motovaultpro/ocr/tests/test_manual_extractor.py

"""Tests for ManualExtractor Gemini-based maintenance schedule extraction.

Covers: normal extraction with subtype mapping, unusual service names,
empty Gemini response, and Gemini call failure.
All GeminiEngine calls are mocked.
"""
from unittest.mock import MagicMock, patch

import pytest

from app.engines.gemini_engine import (
    GeminiProcessingError,
    MaintenanceExtractionResult,
    MaintenanceItem,
)
from app.extractors.manual_extractor import (
    ExtractedSchedule,
    ManualExtractionResult,
    ManualExtractor,
)


# --- Helpers ---


def _make_pdf_bytes(size: int = 1024) -> bytes:
    """Create fake PDF bytes of a given size."""
    header = b"%PDF-1.4 fake"
    return header + b"\x00" * max(0, size - len(header))


def _make_gemini_result(items: list[MaintenanceItem]) -> MaintenanceExtractionResult:
    """Create a mock Gemini extraction result."""
    return MaintenanceExtractionResult(items=items, model="gemini-2.5-flash")


# --- Successful extraction ---


class TestNormalExtraction:
    """Verify normal PDF extraction returns mapped schedules with subtypes."""

    def test_pdf_with_maintenance_schedule_returns_mapped_items(self):
        """Normal: PDF with maintenance schedule returns extracted items with subtypes."""
        items = [
            MaintenanceItem(
                service_name="Engine Oil Change",
                interval_miles=5000,
                interval_months=6,
                details="Use 0W-20 full synthetic oil",
            ),
            MaintenanceItem(
                service_name="Tire Rotation",
                interval_miles=5000,
                interval_months=6,
                details=None,
            ),
            MaintenanceItem(
                service_name="Cabin Filter",
                interval_miles=15000,
                interval_months=12,
                details=None,
            ),
        ]

        extractor = ManualExtractor()
        extractor._engine = MagicMock()
        extractor._engine.extract_maintenance.return_value = _make_gemini_result(items)

        result = extractor.extract(_make_pdf_bytes())

        assert result.success is True
        assert result.error is None
        assert len(result.maintenance_schedules) == 3

        # Oil change should map to Engine Oil subtype
        oil = result.maintenance_schedules[0]
        assert oil.service == "Engine Oil Change"
        assert oil.interval_miles == 5000
        assert oil.interval_months == 6
        assert oil.details == "Use 0W-20 full synthetic oil"
        assert "Engine Oil" in oil.subtypes
        assert oil.confidence > 0.0

        # Tire rotation should map to Tires subtype
        tire = result.maintenance_schedules[1]
        assert tire.service == "Tire Rotation"
        assert "Tires" in tire.subtypes

        # Cabin filter should map to Cabin Air Filter / Purifier
        cabin = result.maintenance_schedules[2]
        assert "Cabin Air Filter / Purifier" in cabin.subtypes

    def test_progress_callbacks_fire_at_intervals(self):
        """Progress callbacks fire at appropriate intervals during processing."""
        items = [
            MaintenanceItem(service_name="Oil Change", interval_miles=5000),
        ]

        extractor = ManualExtractor()
        extractor._engine = MagicMock()
        extractor._engine.extract_maintenance.return_value = _make_gemini_result(items)

        progress_calls: list[tuple[int, str]] = []

        def track_progress(percent: int, message: str) -> None:
            progress_calls.append((percent, message))

        extractor.extract(_make_pdf_bytes(), progress_callback=track_progress)

        # Should have progress calls at 5, 50, 90, 100
        percents = [p for p, _ in progress_calls]
        assert 5 in percents
        assert 50 in percents
        assert 90 in percents
        assert 100 in percents
        # Percents should be non-decreasing
        assert percents == sorted(percents)


# --- Unusual service names ---


class TestUnusualServiceNames:
    """Verify that unusual service names still map to closest subtype."""

    def test_unusual_names_fuzzy_match_to_subtypes(self):
        """Edge: PDF with unusual service names still maps to closest subtype."""
        items = [
            MaintenanceItem(
                service_name="Replace engine air cleaner element",
                interval_miles=30000,
            ),
            MaintenanceItem(
                service_name="Inspect drive belt for cracks",
                interval_miles=60000,
            ),
        ]

        extractor = ManualExtractor()
        extractor._engine = MagicMock()
        extractor._engine.extract_maintenance.return_value = _make_gemini_result(items)

        result = extractor.extract(_make_pdf_bytes())

        assert result.success is True
        assert len(result.maintenance_schedules) == 2

        # "air cleaner element" should fuzzy match to Air Filter Element
        air_filter = result.maintenance_schedules[0]
        assert "Air Filter Element" in air_filter.subtypes

        # "drive belt" should match to Drive Belt
        belt = result.maintenance_schedules[1]
        assert "Drive Belt" in belt.subtypes

    def test_unmapped_service_uses_gemini_name_directly(self):
        """Edge: Service name with no match uses Gemini name and default confidence."""
        items = [
            MaintenanceItem(
                service_name="Recalibrate Quantum Flux Capacitor",
                interval_miles=100000,
            ),
        ]

        extractor = ManualExtractor()
        extractor._engine = MagicMock()
        extractor._engine.extract_maintenance.return_value = _make_gemini_result(items)

        result = extractor.extract(_make_pdf_bytes())

        assert result.success is True
        assert len(result.maintenance_schedules) == 1

        item = result.maintenance_schedules[0]
        assert item.service == "Recalibrate Quantum Flux Capacitor"
        assert item.subtypes == []
        assert item.confidence == ManualExtractor.DEFAULT_CONFIDENCE


# --- Empty response ---


class TestEmptyResponse:
    """Verify handling of empty Gemini responses."""

    def test_empty_gemini_response_returns_empty_schedules(self):
        """Edge: Empty Gemini response returns empty schedules list."""
        extractor = ManualExtractor()
        extractor._engine = MagicMock()
        extractor._engine.extract_maintenance.return_value = _make_gemini_result([])

        result = extractor.extract(_make_pdf_bytes())

        assert result.success is True
        assert result.maintenance_schedules == []
        assert result.error is None
        assert result.processing_time_ms >= 0


# --- Error handling ---


class TestErrorHandling:
    """Verify error handling when Gemini calls fail."""

    def test_gemini_failure_returns_error_result(self):
        """Error: Gemini call failure returns ManualExtractionResult with error."""
        extractor = ManualExtractor()
        extractor._engine = MagicMock()
        extractor._engine.extract_maintenance.side_effect = GeminiProcessingError(
            "Gemini maintenance extraction failed: API quota exceeded"
        )

        result = extractor.extract(_make_pdf_bytes())

        assert result.success is False
        assert result.maintenance_schedules == []
        assert result.error is not None
        assert "quota exceeded" in result.error.lower()

    def test_unexpected_exception_returns_error_result(self):
        """Error: Unexpected exception is caught and returned as error."""
        extractor = ManualExtractor()
        extractor._engine = MagicMock()
        extractor._engine.extract_maintenance.side_effect = RuntimeError(
            "Unexpected failure"
        )

        result = extractor.extract(_make_pdf_bytes())

        assert result.success is False
        assert result.error is not None
        assert "Unexpected failure" in result.error


# --- Job queue integration ---


class TestJobQueueIntegration:
    """Verify the extractor works within the existing job queue flow."""

    def test_extract_returns_all_required_fields(self):
        """The result contains all fields needed by process_manual_job in extract.py."""
        items = [
            MaintenanceItem(service_name="Oil Change", interval_miles=5000),
        ]

        extractor = ManualExtractor()
        extractor._engine = MagicMock()
        extractor._engine.extract_maintenance.return_value = _make_gemini_result(items)

        result = extractor.extract(_make_pdf_bytes())

        # All fields used by process_manual_job must be present
        assert hasattr(result, "success")
        assert hasattr(result, "vehicle_info")
        assert hasattr(result, "maintenance_schedules")
        assert hasattr(result, "raw_tables")
        assert hasattr(result, "processing_time_ms")
        assert hasattr(result, "total_pages")
        assert hasattr(result, "pages_processed")
        assert hasattr(result, "error")

        # Schedules have required fields
        schedule = result.maintenance_schedules[0]
        assert hasattr(schedule, "service")
        assert hasattr(schedule, "interval_miles")
        assert hasattr(schedule, "interval_months")
        assert hasattr(schedule, "details")
        assert hasattr(schedule, "confidence")
        assert hasattr(schedule, "subtypes")