Workflow Testing Guidelines

Overview

The workflow module handles calibration workflow orchestration using Prefect. Tests should verify:

• Task state management: TaskManager, TaskStateManager state transitions
• Execution lifecycle: ExecutionManager, ExecutionService operations
• Scheduling logic: CRScheduler parallel grouping and conflict detection
• Result processing: R² validation, fidelity checks, parameter handling
• Data persistence: File saving, history recording
• Flow helpers: Python Flow Editor API functions
• Context management: Session context thread safety

Test Framework

• Framework: pytest
• Mocking: unittest.mock (MagicMock, patch)
• Fixtures: pytest fixtures with proper scope management
• Database: MongoDB via init_db fixture (shared conftest.py)

---

Test Directory Structure

Tests mirror the source directory structure:

tests/qdash/workflow/
├── __init__.py
├── engine/
│   ├── __init__.py
│   └── calibration/
│       ├── __init__.py
│       ├── task/                          # Task management tests
│       │   ├── __init__.py
│       │   ├── test_manager.py            # TaskManager tests
│       │   ├── test_executor.py           # TaskExecutor tests
│       │   ├── test_state_manager.py      # TaskStateManager tests
│       │   ├── test_result_processor.py   # TaskResultProcessor tests
│       │   └── test_history_recorder.py   # TaskHistoryRecorder tests
│       ├── execution/                     # Execution management tests
│       │   ├── __init__.py
│       │   ├── test_manager.py            # ExecutionManager tests
│       │   ├── test_state_manager.py      # ExecutionStateManager tests
│       │   └── test_service.py            # ExecutionService tests
│       ├── scheduler/                     # Scheduler tests
│       │   ├── __init__.py
│       │   ├── test_cr_scheduler.py       # CRScheduler tests
│       │   └── test_plugins.py            # Scheduler plugins tests
│       └── repository/                    # Repository tests
│           ├── __init__.py
│           └── test_filesystem_impl.py    # Filesystem implementation tests
└── service/                           # Calibration service tests
    ├── __init__.py
    ├── test_config.py                 # CalibServiceConfig tests
    ├── test_factory.py                # Factory tests
    ├── test_context.py                # SessionContext tests
    └── test_session.py                # CalibService tests

Mapping Rules

Source Path	Test Path
src/qdash/workflow/engine/calibration/task/manager.py	tests/qdash/workflow/engine/calibration/task/test_manager.py
src/qdash/workflow/engine/calibration/task/executor.py	tests/qdash/workflow/engine/calibration/task/test_executor.py
src/qdash/workflow/engine/calibration/task/state_manager.py	tests/qdash/workflow/engine/calibration/task/test_state_manager.py
src/qdash/workflow/engine/calibration/execution/manager.py	tests/qdash/workflow/engine/calibration/execution/test_manager.py
src/qdash/workflow/engine/calibration/execution/service.py	tests/qdash/workflow/engine/calibration/execution/test_service.py
src/qdash/workflow/engine/calibration/scheduler/cr_scheduler.py	tests/qdash/workflow/engine/calibration/scheduler/test_cr_scheduler.py
src/qdash/workflow/service/context.py	tests/qdash/workflow/service/test_context.py
src/qdash/workflow/service/session.py	tests/qdash/workflow/service/test_session.py

---

Test File Naming Conventions

File Naming

• Prefix: test_
• Name: Module name being tested
• Extension: .py

# Good
test_task_manager.py
test_cr_scheduler.py
test_flow_helpers.py

# Bad
task_manager_tests.py
TestTaskManager.py

Module Docstrings

Each test file should have a module-level docstring explaining the purpose:

"""Tests for TaskManager.

These tests verify the core behavior of TaskManager to serve as a safety net
during refactoring. Tests cover:
1. Task state management (start_task, end_task, update_task_status)
2. Parameter management (put_input_parameters, put_output_parameters)
3. Figure/raw data saving (save_figures, save_raw_data)
"""

---

Test Class and Method Naming

Class Naming

Group related tests into classes with descriptive names:

# Good - Descriptive class names
class TestTaskStateManagement:
    """Test task state transitions."""

class TestParameterManagement:
    """Test parameter handling."""

class TestFigureAndDataSaving:
    """Test figure and raw data saving."""

class TestBatchOperations:
    """Test batch task operations."""

# Bad
class TestTaskManager:  # Too generic
class Tests:            # Not descriptive

Method Naming

Use snake*case with test* prefix. Include the action and expected outcome:

# Good - Clear action and expectation
def test_start_task_updates_status_to_running(self):
def test_ensure_task_exists_is_idempotent(self):
def test_get_task_raises_for_nonexistent_task(self):
def test_execute_task_with_r2_validation_fail(self):

# Bad
def test_start_task(self):           # Missing expectation
def test1(self):                     # Not descriptive
def test_it_works(self):             # Vague

---

Test Organization Pattern

AAA Pattern (Arrange-Act-Assert)

Always structure tests using the AAA pattern:

def test_start_task_updates_status_to_running(self):
    """Test start_task changes status to RUNNING."""
    # Arrange
    tm = TaskManager(username="test", execution_id="test-001", qids=["0"], calib_dir="/tmp")
    tm._ensure_task_exists("CheckRabi", "qubit", "0")

    # Act
    tm.start_task("CheckRabi", "qubit", "0")

    # Assert
    task = tm.get_task("CheckRabi", "qubit", "0")
    assert task.status == TaskStatusModel.RUNNING
    assert task.start_at != ""

Single Concept Per Test

Each test should verify one concept:

# Good - Single concept
def test_update_task_status_to_completed(self):
    """Test update_task_status_to_completed."""
    tm = TaskManager(username="test", execution_id="test-001", qids=["0"], calib_dir="/tmp")
    tm._ensure_task_exists("CheckRabi", "qubit", "0")

    tm.update_task_status_to_completed("CheckRabi", "Task done", "qubit", "0")

    task = tm.get_task("CheckRabi", "qubit", "0")
    assert task.status == TaskStatusModel.COMPLETED
    assert task.message == "Task done"

def test_update_task_status_to_failed(self):
    """Test update_task_status_to_failed."""
    tm = TaskManager(username="test", execution_id="test-001", qids=["0"], calib_dir="/tmp")
    tm._ensure_task_exists("CheckRabi", "qubit", "0")

    tm.update_task_status_to_failed("CheckRabi", "Error occurred", "qubit", "0")

    task = tm.get_task("CheckRabi", "qubit", "0")
    assert task.status == TaskStatusModel.FAILED
    assert task.message == "Error occurred"

---

Fixture Patterns

Basic Fixtures

Define fixtures for commonly used objects:

@pytest.fixture
def mock_task(self):
    """Create a mock task for testing."""
    task = MagicMock()
    task.get_name.return_value = "CheckRabi"
    task.get_task_type.return_value = "qubit"
    task.is_qubit_task.return_value = True
    task.is_coupling_task.return_value = False
    task.backend = "fake"
    task.r2_threshold = 0.7
    task.name = "CheckRabi"
    return task

@pytest.fixture
def mock_session(self):
    """Create a mock session for testing."""
    session = MagicMock()
    session.name = "fake"
    return session

@pytest.fixture
def mock_execution_manager(self):
    """Create a mock execution manager for testing."""
    em = MagicMock()
    em.execution_id = "test-exec-001"
    em.chip_id = "test-chip"
    em.to_datamodel.return_value = ExecutionModel(...)
    em.update_with_task_manager.return_value = em
    return em

Temporary Directory Fixtures

Use pytest's tmp_path or tempfile for file operations:

@pytest.fixture
def calib_dir(self):
    """Create a temporary calibration directory with required subdirectories."""
    with tempfile.TemporaryDirectory() as tmpdir:
        Path(tmpdir, "task").mkdir()
        Path(tmpdir, "fig").mkdir()
        Path(tmpdir, "raw_data").mkdir()
        yield tmpdir

def test_save_figures_creates_files(self):
    """Test save_figures creates png and json files."""
    with tempfile.TemporaryDirectory() as tmpdir:
        tm = TaskManager(username="test", execution_id="test-001", qids=["0"], calib_dir=tmpdir)
        # ... test code

Mock Class Fixtures

Create mock implementations for complex dependencies:

class MockTask:
    """Mock task for testing."""

    def __init__(
        self,
        name: str = "CheckRabi",
        task_type: str = "qubit",
        r2_threshold: float = 0.7,
        backend: str = "fake",
    ):
        self.name = name
        self._task_type = task_type
        self.r2_threshold = r2_threshold
        self.backend = backend

    def get_name(self) -> str:
        return self.name

    def get_task_type(self) -> str:
        return self._task_type

    def preprocess(self, session, qid):
        return PreProcessResult(input_parameters={"param1": ParameterModel(value=1.0)})

    def run(self, session, qid):
        return RunResult(raw_result={"data": [1, 2, 3]}, r2={"0": 0.95})

    def postprocess(self, session, execution_id, run_result, qid):
        return PostProcessResult(
            output_parameters={"qubit_frequency": ParameterModel(value=5.0)},
            figures=[],
            raw_data=[],
        )

Setup/Teardown Methods

Use setup_method for per-test initialization:

class TestSessionContext:
    """Test SessionContext class."""

    def setup_method(self):
        """Clear session before each test."""
        clear_current_session()

    def test_set_and_get_session(self):
        """Test setting and getting session."""
        context = SessionContext()
        mock_session = MagicMock()
        context.set_session(mock_session)
        result = context.get_session()
        assert result is mock_session

---

Mocking Patterns

Patching Database Dependencies

Use patch for database operations:

def test_execute_task_without_run_result(
    self, init_db, mock_task, mock_session, mock_execution_manager, calib_dir
):
    """Test execute_task when run returns None."""
    tm = TaskManager(
        username="test", execution_id="test-exec-001", qids=["0"], calib_dir=calib_dir
    )

    mock_task.preprocess.return_value = None
    mock_task.run.return_value = None

    with (
        patch("qdash.dbmodel.chip.ChipDocument") as mock_chip_doc,
        patch("qdash.dbmodel.chip_history.ChipHistoryDocument") as mock_chip_history,
    ):
        mock_chip_doc.get_current_chip.return_value = MagicMock()

        em_result, tm_result = tm.execute_task(
            task_instance=mock_task,
            session=mock_session,
            execution_manager=mock_execution_manager,
            qid="0",
        )

    task = tm_result.get_task("CheckRabi", "qubit", "0")
    assert task.status == TaskStatusModel.COMPLETED

Monkeypatching for Module-Level State

Use monkeypatch for global state:

def test_flow_session_attributes(self, monkeypatch):
    """Test that FlowSession initializes with correct attributes."""

    class MockExecutionManager:
        def __init__(self, **kwargs):
            for key, value in kwargs.items():
                setattr(self, key, value)
            self.calib_data_path = kwargs.get("calib_data_path", "")
            self.calib_data = type("obj", (object,), {"qubit": {}, "coupling": {}})()

        def save(self):
            return self

    monkeypatch.setattr(
        "qdash.workflow.helpers.flow_helpers.ExecutionManager",
        MockExecutionManager,
    )

Mocking Method Return Values

Configure mock return values and side effects:

@pytest.fixture
def mock_result_processor(self):
    """Create a mock result processor."""
    processor = MagicMock()
    processor.validate_r2.return_value = True
    processor.process_output_parameters.return_value = {
        "qubit_frequency": ParameterModel(
            value=5.0, execution_id="exec-001", task_id="task-001"
        )
    }
    return processor

def test_execute_task_raises_on_r2_validation_failure(
    self, executor, mock_result_processor, mock_state_manager
):
    """Test execute_task raises on R² validation failure."""
    mock_result_processor.validate_r2.side_effect = R2ValidationError("R² value too low")
    task = MockTask()
    session = MockSession()

    with pytest.raises(ValueError, match="R² value too low"):
        executor.execute_task(task, session, "0")

Patching Class Methods

Use patch.object for class method patching:

@patch.object(CRScheduler, "_load_chip_data")
def test_generate_with_design_based_direction(mock_load, scheduler):
    """Test schedule generation using design-based direction inference."""
    chip_doc = MagicMock()
    chip_doc.qubits = {...}
    mock_load.return_value = chip_doc

    schedule = scheduler.generate(candidate_qubits=["0", "1", "2", "3"])

    assert schedule.metadata["direction_method"] == "design_based"

---

Testing Calibration Engine Components

TaskManager Tests

Test task lifecycle and state management:

class TestTaskStateManagement:
    """Test task state transitions."""

    def test_init_creates_empty_task_containers_for_qids(self):
        """Test TaskManager initialization with qids."""
        qids = ["0", "1", "2"]
        tm = TaskManager(username="test", execution_id="test-001", qids=qids, calib_dir="/tmp")

        for qid in qids:
            assert qid in tm.task_result.qubit_tasks
            assert qid in tm.task_result.coupling_tasks
            assert qid in tm.calib_data.qubit
            assert tm.task_result.qubit_tasks[qid] == []

    def test_ensure_task_exists_adds_new_task(self):
        """Test _ensure_task_exists adds task to container."""
        tm = TaskManager(username="test", execution_id="test-001", qids=["0"], calib_dir="/tmp")

        tm._ensure_task_exists("CheckRabi", "qubit", "0")

        tasks = tm.task_result.qubit_tasks["0"]
        assert len(tasks) == 1
        assert tasks[0].name == "CheckRabi"
        assert tasks[0].status == TaskStatusModel.SCHEDULED

TaskExecutor Tests

Test execution flow and error handling:

class TestTaskExecutorExecuteTask:
    """Test TaskExecutor.execute_task method."""

    def test_execute_task_success(self, executor, mock_state_manager):
        """Test successful task execution."""
        task = MockTask()
        session = MockSession()

        result = executor.execute_task(task, session, "0")

        assert result["success"] is True
        assert result["task_name"] == "CheckRabi"
        mock_state_manager.start_task.assert_called_once_with("CheckRabi", "qubit", "0")
        mock_state_manager.end_task.assert_called_once_with("CheckRabi", "qubit", "0")

    def test_execute_task_always_calls_end_task(self, executor, mock_state_manager):
        """Test execute_task always calls end_task even on failure."""
        task = MockTask()
        task.run = MagicMock(side_effect=RuntimeError("Error"))
        session = MockSession()

        with pytest.raises(TaskExecutionError):
            executor.execute_task(task, session, "0")

        # end_task should always be called (in finally block)
        mock_state_manager.end_task.assert_called_once()

CRScheduler Tests

Test scheduling logic with parameterized tests:

from qdash.workflow.engine.scheduler.cr_utils import group_cr_pairs_by_conflict

@pytest.mark.parametrize(
    "cr_pairs,qid_to_mux,expected_min_groups,reason",
    [
        (["0-1", "2-3"], {"0": 0, "1": 0, "2": 0, "3": 0}, 2, "same MUX conflict"),
        (["0-1", "1-2"], {"0": 0, "1": 0, "2": 0}, 2, "shared qubit conflict"),
        (["0-1", "4-5"], {"0": 0, "1": 0, "4": 1, "5": 1}, 1, "different MUX, no conflicts"),
    ],
)
def test_group_cr_pairs_by_conflict(cr_pairs, qid_to_mux, expected_min_groups, reason):
    """Test CR pair grouping with various conflict scenarios."""
    mux_conflict_map = {}

    groups = group_cr_pairs_by_conflict(cr_pairs, qid_to_mux, mux_conflict_map)

    assert len(groups) >= expected_min_groups, f"Failed: {reason}"

    # Verify no conflicts within groups
    for group in groups:
        qubits_in_group = set()
        for pair in group:
            q1, q2 = pair.split("-")
            assert q1 not in qubits_in_group
            assert q2 not in qubits_in_group
            qubits_in_group.add(q1)
            qubits_in_group.add(q2)

@pytest.mark.parametrize(
    "strategy",
    [
        "largest_first",
        "smallest_last",
        "saturation_largest_first",
    ],
)
def test_coloring_strategies(strategy):
    """Test different coloring strategies produce valid results."""
    cr_pairs = ["0-1", "1-2", "2-3"]
    qid_to_mux = {"0": 0, "1": 0, "2": 0, "3": 0}
    mux_conflict_map = {}

    groups = group_cr_pairs_by_conflict(
        cr_pairs, qid_to_mux, mux_conflict_map, coloring_strategy=strategy
    )

    assert len(groups) >= 2

---

Testing Flow Components

Context Management Tests

Test thread safety and lifecycle:

class TestThreadSafety:
    """Test thread safety of SessionContext."""

    def test_thread_local_isolation(self):
        """Test that sessions are isolated per thread."""
        context = SessionContext()
        results = {}
        barrier = threading.Barrier(2)

        def thread_func(thread_id, session):
            context.set_session(session)
            barrier.wait()  # Sync threads
            results[thread_id] = context.get_session()

        session1 = MagicMock(name="session1")
        session2 = MagicMock(name="session2")

        thread1 = threading.Thread(target=thread_func, args=("t1", session1))
        thread2 = threading.Thread(target=thread_func, args=("t2", session2))

        thread1.start()
        thread2.start()
        thread1.join()
        thread2.join()

        assert results["t1"] is session1
        assert results["t2"] is session2

Context Manager Tests

Test scope-based session management:

def test_session_scope_sets_and_clears(self):
    """Test session_scope context manager."""
    context = SessionContext()
    mock_session = MagicMock()

    with context.session_scope(mock_session) as session:
        assert session is mock_session
        assert context.get_session() is mock_session

    # Session should be cleared after scope
    assert context.get_session() is None

def test_session_scope_restores_previous(self):
    """Test session_scope restores previous session."""
    context = SessionContext()
    session1 = MagicMock(name="session1")
    session2 = MagicMock(name="session2")

    context.set_session(session1)

    with context.session_scope(session2):
        assert context.get_session() is session2

    # Original session should be restored
    assert context.get_session() is session1

---

Testing Helper Functions

Initialization Tests

Test helper function setup:

class TestFlowSessionInitialization:
    """Test FlowSession initialization and basic setup."""

    def test_flow_session_attributes(self, monkeypatch):
        """Test that FlowSession initializes with correct attributes."""
        # Mock dependencies
        monkeypatch.setattr(
            "qdash.workflow.helpers.flow_helpers.ExecutionManager",
            MockExecutionManager,
        )
        monkeypatch.setattr(
            "qdash.workflow.helpers.flow_helpers.create_backend",
            lambda **kwargs: MockSession(),
        )

        session = FlowSession(
            username="test_user",
            execution_id="20240101-001",
            chip_id="chip_1",
            backend="fake",
        )

        assert session.username == "test_user"
        assert session.execution_id == "20240101-001"
        assert session.chip_id == "chip_1"

Global State Tests

Test module-level session management:

class TestGlobalSessionHelpers:
    """Test global session helper functions."""

    def test_init_and_get_session(self, monkeypatch):
        """Test init_calibration and get_session."""
        monkeypatch.setattr(...)

        session1 = init_calibration(
            username="test_user",
            execution_id="20240101-001",
            chip_id="chip_1",
        )

        session2 = get_session()

        assert session1 is session2

    def test_get_session_without_init(self):
        """Test that get_session raises error when no session exists."""
        import qdash.workflow.helpers.flow_helpers as flow_helpers
        flow_helpers._current_session = None

        with pytest.raises(RuntimeError, match="No active calibration session"):
            get_session()

---

Database Testing

Using InMemory Repositories (Recommended)

For most unit tests, use InMemory repository implementations instead of MongoDB. This approach is faster, doesn't require database setup, and provides better isolation.

from qdash.repository.inmemory import (
    InMemoryExecutionRepository,
    InMemoryChipRepository,
    InMemoryCalibrationNoteRepository,
    InMemoryQubitCalibrationRepository,
    InMemoryCouplingCalibrationRepository,
    InMemoryExecutionCounterRepository,
    InMemoryExecutionLockRepository,
    InMemoryUserRepository,
    InMemoryTaskRepository,
)


class TestWithInMemoryRepositories:
    """Test using InMemory repositories."""

    def test_generate_execution_id(self):
        """Test execution ID generation without MongoDB."""
        from qdash.workflow.service.calib_service import generate_execution_id

        counter_repo = InMemoryExecutionCounterRepository()

        id1 = generate_execution_id(
            username="alice",
            chip_id="chip_1",
            project_id="proj-1",
            counter_repo=counter_repo,
        )
        id2 = generate_execution_id(
            username="alice",
            chip_id="chip_1",
            project_id="proj-1",
            counter_repo=counter_repo,
        )

        assert id1.endswith("-000")
        assert id2.endswith("-001")

    def test_scheduler_with_injected_repository(self):
        """Test CRScheduler with InMemory chip repository."""
        from qdash.datamodel.chip import ChipModel
        from qdash.workflow.engine.scheduler import CRScheduler

        chip_repo = InMemoryChipRepository()
        mock_chip = ChipModel(
            chip_id="chip_1",
            username="alice",
            qubits={"0": {}, "1": {}},
            couplings={"0-1": {}},
            # ... other required fields
        )
        chip_repo.add_chip("alice", mock_chip)

        scheduler = CRScheduler(
            username="alice",
            chip_id="chip_1",
            chip_repo=chip_repo,
        )
        # Now scheduler uses InMemory data instead of MongoDB

    def test_service_with_multiple_repositories(self):
        """Test CalibService with multiple InMemory repositories."""
        user_repo = InMemoryUserRepository()
        lock_repo = InMemoryExecutionLockRepository()
        counter_repo = InMemoryExecutionCounterRepository()

        user_repo.add_user("alice", default_project_id="proj-1")

        # Use DI to inject repositories
        # (example pattern, actual implementation may vary)

Benefits of InMemory Testing:

• No MongoDB setup required
• Faster test execution
• Better test isolation (no shared database state)
• Easy to set up specific test scenarios
• clear() method for test setup/teardown

Test Helper Methods:

Each InMemory repository provides helper methods for testing:

# Add test data
repo.add_chip("alice", mock_chip)
repo.add_user("alice", default_project_id="proj-1")
repo.add_tasks("alice", ["CheckFreq", "CheckRabi"])

# Clear all data (useful in setup_method)
repo.clear()

# Get all stored data (for assertions)
results = repo.get_all()

Using init_db Fixture

For tests requiring MongoDB:

class TestExecuteTaskIntegration:
    """Integration tests for execute_task method.

    These tests require MongoDB connection (init_db fixture).
    """

    def test_execute_task_records_to_task_result_history(
        self, init_db, mock_task, mock_session, mock_execution_manager, calib_dir
    ):
        """Test execute_task creates TaskResultHistoryDocument."""
        from qdash.dbmodel.task_result_history import TaskResultHistoryDocument

        tm = TaskManager(
            username="test", execution_id="test-exec-001", qids=["0"], calib_dir=calib_dir
        )

        mock_task.preprocess.return_value = None
        mock_task.run.return_value = None

        with (...):
            em_result, tm_result = tm.execute_task(...)

        # Verify TaskResultHistoryDocument was created
        task = tm_result.get_task("CheckRabi", "qubit", "0")
        doc = TaskResultHistoryDocument.find_one({"task_id": task.task_id}).run()
        assert doc is not None
        assert doc.name == "CheckRabi"

Isolating Database Tests

Mark tests that need database:

@pytest.mark.usefixtures("init_db")
class TestWithDatabase:
    """Tests that require database connection."""
    pass

---

Assertion Patterns

State Assertions

Verify object state changes:

def test_start_task_updates_status_to_running(self):
    """Test start_task changes status to RUNNING."""
    tm = TaskManager(...)
    tm._ensure_task_exists("CheckRabi", "qubit", "0")

    tm.start_task("CheckRabi", "qubit", "0")

    task = tm.get_task("CheckRabi", "qubit", "0")
    assert task.status == TaskStatusModel.RUNNING
    assert task.start_at != ""  # Non-empty timestamp

Exception Assertions

Use pytest.raises with match patterns:

def test_get_task_raises_for_nonexistent_task(self):
    """Test get_task raises ValueError for unknown task."""
    tm = TaskManager(...)

    with pytest.raises(ValueError, match="Task 'Unknown' not found"):
        tm.get_task("Unknown", "qubit", "0")

def test_execute_task_with_r2_validation_fail(self, ...):
    """Test execute_task with failing R² validation."""
    ...
    with pytest.raises(ValueError, match="R² value too low"):
        tm.execute_task(...)

Mock Assertions

Verify mock interactions:

def test_execute_task_success(self, executor, mock_state_manager):
    """Test successful task execution."""
    task = MockTask()
    session = MockSession()

    result = executor.execute_task(task, session, "0")

    # Verify method calls
    mock_state_manager.start_task.assert_called_once_with("CheckRabi", "qubit", "0")
    mock_state_manager.end_task.assert_called_once_with("CheckRabi", "qubit", "0")

def test_execute_task_calls_preprocess(self, executor, mock_state_manager):
    """Test execute_task calls preprocess and stores input parameters."""
    task = MockTask()
    session = MockSession()

    executor.execute_task(task, session, "0")

    mock_state_manager.put_input_parameters.assert_called_once()
    call_args = mock_state_manager.put_input_parameters.call_args
    assert call_args[0][0] == "CheckRabi"
    assert call_args[0][1] == {"param1": 1.0}

File Assertions

Verify file operations:

def test_save_figures_creates_files(self):
    """Test save_figures creates png and json files."""
    with tempfile.TemporaryDirectory() as tmpdir:
        tm = TaskManager(...)
        fig = go.Figure(data=[go.Scatter(x=[1, 2, 3], y=[4, 5, 6])])

        tm.save_figures([fig], "CheckRabi", "qubit", qid="0")

        task = tm.get_task("CheckRabi", "qubit", "0")

        # Check paths are recorded
        assert len(task.figure_path) == 1
        assert len(task.json_figure_path) == 1

        # Check files exist
        png_path = Path(task.figure_path[0])
        json_path = Path(task.json_figure_path[0])
        assert png_path.exists()
        assert json_path.exists()

---

Coverage Guidelines

Minimum Coverage Targets

Component	Target Coverage
Task State Management	90%
Task Execution	85%
Result Processing	90%
CR Scheduler	85%
Flow Helpers	80%
Context Management	90%

Required Test Categories

Each component should have tests for:

1. Initialization: Constructor behavior, default values
2. Happy Path: Normal operation scenarios
3. Error Handling: Exception cases, validation failures
4. Edge Cases: Empty inputs, boundary conditions
5. State Transitions: Lifecycle operations

Critical Paths to Test

• Task status transitions (SCHEDULED → RUNNING → COMPLETED/FAILED/CANCELLED)
• Cancellation handling (on_cancellation hook, status updates, lock release)
• R² validation and rollback behavior
• Fidelity validation (> 100% rejection)
• Parameter propagation (input → output → calib_data)
• File saving with conflict resolution
• Thread-safe context management

---

Common Pitfalls

1. Not Cleaning Up Global State

# Bad - Global state persists between tests
def test_session1(self):
    init_calibration(...)

def test_session2(self):
    get_session()  # May get session from test_session1!

# Good - Clean up in setup
class TestSession:
    def setup_method(self):
        clear_current_session()

    def test_session1(self):
        init_calibration(...)

2. Using Real Database Without Fixture

# Bad - No database fixture
def test_db_operation(self):
    doc = TaskResultHistoryDocument.find_one({})  # Fails without DB

# Good - Use init_db fixture
def test_db_operation(self, init_db):
    doc = TaskResultHistoryDocument.find_one({})

3. Insufficient Mock Configuration

# Bad - Mock not fully configured
mock_task = MagicMock()
executor.execute_task(mock_task, session, "0")  # May fail unpredictably

# Good - Configure all required attributes
mock_task = MagicMock()
mock_task.get_name.return_value = "CheckRabi"
mock_task.get_task_type.return_value = "qubit"
mock_task.is_qubit_task.return_value = True
mock_task.r2_threshold = 0.7

4. Not Testing Error Recovery

# Bad - Only test success
def test_execute_task(self):
    result = executor.execute_task(task, session, "0")
    assert result["success"] is True

# Good - Test error handling and cleanup
def test_execute_task_error_recovery(self):
    task.run.side_effect = RuntimeError("Error")

    with pytest.raises(TaskExecutionError):
        executor.execute_task(task, session, "0")

    # Verify cleanup was performed
    mock_state_manager.end_task.assert_called_once()

5. Hardcoded Paths in Tests

# Bad - Hardcoded paths
def test_save_data(self):
    tm = TaskManager(..., calib_dir="/tmp/test")  # May conflict

# Good - Use temporary directories
def test_save_data(self):
    with tempfile.TemporaryDirectory() as tmpdir:
        tm = TaskManager(..., calib_dir=tmpdir)

---

Running Tests

Run All Workflow Tests

pytest tests/qdash/workflow/ -v

Run Specific Test File

pytest tests/qdash/workflow/engine/calibration/test_task_manager.py -v

Run Specific Test Class

pytest tests/qdash/workflow/engine/calibration/test_task_manager.py::TestTaskStateManagement -v

Run With Coverage

pytest tests/qdash/workflow/ --cov=src/qdash/workflow --cov-report=html

Run Parameterized Tests Only

pytest tests/qdash/workflow/engine/calibration/test_cr_scheduler.py -v -k "parametrize"