UI test automation for Totebox, a small Kotlin shopping app, using Python, pytest, Appium (UiAutomator2) and the Page Object pattern. It is the Appium counterpart to the Espresso suite that lives with the app's source, and covers the same flows: 50 tests, all passing.
pytest ---> Appium server (127.0.0.1:4723) ---> Android emulator/device ---> Totebox
apps/totebox-debug.apk the app under test (debug build, bundled so a clone runs as-is)
config/capabilities.py Appium session options, env-overridable
pages/ page objects: base_page, toolbar, login, products, product_detail,
cart (cart + saved for later tabs), checkout, order_complete, settings
tests/ conftest.py (driver, restart, failure artifacts), data.py, test_*.py
artifacts/ screenshot + UI dump + logcat of each failed test (git-ignored)
- Android SDK with an emulator (API 26+) or a device,
adb devicesshowing it. - Appium 2/3 with the driver:
appium driver install uiautomator2. - Python environment:
python3 -m venv .venv && source .venv/bin/activate pip install -r requirements.txt
appium & # in another terminal
pytest # everything (50 tests)
pytest -m smoke # critical path only
pytest -m "regression and not crash"
pytest tests/test_cart.py -k shippingEnvironment overrides: APPIUM_SERVER_URL, ANDROID_DEVICE_NAME, ANDROID_UDID,
ANDROID_PLATFORM_VERSION, ANDROID_APP_PATH (use another build of the app).
The APK is bundled, so nothing else is needed. To test a newer build, get the app source from
totebox-espresso-test, run
./gradlew assembleDebug, and point ANDROID_APP_PATH at app/build/outputs/apk/debug/app-debug.apk
(or copy it to apps/totebox-debug.apk).
GitHub Actions starts an API 34 emulator, installs Appium 3.9.0 with the UiAutomator2 driver, and runs
the smoke tests on every push and pull request. Run the workflow manually to pick another marker
or to run all 50 tests. The raw Allure results are uploaded as a build artifact.
allure-pytest writes raw results to allure-results/; view them with the Allure CLI
(allure serve allure-results). Failed tests attach a screenshot, the UI hierarchy and the
device log; the same files are saved in artifacts/.
- One session, fresh app per test. Starting an Appium session is the slow part, so the
session is shared. Totebox keeps all state in memory, so the
driverfixture kills the app and starts the login screen in a new task before every test: each test starts logged out with an empty cart and is independent of order. - Locators are resource IDs (
rid("login_button")) and accessibility labels. Product-row buttons are found by the labels the app gives them, e.g.Add Canvas Tote Bag to cart. Material field errors are located as children of each field's layout. - No sleeps. Everything uses explicit waits. The product list waits out the app's simulated network delay; toasts and snackbars are read while they are visible.
- Swipe to place order is a
mobile: dragGesturestarting on the slider's handle; a half swipe is a test that the order is not placed.
| File | Covers |
|---|---|
test_login.py |
login + toast, guest, helper text, required fields, email format, password rules, locked account |
test_products.py |
list loads, three sort orders, add to cart + badge, save for later, crash item |
test_product_detail.py |
quantity stepper limits, add to cart snackbar + "View cart", 10-item limit |
test_cart.py |
totals, shipping fee and free-shipping threshold, remove, empty state, start shopping |
test_saved_for_later.py |
save, move back, delete, product returns to the list |
test_checkout.py |
card and pay-on-delivery orders, half swipe, required fields, ZIP and card rules |
test_guest_checkout.py |
guest must log in, cart is kept, back returns to the cart |
test_settings.py |
account line, guest mode, network speed & load time, Home menu, guest Log in, log out |
