eBay E2E Test Suite

End-to-end tests against ebay.com — search, pagination, add-to-cart, and cart-total verification.

automation-exercise.md · Exercise 5: ReadMeAIBugs.md

⚡ Technologies

Philosophy

Thin tests, thick workflows. Test files assert outcomes; multi-step flows live in workflows/ and page interactions in pages/. Each layer has one job.

Data-driven by default. Search, cart, and pagination cases load from JSON. Add a row to tests/data/*.json — no code changes required.

Guest first, login when needed. Most E2E runs use a guest session with Ship-to-US cookies for USD pricing. Login tests and --with-login replay a saved session — programmatic sign-in is unreliable against eBay's CAPTCHA and bot detection.

Small enough to navigate. One browser per worker, isolated contexts per test, environment config in a single JSON file per profile. Reports, traces, and screenshots attach to Allure on failure.

Requirements

Tool	Install
uv	`curl -LsSf https://astral.sh/uv/install.sh \| sh`
Python 3.13+	managed by uv
Allure CLI	`brew install allure`
eBay account	login tests and `--with-login` only

uv sync                                          # creates .venv
source .venv/bin/activate                        # optional — uv run works without it
uv run python -m scripts.healthcheck             # optional preflight

Quick start

cd /path/to/Ebay
uv sync
uv run playwright install chromium
uv run pytest tests/test_search_and_cart.py
allure serve reports/allure-results

Login tests and --with-login E2E need a one-time manual bootstrap:

uv run python -m scripts.bootstrap_ebay_auth   # saves secured_env_files/ebay-auth.json
uv run pytest tests/test_login.py

What it supports

Search & cart E2E — price-filtered search, add listings with variant/quantity handling, verify cart subtotal and budget cap (TC001–TC003, data-driven)
Pagination — verify page 2 returns mostly new listings (PG001)
Session replay — login validation via saved Playwright storage state
Parallel runs — pytest-xdist, one worker per DDT case
Environment profiles — staging / preprod / prod config folders (retries, traces, video, slow_mo)
Allure reporting — steps, parameters, screenshots, traces; live [search] / [cart] / [verify] logs in terminal
CI-ready — guest runs need no secrets; login runs restore ebay-auth.json from a GitHub Actions secret

Running tests

Default pytest options (in pyproject.toml): -s, -v, Allure to reports/allure-results, live INFO logging.

Test catalog

File	Markers	Cases	Verifies
`test_search_and_cart.py`	`search`, `cart`, `regression`	TC001–TC003	Search → filter → cart → subtotal
`test_search_pagination.py`	`search`, `pagination`	PG001	Page 2 mostly new listings
`test_login.py`	`login`	1	Saved session replays
`test_bootstrap_auth.py`	`bootstrap`	1	Manual sign-in (excluded from default runs)

Search & cart

Data from tests/data/test_data.json.

Case	Query	Price	Listings	Qty
TC001	wireless earbuds	$15–$60	3	1
TC002	phone case	$10–$40	2	2
TC003	vintage watch	$5–$50	4	1

uv run pytest tests/test_search_and_cart.py
uv run pytest tests/test_search_and_cart.py -n 3              # parallel
uv run pytest tests/test_search_and_cart.py -k TC001
HEADLESS=false uv run pytest tests/test_search_and_cart.py -k TC001
uv run pytest tests/test_search_and_cart.py --with-login

Pagination

uv run pytest tests/test_search_pagination.py
uv run pytest -m pagination

Login & bootstrap

Login tests auto-skip when secured_env_files/ebay-auth.json is missing.

uv run python -m scripts.bootstrap_ebay_auth
uv run python -m scripts.bootstrap_ebay_auth --force
HEADLESS=false uv run pytest -m bootstrap -s
uv run pytest tests/test_login.py

Common patterns

uv run pytest tests/ -m "not login"                # CI-style (no auth)
uv run pytest tests/                               # full suite
TEST_ENV=staging uv run pytest tests/ -m "not login"
HEADLESS=false uv run pytest tests/test_search_and_cart.py

Markers: search · cart · pagination · login · bootstrap · regression · smoke

Environment profiles (TEST_ENV, default preprod):

Profile	`retry_count`	`trace_on_failure`	`record_video`	`slow_mo`
`staging`	3	yes	yes	0
`preprod`	1	no	no	0
`prod`	1	yes	no	50 ms

Shell overrides: HEADLESS, SLOW_MO, WITH_LOGIN (same as --with-login), DEFAULT_TIMEOUT (max 15 s).

Architecture

JSON data → tests/test_*.py → workflows/ → pages/ → ebay.com
                              ↘ utilities/ (config, auth, shipping, logging)

flowchart LR
  J[tests/data/*.json] --> T[test_*.py]
  T --> W[workflows]
  W --> P[pages]
  W --> U[utilities]
  P --> L[locators/]

Fixture lifecycle (root conftest.py):

Scope	What
session	Playwright, Chromium, Ship-to-US storage state
function	Isolated context + page; failure screenshots/traces → Allure

Guest tests reuse secured_env_files/shipping-us.json (ZIP 95621) for USD pricing. Login tests and --with-login load ebay-auth.json instead. Auto-retry comes from retry_count in config via pytest-rerunfailures.

Key files

Path	Role
`conftest.py`	Browser fixtures, `--with-login`, Allure hooks, auto-retry
`tests/conftest.py`	DDT parametrization from JSON, login guard for search/cart
`workflows/search_items.py`	Search, price filter, collect URLs, CSV log
`workflows/add_items_to_cart.py`	Variants, quantity, add-to-cart
`workflows/assert_cart_total.py`	Subtotal verification and budget cap
`workflows/paginate_search.py`	Multi-page listing ID collection
`pages/`	Page Object Model — one class per eBay screen
`pages/locators/`	CSS selectors split by page
`utilities/config_loader.py`	Load `resources/<env>/config.json`
`utilities/auth_storage.py`	Save/load `ebay-auth.json`
`utilities/shipping_storage.py`	Ship-to-US session bootstrap
`scripts/bootstrap_ebay_auth.py`	Headed manual sign-in
`scripts/healthcheck.py`	Preflight before bootstrap
`resources/<env>/config.json`	Per-env browser and reporting settings
`pyproject.toml`	Dependencies, pytest config, markers

Output artifacts (git-ignored)

Path	Contents
`reports/allure-results/`	Allure raw results
`reports/traces/`	Playwright traces — trace.playwright.dev
`reports/videos/`	WebM when `record_video: true`
`downloads/search_results.csv`	Search URL log
`screenshots/`	PNG captures
`secured_env_files/`	`ebay-auth.json`, `shipping-us.json` (never commit)

Running in CI

Guest tests need no credentials. Login tests and --with-login require secured_env_files/ebay-auth.json restored from a GitHub Actions secret.

Store `ebay-auth.json` as a secured secret

Bootstrap locally: uv run python -m scripts.bootstrap_ebay_auth
Encode: base64 -i secured_env_files/ebay-auth.json | pbcopy
GitHub → Settings → Secrets and variables → Actions → secret EBAY_AUTH_JSON_B64
Refresh when the session expires (sign-out, password change, replay failures)

Use an environment secret for stricter access. Never commit the session file.

Workflow

name: eBay E2E

on:
  push:
    branches: [main]
  pull_request:

jobs:
  e2e:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: astral-sh/setup-uv@v4
        with:
          python-version: "3.13"
      - run: uv sync
      - run: uv run playwright install --with-deps chromium

      - name: Restore eBay auth session
        if: ${{ secrets.EBAY_AUTH_JSON_B64 != '' }}
        run: |
          mkdir -p secured_env_files
          echo "${{ secrets.EBAY_AUTH_JSON_B64 }}" | base64 -d > secured_env_files/ebay-auth.json
          test -s secured_env_files/ebay-auth.json

      - name: Run tests
        run: |
          if [ -f secured_env_files/ebay-auth.json ]; then
            uv run pytest tests/
          else
            uv run pytest tests/ -m "not login"
          fi
        env:
          TEST_ENV: preprod
          HEADLESS: "true"

      - uses: actions/upload-artifact@v4
        if: always()
        with:
          name: allure-results
          path: reports/allure-results/

For authenticated search/cart only (skip test_login.py):

- run: uv run pytest tests/test_search_and_cart.py --with-login
  env:
    TEST_ENV: preprod
    HEADLESS: "true"

FAQ

Why session replay instead of programmatic login? eBay sign-in includes CAPTCHA, 2FA, and bot detection. A headed bootstrap once, then replay in CI, is the reliable approach.

Why guest by default? Search and cart flows work without auth. Guest runs need no secrets and are faster to set up in CI.

How do I add a new test case? Edit tests/data/test_data.json or pagination_data.json. Pytest picks it up via parametrization in tests/conftest.py.

How do I debug a failure? Run headed: HEADLESS=false uv run pytest … -k TC001. Check Allure attachments (screenshots, traces). Open trace zips at trace.playwright.dev. Enable trace_on_failure: true in your env config for richer artifacts.

Why first variant, not random? Random size/color picks often change price or select premium options, breaking price-filter and budget assertions. Random selection can be accounted for — e.g. re-read the listing price after each variant pick and skip or adjust when it falls outside the filter range.

Why Ship-to-US cookies? eBay shows different prices by shipping destination. Session-scoped ZIP 95621 cookies keep listings in USD across tests. Multi-currency support and conversion (parse locale-specific prices, normalize to a target currency before assertions) can be implemented if needed.

How does --with-login work? Root conftest.py registers the flag and loads ebay-auth.json into browser storage_state. tests/conftest.py calls ensure_logged_in() before search/cart tests and fails fast if the session is invalid.

Limitations

Topic	Note
Selectors	CSS and ARIA preferred; XPath in fallback paths only
Currency	USD only; foreign-currency listings skipped
Env profiles	Same URL (`ebay.com`); differ in retries, traces, video, slow_mo
Cart budget	`subtotal ≤ budget_per_item × total_units`; optional `min_total` / `max_total` in JSON
Quantity	Set in JSON; product page tries select/input, falls back to repeated add-to-cart clicks

Dependency management

uv add <package>
uv lock --upgrade && uv sync
uv pip list

Dependencies are in pyproject.toml, pinned at major version.

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
.claude		.claude
.github/workflows		.github/workflows
docs		docs
pages		pages
resources		resources
scripts		scripts
secured_env_files		secured_env_files
tests		tests
utilities		utilities
workflows		workflows
.aiexclude		.aiexclude
.aiignore		.aiignore
.claudeignore		.claudeignore
.cursorignore		.cursorignore
.flake8		.flake8
.geminiignore		.geminiignore
.gitignore		.gitignore
.llmignore		.llmignore
.python-version		.python-version
README.md		README.md
conftest.py		conftest.py
pyproject.toml		pyproject.toml
requirements.txt		requirements.txt
uv.lock		uv.lock

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

eBay E2E Test Suite

⚡ Technologies

Philosophy

Requirements

Quick start

What it supports

Running tests

Test catalog

Search & cart

Pagination

Login & bootstrap

Common patterns

Architecture

Key files

Output artifacts (git-ignored)

Running in CI

Store `ebay-auth.json` as a secured secret

Workflow

FAQ

Limitations

Dependency management

About

Uh oh!

Releases

Packages

Uh oh!

Contributors

Uh oh!

Languages

Folders and files

Latest commit

History

Repository files navigation

eBay E2E Test Suite

⚡ Technologies

Philosophy

Requirements

Quick start

What it supports

Running tests

Test catalog

Search & cart

Pagination

Login & bootstrap

Common patterns

Architecture

Key files

Output artifacts (git-ignored)

Running in CI

Store ebay-auth.json as a secured secret

Workflow

FAQ

Limitations

Dependency management

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Store `ebay-auth.json` as a secured secret

Packages