erp-nutzfahrzeuge/backend/tests/test_ocr.py

"""Tests for OCR module: upload, results, apply, and processing with mocked OpenRouter."""

import io
import uuid
from unittest.mock import AsyncMock, MagicMock, patch

import pytest
import pytest_asyncio
from httpx import ASGITransport, AsyncClient
from sqlalchemy import select
from sqlalchemy.ext.asyncio import AsyncSession

from app.database import get_db
from app.main import app
from app.models.ocr_result import OCRResult, OCRStatus
from app.models.vehicle import Vehicle
from app.services import ocr_service

# Ensure all models are registered with Base.metadata
from app.models import user, vehicle as vehicle_model, ocr_result  # noqa: F401


@pytest_asyncio.fixture
async def test_vehicle(db_session: AsyncSession) -> Vehicle:
    """Create a test vehicle for OCR linking."""
    vehicle = Vehicle(
        make="Mercedes",
        model="Actros",
        fin="WDB9066351L123456",
        year=2020,
        power_kw=350,
        fuel_type="Diesel",
        condition="used",
        availability="available",
        price=45000,
        vehicle_type="lkw",
    )
    db_session.add(vehicle)
    await db_session.commit()
    await db_session.refresh(vehicle)
    return vehicle


def _make_mock_openrouter_response(
    confidence: float = 0.85,
    brand: str = "Mercedes",
    model: str = "Actros",
    vin: str = "WDB9066351L123456",
    first_registration: str = "15.03.2020",
    mileage: int = 120000,
    power_kw: int = 350,
    fuel_type: str = "Diesel",
) -> dict:
    """Build a mock OpenRouter response dict."""
    return {
        "structured_data": {
            "brand": brand,
            "model": model,
            "vin": vin,
            "first_registration": first_registration,
            "mileage": mileage,
            "power_kw": power_kw,
            "fuel_type": fuel_type,
        },
        "confidence_score": confidence,
        "raw_text": f'{{"brand": "{brand}", "model": "{model}", "vin": "{vin}", "confidence_score": {confidence}}}',
    }


class TestOCRService:
    """Unit tests for ocr_service functions."""

    @pytest.mark.asyncio
    async def test_validate_mime_type_valid(self):
        assert ocr_service.validate_mime_type("image/png") is True
        assert ocr_service.validate_mime_type("image/jpeg") is True
        assert ocr_service.validate_mime_type("image/webp") is True

    @pytest.mark.asyncio
    async def test_validate_mime_type_invalid(self):
        assert ocr_service.validate_mime_type("application/pdf") is False
        assert ocr_service.validate_mime_type("text/plain") is False
        assert ocr_service.validate_mime_type("") is False

    @pytest.mark.asyncio
    async def test_validate_file_size_valid(self):
        # 1 MB should be valid (limit is 50 MB)
        assert ocr_service.validate_file_size(1024 * 1024) is True

    @pytest.mark.asyncio
    async def test_validate_file_size_too_large(self):
        # 51 MB should be invalid
        assert ocr_service.validate_file_size(51 * 1024 * 1024) is False

    @pytest.mark.asyncio
    async def test_upload_file_success(self, db_session: AsyncSession, tmp_path):
        """Test uploading a valid image file creates an OCRResult."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            file_bytes = b"fake-image-data"
            result = await ocr_service.upload_file(
                db=db_session,
                file_bytes=file_bytes,
                file_name="scan.png",
                mime_type="image/png",
            )
            assert result.id is not None
            assert result.status == OCRStatus.pending
            assert result.file_name == "scan.png"
            assert result.mime_type == "image/png"
            assert result.file_path.endswith(".png")

    @pytest.mark.asyncio
    async def test_upload_file_invalid_mime(self, db_session: AsyncSession):
        """Test uploading with invalid MIME type raises ValueError."""
        with pytest.raises(ValueError, match="Invalid MIME type"):
            await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"data",
                file_name="doc.pdf",
                mime_type="application/pdf",
            )

    @pytest.mark.asyncio
    async def test_upload_file_too_large(self, db_session: AsyncSession):
        """Test uploading a file that exceeds size limit raises ValueError."""
        large_bytes = b"x" * (51 * 1024 * 1024)
        with pytest.raises(ValueError, match="File size exceeds"):
            await ocr_service.upload_file(
                db=db_session,
                file_bytes=large_bytes,
                file_name="big.png",
                mime_type="image/png",
            )

    @pytest.mark.asyncio
    async def test_get_result_not_found(self, db_session: AsyncSession):
        """Test getting a non-existent OCR result returns None."""
        result = await ocr_service.get_result(db_session, uuid.uuid4())
        assert result is None

    @pytest.mark.asyncio
    async def test_list_results_empty(self, db_session: AsyncSession):
        """Test listing OCR results when none exist."""
        items, total = await ocr_service.list_results(db_session)
        assert total == 0
        assert items == []

    @pytest.mark.asyncio
    async def test_list_results_with_data(self, db_session: AsyncSession, tmp_path):
        """Test listing OCR results with data."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"img1",
                file_name="scan1.png",
                mime_type="image/png",
            )
            await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"img2",
                file_name="scan2.png",
                mime_type="image/png",
            )
            await db_session.commit()

        items, total = await ocr_service.list_results(db_session)
        assert total == 2
        assert len(items) == 2

    @pytest.mark.asyncio
    async def test_list_results_filter_by_vehicle(
        self, db_session: AsyncSession, test_vehicle: Vehicle, tmp_path
    ):
        """Test listing OCR results filtered by vehicle_id."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"img1",
                file_name="scan1.png",
                mime_type="image/png",
                vehicle_id=test_vehicle.id,
            )
            await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"img2",
                file_name="scan2.png",
                mime_type="image/png",
            )
            await db_session.commit()

        items, total = await ocr_service.list_results(
            db_session, vehicle_id=test_vehicle.id
        )
        assert total == 1
        assert len(items) == 1
        assert items[0].vehicle_id == test_vehicle.id

    @pytest.mark.asyncio
    async def test_process_ocr_high_confidence(
        self, db_session: AsyncSession, tmp_path
    ):
        """Test OCR processing with high confidence sets status to completed."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            ocr_result = await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"fake-image",
                file_name="scan.png",
                mime_type="image/png",
            )
            await db_session.commit()

        mock_response = _make_mock_openrouter_response(confidence=0.92)
        with patch(
            "app.services.ocr_service.perform_ocr",
            new_callable=AsyncMock,
            return_value=mock_response,
        ):
            result = await ocr_service.process_ocr(db_session, ocr_result.id)

        assert result.status == OCRStatus.completed
        assert result.confidence_score == 0.92
        assert result.structured_data is not None
        assert result.structured_data["brand"] == "Mercedes"
        assert result.structured_data["model"] == "Actros"
        assert result.structured_data["vin"] == "WDB9066351L123456"

    @pytest.mark.asyncio
    async def test_process_ocr_low_confidence(self, db_session: AsyncSession, tmp_path):
        """Test OCR processing with low confidence sets status to manual_review."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            ocr_result = await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"fake-image",
                file_name="scan.png",
                mime_type="image/png",
            )
            await db_session.commit()

        mock_response = _make_mock_openrouter_response(confidence=0.45)
        with patch(
            "app.services.ocr_service.perform_ocr",
            new_callable=AsyncMock,
            return_value=mock_response,
        ):
            result = await ocr_service.process_ocr(db_session, ocr_result.id)

        assert result.status == OCRStatus.manual_review
        assert result.confidence_score == 0.45

    @pytest.mark.asyncio
    async def test_process_ocr_openrouter_failure(
        self, db_session: AsyncSession, tmp_path
    ):
        """Test OCR processing when OpenRouter fails sets status to failed."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            ocr_result = await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"fake-image",
                file_name="scan.png",
                mime_type="image/png",
            )
            await db_session.commit()

        with patch(
            "app.services.ocr_service.perform_ocr",
            new_callable=AsyncMock,
            side_effect=Exception("OpenRouter API unavailable"),
        ):
            result = await ocr_service.process_ocr(db_session, ocr_result.id)

        assert result.status == OCRStatus.failed
        assert result.error_message is not None
        assert "OpenRouter API unavailable" in result.error_message

    @pytest.mark.asyncio
    async def test_apply_to_vehicle_success(
        self, db_session: AsyncSession, test_vehicle: Vehicle, tmp_path
    ):
        """Test applying OCR data to a vehicle."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            ocr_result = await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"fake-image",
                file_name="scan.png",
                mime_type="image/png",
                vehicle_id=test_vehicle.id,
            )
            # Set structured data manually
            ocr_result.structured_data = {
                "brand": "MAN",
                "model": "TGX",
                "vin": "WDB9066351L123456",
                "first_registration": "15.03.2020",
                "mileage": 85000,
                "power_kw": 400,
                "fuel_type": "Diesel",
            }
            ocr_result.confidence_score = 0.88
            ocr_result.status = OCRStatus.completed
            await db_session.commit()

        ocr, vehicle, updated_fields = await ocr_service.apply_to_vehicle(
            db_session, ocr_result.id
        )

        assert vehicle.make == "MAN"
        assert vehicle.model == "TGX"
        assert vehicle.mileage_km == 85000
        assert vehicle.power_kw == 400
        assert vehicle.fuel_type == "Diesel"
        assert "make" in updated_fields
        assert "model" in updated_fields
        assert "mileage_km" in updated_fields

    @pytest.mark.asyncio
    async def test_apply_to_vehicle_no_vehicle(
        self, db_session: AsyncSession, tmp_path
    ):
        """Test applying OCR data when no vehicle is linked."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            ocr_result = await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"fake-image",
                file_name="scan.png",
                mime_type="image/png",
            )
            ocr_result.structured_data = {"brand": "MAN"}
            await db_session.commit()

        with pytest.raises(ValueError, match="No vehicle linked"):
            await ocr_service.apply_to_vehicle(db_session, ocr_result.id)

    @pytest.mark.asyncio
    async def test_apply_to_vehicle_no_data(
        self, db_session: AsyncSession, test_vehicle: Vehicle, tmp_path
    ):
        """Test applying OCR data when no structured data exists."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            ocr_result = await ocr_service.upload_file(
                db=db_session,
                file_bytes=b"fake-image",
                file_name="scan.png",
                mime_type="image/png",
                vehicle_id=test_vehicle.id,
            )
            await db_session.commit()

        with pytest.raises(ValueError, match="No structured data"):
            await ocr_service.apply_to_vehicle(db_session, ocr_result.id)


class TestOCRRouter:
    """Integration tests for OCR API endpoints."""

    @pytest_asyncio.fixture
    async def ocr_client(self, test_session_factory, admin_token):
        """HTTP client with DB override and admin auth."""

        async def _override_get_db():
            async with test_session_factory() as session:
                try:
                    yield session
                    await session.commit()
                except Exception:
                    await session.rollback()
                    raise
                finally:
                    await session.close()

        app.dependency_overrides[get_db] = _override_get_db
        transport = ASGITransport(app=app)
        async with AsyncClient(transport=transport, base_url="http://test") as ac:
            ac.headers.update({"Authorization": f"Bearer {admin_token}"})
            yield ac
        app.dependency_overrides.clear()

    @pytest.mark.asyncio
    async def test_upload_success(self, ocr_client: AsyncClient, tmp_path):
        """POST /api/v1/ocr/upload with valid image returns 202."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            # Patch background task to avoid actual processing
            with patch("app.routers.ocr.run_ocr_processing"):
                response = await ocr_client.post(
                    "/api/v1/ocr/upload",
                    files={
                        "file": ("scan.png", io.BytesIO(b"fake-image"), "image/png")
                    },
                )
        assert response.status_code == 202
        data = response.json()
        assert "ocr_result_id" in data
        assert data["status"] == "pending"
        assert data["message"] == "OCR processing queued"

    @pytest.mark.asyncio
    async def test_upload_no_file(self, ocr_client: AsyncClient):
        """POST /api/v1/ocr/upload without file returns 422."""
        response = await ocr_client.post("/api/v1/ocr/upload")
        assert response.status_code == 422

    @pytest.mark.asyncio
    async def test_upload_invalid_mime(self, ocr_client: AsyncClient, tmp_path):
        """POST /api/v1/ocr/upload with invalid MIME type returns 422."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            response = await ocr_client.post(
                "/api/v1/ocr/upload",
                files={"file": ("doc.pdf", io.BytesIO(b"fake-pdf"), "application/pdf")},
            )
        assert response.status_code == 422
        data = response.json()
        assert data["detail"]["error"]["code"] == "INVALID_MIME_TYPE"

    @pytest.mark.asyncio
    async def test_get_result_success(self, ocr_client: AsyncClient, tmp_path):
        """GET /api/v1/ocr/results/:id returns 200 with result data."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            with patch("app.routers.ocr.run_ocr_processing"):
                upload_resp = await ocr_client.post(
                    "/api/v1/ocr/upload",
                    files={
                        "file": ("scan.png", io.BytesIO(b"fake-image"), "image/png")
                    },
                )
            result_id = upload_resp.json()["ocr_result_id"]

            response = await ocr_client.get(f"/api/v1/ocr/results/{result_id}")
        assert response.status_code == 200
        data = response.json()
        assert data["id"] == result_id
        assert data["status"] == "pending"

    @pytest.mark.asyncio
    async def test_get_result_not_found(self, ocr_client: AsyncClient):
        """GET /api/v1/ocr/results/:nonexistent returns 404."""
        fake_id = uuid.uuid4()
        response = await ocr_client.get(f"/api/v1/ocr/results/{fake_id}")
        assert response.status_code == 404

    @pytest.mark.asyncio
    async def test_list_results(self, ocr_client: AsyncClient, tmp_path):
        """GET /api/v1/ocr/results returns 200 with list."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            with patch("app.routers.ocr.run_ocr_processing"):
                for i in range(3):
                    await ocr_client.post(
                        "/api/v1/ocr/upload",
                        files={
                            "file": (
                                f"scan{i}.png",
                                io.BytesIO(b"fake-image"),
                                "image/png",
                            )
                        },
                    )

            response = await ocr_client.get("/api/v1/ocr/results")
        assert response.status_code == 200
        data = response.json()
        assert data["total"] >= 3
        assert len(data["items"]) >= 3

    @pytest.mark.asyncio
    async def test_list_results_filter_vehicle(
        self, ocr_client: AsyncClient, test_vehicle: Vehicle, tmp_path
    ):
        """GET /api/v1/ocr/results?vehicle_id=X returns filtered list."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            with patch("app.routers.ocr.run_ocr_processing"):
                await ocr_client.post(
                    "/api/v1/ocr/upload",
                    files={"file": ("scan1.png", io.BytesIO(b"img1"), "image/png")},
                    data={"vehicle_id": str(test_vehicle.id)},
                )
                await ocr_client.post(
                    "/api/v1/ocr/upload",
                    files={"file": ("scan2.png", io.BytesIO(b"img2"), "image/png")},
                )

            response = await ocr_client.get(
                f"/api/v1/ocr/results?vehicle_id={test_vehicle.id}"
            )
        assert response.status_code == 200
        data = response.json()
        assert data["total"] == 1

    @pytest.mark.asyncio
    async def test_apply_to_vehicle_endpoint(
        self, ocr_client: AsyncClient, test_vehicle: Vehicle, tmp_path
    ):
        """POST /api/v1/ocr/results/:id/apply returns 200 and updates vehicle."""
        with patch.object(ocr_service.settings, "UPLOAD_DIR", str(tmp_path)):
            with patch("app.routers.ocr.run_ocr_processing"):
                upload_resp = await ocr_client.post(
                    "/api/v1/ocr/upload",
                    files={
                        "file": ("scan.png", io.BytesIO(b"fake-image"), "image/png")
                    },
                    data={"vehicle_id": str(test_vehicle.id)},
                )
            result_id = upload_resp.json()["ocr_result_id"]

            # Manually set structured data via direct DB session
            from tests.conftest import TEST_DATABASE_URL
            from sqlalchemy.ext.asyncio import create_async_engine, async_sessionmaker

            engine = create_async_engine(TEST_DATABASE_URL)
            factory = async_sessionmaker(
                engine, class_=AsyncSession, expire_on_commit=False
            )
            async with factory() as session:
                stmt = select(OCRResult).where(OCRResult.id == uuid.UUID(result_id))
                res = await session.execute(stmt)
                ocr = res.scalar_one()
                ocr.structured_data = {
                    "brand": "Volvo",
                    "model": "FH16",
                    "vin": "WDB9066351L123456",
                    "first_registration": "20.01.2021",
                    "mileage": 200000,
                    "power_kw": 500,
                    "fuel_type": "Diesel",
                }
                ocr.confidence_score = 0.9
                ocr.status = OCRStatus.completed
                await session.commit()
            await engine.dispose()

            response = await ocr_client.post(f"/api/v1/ocr/results/{result_id}/apply")
        assert response.status_code == 200
        data = response.json()
        assert data["vehicle_id"] == str(test_vehicle.id)
        assert "make" in data["updated_fields"]


class TestOpenRouterClient:
    """Tests for the OpenRouter API client utility."""

    def test_parse_response_valid_json(self):
        """Test parsing a valid JSON response."""
        from app.utils.openrouter import _parse_response

        raw = (
            '{"brand": "BMW", "model": "X5", "vin": "ABC123", "confidence_score": 0.9}'
        )
        result = _parse_response(raw)
        assert result["structured_data"]["brand"] == "BMW"
        assert result["confidence_score"] == 0.9

    def test_parse_response_markdown_fenced(self):
        """Test parsing a markdown-fenced JSON response."""
        from app.utils.openrouter import _parse_response

        raw = '```json\n{"brand": "Audi", "model": "A4", "confidence_score": 0.85}\n```'
        result = _parse_response(raw)
        assert result["structured_data"]["brand"] == "Audi"
        assert result["confidence_score"] == 0.85

    def test_parse_response_with_text_around(self):
        """Test parsing JSON embedded in text."""
        from app.utils.openrouter import _parse_response

        raw = 'Here is the result: {"brand": "VW", "model": "Golf", "confidence_score": 0.7} done.'
        result = _parse_response(raw)
        assert result["structured_data"]["brand"] == "VW"
        assert result["confidence_score"] == 0.7

    def test_parse_response_invalid(self):
        """Test parsing an invalid response returns defaults."""
        from app.utils.openrouter import _parse_response

        result = _parse_response("not json at all")
        assert result["structured_data"] == {}
        assert result["confidence_score"] == 0.0

    def test_parse_response_clamps_confidence(self):
        """Test that confidence score is clamped to 0.0-1.0."""
        from app.utils.openrouter import _parse_response

        result = _parse_response('{"brand": "X", "confidence_score": 1.5}')
        assert result["confidence_score"] == 1.0

        result = _parse_response('{"brand": "X", "confidence_score": -0.5}')
        assert result["confidence_score"] == 0.0

    def test_parse_response_all_expected_fields(self):
        """Test that all expected fields are present in structured_data."""
        from app.utils.openrouter import _parse_response, EXPECTED_FIELDS

        raw = '{"brand": "M", "model": "A", "vin": "V", "first_registration": "01.01.2020", "mileage": 100, "power_kw": 200, "fuel_type": "D", "confidence_score": 0.8}'
        result = _parse_response(raw)
        for field in EXPECTED_FIELDS:
            assert field in result["structured_data"]

    @pytest.mark.asyncio
    async def test_perform_ocr_no_api_key(self):
        """Test perform_ocr raises ValueError when no API key is configured."""
        from app.utils.openrouter import perform_ocr

        with patch("app.utils.openrouter.settings") as mock_settings:
            mock_settings.OPENROUTER_API_KEY = ""
            mock_settings.OPENROUTER_OCR_MODEL = "test-model"
            mock_settings.OPENROUTER_BASE_URL = "https://test.example.com"
            with pytest.raises(ValueError, match="OPENROUTER_API_KEY"):
                await perform_ocr(b"image", "image/png")

    @pytest.mark.asyncio
    async def test_perform_ocr_mocked_httpx(self):
        """Test perform_ocr with mocked httpx client."""
        from app.utils.openrouter import perform_ocr

        mock_response = MagicMock()
        mock_response.json.return_value = {
            "choices": [
                {
                    "message": {
                        "content": '{"brand": "Test", "model": "Model", "vin": "VIN123", "confidence_score": 0.95}'
                    }
                }
            ]
        }
        mock_response.raise_for_status = MagicMock()

        with patch("app.utils.openrouter.settings") as mock_settings:
            mock_settings.OPENROUTER_API_KEY = "test-key"
            mock_settings.OPENROUTER_OCR_MODEL = "test-model"
            mock_settings.OPENROUTER_BASE_URL = "https://test.example.com"

            with patch("httpx.AsyncClient") as mock_client_class:
                mock_client = AsyncMock()
                mock_client.post = AsyncMock(return_value=mock_response)
                mock_client.__aenter__ = AsyncMock(return_value=mock_client)
                mock_client.__aexit__ = AsyncMock(return_value=None)
                mock_client_class.return_value = mock_client

                result = await perform_ocr(b"image", "image/png")

        assert result["structured_data"]["brand"] == "Test"
        assert result["confidence_score"] == 0.95