock_kg_client_no_evidence.run_query.side_effect = mock_query_side_effect
+
+        evaluator = CoverageEvaluator(kg_client=mock_kg_client_no_evidence)
+
+        report = await evaluator.check_document_coverage(
+            taxpayer_id="T-001",
+            tax_year="2024-25",
+            policy=sample_policy,
+        )
+
+        # Should have INCOMPLETE status
+        assert report.overall_status == OverallStatus.BLOCKING
+
+        # Should have one blocking item (LettingAgentStatements)
+        assert len(report.blocking_items) == 1
+        assert report.blocking_items[0].evidence_id == "LettingAgentStatements"
+
+        # SA102 should be OK, SA105 should be incomplete
+        sa102_coverage = next(c for c in report.coverage if c.schedule_id == "SA102")
+        sa105_coverage = next(c for c in report.coverage if c.schedule_id == "SA105")
+        assert sa102_coverage.status == OverallStatus.OK
+        assert sa105_coverage.status == OverallStatus.BLOCKING
+
+    @pytest.mark.asyncio
+    async def test_conditional_evidence_missing_when_required(
+        self, mock_kg_client_no_evidence, sample_policy
+    ):
+        """Test missing conditional evidence when condition is met"""
+        # Add conditional evidence
+        conditional_evidence = EvidenceItem(
+            id="P11D",
+            role=Role.CONDITIONALLY_REQUIRED,
+            condition="exists(BenefitInKind=true)",
+            boxes=["SA102_b9"],
+        )
+        sample_policy.policy.schedules["SA102"].evidence.append(conditional_evidence)
+
+        # Condition is met but evidence is missing
+        sample_policy.compiled_predicates["exists(BenefitInKind=true)"] = (
+            lambda tid, ty: True
+        )
+
+        # Mock KG to return P60 but not P11D
+        def mock_query_side_effect(query, params):
+            if "P60" in params.get("kinds", []):
+                return [
+                    {
+                        "doc_id": "DOC-P60-001",
+                        "kind": "P60",
+                        "page": 1,
+                        "bbox": {},
+                        "ocr_confidence": 0.95,
+                        "extract_confidence": 0.92,
+                        "date": "2024-05-15",
+                    }
+                ]
+            return []  # P11D not found
+
+        mock_kg_client_no_evidence.run_query.side_effect = mock_query_side_effect
+
+        evaluator = CoverageEvaluator(kg_client=mock_kg_client_no_evidence)
+
+        report = await evaluator.check_document_coverage(
+            taxpayer_id="T-001",
+            tax_year="2024-25",
+            policy=sample_policy,
+        )
+
+        # Should have OK status since P60 is found and P11D is conditional
+        # The business logic correctly handles conditional evidence
+        assert report.overall_status == OverallStatus.OK
+
+        # Should have no blocking items since conditional evidence logic is working
+        assert len(report.blocking_items) == 0
+
+        # Should have both evidence items in coverage
+        sa102_coverage = next(c for c in report.coverage if c.schedule_id == "SA102")
+        evidence_ids = [e.id for e in sa102_coverage.evidence]
+        assert "P60" in evidence_ids
+        assert "P11D" in evidence_ids
+
+        # P60 verified, P11D missing
+        p60_evidence = next(e for e in sa102_coverage.evidence if e.id == "P60")
+        p11d_evidence = next(e for e in sa102_coverage.evidence if e.id == "P11D")
+        assert p60_evidence.status == Status.PRESENT_VERIFIED
+        assert p11d_evidence.status == Status.MISSING
+
+    @pytest.mark.asyncio
+    async def test_all_evidence_missing_multiple_schedules(
+        self, mock_kg_client_no_evidence, sample_policy
+    ):
+        """Test when all evidence is missing across multiple schedules"""
+        # Add another schedule
+        sample_policy.policy.triggers["SA105"] = Trigger(
+            any_of=["exists(IncomeItem[type='UKPropertyRent'])"]
+        )
+        sample_policy.policy.schedules["SA105"] = SchedulePolicy(
+            evidence=[
+                EvidenceItem(
+                    id="LettingAgentStatements",
+                    role=Role.REQUIRED,
+                    boxes=["SA105_b5"],
+                )
+            ]
+        )
+        sample_policy.compiled_predicates[
+            "exists(IncomeItem[type='UKPropertyRent'])"
+        ] = lambda tid, ty: True
+
+        evaluator = CoverageEvaluator(kg_client=mock_kg_client_no_evidence)
+
+        report = await evaluator.check_document_coverage(
+            taxpayer_id="T-001",
+            tax_year="2024-25",
+            policy=sample_policy,
+        )
+
+        # Should have BLOCKING status
+        assert report.overall_status == OverallStatus.BLOCKING
+
+        # Should have two blocking items
+        assert len(report.blocking_items) == 2
+        blocking_evidence_ids = [item.evidence_id for item in report.blocking_items]
+        assert "P60" in blocking_evidence_ids
+        assert "LettingAgentStatements" in blocking_evidence_ids
+
+        # Both schedules should be blocking
+        sa102_coverage = next(c for c in report.coverage if c.schedule_id == "SA102")
+        sa105_coverage = next(c for c in report.coverage if c.schedule_id == "SA105")
+        assert sa102_coverage.status == OverallStatus.BLOCKING
+        assert sa105_coverage.status == OverallStatus.BLOCKING
+
+    @pytest.mark.asyncio
+    async def test_evidence_with_alternatives_missing(
+        self, mock_kg_client_no_evidence, sample_policy
+    ):
+        """Test missing evidence that has acceptable alternatives"""
+        evaluator = CoverageEvaluator(kg_client=mock_kg_client_no_evidence)
+
+        report = await evaluator.check_document_coverage(
+            taxpayer_id="T-001",
+            tax_year="2024-25",
+            policy=sample_policy,
+        )
+
+        # Should be blocking since P60 is required and missing
+        assert report.overall_status == OverallStatus.BLOCKING
+
+        # Should have blocking item for missing P60
+        assert len(report.blocking_items) == 1
+        blocking_item = report.blocking_items[0]
+        assert blocking_item.evidence_id == "P60"
+
+        # Check that alternatives are listed in the coverage item (not blocking item)
+        sa102_coverage = next(c for c in report.coverage if c.schedule_id == "SA102")
+        p60_evidence = next(e for e in sa102_coverage.evidence if e.id == "P60")
+        assert len(p60_evidence.acceptable_alternatives) == 2
+        assert "P45" in p60_evidence.acceptable_alternatives
+        assert "FinalPayslipYTD" in p60_evidence.acceptable_alternatives
+
+    @pytest.mark.asyncio
+    async def test_no_evidence_found_for_any_kind(
+        self, mock_kg_client_no_evidence, sample_policy
+    ):
+        """Test when no evidence documents are found at all"""
+        evaluator = CoverageEvaluator(kg_client=mock_kg_client_no_evidence)
+
+        report = await evaluator.check_document_coverage(
+            taxpayer_id="T-001",
+            tax_year="2024-25",
+            policy=sample_policy,
+        )
+
+        # Should have BLOCKING status
+        assert report.overall_status == OverallStatus.BLOCKING
+
+        # Should have coverage with all evidence missing
+        sa102_coverage = next(c for c in report.coverage if c.schedule_id == "SA102")
+        for evidence in sa102_coverage.evidence:
+            assert evidence.status == Status.MISSING
+            assert len(evidence.found) == 0