learningcircuit--local-deep-research/tests/benchmarks/web_api/test_benchmark_routes_coverage.py

"""
Coverage tests for benchmarks/web_api/benchmark_routes.py

Targets the uncovered code paths: route handler bodies, branching on
provider types, evaluation config sources, dataset validation,
error-handling except blocks, search-metric aggregation logic, and
export/delete/cancel/validate/search-quality endpoints.
"""

import enum
from contextlib import contextmanager
from datetime import datetime, UTC
from unittest.mock import MagicMock, patch

from flask import Blueprint, Flask


# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------


def _make_app():
    """Create a minimal Flask app with the benchmark blueprint registered."""
    # Must import *after* patching decorators is no longer needed because
    # the blueprint is registered at import time.  Instead we patch the
    # auth check at the decorator level so the routes execute normally.
    from local_deep_research.benchmarks.web_api.benchmark_routes import (
        benchmark_bp,
    )

    app = Flask(__name__)
    app.config["SECRET_KEY"] = "test-secret"
    app.config["TESTING"] = True
    app.config["WTF_CSRF_ENABLED"] = False

    # Stub auth blueprint so url_for("auth.login") resolves
    auth_bp = Blueprint("auth", __name__, url_prefix="/auth")

    @auth_bp.route("/login")
    def login():
        return "login"

    app.register_blueprint(auth_bp)

    # Only register if not already registered (avoid duplicate blueprint error)
    if "benchmark" not in app.blueprints:
        app.register_blueprint(benchmark_bp)

    return app


def _fake_settings(overrides=None):
    """Return a dict of settings and a mock SettingsManager."""
    defaults = {
        "search.iterations": 8,
        "search.questions_per_iteration": 5,
        "search.tool": "searxng",
        "search.search_strategy": "focused_iteration",
        "llm.model": "gpt-4",
        "llm.provider": "openai_endpoint",
        "llm.temperature": 0.7,
        "llm.max_tokens": 30000,
        "llm.context_window_unrestricted": True,
        "llm.context_window_size": 128000,
        "llm.local_context_window_size": 4096,
        "llm.openai_endpoint.url": "http://localhost:8080",
        "llm.openai_endpoint.api_key": "sk-test",
        "llm.openai.api_key": "sk-openai",
        "llm.anthropic.api_key": "sk-anthropic",
        "benchmark.evaluation.provider": "openai_endpoint",
        "benchmark.evaluation.model": "anthropic/claude-3.7-sonnet",
        "benchmark.evaluation.temperature": 0,
        "benchmark.evaluation.endpoint_url": "https://openrouter.ai/api/v1",
    }
    if overrides:
        defaults.update(overrides)

    mgr = MagicMock()
    mgr.get_setting.side_effect = lambda key, default=None: defaults.get(
        key, default
    )
    return defaults, mgr


@contextmanager
def _patch_auth_and_db(settings_overrides=None):
    """Context manager that patches login_required, db_manager, limiter,
    get_user_db_session, and SettingsManager so route bodies execute."""
    _, mgr = _fake_settings(settings_overrides)
    mock_db_session = MagicMock()

    with (
        patch(
            "local_deep_research.web.auth.decorators.db_manager"
        ) as mock_db_mgr,
        patch(
            "local_deep_research.benchmarks.web_api.benchmark_routes.get_user_db_session"
        ) as mock_get_session,
        patch(
            "local_deep_research.database.session_context.get_user_db_session"
        ) as mock_get_session2,
        patch(
            "local_deep_research.benchmarks.web_api.benchmark_routes.SettingsManager",
            return_value=mgr,
        ),
        patch(
            "local_deep_research.settings.SettingsManager",
            return_value=mgr,
        ),
        patch(
            "local_deep_research.benchmarks.web_api.benchmark_routes.benchmark_service"
        ) as mock_svc,
    ):
        mock_db_mgr.is_user_connected.return_value = True

        @contextmanager
        def _session_ctx(username, password=None):
            yield mock_db_session

        mock_get_session.side_effect = _session_ctx
        mock_get_session2.side_effect = _session_ctx
        yield mock_svc, mgr, mock_db_session


def _make_routed_query(
    *,
    runs=None,
    avg_processing=None,
    results=None,
    search_calls=None,
    search_calls_exc=None,
):
    """Side-effect for mock_db.query that routes by model class.

    After the SearchTracker singleton removal, production code queries
    SearchCall directly on the same session.  Tests must therefore route
    ``session.query(Model)`` to the right mock chain depending on the
    model class name.
    """

    def _route(model, *args):
        chain = MagicMock()
        chain.filter.return_value = chain
        chain.order_by.return_value = chain
        chain.limit.return_value = chain

        if not isinstance(model, type):
            # func.avg(...) — not a class
            chain.scalar.return_value = avg_processing
            return chain

        name = getattr(model, "__name__", "")
        if "BenchmarkRun" in name:
            chain.all.return_value = runs or []
        elif "SearchCall" in name:
            if search_calls_exc:
                raise search_calls_exc
            chain.all.return_value = search_calls or []
        else:
            # BenchmarkResult or anything else
            chain.all.return_value = results or []
        return chain

    return _route


# Fake enum for BenchmarkStatus
class _FakeStatus(enum.Enum):
    IN_PROGRESS = "in_progress"
    COMPLETED = "completed"
    FAILED = "failed"
    CANCELLED = "cancelled"


class _FakeDatasetType(enum.Enum):
    SIMPLEQA = "simpleqa"
    BROWSECOMP = "browsecomp"


# ---------------------------------------------------------------------------
# index route
# ---------------------------------------------------------------------------


class TestIndex:
    def test_index_renders_template(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.benchmarks.web_api.benchmark_routes.render_template_with_defaults",
                return_value="<html>ok</html>",
            ) as mock_render:
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = client.get("/benchmark/")
                    assert resp.status_code == 200
                    mock_render.assert_called_once()
                    call_kwargs = mock_render.call_args
                    assert (
                        "eval_settings" in call_kwargs.kwargs
                        or "eval_settings"
                        in (call_kwargs[1] if len(call_kwargs) > 1 else {})
                    )


class TestResults:
    def test_results_page(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.benchmarks.web_api.benchmark_routes.render_template_with_defaults",
                return_value="<html>results</html>",
            ):
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = client.get("/benchmark/results")
                    assert resp.status_code == 200


# ---------------------------------------------------------------------------
# start_benchmark route
# ---------------------------------------------------------------------------


class TestStartBenchmark:
    def _post_start(self, client, json_data):
        return client.post(
            "/benchmark/api/start",
            json=json_data,
            content_type="application/json",
        )

    def test_start_no_json_body(self):
        """No JSON body triggers require_json_body decorator -> 400."""
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.post("/benchmark/api/start", data="not json")
                assert resp.status_code == 400

    def test_start_empty_datasets(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(client, {"datasets_config": {}})
                    assert resp.status_code == 400

    def test_start_datasets_all_zero_count(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 0}}},
                    )
                    assert resp.status_code == 400

    def test_start_success_openai_endpoint_provider(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = "pw123"
                mock_svc.create_benchmark_run.return_value = 42
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                        sess["session_id"] = "sid1"
                    resp = self._post_start(
                        client,
                        {
                            "run_name": "test run",
                            "datasets_config": {"simpleqa": {"count": 5}},
                        },
                    )
                    assert resp.status_code == 200
                    data = resp.get_json()
                    assert data["success"] is True
                    assert data["benchmark_run_id"] == 42

    def test_start_success_openai_provider(self):
        app = _make_app()
        with _patch_auth_and_db({"llm.provider": "openai"}) as (
            mock_svc,
            mgr,
            _,
        ):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 10
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 2}}},
                    )
                    assert resp.status_code == 200

    def test_start_success_anthropic_provider(self):
        app = _make_app()
        with _patch_auth_and_db({"llm.provider": "anthropic"}) as (
            mock_svc,
            mgr,
            _,
        ):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 11
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 200

    def test_start_with_evaluation_config_in_data(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 99
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(
                        client,
                        {
                            "datasets_config": {"simpleqa": {"count": 3}},
                            "evaluation_config": {
                                "provider": "openai",
                                "model_name": "gpt-4",
                            },
                        },
                    )
                    assert resp.status_code == 200
                    # Verify evaluation_config was passed through
                    call_kwargs = mock_svc.create_benchmark_run.call_args
                    assert (
                        call_kwargs.kwargs["evaluation_config"]["provider"]
                        == "openai"
                    )

    def test_start_eval_provider_openai(self):
        """Evaluation provider openai branch."""
        app = _make_app()
        with _patch_auth_and_db(
            {
                "benchmark.evaluation.provider": "openai",
            }
        ) as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 50
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 200

    def test_start_eval_provider_anthropic(self):
        """Evaluation provider anthropic branch."""
        app = _make_app()
        with _patch_auth_and_db(
            {
                "benchmark.evaluation.provider": "anthropic",
            }
        ) as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 51
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 200

    def test_start_benchmark_fails(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 77
                mock_svc.start_benchmark.return_value = False
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 5}}},
                    )
                    assert resp.status_code == 500
                    assert resp.get_json()["success"] is False

    def test_start_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.side_effect = RuntimeError("boom")
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_start(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 5}}},
                    )
                    assert resp.status_code == 500


# ---------------------------------------------------------------------------
# get_running_benchmark
# ---------------------------------------------------------------------------


class TestGetRunningBenchmark:
    def test_running_found(self):
        app = _make_app()
        mock_run = MagicMock()
        mock_run.id = 1
        mock_run.run_name = "Run 1"
        mock_run.total_examples = 10
        mock_run.completed_examples = 3

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            with (
                patch(
                    "local_deep_research.benchmarks.web_api.benchmark_routes.BenchmarkRun",
                    create=True,
                ),
                patch(
                    "local_deep_research.benchmarks.web_api.benchmark_routes.BenchmarkStatus",
                    create=True,
                ) as MockBS,
            ):
                MockBS.IN_PROGRESS = _FakeStatus.IN_PROGRESS
                mock_query = MagicMock()
                mock_db.query.return_value = mock_query
                mock_query.filter.return_value = mock_query
                mock_query.order_by.return_value = mock_query
                mock_query.first.return_value = mock_run

                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = client.get("/benchmark/api/running")
                    assert resp.status_code == 200
                    data = resp.get_json()
                    assert data["success"] is True
                    assert data["benchmark_run_id"] == 1

    def test_no_running(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            with (
                patch(
                    "local_deep_research.benchmarks.web_api.benchmark_routes.BenchmarkRun",
                    create=True,
                ),
                patch(
                    "local_deep_research.benchmarks.web_api.benchmark_routes.BenchmarkStatus",
                    create=True,
                ) as MockBS,
            ):
                MockBS.IN_PROGRESS = _FakeStatus.IN_PROGRESS
                mock_query = MagicMock()
                mock_db.query.return_value = mock_query
                mock_query.filter.return_value = mock_query
                mock_query.order_by.return_value = mock_query
                mock_query.first.return_value = None

                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = client.get("/benchmark/api/running")
                    assert resp.status_code == 200
                    data = resp.get_json()
                    assert data["success"] is False

    def test_running_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            # Force exception in the route
            mock_db.query.side_effect = RuntimeError("db error")
            with (
                patch(
                    "local_deep_research.benchmarks.web_api.benchmark_routes.BenchmarkRun",
                    create=True,
                ),
                patch(
                    "local_deep_research.benchmarks.web_api.benchmark_routes.BenchmarkStatus",
                    create=True,
                ),
            ):
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = client.get("/benchmark/api/running")
                    assert resp.status_code == 500


# ---------------------------------------------------------------------------
# get_benchmark_status
# ---------------------------------------------------------------------------


class TestGetBenchmarkStatus:
    def test_status_found(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            mock_svc.get_benchmark_status.return_value = {
                "completed_examples": 5,
                "overall_accuracy": 0.8,
                "avg_time_per_example": 12.5,
                "estimated_time_remaining": 60,
            }
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/status/1")
                assert resp.status_code == 200
                data = resp.get_json()
                assert data["success"] is True
                assert data["status"]["completed_examples"] == 5

    def test_status_not_found(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            mock_svc.get_benchmark_status.return_value = None
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/status/999")
                assert resp.status_code == 404

    def test_status_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            mock_svc.get_benchmark_status.side_effect = RuntimeError("boom")
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/status/1")
                assert resp.status_code == 500


# ---------------------------------------------------------------------------
# cancel_benchmark
# ---------------------------------------------------------------------------


class TestCancelBenchmark:
    def test_cancel_success(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            mock_svc.cancel_benchmark.return_value = True
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.post("/benchmark/api/cancel/1")
                assert resp.status_code == 200
                assert resp.get_json()["success"] is True

    def test_cancel_failure(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            mock_svc.cancel_benchmark.return_value = False
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.post("/benchmark/api/cancel/1")
                assert resp.status_code == 500

    def test_cancel_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            mock_svc.cancel_benchmark.side_effect = RuntimeError("oops")
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.post("/benchmark/api/cancel/1")
                assert resp.status_code == 500


# ---------------------------------------------------------------------------
# get_benchmark_history
# ---------------------------------------------------------------------------


class TestGetBenchmarkHistory:
    def _make_run(self, run_id, status_val="completed", run_name=None):
        run = MagicMock()
        run.id = run_id
        run.run_name = run_name
        run.created_at = datetime(2025, 1, 1, tzinfo=UTC)
        # Provenance fields added in migration 0014 — None mimics a pre-0014
        # row, so the existing history tests cover the back-compat path.
        run.start_time = None
        run.ldr_version = None
        run.total_examples = 10
        run.completed_examples = 8
        run.overall_accuracy = 0.75
        run.status = MagicMock()
        run.status.value = status_val
        run.search_config = {"tool": "searxng"}
        run.evaluation_config = {"provider": "openai"}
        run.datasets_config = {"simpleqa": {"count": 10}}
        return run

    def test_history_empty(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_query = MagicMock()
            mock_db.query.return_value = mock_query
            mock_query.order_by.return_value = mock_query
            mock_query.limit.return_value = mock_query
            mock_query.all.return_value = []

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/history")
                assert resp.status_code == 200
                assert resp.get_json()["runs"] == []

    def test_history_with_runs_and_avg_processing_time(self):
        app = _make_app()
        run = self._make_run(1, run_name=None)

        mock_result = MagicMock()
        mock_result.research_id = "res-1"
        mock_search_call = MagicMock()
        mock_search_call.research_id = "res-1"
        mock_search_call.results_count = 20

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_routed_query(
                runs=[run],
                avg_processing=15.5,
                results=[mock_result],
                search_calls=[mock_search_call],
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/history")
                assert resp.status_code == 200
                data = resp.get_json()
                assert data["success"] is True
                assert len(data["runs"]) == 1
                # run_name falls back to "Benchmark #{id}"
                assert "Benchmark #1" in data["runs"][0]["run_name"]

    def test_history_avg_processing_time_none(self):
        """Branch: avg_result is None."""
        app = _make_app()
        run = self._make_run(2, run_name="Named Run")

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_routed_query(
                runs=[run],
                avg_processing=None,
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/history")
                assert resp.status_code == 200
                data = resp.get_json()
                assert data["runs"][0]["avg_processing_time"] is None

    def test_history_search_metrics_exception(self):
        """Exception in search metrics calculation logged as warning."""
        app = _make_app()
        run = self._make_run(3)

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_routed_query(
                runs=[run],
                avg_processing=10.0,
                search_calls_exc=RuntimeError("no tracker"),
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/history")
                assert resp.status_code == 200

    def test_history_avg_time_exception(self):
        """Exception in avg processing time calculation."""
        app = _make_app()
        run = self._make_run(4)

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_query = MagicMock()
            mock_db.query.return_value = mock_query
            mock_query.order_by.return_value = mock_query
            mock_query.limit.return_value = mock_query
            mock_query.all.return_value = [run]
            mock_query.filter.side_effect = RuntimeError("avg fail")

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/history")
                # Should still return 200 with avg_processing_time=None
                assert resp.status_code == 200

    def test_history_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = RuntimeError("db fail")
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/history")
                assert resp.status_code == 500

    def test_history_no_research_ids(self):
        """Branch where research_ids list is empty."""
        app = _make_app()
        run = self._make_run(5)

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_query = MagicMock()
            mock_db.query.return_value = mock_query
            mock_query.order_by.return_value = mock_query
            mock_query.limit.return_value = mock_query
            mock_query.all.side_effect = [[run], []]  # runs, then results
            mock_query.filter.return_value = mock_query
            mock_query.scalar.return_value = 5.0

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/history")
                assert resp.status_code == 200


# ---------------------------------------------------------------------------
# get_benchmark_results
# ---------------------------------------------------------------------------


class TestGetBenchmarkResults:
    def _make_result(self, example_id="ex1", research_id="r1", completed=True):
        r = MagicMock()
        r.example_id = example_id
        r.dataset_type = _FakeDatasetType.SIMPLEQA
        r.question = "What is X?"
        r.correct_answer = "42"
        r.extracted_answer = "42"
        r.response = "The answer is 42"
        r.is_correct = True
        r.confidence = 0.95
        r.grader_response = "Correct"
        r.processing_time = 10.5
        r.sources = ["source1"]
        r.research_id = research_id
        r.completed_at = datetime(2025, 1, 1, tzinfo=UTC) if completed else None
        return r

    def test_results_success(self):
        app = _make_app()
        result = self._make_result()

        mock_search_call = MagicMock()
        mock_search_call.research_id = "r1"
        mock_search_call.results_count = 15

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_routed_query(
                results=[result],
                search_calls=[mock_search_call],
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1")
                assert resp.status_code == 200
                data = resp.get_json()
                assert data["success"] is True
                assert len(data["results"]) == 1
                assert data["results"][0]["search_result_count"] == 15

    def test_results_with_limit_param(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_routed_query()

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1?limit=5")
                assert resp.status_code == 200

    def test_results_no_completed_at(self):
        """Result with completed_at = None."""
        app = _make_app()
        result = self._make_result(completed=False)

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_routed_query(
                results=[result],
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1")
                assert resp.status_code == 200
                assert resp.get_json()["results"][0]["completed_at"] is None

    def test_results_no_research_id(self):
        """Result with research_id = None -> search_result_count = 0."""
        app = _make_app()
        result = self._make_result(research_id=None)

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_routed_query(
                results=[result],
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1")
                assert resp.status_code == 200
                assert resp.get_json()["results"][0]["search_result_count"] == 0

    def test_results_search_tracker_exception(self):
        """Exception fetching search metrics does not break the route."""
        app = _make_app()
        result = self._make_result()

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_routed_query(
                results=[result],
                search_calls_exc=RuntimeError("no tracker"),
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1")
                assert resp.status_code == 200

    def test_results_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_svc.sync_pending_results.side_effect = RuntimeError("fail")
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1")
                assert resp.status_code == 500


# ---------------------------------------------------------------------------
# export_benchmark_results
# ---------------------------------------------------------------------------


def _make_export_query_router(*, run=None, results=None):
    """Side-effect for mock_db.query that routes by model class for the
    /export endpoint.

    The endpoint runs two queries:
      1. session.query(BenchmarkRun).options(...).filter(...).one_or_none()
      2. session.query(BenchmarkResult).options(...).filter(...).order_by(...).all()
    """

    def _route(model, *args):
        chain = MagicMock()
        chain.options.return_value = chain
        chain.filter.return_value = chain
        chain.order_by.return_value = chain
        name = getattr(model, "__name__", "")
        if "BenchmarkRun" in name:
            chain.one_or_none.return_value = run
        else:
            chain.all.return_value = results or []
        return chain

    return _route


def _make_run_mock(
    *, ldr_version=None, start_time=None, settings_snapshot=None
):
    run = MagicMock()
    run.ldr_version = ldr_version
    run.start_time = start_time
    run.created_at = datetime(2025, 1, 1, tzinfo=UTC)
    run.settings_snapshot = settings_snapshot
    return run


class TestExportBenchmarkResults:
    def test_export_success(self):
        app = _make_app()
        mock_result = MagicMock()
        mock_result.example_id = "ex1"
        mock_result.dataset_type = _FakeDatasetType.SIMPLEQA
        mock_result.question = "Q?"
        mock_result.correct_answer = "A"
        mock_result.extracted_answer = "A"
        mock_result.is_correct = True
        mock_result.confidence = 0.9
        mock_result.processing_time = 5.0
        mock_result.completed_at = datetime(2025, 1, 1, tzinfo=UTC)

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_export_query_router(
                run=_make_run_mock(ldr_version="1.6.10"),
                results=[mock_result],
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1/export")
                assert resp.status_code == 200
                data = resp.get_json()
                assert data["success"] is True
                assert len(data["results"]) == 1
                assert "full_response" not in data["results"][0]

    def test_export_no_completed_at(self):
        app = _make_app()
        mock_result = MagicMock()
        mock_result.example_id = "ex2"
        mock_result.dataset_type = _FakeDatasetType.BROWSECOMP
        mock_result.question = "Q2?"
        mock_result.correct_answer = "B"
        mock_result.extracted_answer = "B"
        mock_result.is_correct = False
        mock_result.confidence = 0.5
        mock_result.processing_time = 3.0
        mock_result.completed_at = None

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = _make_export_query_router(
                run=_make_run_mock(),
                results=[mock_result],
            )

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1/export")
                assert resp.status_code == 200
                assert resp.get_json()["results"][0]["completed_at"] is None

    def test_export_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = RuntimeError("fail")
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/results/1/export")
                assert resp.status_code == 500


# ---------------------------------------------------------------------------
# get_saved_configs
# ---------------------------------------------------------------------------


class TestGetSavedConfigs:
    def test_configs_success(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/configs")
                assert resp.status_code == 200
                data = resp.get_json()
                assert data["success"] is True
                assert len(data["configs"]) == 2
                assert data["configs"][0]["name"] == "Quick Test"


# ---------------------------------------------------------------------------
# start_benchmark_simple
# ---------------------------------------------------------------------------


class TestStartBenchmarkSimple:
    def _post_simple(self, client, json_data):
        return client.post(
            "/benchmark/api/start-simple",
            json=json_data,
            content_type="application/json",
        )

    def test_simple_no_json(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.post("/benchmark/api/start-simple", data="bad")
                assert resp.status_code == 400

    def test_simple_empty_datasets(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = self._post_simple(client, {"datasets_config": {}})
                assert resp.status_code == 400

    def test_simple_success_openai_endpoint(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = "pw"
                mock_svc.create_benchmark_run.return_value = 100
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                        sess["session_id"] = "sid"
                    resp = self._post_simple(
                        client,
                        {
                            "run_name": "simple test",
                            "datasets_config": {"simpleqa": {"count": 3}},
                        },
                    )
                    assert resp.status_code == 200
                    assert resp.get_json()["success"] is True

    def test_simple_openai_provider(self):
        app = _make_app()
        with _patch_auth_and_db({"llm.provider": "openai"}) as (
            mock_svc,
            mgr,
            _,
        ):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 101
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_simple(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 200

    def test_simple_anthropic_provider(self):
        app = _make_app()
        with _patch_auth_and_db({"llm.provider": "anthropic"}) as (
            mock_svc,
            mgr,
            _,
        ):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 102
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_simple(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 200

    def test_simple_eval_openai_provider(self):
        app = _make_app()
        with _patch_auth_and_db(
            {
                "benchmark.evaluation.provider": "openai",
            }
        ) as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 103
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_simple(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 200

    def test_simple_eval_anthropic_provider(self):
        app = _make_app()
        with _patch_auth_and_db(
            {
                "benchmark.evaluation.provider": "anthropic",
            }
        ) as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 104
                mock_svc.start_benchmark.return_value = True
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_simple(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 200

    def test_simple_start_fails(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.return_value = 200
                mock_svc.start_benchmark.return_value = False
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_simple(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 500

    def test_simple_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with patch(
                "local_deep_research.database.session_passwords.session_password_store"
            ) as mock_pw:
                mock_pw.get_session_password.return_value = None
                mock_svc.create_benchmark_run.side_effect = RuntimeError("boom")
                with app.test_client() as client:
                    with client.session_transaction() as sess:
                        sess["username"] = "testuser"
                    resp = self._post_simple(
                        client,
                        {"datasets_config": {"simpleqa": {"count": 1}}},
                    )
                    assert resp.status_code == 500


# ---------------------------------------------------------------------------
# validate_config
# ---------------------------------------------------------------------------


class TestValidateConfig:
    def _post_validate(self, client, json_data):
        return client.post(
            "/benchmark/api/validate-config",
            json=json_data,
            content_type="application/json",
        )

    def test_validate_valid(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = self._post_validate(
                    client,
                    {
                        "search_config": {
                            "search_tool": "searxng",
                            "search_strategy": "focused_iteration",
                        },
                        "datasets_config": {"simpleqa": {"count": 10}},
                    },
                )
                assert resp.status_code == 200
                data = resp.get_json()
                assert data["valid"] is True
                assert data["total_examples"] == 10

    def test_validate_no_data(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                # Send non-dict JSON (list)
                resp = client.post(
                    "/benchmark/api/validate-config",
                    data="[]",
                    content_type="application/json",
                )
                assert resp.status_code == 200
                data = resp.get_json()
                assert data["valid"] is False

    def test_validate_missing_search_tool(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = self._post_validate(
                    client,
                    {
                        "search_config": {
                            "search_strategy": "focused_iteration"
                        },
                        "datasets_config": {"simpleqa": {"count": 10}},
                    },
                )
                data = resp.get_json()
                assert data["valid"] is False
                assert any("Search tool" in e for e in data["errors"])

    def test_validate_missing_search_strategy(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = self._post_validate(
                    client,
                    {
                        "search_config": {"search_tool": "searxng"},
                        "datasets_config": {"simpleqa": {"count": 10}},
                    },
                )
                data = resp.get_json()
                assert data["valid"] is False

    def test_validate_no_datasets(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = self._post_validate(
                    client,
                    {
                        "search_config": {
                            "search_tool": "searxng",
                            "search_strategy": "focused_iteration",
                        },
                        "datasets_config": {},
                    },
                )
                data = resp.get_json()
                assert data["valid"] is False
                assert any("dataset" in e.lower() for e in data["errors"])

    def test_validate_zero_total_examples(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = self._post_validate(
                    client,
                    {
                        "search_config": {
                            "search_tool": "searxng",
                            "search_strategy": "focused_iteration",
                        },
                        "datasets_config": {"simpleqa": {"count": 0}},
                    },
                )
                data = resp.get_json()
                assert data["valid"] is False

    def test_validate_large_count_accepted(self):
        """Large example counts should pass validation (no artificial cap)."""
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = self._post_validate(
                    client,
                    {
                        "search_config": {
                            "search_tool": "searxng",
                            "search_strategy": "focused_iteration",
                        },
                        "datasets_config": {"simpleqa": {"count": 1001}},
                    },
                )
                data = resp.get_json()
                assert data["valid"] is True
                assert data["total_examples"] == 1001

    def test_validate_datasets_config_with_non_dict_values(self):
        """datasets_config with valid structure but non-integer count."""
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, _):
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = self._post_validate(
                    client,
                    {
                        "search_config": {
                            "search_tool": "searxng",
                            "search_strategy": "focused_iteration",
                        },
                        "datasets_config": {"simpleqa": {}},
                    },
                )
                data = resp.get_json()
                # count defaults to 0, so total_examples = 0
                assert data["valid"] is False
                assert data["total_examples"] == 0


# ---------------------------------------------------------------------------
# get_search_quality
# ---------------------------------------------------------------------------


class TestGetSearchQuality:
    """Tests for /benchmark/api/search-quality.

    The route reads RateLimitEstimate rows from the user DB and maps each
    to {engine_type, total_attempts, success_rate (0-100 scale), status
    (EXCELLENT/GOOD/CAUTION/WARNING/CRITICAL)}.
    """

    @staticmethod
    def _make_estimate(engine_type, success_rate, total_attempts=100):
        est = MagicMock()
        est.engine_type = engine_type
        est.total_attempts = total_attempts
        est.success_rate = success_rate
        return est

    @staticmethod
    def _fetch(estimates):
        """Run the route with the given estimates and return (resp, json)."""
        app = _make_app()
        with _patch_auth_and_db() as (_svc, _mgr, mock_db):
            mock_db.query.return_value.all.return_value = estimates
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/search-quality")
                return resp, resp.get_json()

    def test_requires_authentication(self):
        app = _make_app()
        with _patch_auth_and_db():
            with app.test_client() as client:
                resp = client.get("/benchmark/api/search-quality")
                assert resp.status_code in (401, 302)

    def test_status_tier_excellent(self):
        resp, data = self._fetch([self._make_estimate("pubmed", 0.95)])
        assert resp.status_code == 200
        assert data["success"] is True
        assert "timestamp" in data
        row = data["search_quality"][0]
        assert row["engine_type"] == "pubmed"
        assert row["status"] == "EXCELLENT"
        assert row["success_rate"] == 95.0
        assert row["total_attempts"] == 100

    def test_status_tier_good(self):
        _resp, data = self._fetch([self._make_estimate("google", 0.90)])
        assert data["search_quality"][0]["status"] == "GOOD"
        assert data["search_quality"][0]["success_rate"] == 90.0

    def test_status_tier_caution(self):
        _resp, data = self._fetch([self._make_estimate("google", 0.75)])
        assert data["search_quality"][0]["status"] == "CAUTION"

    def test_status_tier_warning(self):
        _resp, data = self._fetch([self._make_estimate("google", 0.50)])
        assert data["search_quality"][0]["status"] == "WARNING"

    def test_status_tier_critical(self):
        _resp, data = self._fetch([self._make_estimate("google", 0.30)])
        assert data["search_quality"][0]["status"] == "CRITICAL"

    def test_success_rate_unit_conversion(self):
        # success_rate is stored as 0-1 in the DB; route multiplies by 100.
        _resp, data = self._fetch([self._make_estimate("google", 0.873)])
        assert data["search_quality"][0]["success_rate"] == 87.3

    def test_multiple_engines(self):
        resp, data = self._fetch(
            [
                self._make_estimate("bing", 0.6),
                self._make_estimate("google", 0.95),
            ]
        )
        assert resp.status_code == 200
        assert len(data["search_quality"]) == 2
        assert {r["engine_type"] for r in data["search_quality"]} == {
            "bing",
            "google",
        }

    def test_empty_engines(self):
        resp, data = self._fetch([])
        assert resp.status_code == 200
        assert data["success"] is True
        assert data["search_quality"] == []

    def test_shape_excludes_legacy_fields(self):
        # Regression guard: the old get_search_quality_stats shape included
        # recent_avg_results / min_recent_results / max_recent_results /
        # sample_size. Those are gone (the underlying data lived only in
        # the per-request in-memory tracker). Asserting their absence
        # protects the benchmark.html JS from accidentally depending on
        # them again.
        _resp, data = self._fetch([self._make_estimate("google", 0.95)])
        assert set(data["search_quality"][0].keys()) == {
            "engine_type",
            "total_attempts",
            "success_rate",
            "status",
        }

    def test_search_quality_exception(self):
        # The route now reads RateLimitEstimate rows from the user DB
        # (no get_tracker call), so a DB error is what surfaces a 500.
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = RuntimeError("db error")
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.get("/benchmark/api/search-quality")
                assert resp.status_code == 500


# ---------------------------------------------------------------------------
# delete_benchmark_run
# ---------------------------------------------------------------------------


class TestDeleteBenchmarkRun:
    def test_delete_success(self):
        app = _make_app()
        mock_run = MagicMock()
        mock_run.id = 1
        mock_run.status = MagicMock()
        mock_run.status.value = "completed"

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_query = MagicMock()
            mock_db.query.return_value = mock_query
            mock_query.filter.return_value = mock_query
            mock_query.first.return_value = mock_run
            mock_query.delete.return_value = None

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.delete("/benchmark/api/delete/1")
                assert resp.status_code == 200
                assert resp.get_json()["success"] is True

    def test_delete_not_found(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_query = MagicMock()
            mock_db.query.return_value = mock_query
            mock_query.filter.return_value = mock_query
            mock_query.first.return_value = None

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.delete("/benchmark/api/delete/999")
                assert resp.status_code == 404

    def test_delete_in_progress(self):
        app = _make_app()
        mock_run = MagicMock()
        mock_run.status = MagicMock()
        mock_run.status.value = "in_progress"

        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_query = MagicMock()
            mock_db.query.return_value = mock_query
            mock_query.filter.return_value = mock_query
            mock_query.first.return_value = mock_run

            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.delete("/benchmark/api/delete/1")
                assert resp.status_code == 400
                assert (
                    "running" in resp.get_json()["error"].lower()
                    or "cancel" in resp.get_json()["error"].lower()
                )

    def test_delete_exception(self):
        app = _make_app()
        with _patch_auth_and_db() as (mock_svc, mgr, mock_db):
            mock_db.query.side_effect = RuntimeError("db error")
            with app.test_client() as client:
                with client.session_transaction() as sess:
                    sess["username"] = "testuser"
                resp = client.delete("/benchmark/api/delete/1")
                assert resp.status_code == 500