Skip to content

~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.

57.5 KB
Raw Blame
"""
Tests for benchmarks/web_api/benchmark_service.py

Tests cover:
- BenchmarkQueueTracker functionality
- BenchmarkService initialization and methods
- Config hash generation
"""

from datetime import datetime, timedelta, UTC
from unittest.mock import Mock, patch

import pytest


class TestBenchmarkTaskStatus:
    """Tests for BenchmarkTaskStatus enum."""

    def test_status_values(self):
        """Test that status enum has expected values."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkTaskStatus,
        )

        assert BenchmarkTaskStatus.QUEUED.value == "queued"
        assert BenchmarkTaskStatus.PROCESSING.value == "processing"
        assert BenchmarkTaskStatus.COMPLETED.value == "completed"
        assert BenchmarkTaskStatus.FAILED.value == "failed"
        assert BenchmarkTaskStatus.CANCELLED.value == "cancelled"


class TestBenchmarkQueueTracker:
    """Tests for BenchmarkQueueTracker class."""

    def test_add_task(self):
        """Test adding a task to the tracker."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-123", "testuser", "benchmark")

        task = tracker.get_task_status("task-123")
        assert task is not None
        assert task["username"] == "testuser"
        assert task["task_type"] == "benchmark"
        assert task["status"] == BenchmarkTaskStatus.QUEUED.value

    def test_update_task_status(self):
        """Test updating task status."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-123", "testuser")
        tracker.update_task_status("task-123", BenchmarkTaskStatus.PROCESSING)

        task = tracker.get_task_status("task-123")
        assert task["status"] == BenchmarkTaskStatus.PROCESSING.value

    def test_update_nonexistent_task(self):
        """Test updating status of non-existent task logs warning."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        # Should not raise, just log warning
        tracker.update_task_status("nonexistent", BenchmarkTaskStatus.COMPLETED)

        task = tracker.get_task_status("nonexistent")
        assert task is None

    def test_remove_task(self):
        """Test removing a task from tracker."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-123", "testuser")
        tracker.remove_task("task-123")

        task = tracker.get_task_status("task-123")
        assert task is None

    def test_remove_nonexistent_task(self):
        """Test removing non-existent task doesn't raise."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
        )

        tracker = BenchmarkQueueTracker()
        # Should not raise
        tracker.remove_task("nonexistent")

    def test_cleanup_completed_tasks(self):
        """Test cleanup of old completed tasks."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-old", "testuser")
        tracker.update_task_status("task-old", BenchmarkTaskStatus.COMPLETED)

        # Manually set updated_at to be old
        tracker.tasks["task-old"]["updated_at"] = datetime.now(UTC) - timedelta(
            hours=2
        )

        tracker.add_task("task-new", "testuser")
        tracker.update_task_status("task-new", BenchmarkTaskStatus.COMPLETED)

        # Cleanup with 1 hour max age
        tracker.cleanup_completed_tasks(max_age_seconds=3600)

        # Old task should be removed
        assert tracker.get_task_status("task-old") is None
        # New task should remain
        assert tracker.get_task_status("task-new") is not None

    def test_cleanup_skips_processing_tasks(self):
        """Test that cleanup doesn't remove processing tasks."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-processing", "testuser")
        tracker.update_task_status(
            "task-processing", BenchmarkTaskStatus.PROCESSING
        )

        # Set old timestamp
        tracker.tasks["task-processing"]["updated_at"] = datetime.now(
            UTC
        ) - timedelta(hours=2)

        tracker.cleanup_completed_tasks(max_age_seconds=3600)

        # Processing task should remain
        assert tracker.get_task_status("task-processing") is not None


class TestBenchmarkServiceInit:
    """Tests for BenchmarkService initialization."""

    def test_init_default(self):
        """Test default initialization."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        # Mock socket service
        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            assert service.active_runs == {}
            assert service.socket_service is not None
            assert service.queue_tracker is not None

    def test_init_with_custom_socket(self):
        """Test initialization with custom socket service."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        assert service.socket_service is mock_socket

    def test_socket_service_delegates_to_socketio_asgi(self):
        """The default SocketIOService adapter must forward benchmark
        progress events to the FastAPI-native emit path — an earlier port
        wired it to a no-op stub, silently dropping every benchmark
        progress / completion / rate-limit WebSocket event."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        service = BenchmarkService()

        with patch(
            "local_deep_research.web.services.socketio_asgi"
            ".emit_to_subscribers",
            return_value=True,
        ) as emit:
            result = service.socket_service.emit_to_subscribers(
                "research_progress", 42, {"progress": 50}, owner="user1"
            )

        assert result is True
        # ``owner`` is forwarded: subscriptions are keyed by
        # (owner, run id) because a benchmark id is only unique within one
        # user's database (ADR-0009), so an emit without it reaches nobody.
        emit.assert_called_once_with(
            "research_progress",
            42,
            {"progress": 50},
            owner="user1",
            enable_logging=True,
        )

    def test_socket_service_emit_to_room_delegates(self):
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        service = BenchmarkService()

        with patch(
            "local_deep_research.web.services.socketio_asgi.emit_socket_event",
            return_value=True,
        ) as emit:
            result = service.socket_service.emit_to_room(
                "benchmark_update", {"x": 1}, room="sid-1"
            )

        assert result is True
        emit.assert_called_once_with("benchmark_update", {"x": 1}, room="sid-1")


class TestBenchmarkServiceConfigHash:
    """Tests for config hash generation."""

    def test_generate_config_hash(self):
        """Test generating config hash."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            config = {
                "iterations": 2,
                "questions_per_iteration": 3,
                "search_strategy": "iterdrag",
            }

            hash1 = service.generate_config_hash(config)
            hash2 = service.generate_config_hash(config)

            # Same config should produce same hash
            assert hash1 == hash2
            assert len(hash1) == 8  # Short hash

    def test_different_configs_different_hashes(self):
        """Test that different configs produce different hashes."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            config1 = {"iterations": 2}
            config2 = {"iterations": 3}

            hash1 = service.generate_config_hash(config1)
            hash2 = service.generate_config_hash(config2)

            assert hash1 != hash2


class TestBenchmarkServiceActiveRuns:
    """Tests for active run tracking."""

    def test_track_active_run(self):
        """Test tracking an active benchmark run."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            service.active_runs[("testuser", 1)] = {
                "run_id": 1,
                "status": "running",
                "progress": 50,
            }

            assert ("testuser", 1) in service.active_runs
            assert service.active_runs[("testuser", 1)]["progress"] == 50

    def test_rate_limit_tracking(self):
        """Test rate limit tracking per run."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            service.rate_limit_detected[("alice", 1)] = True
            service.rate_limit_detected[("bob", 1)] = False

            assert service.rate_limit_detected[("alice", 1)] is True
            assert service.rate_limit_detected[("bob", 1)] is False


class TestBenchmarkQueueTrackerThreadSafety:
    """Tests for thread safety of BenchmarkQueueTracker."""

    def test_concurrent_add_and_update(self):
        """Test concurrent access to tracker."""
        import threading
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        errors = []

        def add_tasks():
            try:
                for i in range(100):
                    tracker.add_task(
                        f"task-{threading.current_thread().name}-{i}", "user"
                    )
            except Exception as e:
                errors.append(e)

        def update_tasks():
            try:
                for i in range(100):
                    tracker.update_task_status(
                        f"task-{threading.current_thread().name}-{i}",
                        BenchmarkTaskStatus.COMPLETED,
                    )
            except Exception as e:
                errors.append(e)

        threads = []
        for i in range(4):
            t = threading.Thread(target=add_tasks, name=f"add-{i}")
            threads.append(t)
            t = threading.Thread(target=update_tasks, name=f"update-{i}")
            threads.append(t)

        for t in threads:
            t.start()
        for t in threads:
            t.join()

        # No errors should have occurred
        assert len(errors) == 0


class TestBenchmarkServiceQueueTrackerIntegration:
    """Tests for queue tracker integration with service."""

    def test_service_uses_queue_tracker(self):
        """Test that service uses queue tracker."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
            BenchmarkTaskStatus,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            # Use the tracker through the service
            service.queue_tracker.add_task("test-task", "testuser", "benchmark")

            task = service.queue_tracker.get_task_status("test-task")
            assert task is not None
            assert task["status"] == BenchmarkTaskStatus.QUEUED.value


class TestBenchmarkServiceQueryHash:
    """Tests for query hash generation."""

    def test_generate_query_hash(self):
        """Test generating query hash."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            hash1 = service.generate_query_hash(
                "What is AI?", "simpleqa", "ex1"
            )
            hash2 = service.generate_query_hash(
                "What is AI?", "simpleqa", "ex1"
            )

            # Same question, dataset and example should produce same hash
            assert hash1 == hash2
            assert len(hash1) == 32  # MD5 hex digest

    def test_different_example_ids_different_hashes(self):
        """#4860: distinct examples with identical question text must not
        collide, otherwise the sync dedup drops all but the first row."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            hash1 = service.generate_query_hash(
                "Same question", "simpleqa", "ex1"
            )
            hash2 = service.generate_query_hash(
                "Same question", "simpleqa", "ex2"
            )

            assert hash1 != hash2

    def test_different_questions_different_hashes(self):
        """Test that different questions produce different hashes."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            hash1 = service.generate_query_hash("Question 1", "simpleqa", "ex1")
            hash2 = service.generate_query_hash("Question 2", "simpleqa", "ex1")

            assert hash1 != hash2

    def test_different_datasets_different_hashes(self):
        """Test that different datasets produce different hashes."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            hash1 = service.generate_query_hash(
                "Same question", "simpleqa", "ex1"
            )
            hash2 = service.generate_query_hash(
                "Same question", "browsecomp", "ex1"
            )

            assert hash1 != hash2

    def test_query_hash_strips_whitespace(self):
        """Test that query hash strips whitespace from question."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.SocketIOService"
        ) as mock_socket:
            mock_socket.return_value = Mock()
            service = BenchmarkService()

            hash1 = service.generate_query_hash(
                "  Question  ", "simpleqa", "ex1"
            )
            hash2 = service.generate_query_hash("Question", "simpleqa", "ex1")

            assert hash1 == hash2


class TestBenchmarkServiceProgressUpdate:
    """Tests for progress update functionality."""

    def test_send_progress_update(self):
        """Test sending progress update."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        service._send_progress_update(
            benchmark_run_id=1,
            completed=5,
            total=10,
            username="testuser",
        )

        # Should have called emit_to_subscribers
        mock_socket.emit_to_subscribers.assert_called_once()
        call_args = mock_socket.emit_to_subscribers.call_args

        assert call_args[0][0] == "research_progress"
        assert call_args[0][1] == 1

    def test_send_progress_update_handles_zero_total(self):
        """Test progress update handles zero total gracefully."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        # Should not raise division by zero
        service._send_progress_update(
            benchmark_run_id=1,
            completed=0,
            total=0,
            username="testuser",
        )

        mock_socket.emit_to_subscribers.assert_called_once()


class TestBenchmarkQueueTrackerCleanup:
    """Additional tests for queue tracker cleanup."""

    def test_cleanup_failed_tasks(self):
        """Test cleanup of old failed tasks."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-failed", "testuser")
        tracker.update_task_status("task-failed", BenchmarkTaskStatus.FAILED)

        # Set old timestamp
        tracker.tasks["task-failed"]["updated_at"] = datetime.now(
            UTC
        ) - timedelta(hours=2)

        tracker.cleanup_completed_tasks(max_age_seconds=3600)

        # Failed task should be removed
        assert tracker.get_task_status("task-failed") is None

    def test_cleanup_cancelled_tasks(self):
        """Test cleanup of old cancelled tasks."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-cancelled", "testuser")
        tracker.update_task_status(
            "task-cancelled", BenchmarkTaskStatus.CANCELLED
        )

        # Set old timestamp
        tracker.tasks["task-cancelled"]["updated_at"] = datetime.now(
            UTC
        ) - timedelta(hours=2)

        tracker.cleanup_completed_tasks(max_age_seconds=3600)

        # Cancelled task should be removed
        assert tracker.get_task_status("task-cancelled") is None

    def test_cleanup_skips_queued_tasks(self):
        """Test that cleanup doesn't remove queued tasks."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-queued", "testuser")
        # Keep as QUEUED (default)

        # Set old timestamp
        tracker.tasks["task-queued"]["created_at"] = datetime.now(
            UTC
        ) - timedelta(hours=2)

        tracker.cleanup_completed_tasks(max_age_seconds=3600)

        # Queued task should remain
        assert tracker.get_task_status("task-queued") is not None

    def test_cleanup_uses_created_at_if_no_updated_at(self):
        """Test that cleanup uses created_at if updated_at is missing."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkQueueTracker,
            BenchmarkTaskStatus,
        )

        tracker = BenchmarkQueueTracker()
        tracker.add_task("task-old", "testuser")
        tracker.update_task_status("task-old", BenchmarkTaskStatus.COMPLETED)

        # Remove updated_at and set old created_at
        if "updated_at" in tracker.tasks["task-old"]:
            del tracker.tasks["task-old"]["updated_at"]
        tracker.tasks["task-old"]["created_at"] = datetime.now(UTC) - timedelta(
            hours=2
        )

        tracker.cleanup_completed_tasks(max_age_seconds=3600)

        # Task should be removed based on created_at
        assert tracker.get_task_status("task-old") is None


class TestBenchmarkServiceCancelBenchmark:
    """Tests for cancel benchmark functionality."""

    def test_cancel_sets_status_cancelled(self):
        """Test that cancel sets status to cancelled."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        # Add an active run owned by the cancelling user. active_runs is keyed
        # by (username, id), so cancel only ever finds — and stops — the
        # in-memory run belonging to the requesting user.
        service.active_runs[("testuser", 1)] = {
            "status": "running",
            "data": {"username": "testuser"},
        }

        with patch.object(service, "update_benchmark_status"):
            service.cancel_benchmark(1, username="testuser")

            assert service.active_runs[("testuser", 1)]["status"] == "cancelled"

    def test_cancel_returns_true_on_success(self):
        """Test that cancel returns True on success."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        service.active_runs[("testuser", 1)] = {"status": "running"}

        with patch.object(service, "update_benchmark_status"):
            result = service.cancel_benchmark(1, username="testuser")

            assert result is True


class TestBenchmarkServiceSyncResults:
    """Tests for sync pending results functionality."""

    def test_sync_pending_results_returns_zero_for_unknown_run(self):
        """Test that sync returns 0 for unknown benchmark run."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        result = service.sync_pending_results(99999, username="testuser")

        assert result == 0


class TestBenchmarkServiceCreateBenchmarkRun:
    """Tests for create_benchmark_run functionality."""

    def test_create_benchmark_run_success(self):
        """Test creating a benchmark run in the database."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        # Mock the database session - patch at the source module
        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            # Mock the BenchmarkRun model to capture the created object
            created_run = Mock()
            created_run.id = 1

            def add_side_effect(run):
                run.id = 1

            mock_session.add.side_effect = add_side_effect
            mock_session.commit = Mock()

            search_config = {"iterations": 2, "search_strategy": "iterdrag"}
            evaluation_config = {"model_name": "test-model"}
            datasets_config = {"simpleqa": {"count": 10}}

            run_id = service.create_benchmark_run(
                run_name="Test Run",
                search_config=search_config,
                evaluation_config=evaluation_config,
                datasets_config=datasets_config,
                username="testuser",
            )

            assert run_id == 1
            mock_session.add.assert_called_once()
            mock_session.commit.assert_called_once()

    def test_create_benchmark_run_generates_config_hash(self):
        """Test that create_benchmark_run generates config hash."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            def capture_run(run):
                run.id = 1
                assert run.config_hash is not None
                # Config hash is first 8 chars of MD5 hexdigest (see generate_config_hash)
                assert len(run.config_hash) == 8

            mock_session.add.side_effect = capture_run

            service.create_benchmark_run(
                run_name="Test",
                search_config={"iterations": 2},
                evaluation_config={},
                datasets_config={"simpleqa": {"count": 5}},
            )

    def test_create_benchmark_run_calculates_total_examples(self):
        """Test that total_examples is calculated correctly."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            def capture_run(run):
                run.id = 1
                assert run.total_examples == 25  # 10 + 15

            mock_session.add.side_effect = capture_run

            service.create_benchmark_run(
                run_name="Test",
                search_config={},
                evaluation_config={},
                datasets_config={
                    "simpleqa": {"count": 10},
                    "browsecomp": {"count": 15},
                },
            )

    def test_create_benchmark_run_handles_db_error(self):
        """Test that create_benchmark_run handles database errors."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_session.commit.side_effect = Exception("Database error")
            mock_get_session.return_value = mock_session

            with pytest.raises(Exception, match="Database error"):
                service.create_benchmark_run(
                    run_name="Test",
                    search_config={},
                    evaluation_config={},
                    datasets_config={"simpleqa": {"count": 5}},
                )

            mock_session.rollback.assert_called_once()


class TestBenchmarkServiceStartBenchmark:
    """Tests for start_benchmark functionality."""

    def test_start_benchmark_creates_thread(self):
        """Test that start_benchmark creates a background thread."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            # Mock the benchmark run query
            mock_run = Mock()
            mock_run.id = 1
            mock_run.config_hash = "abc12345"
            mock_run.datasets_config = {"simpleqa": {"count": 2}}
            mock_run.search_config = {}
            mock_run.evaluation_config = {}
            mock_session.query.return_value.filter.return_value.first.return_value = mock_run

            # Mock SettingsManager
            with patch(
                "local_deep_research.settings.SettingsManager"
            ) as mock_settings_mgr:
                mock_settings_mgr.return_value.get_all_settings.return_value = {}

                # Mock the request-context session id (FastAPI contextvar)
                with patch(
                    "local_deep_research.utilities.request_context.get_current_session_id",
                    return_value="test-session",
                ):
                    with patch(
                        "local_deep_research.database.session_passwords.session_password_store"
                    ) as mock_password_store:
                        mock_password_store.get_session_password.return_value = "test-password"

                        # Mock the thread execution
                        with patch.object(
                            service,
                            "_run_benchmark_thread",
                            return_value=None,
                        ):
                            result = service.start_benchmark(
                                1, username="testuser", user_password="test"
                            )

                            assert result is True
                            assert ("testuser", 1) in service.active_runs
                            assert (
                                service.active_runs[("testuser", 1)]["status"]
                                == "running"
                            )

    def test_start_benchmark_stores_data_in_memory(self):
        """Test that start_benchmark stores benchmark data in memory."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            mock_run = Mock()
            mock_run.id = 1
            mock_run.config_hash = "abc12345"
            mock_run.datasets_config = {"simpleqa": {"count": 2}}
            mock_run.search_config = {"iterations": 2}
            mock_run.evaluation_config = {"model_name": "test"}
            mock_session.query.return_value.filter.return_value.first.return_value = mock_run

            with patch(
                "local_deep_research.settings.SettingsManager"
            ) as mock_settings_mgr:
                mock_settings_mgr.return_value.get_all_settings.return_value = {
                    "key": "value"
                }

                with patch(
                    "local_deep_research.utilities.request_context.get_current_session_id",
                    return_value="test-session",
                ):
                    with patch(
                        "local_deep_research.database.session_passwords.session_password_store"
                    ):
                        with patch.object(
                            service,
                            "_run_benchmark_thread",
                            return_value=None,
                        ):
                            service.start_benchmark(1, username="testuser")

                            assert (
                                "data" in service.active_runs[("testuser", 1)]
                            )
                            assert (
                                service.active_runs[("testuser", 1)]["data"][
                                    "benchmark_run_id"
                                ]
                                == 1
                            )
                            assert (
                                service.active_runs[("testuser", 1)]["data"][
                                    "username"
                                ]
                                == "testuser"
                            )

    def test_start_benchmark_two_users_same_id_coexist(self):
        """Two different users starting a run with the SAME per-user id must not
        clobber each other in the process-global active_runs -- the exact write
        collision the composite (username, id) key prevents. Drives the real
        start_benchmark write path for both users (thread spawn mocked)."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        service = BenchmarkService(socket_service=Mock())

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            mock_run = Mock()
            mock_run.id = 1  # the same per-user id for both users
            mock_run.config_hash = "abc12345"
            mock_run.datasets_config = {"simpleqa": {"count": 1}}
            mock_run.search_config = {"iterations": 1}
            mock_run.evaluation_config = {"model_name": "test"}
            mock_session.query.return_value.filter.return_value.first.return_value = mock_run

            with patch(
                "local_deep_research.settings.SettingsManager"
            ) as mock_settings_mgr:
                mock_settings_mgr.return_value.get_all_settings.return_value = {}
                # main patches flask.session here; this branch resolves the
                # session id through utilities.request_context instead, and
                # start_benchmark imports it at function scope, so the source
                # module is the right patch target.
                with patch(
                    "local_deep_research.utilities.request_context.get_current_session_id",
                    return_value="sid",
                ):
                    with patch(
                        "local_deep_research.database.session_passwords.session_password_store"
                    ):
                        with patch.object(
                            service, "_run_benchmark_thread", return_value=None
                        ):
                            assert service.start_benchmark(1, username="alice")
                            assert service.start_benchmark(1, username="bob")

        # Both users' id-1 runs live under distinct composite keys; the second
        # start did not overwrite the first.
        assert len(service.active_runs) == 2
        assert service.active_runs[("alice", 1)]["data"]["username"] == "alice"
        assert service.active_runs[("bob", 1)]["data"]["username"] == "bob"

    def test_start_benchmark_handles_not_found(self):
        """Test that start_benchmark handles benchmark not found."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            # Return None for the benchmark run
            mock_session.query.return_value.filter.return_value.first.return_value = None

            result = service.start_benchmark(999, username="testuser")

            assert result is False

    def test_start_benchmark_cleans_up_active_runs_on_spawn_failure(self):
        """If Thread.start raises after active_runs was populated, the stale
        entry must be removed so later cancel/status calls don't see a
        thread-less zombie run."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )
        from local_deep_research.database.models import BenchmarkStatus

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            mock_run = Mock()
            mock_run.id = 42
            mock_run.config_hash = "hash42"
            mock_run.datasets_config = {}
            mock_run.search_config = {}
            mock_run.evaluation_config = {}
            mock_session.query.return_value.filter.return_value.first.return_value = mock_run

            with patch("local_deep_research.settings.SettingsManager") as sm:
                sm.return_value.get_all_settings.return_value = {}
                with (
                    patch(
                        "local_deep_research.utilities.request_context.get_current_session_id",
                        return_value="sid",
                    ),
                    patch(
                        "local_deep_research.database.session_passwords.session_password_store"
                    ),
                    patch("threading.Thread") as mock_thread_cls,
                ):
                    mock_thread = Mock()
                    mock_thread.start.side_effect = RuntimeError(
                        "thread spawn failed"
                    )
                    mock_thread_cls.return_value = mock_thread

                    result = service.start_benchmark(
                        42, username="testuser", user_password="pw"
                    )

            assert result is False
            # active_runs must not retain the zombie entry.
            assert ("testuser", 42) not in service.active_runs
            # BenchmarkRun.status was flipped to FAILED by the error handler.
            assert mock_run.status == BenchmarkStatus.FAILED


class TestBenchmarkServiceProcessTask:
    """Tests for _process_benchmark_task functionality."""

    def test_process_benchmark_task_success(self):
        """Test successful processing of a benchmark task."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        task = {
            "benchmark_run_id": 1,
            "example_id": "ex1",
            "dataset_type": "simpleqa",
            "question": "What is 2+2?",
            "correct_answer": "4",
            "query_hash": "hash123",
            "task_index": 0,
        }

        search_config = {"iterations": 1}
        evaluation_config = {}

        with patch(
            "local_deep_research.config.thread_settings.get_settings_context"
        ) as mock_get_ctx:
            mock_ctx = Mock()
            mock_ctx.snapshot = {}
            mock_get_ctx.return_value = mock_ctx

            with patch(
                "local_deep_research.benchmarks.web_api.benchmark_service.format_query"
            ) as mock_format:
                mock_format.return_value = "formatted query"

                with patch(
                    "local_deep_research.benchmarks.web_api.benchmark_service.quick_summary"
                ) as mock_summary:
                    mock_summary.return_value = {
                        "summary": "The answer is 4.",
                        "sources": [],
                    }

                    with patch(
                        "local_deep_research.benchmarks.web_api.benchmark_service.extract_answer_from_response"
                    ) as mock_extract:
                        mock_extract.return_value = {
                            "extracted_answer": "4",
                            "confidence": "100",
                        }

                        with patch(
                            "local_deep_research.benchmarks.web_api.benchmark_service.grade_single_result"
                        ) as mock_grade:
                            mock_grade.return_value = {
                                "is_correct": True,
                                "graded_confidence": "100",
                                "grader_response": "Correct!",
                            }

                            result = service._process_benchmark_task(
                                task, search_config, evaluation_config
                            )

                            assert result["response"] == "The answer is 4."
                            assert result["is_correct"] is True
                            assert result["query_hash"] == "hash123"

    def test_process_benchmark_task_handles_research_error(self):
        """Test handling of research errors in task processing."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        task = {
            "benchmark_run_id": 1,
            "example_id": "ex1",
            "dataset_type": "simpleqa",
            "question": "What is 2+2?",
            "correct_answer": "4",
            "query_hash": "hash123",
            "task_index": 0,
        }

        with patch(
            "local_deep_research.config.thread_settings.get_settings_context"
        ) as mock_get_ctx:
            mock_ctx = Mock()
            mock_ctx.snapshot = {}
            mock_get_ctx.return_value = mock_ctx

            with patch(
                "local_deep_research.benchmarks.web_api.benchmark_service.format_query"
            ) as mock_format:
                mock_format.return_value = "formatted query"

                with patch(
                    "local_deep_research.benchmarks.web_api.benchmark_service.quick_summary"
                ) as mock_summary:
                    mock_summary.side_effect = Exception("Research failed")

                    result = service._process_benchmark_task(task, {}, {})

                    assert "research_error" in result
                    assert "Research failed" in result["research_error"]

    def test_process_benchmark_task_handles_evaluation_error(self):
        """Test handling of evaluation errors in task processing."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        task = {
            "benchmark_run_id": 1,
            "example_id": "ex1",
            "dataset_type": "simpleqa",
            "question": "What is 2+2?",
            "correct_answer": "4",
            "query_hash": "hash123",
            "task_index": 0,
        }

        with patch(
            "local_deep_research.config.thread_settings.get_settings_context"
        ) as mock_get_ctx:
            mock_ctx = Mock()
            mock_ctx.snapshot = {}
            mock_get_ctx.return_value = mock_ctx

            with patch(
                "local_deep_research.benchmarks.web_api.benchmark_service.format_query"
            ) as mock_format:
                mock_format.return_value = "formatted query"

                with patch(
                    "local_deep_research.benchmarks.web_api.benchmark_service.quick_summary"
                ) as mock_summary:
                    mock_summary.return_value = {
                        "summary": "Answer",
                        "sources": [],
                    }

                    with patch(
                        "local_deep_research.benchmarks.web_api.benchmark_service.extract_answer_from_response"
                    ) as mock_extract:
                        mock_extract.return_value = {
                            "extracted_answer": "4",
                            "confidence": "100",
                        }

                        with patch(
                            "local_deep_research.benchmarks.web_api.benchmark_service.grade_single_result"
                        ) as mock_grade:
                            mock_grade.side_effect = Exception("Grading failed")

                            result = service._process_benchmark_task(
                                task, {}, {}
                            )

                            assert result["is_correct"] is None
                            assert "evaluation_error" in result


class TestBenchmarkServiceGetBenchmarkStatus:
    """Tests for get_benchmark_status functionality."""

    def test_get_benchmark_status_returns_none_for_unknown(self):
        """Test that get_benchmark_status returns None for unknown run."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            mock_session.query.return_value.filter.return_value.first.return_value = None

            result = service.get_benchmark_status(999)

            assert result is None

    def test_get_benchmark_status_method_exists(self):
        """Test that get_benchmark_status method exists and is callable."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        assert hasattr(service, "get_benchmark_status")
        assert callable(service.get_benchmark_status)

    def test_get_benchmark_status_returns_dict_or_none(self):
        """Test that get_benchmark_status returns dict or None."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )
        import inspect

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        # Verify function signature
        sig = inspect.signature(service.get_benchmark_status)
        params = list(sig.parameters.keys())
        assert "benchmark_run_id" in params

    def test_get_benchmark_status_does_not_leak_raw_error(self):
        """A failed run's raw error_message (which can carry server-level LLM
        endpoints / filesystem paths) must NOT be returned by
        get_benchmark_status — it is replaced with a generic message (CWE-209).
        """
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
            BenchmarkStatus,
            _GENERIC_BENCHMARK_ERROR,
        )

        secret = "Connection refused to http://internal-llm:8000/v1 /srv/secret"
        mock_run = Mock()
        mock_run.id = 1
        mock_run.run_name = "run"
        mock_run.status = BenchmarkStatus.FAILED
        mock_run.completed_examples = 0
        mock_run.total_examples = 10
        mock_run.failed_examples = 1
        mock_run.overall_accuracy = None
        mock_run.processing_rate = None
        mock_run.created_at = None
        mock_run.start_time = None  # skips the timing-calculation block
        mock_run.end_time = None
        mock_run.error_message = secret

        service = BenchmarkService(socket_service=Mock())

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            q = mock_session.query.return_value
            q.filter.return_value.first.return_value = mock_run
            q.filter.return_value.all.return_value = []
            q.filter.return_value.filter.return_value.all.return_value = []

            result = service.get_benchmark_status(1)

        assert result is not None
        assert result["error_message"] == _GENERIC_BENCHMARK_ERROR
        assert secret not in str(result)
        assert "internal-llm" not in str(result)


class TestBenchmarkServiceTaskQueue:
    """Tests for task queue creation."""

    def test_create_task_queue_creates_tasks(self):
        """Test that _create_task_queue creates tasks correctly."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        datasets_config = {"simpleqa": {"count": 3}}

        # Mock load_dataset
        with patch(
            "local_deep_research.benchmarks.web_api.benchmark_service.load_dataset"
        ) as mock_load:
            mock_load.return_value = [
                {"id": "1", "problem": "Q1", "answer": "A1"},
                {"id": "2", "problem": "Q2", "answer": "A2"},
                {"id": "3", "problem": "Q3", "answer": "A3"},
            ]

            tasks = service._create_task_queue(
                datasets_config=datasets_config,
                benchmark_run_id=1,
            )

            assert len(tasks) == 3
            assert tasks[0]["question"] == "Q1"
            assert tasks[0]["correct_answer"] == "A1"
            assert tasks[0]["benchmark_run_id"] == 1


class TestBenchmarkServiceUpdateStatus:
    """Tests for update_benchmark_status functionality."""

    def test_update_benchmark_status_updates_db(self):
        """Test that update_benchmark_status updates the database."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )
        from local_deep_research.database.models.benchmark import (
            BenchmarkStatus,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            mock_run = Mock()
            mock_run.status = BenchmarkStatus.PENDING
            mock_run.start_time = None
            mock_run.end_time = None
            mock_session.query.return_value.filter.return_value.first.return_value = mock_run

            service.update_benchmark_status(
                1, BenchmarkStatus.IN_PROGRESS, username="testuser"
            )

            assert mock_run.status == BenchmarkStatus.IN_PROGRESS
            mock_session.commit.assert_called_once()

    def test_update_benchmark_status_sets_start_time(self):
        """Test that start_time is set when transitioning to IN_PROGRESS."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )
        from local_deep_research.database.models.benchmark import (
            BenchmarkStatus,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            mock_run = Mock()
            mock_run.status = BenchmarkStatus.PENDING
            mock_run.start_time = None
            mock_run.end_time = None
            mock_session.query.return_value.filter.return_value.first.return_value = mock_run

            service.update_benchmark_status(1, BenchmarkStatus.IN_PROGRESS)

            assert mock_run.start_time is not None

    def test_update_benchmark_status_sets_end_time_on_completion(self):
        """Test that end_time is set when transitioning to COMPLETED."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )
        from local_deep_research.database.models.benchmark import (
            BenchmarkStatus,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            mock_run = Mock()
            mock_run.status = BenchmarkStatus.IN_PROGRESS
            mock_run.start_time = datetime.now(UTC)
            mock_run.end_time = None
            mock_session.query.return_value.filter.return_value.first.return_value = mock_run

            service.update_benchmark_status(1, BenchmarkStatus.COMPLETED)

            assert mock_run.end_time is not None

    def test_update_benchmark_status_stores_error_message(self):
        """Test that error message is stored when provided."""
        from local_deep_research.benchmarks.web_api.benchmark_service import (
            BenchmarkService,
        )
        from local_deep_research.database.models.benchmark import (
            BenchmarkStatus,
        )

        mock_socket = Mock()
        service = BenchmarkService(socket_service=mock_socket)

        with patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session:
            mock_session = Mock()
            mock_session.__enter__ = Mock(return_value=mock_session)
            mock_session.__exit__ = Mock(return_value=False)
            mock_get_session.return_value = mock_session

            mock_run = Mock()
            mock_run.status = BenchmarkStatus.IN_PROGRESS
            mock_run.start_time = datetime.now(UTC)
            mock_run.end_time = None
            mock_run.error_message = None
            mock_session.query.return_value.filter.return_value.first.return_value = mock_run

            service.update_benchmark_status(
                1,
                BenchmarkStatus.FAILED,
                error_message="Test error",
            )

            assert mock_run.error_message == "Test error"


class TestSessionRollbackExceptionHandling:
    """Tests for session rollback exception handling (PR #2029).

    PR #2029 changed bare `except:` to `except Exception:` in the
    session.rollback() error handler in sync_pending_results.
    """

    def test_rollback_exception_is_caught(self):
        """Exception during rollback is caught and doesn't propagate."""
        mock_session = Mock()
        mock_session.rollback.side_effect = Exception("Rollback failed")

        # Simulate the pattern from sync_pending_results
        caught = False
        try:
            mock_session.rollback()
        except Exception:
            caught = True

        assert caught

    def test_rollback_success_no_exception(self):
        """Successful rollback raises no exception."""
        mock_session = Mock()
        mock_session.rollback.return_value = None

        error = None
        try:
            mock_session.rollback()
        except Exception as e:
            error = e

        assert error is None