Python Playwright AI

UI test automation framework built with Python 3.14+, Playwright, Pytest, and Allure.

Targets DemoQA using the Page Object Model pattern with strongly-typed configuration and modern tooling.

Features

• Page Object Model -- all UI interactions abstracted into page objects with method chaining (Self return types)
• Allure Reporting -- epics, features, stories, severity levels, step-level tracing, failure screenshots, environment metadata
• Pydantic Settings -- typed configuration with validation, env var overrides (PW_*), .env file support
• Parallel Execution -- pytest-xdist with configurable thread count (PW_THREADS)
• Pytest Markers -- smoke for critical path, regression for full suite, parametrize for data-driven tests
• Faker -- dynamic test data generation via session-scoped fixture
• Playwright -- auto-waiting, tracing, screenshots, and video on failure
• Code Quality -- Ruff (linting + import sorting), Black (formatting), Mypy (type checking)

Project Structure

tests/
  conftest.py                         # Fixtures, Playwright config, Allure hooks
  framework/
    base_test.py                      # BaseTest class (injects page fixture)
    config/
      settings.py                     # PlaywrightConfig (pydantic-settings)
    pages/
      base_page.py                    # BasePage with navigate() and @allure.step
      alerts_frames_windows/
        alerts_page.py                # AlertsPage
        modal_dialogs_page.py         # ModalDialogsPage
      book_store/
        book_store_page.py            # BookStorePage
        login_page.py                 # BookStoreLoginPage
      elements/
        check_box_page.py             # CheckBoxPage
        text_box_page.py              # TextBoxPage
      forms/
        practice_form_page.py         # PracticeFormPage
      interactions/
        droppable_page.py             # DroppablePage
        selectable_page.py            # SelectablePage
      widgets/
        date_picker_page.py           # DatePickerPage
        slider_page.py                # SliderPage
  tests/
    test_alerts_frames_windows.py     # Alerts and modal dialog tests
    test_book_store.py                # Login and book search tests
    test_elements.py                  # Text box and checkbox tests
    test_forms.py                     # Practice form tests (parametrized)
    test_interactions.py              # Drag-and-drop and selectable tests
    test_widgets.py                   # Date picker and slider tests

Setup

Requires uv as the package manager.

# Install uv (macOS/Linux)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install dependencies
uv sync

# Install Playwright browsers
uv run playwright install chromium

Configuration

Managed via PlaywrightConfig in tests/framework/config/settings.py using pydantic-settings.

Priority order: OS env variables (PW_*) > .env file > pyproject.toml > defaults.

Variable	Default	Description
PW_BROWSER	chromium	Browser engine (chromium, firefox, webkit)
PW_HEADLESS	true	Run in headless mode
PW_BASE_URL	https://demoqa.com	Base URL for all tests
PW_TIMEOUT	30000	Default timeout in milliseconds
PW_VIEWPORT_WIDTH	1280	Browser viewport width (320-3840)
PW_VIEWPORT_HEIGHT	720	Browser viewport height (240-2160)
PW_SLOWMO	0	Slow down actions by ms
PW_TRACING	retain-on-failure	Playwright tracing mode
PW_SCREENSHOT	only-on-failure	Screenshot capture mode
PW_VIDEO	retain-on-failure	Video recording mode
PW_OUTPUT_DIR	test-results	Output directory for artifacts
PW_FULLSCREEN	false	Run in fullscreen mode
PW_THREADS	3	Number of parallel xdist workers (0 to disable)

Override via environment:

PW_HEADLESS=false PW_BROWSER=firefox uv run pytest

Running Tests

# Run all tests (3 parallel workers by default)
uv run pytest

# Run smoke tests only
uv run pytest -m smoke

# Run regression suite
uv run pytest -m regression

# Run specific test file
uv run pytest tests/tests/test_elements.py

# Run by keyword
uv run pytest -k "test_search"

# Run headed (visible browser) with 3 parallel workers
PW_HEADLESS=false uv run pytest -n=3

# Run with custom number of parallel workers
PW_THREADS=5 uv run pytest

# Run sequentially (disable parallel execution)
uv run pytest -n=0

Allure Reports

Allure results are generated automatically in allure-results/ on every test run.

View the report:

allure serve allure-results

The report includes:

• Epic/Feature/Story hierarchy matching the DemoQA application structure
• Severity levels (CRITICAL, NORMAL) on each test
• Step-level tracing from page object @allure.step() decorators
• Failure screenshots attached automatically
• Environment metadata (browser, base URL, Python version, viewport)
• Defect categories (element interaction failures, assertion failures, infrastructure issues)

CI/CD

GitHub Actions workflow (.github/workflows/ci.yml) runs on every push and PR to main:

1. Lint & Type Check -- Ruff, Black, Mypy
2. Tests -- parallel matrix across Chromium and Firefox
3. Allure Report -- generated and deployed to GitHub Pages (main branch only)

Test artifacts (screenshots, traces, videos) are uploaded on failure. Allure results are uploaded for all runs.

Claude Code Integration

claude.yml implements a two-phase AI-assisted workflow via Claude Code Action:

Phase 1 — Analyze (read-only)

Triggered automatically on PRs, or when a PR review or issue body contains @claude. Claude reads the diff or issue, identifies problems and improvements, and posts a summary comment ending with:

"Reply @claude apply to approve and apply these changes."

Phase 2 — Apply

Triggered when @claude appears in an issue comment or PR review comment. Claude applies the proposed changes directly to the branch.

Both phases use claude-sonnet-4-6 with up to 25 turns.

Code Quality

# Lint
uv run ruff check tests/

# Format
uv run black tests/

# Type check
uv run mypy tests/