ai-tax-agent/tests/unit/test_event_schemas.py

"""Tests for event schema validation."""

import pytest
from pydantic import ValidationError

from libs.events.topics import EventTopics
from libs.schemas.events import (
    EVENT_SCHEMA_MAP,
    CalculationReadyEventData,
    DocumentExtractedEventData,
    DocumentIngestedEventData,
    DocumentOCRReadyEventData,
    FirmSyncCompletedEventData,
    FormFilledEventData,
    HMRCSubmittedEventData,
    KGUpsertedEventData,
    KGUpsertReadyEventData,
    RAGIndexedEventData,
    ReviewCompletedEventData,
    ReviewRequestedEventData,
    get_schema_for_topic,
    validate_event_data,
)


class TestDocumentIngestedEventData:
    """Test DocumentIngestedEventData schema."""

    def test_valid_event(self) -> None:
        """Test creating a valid document ingested event."""
        data = DocumentIngestedEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            filename="invoice_2024.pdf",
            mime_type="application/pdf",
            size_bytes=102400,
            checksum_sha256="a" * 64,
            kind="invoice",
            source="manual_upload",
            storage_path="raw-documents/2024/invoice_2024.pdf",
        )
        assert data.doc_id == "01H8Y9Z5M3K7N2P4Q6R8T0V1W3"
        assert data.size_bytes == 102400
        assert len(data.checksum_sha256) == 64

    def test_invalid_checksum(self) -> None:
        """Test invalid SHA-256 checksum."""
        with pytest.raises(ValidationError) as exc_info:
            DocumentIngestedEventData(
                doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
                filename="test.pdf",
                mime_type="application/pdf",
                size_bytes=1024,
                checksum_sha256="invalid",  # Too short
                kind="invoice",
                source="manual_upload",
                storage_path="path/to/file",
            )
        assert "Invalid SHA-256 checksum format" in str(exc_info.value)

    def test_negative_size(self) -> None:
        """Test negative file size validation."""
        with pytest.raises(ValidationError):
            DocumentIngestedEventData(
                doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
                filename="test.pdf",
                mime_type="application/pdf",
                size_bytes=-1,  # Negative size
                checksum_sha256="a" * 64,
                kind="invoice",
                source="manual_upload",
                storage_path="path/to/file",
            )

    def test_immutable(self) -> None:
        """Test that event data is immutable."""
        data = DocumentIngestedEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            filename="test.pdf",
            mime_type="application/pdf",
            size_bytes=1024,
            checksum_sha256="a" * 64,
            kind="invoice",
            source="manual_upload",
            storage_path="path/to/file",
        )
        with pytest.raises(ValidationError):
            data.filename = "changed.pdf"  # Should raise because frozen=True


class TestDocumentOCRReadyEventData:
    """Test DocumentOCRReadyEventData schema."""

    def test_valid_event(self) -> None:
        """Test creating a valid OCR ready event."""
        data = DocumentOCRReadyEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            ocr_engine="tesseract",
            page_count=3,
            confidence_avg=0.95,
            text_length=5000,
            layout_detected=True,
            languages_detected=["en"],
            processing_time_ms=1500,
            storage_path="ocr-results/doc_123.json",
        )
        assert data.ocr_engine == "tesseract"
        assert data.confidence_avg == 0.95
        assert 0.0 <= data.confidence_avg <= 1.0

    def test_invalid_confidence(self) -> None:
        """Test invalid confidence score."""
        with pytest.raises(ValidationError):
            DocumentOCRReadyEventData(
                doc_id="123",
                ocr_engine="tesseract",
                page_count=1,
                confidence_avg=1.5,  # > 1.0
                text_length=100,
                layout_detected=True,
                processing_time_ms=1000,
                storage_path="path",
            )

    def test_invalid_ocr_engine(self) -> None:
        """Test invalid OCR engine value."""
        with pytest.raises(ValidationError):
            DocumentOCRReadyEventData(
                doc_id="123",
                ocr_engine="invalid_engine",  # Not in allowed values
                page_count=1,
                confidence_avg=0.9,
                text_length=100,
                layout_detected=True,
                processing_time_ms=1000,
                storage_path="path",
            )


class TestDocumentExtractedEventData:
    """Test DocumentExtractedEventData schema."""

    def test_valid_event(self) -> None:
        """Test creating a valid extraction event."""
        data = DocumentExtractedEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            extraction_id="extr_123",
            strategy="hybrid",
            fields_extracted=15,
            confidence_avg=0.88,
            calibrated_confidence=0.91,
            model_name="gpt-4",
            processing_time_ms=3000,
            storage_path="extractions/extr_123.json",
        )
        assert data.strategy == "hybrid"
        assert data.model_name == "gpt-4"

    def test_valid_without_model(self) -> None:
        """Test extraction event without model (rules-based)."""
        data = DocumentExtractedEventData(
            doc_id="123",
            extraction_id="extr_456",
            strategy="rules",
            fields_extracted=10,
            confidence_avg=0.95,
            calibrated_confidence=0.93,
            model_name=None,  # No model for rules-based
            processing_time_ms=500,
            storage_path="path",
        )
        assert data.model_name is None
        assert data.strategy == "rules"


class TestKGEvents:
    """Test Knowledge Graph event schemas."""

    def test_kg_upsert_ready(self) -> None:
        """Test KG upsert ready event."""
        data = KGUpsertReadyEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            entity_count=25,
            relationship_count=40,
            tax_year="2024-25",
            taxpayer_id="TP-001",
            normalization_id="norm_123",
            storage_path="normalized/norm_123.json",
        )
        assert data.entity_count == 25
        assert data.tax_year == "2024-25"

    def test_kg_upserted(self) -> None:
        """Test KG upserted event."""
        data = KGUpsertedEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            entities_created=10,
            entities_updated=5,
            relationships_created=20,
            relationships_updated=10,
            shacl_violations=0,
            processing_time_ms=2000,
            success=True,
            error_message=None,
        )
        assert data.success is True
        assert data.shacl_violations == 0

    def test_kg_upserted_with_violations(self) -> None:
        """Test KG upserted event with SHACL violations."""
        data = KGUpsertedEventData(
            doc_id="123",
            entities_created=5,
            entities_updated=0,
            relationships_created=8,
            relationships_updated=0,
            shacl_violations=3,
            processing_time_ms=1500,
            success=False,
            error_message="SHACL validation failed: Missing required property",
        )
        assert data.success is False
        assert data.shacl_violations == 3
        assert data.error_message is not None


class TestRAGIndexedEventData:
    """Test RAG indexed event schema."""

    def test_valid_event(self) -> None:
        """Test creating a valid RAG indexed event."""
        data = RAGIndexedEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            collection_name="firm_knowledge",
            chunks_indexed=45,
            embedding_model="bge-small-en-v1.5",
            pii_detected=True,
            pii_redacted=True,
            processing_time_ms=5000,
            storage_path="chunks/doc_123.json",
        )
        assert data.pii_detected is True
        assert data.pii_redacted is True
        assert data.chunks_indexed == 45


class TestCalculationReadyEventData:
    """Test calculation ready event schema."""

    def test_valid_event(self) -> None:
        """Test creating a valid calculation event."""
        data = CalculationReadyEventData(
            taxpayer_id="TP-001",
            tax_year="2024-25",
            schedule_id="SA103",
            calculation_id="calc_789",
            boxes_computed=50,
            total_income=85000.50,
            total_tax=18500.25,
            confidence=0.92,
            evidence_count=15,
            processing_time_ms=2500,
            storage_path="calculations/calc_789.json",
        )
        assert data.schedule_id == "SA103"
        assert data.total_income == 85000.50
        assert data.total_tax == 18500.25

    def test_valid_without_totals(self) -> None:
        """Test calculation event without totals (partial calculation)."""
        data = CalculationReadyEventData(
            taxpayer_id="TP-001",
            tax_year="2024-25",
            schedule_id="SA102",
            calculation_id="calc_456",
            boxes_computed=20,
            total_income=None,
            total_tax=None,
            confidence=0.85,
            evidence_count=10,
            processing_time_ms=1000,
            storage_path="calculations/calc_456.json",
        )
        assert data.total_income is None
        assert data.total_tax is None


class TestFormFilledEventData:
    """Test form filled event schema."""

    def test_valid_event(self) -> None:
        """Test creating a valid form filled event."""
        data = FormFilledEventData(
            taxpayer_id="TP-001",
            tax_year="2024-25",
            form_id="SA100",
            fields_filled=75,
            pdf_size_bytes=524288,
            storage_path="forms/SA100_filled.pdf",
            evidence_bundle_path="evidence/bundle_123.zip",
            checksum_sha256="b" * 64,
        )
        assert data.form_id == "SA100"
        assert data.evidence_bundle_path is not None


class TestHMRCSubmittedEventData:
    """Test HMRC submitted event schema."""

    def test_successful_submission(self) -> None:
        """Test successful HMRC submission."""
        data = HMRCSubmittedEventData(
            taxpayer_id="TP-001",
            tax_year="2024-25",
            submission_id="sub_999",
            hmrc_reference="HMRC-REF-12345",
            submission_type="sandbox",
            success=True,
            status_code=200,
            error_message=None,
            processing_time_ms=3000,
        )
        assert data.success is True
        assert data.hmrc_reference is not None

    def test_failed_submission(self) -> None:
        """Test failed HMRC submission."""
        data = HMRCSubmittedEventData(
            taxpayer_id="TP-001",
            tax_year="2024-25",
            submission_id="sub_888",
            hmrc_reference=None,
            submission_type="live",
            success=False,
            status_code=400,
            error_message="Invalid UTR number",
            processing_time_ms=1500,
        )
        assert data.success is False
        assert data.error_message is not None

    def test_invalid_submission_type(self) -> None:
        """Test invalid submission type."""
        with pytest.raises(ValidationError):
            HMRCSubmittedEventData(
                taxpayer_id="TP-001",
                tax_year="2024-25",
                submission_id="sub_777",
                hmrc_reference=None,
                submission_type="invalid",  # Not in allowed values
                success=False,
                status_code=None,
                error_message=None,
                processing_time_ms=1000,
            )


class TestReviewEvents:
    """Test review event schemas."""

    def test_review_requested(self) -> None:
        """Test review requested event."""
        data = ReviewRequestedEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            review_type="extraction",
            priority="high",
            reason="Low confidence extraction (0.65)",
            assigned_to="reviewer@example.com",
            due_date="2024-12-01T10:00:00Z",
            metadata={"extraction_id": "extr_123"},
        )
        assert data.priority == "high"
        assert data.review_type == "extraction"

    def test_review_completed(self) -> None:
        """Test review completed event."""
        data = ReviewCompletedEventData(
            doc_id="01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            review_id="rev_456",
            reviewer="reviewer@example.com",
            decision="approved",
            changes_made=3,
            comments="Fixed vendor name and amount",
            review_duration_seconds=180,
        )
        assert data.decision == "approved"
        assert data.changes_made == 3


class TestFirmSyncCompletedEventData:
    """Test firm sync completed event schema."""

    def test_successful_sync(self) -> None:
        """Test successful firm sync."""
        data = FirmSyncCompletedEventData(
            firm_id="FIRM-001",
            connector_type="xero",
            sync_id="sync_123",
            records_synced=150,
            records_created=50,
            records_updated=100,
            records_failed=0,
            success=True,
            error_message=None,
            processing_time_ms=10000,
        )
        assert data.success is True
        assert data.records_failed == 0

    def test_partial_sync_failure(self) -> None:
        """Test sync with some failures."""
        data = FirmSyncCompletedEventData(
            firm_id="FIRM-002",
            connector_type="sage",
            sync_id="sync_456",
            records_synced=90,
            records_created=30,
            records_updated=60,
            records_failed=10,
            success=True,  # Overall success despite some failures
            error_message="10 records failed validation",
            processing_time_ms=15000,
        )
        assert data.records_failed == 10
        assert data.error_message is not None


class TestSchemaMapping:
    """Test schema mapping and validation utilities."""

    def test_all_topics_have_schemas(self) -> None:
        """Test that all topics in EventTopics have corresponding schemas."""
        topic_values = {
            getattr(EventTopics, attr)
            for attr in dir(EventTopics)
            if not attr.startswith("_")
        }
        schema_topics = set(EVENT_SCHEMA_MAP.keys())

        # All event topics should have schemas
        missing_schemas = topic_values - schema_topics
        assert not missing_schemas, f"Missing schemas for topics: {missing_schemas}"

    def test_validate_event_data(self) -> None:
        """Test validate_event_data function."""
        valid_data = {
            "doc_id": "01H8Y9Z5M3K7N2P4Q6R8T0V1W3",
            "filename": "test.pdf",
            "mime_type": "application/pdf",
            "size_bytes": 1024,
            "checksum_sha256": "a" * 64,
            "kind": "invoice",
            "source": "manual_upload",
            "storage_path": "path/to/file",
        }

        result = validate_event_data("doc.ingested", valid_data)
        assert isinstance(result, DocumentIngestedEventData)
        assert result.doc_id == "01H8Y9Z5M3K7N2P4Q6R8T0V1W3"

    def test_validate_unknown_topic(self) -> None:
        """Test validation with unknown topic."""
        with pytest.raises(ValueError, match="Unknown event topic"):
            validate_event_data("unknown.topic", {})

    def test_validate_invalid_data(self) -> None:
        """Test validation with invalid data."""
        invalid_data = {
            "doc_id": "123",
            "filename": "test.pdf",
            # Missing required fields
        }

        with pytest.raises(ValidationError):
            validate_event_data("doc.ingested", invalid_data)

    def test_get_schema_for_topic(self) -> None:
        """Test get_schema_for_topic function."""
        schema = get_schema_for_topic("doc.ingested")
        assert schema == DocumentIngestedEventData

    def test_get_schema_unknown_topic(self) -> None:
        """Test get_schema_for_topic with unknown topic."""
        with pytest.raises(ValueError, match="Unknown event topic"):
            get_schema_for_topic("unknown.topic")

    def test_schema_prevents_extra_fields(self) -> None:
        """Test that schemas prevent extra fields (extra='forbid')."""
        with pytest.raises(ValidationError) as exc_info:
            DocumentIngestedEventData(
                doc_id="123",
                filename="test.pdf",
                mime_type="application/pdf",
                size_bytes=1024,
                checksum_sha256="a" * 64,
                kind="invoice",
                source="manual_upload",
                storage_path="path",
                unexpected_field="should_fail",  # Extra field
            )
        assert "Extra inputs are not permitted" in str(exc_info.value)