testing-strategies

name	testing-strategies
description	Comprehensive testing patterns and examples for Dagster, dbt, FastAPI, and React in the economic data project.

Testing Strategies for Economic Data Project

This skill provides comprehensive testing patterns and examples for all components of the economic data project: Dagster (orchestration), dbt (transformations), FastAPI (backend), and React (frontend).

Quick Reference

Component	Framework	Run Command	Location
Dagster	pytest	make test or cd macro_agents && uv run pytest tests/ -v	macro_agents/tests/
dbt	dbt test	dbt test or dbt test --select model_name	dbt_project/tests/ + schema.yml
Backend API	pytest	cd api && PYTHONPATH=.. uv run pytest tests/ -v	api/tests/
Frontend	vitest	cd frontend && npm test -- --run	frontend/tests/

---

1. Dagster Testing

Test Categories

1. Unit Tests - Test individual resources and functions
2. Integration Tests - Test Dagster definitions load correctly
3. Schedule/Sensor Tests - Test automation configuration
4. Data Validation Tests - Test data schemas and integrity

Resource Unit Tests

Test Dagster resources in isolation with mocked dependencies:

import pytest
from unittest.mock import Mock, patch
from macro_agents.defs.resources.motherduck import MotherDuckResource

class TestMotherDuckResource:
    def test_initialization_dev_environment(self):
        resource = MotherDuckResource(
            md_token="test_token",
            environment="dev",
            local_path="test.duckdb"
        )
        assert resource.environment == "dev"
        assert resource.db_connection == "test.duckdb"

    def test_initialization_prod_environment(self):
        resource = MotherDuckResource(
            md_token="test_token",
            environment="prod",
            md_database="test_db"
        )
        assert resource.environment == "prod"
        assert resource.db_connection == "md:?motherduck_token=test_token"

Testing with Temporary Database

Use tempfile for tests requiring actual database operations:

import tempfile
import os
import polars as pl

def test_table_exists(self):
    with tempfile.NamedTemporaryFile(suffix=".duckdb", delete=False) as tmp_file:
        resource = MotherDuckResource(
            md_token="test_token",
            environment="dev",
            local_path=tmp_file.name
        )

        assert not resource.table_exists("non_existent_table")

        test_df = pl.DataFrame({"id": [1, 2, 3]})
        resource.drop_create_duck_db_table("test_table", test_df)
        assert resource.table_exists("test_table")

        os.unlink(tmp_file.name)

Integration Tests - Definitions Load

Verify all Dagster definitions load without errors:

from macro_agents.definitions import defs

class TestDagsterDefinitions:
    def test_definitions_load(self):
        assert defs is not None
        assert len(defs.assets) > 0
        assert len(defs.resources) > 0

    def test_all_resources_are_configurable(self):
        for resource_key, resource in defs.resources.items():
            assert hasattr(resource, "model_config")
            assert hasattr(type(resource), "model_fields")

Schedule and Sensor Tests

Test schedule configuration and timing:

from macro_agents.defs.replication import weekly_replication_schedule

class TestSchedules:
    def test_weekly_replication_schedule_configuration(self):
        assert weekly_replication_schedule.name == "weekly_replication_schedule"
        assert weekly_replication_schedule.cron_schedule == "0 2 * * 0"
        assert weekly_replication_schedule.execution_timezone == "America/New_York"

    def test_schedule_timing_consistency(self):
        replication_hour = 2  # 2 AM EST
        assert replication_hour <= 6, "Weekly replication should run early in the morning"

Mocking External Services

Use patch for external API calls and credentials:

from unittest.mock import Mock, patch

def test_sling_resource_setup_with_credentials(self):
    test_creds = {
        "type": "service_account",
        "project_id": "test-project",
        "private_key": "-----BEGIN PRIVATE KEY-----\ntest\n-----END PRIVATE KEY-----\n",
    }
    creds_json = json.dumps(test_creds)
    mock_context = Mock()
    mock_context.log = Mock()

    with patch.dict(os.environ, {"SLING_GOOGLE_APPLICATION_CREDENTIALS": creds_json}):
        with patch("macro_agents.defs.replication.sling.SlingConnectionResource") as mock_conn:
            resource = SlingResourceWithCredentials()
            resource.setup_for_execution(mock_context)
            assert hasattr(resource, "_sling_resource")

Skip Tests When Environment Not Available

Use pytest.mark.skipif for optional integration tests:

@pytest.mark.skipif(
    not os.getenv("MOTHERDUCK_TOKEN"),
    reason="MOTHERDUCK_TOKEN not set"
)
def test_motherduck_connection_parameters(self):
    token = os.getenv("MOTHERDUCK_TOKEN")
    assert len(token) > 0

---

2. dbt Testing

Test Types

1. Schema Tests - Defined in schema.yml files (not_null, unique, accepted_values)
2. Data Tests - Custom SQL in tests/ folder
3. dbt Project Validation - Python tests that run dbt commands

Schema Tests in schema.yml

version: 2

models:
  - name: major_indicies_summary
    description: "Performance analysis of major stock market indices"
    columns:
      - name: symbol
        tests:
          - not_null
          - unique
      - name: time_period
        tests:
          - not_null
          - accepted_values:
              values: ['12_weeks', '6_months', '1_year', '5_years']
      - name: total_return_pct
        tests:
          - not_null

Custom Data Tests

Create SQL files in dbt_project/tests/ that return failing rows:

-- tests/test_forward_returns_not_zero.sql
{{ config(severity='warn') }}

{% set models_to_test = [
    'currency_analysis_return',
    'major_indicies_analysis_return',
] %}

{% for model in models_to_test %}
    SELECT
        symbol,
        month_date,
        pct_change_q1_forward,
        'zero_forward_return' as issue_type
    FROM {{ ref(model) }}
    WHERE ABS(pct_change_q1_forward) < 0.01
        AND pct_change_q1_forward IS NOT NULL

    {% if not loop.last %}
    UNION ALL
    {% endif %}
{% endfor %}

Data Completeness Tests

-- tests/test_weekly_data_completeness.sql
WITH weekly_counts AS (
    SELECT
        DATE_TRUNC('week', date) AS week_start,
        COUNT(*) AS record_count
    FROM {{ ref('stg_us_sectors') }}
    WHERE date >= CURRENT_DATE - INTERVAL '12 weeks'
    GROUP BY DATE_TRUNC('week', date)
)
SELECT week_start
FROM weekly_counts
WHERE record_count < 10

Python Tests for dbt Validation

Test dbt project from Python (runs in Dagster test suite):

import subprocess
from pathlib import Path

class TestDbtProjectValidation:
    @pytest.fixture
    def dbt_project_dir(self):
        return Path(__file__).parent.parent.parent / "dbt_project"

    def test_dbt_parse_succeeds(self, dbt_project_dir):
        original_cwd = os.getcwd()
        try:
            os.chdir(dbt_project_dir)
            result = subprocess.run(
                ["dbt", "parse", "--target", "local"],
                capture_output=True,
                text=True,
                timeout=60,
            )
            if result.returncode != 0:
                pytest.fail(f"dbt parse failed:\n{result.stderr}")
        finally:
            os.chdir(original_cwd)

    def test_dbt_list_models_succeeds(self, dbt_project_dir):
        original_cwd = os.getcwd()
        try:
            os.chdir(dbt_project_dir)
            result = subprocess.run(
                ["dbt", "list", "--target", "local", "--resource-type", "model"],
                capture_output=True,
                text=True,
            )
            models = [l.strip() for l in result.stdout.split("\n") if l.strip()]
            assert len(models) > 0, "No models found"
        finally:
            os.chdir(original_cwd)

---

3. FastAPI Backend Testing

Test Setup (conftest.py)

import pytest
from fastapi.testclient import TestClient
from api.main import app

@pytest.fixture
def client():
    return TestClient(app)

@pytest.fixture
def temp_duckdb_file():
    with tempfile.NamedTemporaryFile(suffix=".duckdb", delete=False) as tmp_file:
        yield tmp_file.name
        if os.path.exists(tmp_file.name):
            os.unlink(tmp_file.name)

@pytest.fixture
def mock_motherduck_resource(temp_duckdb_file):
    return MotherDuckResource(
        md_token="test_token",
        environment="dev",
        local_path=temp_duckdb_file,
    )

API Endpoint Tests

from fastapi.testclient import TestClient
from api.main import app

client = TestClient(app)

def test_get_market_summaries():
    response = client.get("/api/markets/summary")
    assert response.status_code in [200, 500]  # 500 if no DB
    if response.status_code == 200:
        data = response.json()
        assert "results" in data
        assert "count" in data
        assert isinstance(data["results"], list)

def test_get_market_summaries_with_category():
    response = client.get("/api/markets/summary?category=sector")
    assert response.status_code in [200, 500]

def test_get_market_summaries_invalid_limit():
    response = client.get("/api/markets/summary?limit=0")
    assert response.status_code == 400

    response = client.get("/api/markets/summary?limit=1001")
    assert response.status_code == 400

Testing Validation Errors

def test_invalid_limit_returns_400():
    response = client.get("/api/markets/summary?limit=0")
    assert response.status_code == 400
    assert "detail" in response.json()

Service Layer Tests

def test_data_service_initialization():
    service = DataService()
    assert service.motherduck is not None

def test_get_market_summaries_empty():
    service = DataService()
    results = service.get_market_summaries()
    assert isinstance(results, list)

---

4. React Frontend Testing

Test Setup (tests/setup.ts)

import { expect, afterEach } from 'vitest';
import { cleanup } from '@testing-library/react';
import * as matchers from '@testing-library/jest-dom/matchers';

expect.extend(matchers);

afterEach(() => {
  cleanup();
});

Component Tests with Providers

Wrap components with required providers:

import { describe, it, expect, vi } from 'vitest';
import { render, screen } from '@testing-library/react';
import { QueryClient, QueryClientProvider } from '@tanstack/react-query';
import { ThemeProvider } from '../theme/ThemeContext';

function renderWithProviders(component: React.ReactElement) {
  const queryClient = new QueryClient({
    defaultOptions: {
      queries: { retry: false },
    },
  });

  return render(
    <QueryClientProvider client={queryClient}>
      <ThemeProvider>{component}</ThemeProvider>
    </QueryClientProvider>
  );
}

Mocking API Hooks

import * as apiHooks from '../hooks/useApi';

vi.mock('../hooks/useApi');

describe('Dashboard', () => {
  it('should render loading state', () => {
    vi.mocked(apiHooks.useLatestEconomyState).mockReturnValue({
      data: undefined,
      isLoading: true,
      error: null,
    } as ReturnType<typeof apiHooks.useLatestEconomyState>);

    renderWithProviders(<Dashboard />);
    expect(screen.getAllByText('Loading...').length).toBeGreaterThan(0);
  });

  it('should render dashboard with data', async () => {
    const mockData = {
      analysis_date: '2024-01-01',
      model_name: 'test-model',
      analysis_content: 'Test content',
    };

    vi.mocked(apiHooks.useLatestEconomyState).mockReturnValue({
      data: mockData,
      isLoading: false,
      error: null,
    } as ReturnType<typeof apiHooks.useLatestEconomyState>);

    renderWithProviders(<Dashboard />);
    await waitFor(() => {
      expect(screen.getByText('Latest Economy State')).toBeInTheDocument();
    });
  });
});

Testing User Interactions

import { fireEvent } from '@testing-library/react';

it('should call onRangeChange when a button is clicked', () => {
  const onRangeChange = vi.fn();
  render(<TimeRangeSelector selectedRange="1Y" onRangeChange={onRangeChange} />);

  fireEvent.click(screen.getByText('2Y'));
  expect(onRangeChange).toHaveBeenCalledWith('2Y');
});

Testing with Theme Provider

import { ThemeProvider } from '../../theme/ThemeContext';

function renderWithTheme(component: React.ReactElement) {
  return render(<ThemeProvider>{component}</ThemeProvider>);
}

describe('TimeSeriesChart', () => {
  const mockData = [
    { date: '2024-01-01', value: 100 },
    { date: '2024-01-02', value: 105 },
  ];

  it('should render chart with data', () => {
    renderWithTheme(<TimeSeriesChart data={mockData} title="Test Chart" />);
    expect(screen.getByText('Test Chart')).toBeInTheDocument();
  });

  it('should render chart without title', () => {
    const { container } = renderWithTheme(<TimeSeriesChart data={mockData} />);
    expect(container.querySelector('h3')).not.toBeInTheDocument();
  });
});

Testing Async Effects and Cleanup

Test components with useEffect cleanup to ensure no memory leaks:

import { render, screen, waitFor, act } from '@testing-library/react';

describe('DataLoader (async useEffect)', () => {
  beforeEach(() => {
    vi.useFakeTimers();
  });

  afterEach(() => {
    vi.useRealTimers();
  });

  it('should not update state after unmount', async () => {
    const consoleSpy = vi.spyOn(console, 'error');

    const mockFetch = vi.fn().mockImplementation(
      () => new Promise(resolve => setTimeout(() => resolve({ data: 'test' }), 1000))
    );
    vi.mocked(api.fetchData).mockImplementation(mockFetch);

    const { unmount } = renderWithProviders(<DataLoader />);

    // Unmount before the fetch resolves
    unmount();

    // Fast-forward past the timeout
    await act(async () => {
      vi.advanceTimersByTime(1500);
    });

    // Should not have any "Can't perform state update on unmounted component" warnings
    expect(consoleSpy).not.toHaveBeenCalled();
  });

  it('should cancel previous request when dependencies change', async () => {
    const abortSpy = vi.fn();
    global.AbortController = vi.fn().mockImplementation(() => ({
      signal: {},
      abort: abortSpy,
    }));

    const { rerender } = renderWithProviders(<DataLoader id="1" />);

    // Trigger a re-render with new props before first fetch completes
    rerender(
      <QueryClientProvider client={queryClient}>
        <DataLoader id="2" />
      </QueryClientProvider>
    );

    // First request should have been aborted
    expect(abortSpy).toHaveBeenCalled();
  });

  it('should handle loading, success, and error states', async () => {
    vi.mocked(api.fetchData).mockResolvedValueOnce({ items: [] });

    renderWithProviders(<DataLoader />);

    // Loading state
    expect(screen.getByTestId('loading-spinner')).toBeInTheDocument();

    await waitFor(() => {
      // Success state
      expect(screen.queryByTestId('loading-spinner')).not.toBeInTheDocument();
      expect(screen.getByText('No items found')).toBeInTheDocument();
    });
  });
});

Testing Custom Hooks with Async Operations

import { renderHook, waitFor, act } from '@testing-library/react';

describe('useAsyncData hook', () => {
  it('should return data after fetch completes', async () => {
    vi.mocked(api.getData).mockResolvedValueOnce({ result: 'test' });

    const { result } = renderHook(() => useAsyncData('test-id'));

    expect(result.current.isLoading).toBe(true);

    await waitFor(() => {
      expect(result.current.isLoading).toBe(false);
      expect(result.current.data).toEqual({ result: 'test' });
    });
  });

  it('should handle errors gracefully', async () => {
    vi.mocked(api.getData).mockRejectedValueOnce(new Error('Network error'));

    const { result } = renderHook(() => useAsyncData('test-id'));

    await waitFor(() => {
      expect(result.current.isLoading).toBe(false);
      expect(result.current.error).toBe('Network error');
    });
  });

  it('should cleanup on unmount', async () => {
    const mockAbort = vi.fn();
    global.AbortController = vi.fn().mockImplementation(() => ({
      signal: { aborted: false },
      abort: mockAbort,
    }));

    const { unmount } = renderHook(() => useAsyncData('test-id'));

    unmount();

    expect(mockAbort).toHaveBeenCalled();
  });
});

---

5. Running Tests

All Tests (via Makefile)

make test

Dagster Tests

cd macro_agents
uv run pytest tests/ -v

# Run specific test file
uv run pytest tests/test_resources.py -v

# Run specific test class
uv run pytest tests/test_resources.py::TestMotherDuckResource -v

# Run with coverage
uv run pytest tests/ --cov=macro_agents --cov-report=html

dbt Tests

cd dbt_project

# Run all tests
dbt test

# Run tests for specific model
dbt test --select major_indicies_summary

# Run data tests only
dbt test --select test_type:data

# Run schema tests only
dbt test --select test_type:schema

API Tests

cd api
PYTHONPATH=.. uv run pytest tests/ -v

# Run specific test file
uv run pytest tests/test_markets.py -v

Frontend Tests

cd frontend

# Run tests (watch mode)
npm test

# Run tests once
npm test -- --run

# Run with UI
npm run test:ui

---

6. Best Practices

General

1. Test one thing per test - Each test should verify a single behavior
2. Use descriptive test names - Name should explain what is being tested
3. Arrange-Act-Assert - Structure tests clearly
4. Clean up resources - Use fixtures with cleanup for temp files/databases

Dagster

1. Mock external services - Never call real APIs in unit tests
2. Test resource initialization - Verify both dev and prod configurations
3. Test definitions load - Catch import/config errors early
4. Use pytest.mark.skipif - Skip integration tests when credentials unavailable

dbt

1. Add schema tests for every model - At minimum: not_null on key columns
2. Use custom tests for business logic - Data completeness, value ranges
3. Set severity levels - Use severity='warn' for non-blocking issues
4. Test incrementally - Run dbt test --select model_name during development

FastAPI

1. Test both success and error cases - Include validation errors
2. Use TestClient - Simulates HTTP requests without running server
3. Accept multiple status codes - [200, 500] for endpoints that depend on DB state
4. Test with fixtures - Use conftest.py for shared test setup

React

1. Mock API hooks - Don't make real API calls in tests
2. Use providers - Wrap with QueryClientProvider, ThemeProvider
3. Test loading/error states - Cover all UI states
4. Prefer user-centric queries - Use getByText, getByRole over querySelector