← Back to projects
Case study · Web · BDD

e-bebek E2E test automation

Search, cart, and session scenarios reshaped around observed behavior on a live e-commerce site.

01 / Context and contribution

Scope of the work

A public end-to-end testing project covering sign-in, search, cart, and session flows on e-bebek.com. It is not presented as an official company project.

Scenario design, Page Object structure, and run isolation.

02 / Approach

Comparing expectations with product behavior

The repository README records that an unmatched search displays suggested products instead of an empty result. The negative scenario therefore checks whether returned product names relate to the search term rather than expecting zero results.

When an unregistered account signs in, the product routes the user to registration instead of displaying a conventional error. The related sign-in scenario was adapted to this behavior.

03 / Approach

Reading cart totals correctly

According to the README, the cart price field shows a unit price for one item and a line total after quantity increases. Multiplying that value by quantity again produced an incorrect subtotal. The Page Object and cart steps were changed to read line totals directly and compare them with the summary.

Values are read from the live page instead of using fixed prices. This aims to keep comparisons independent of price changes; it does not verify whether a commercial price is correct.

04 / Approach

Session and run isolation

The repository contains Gherkin scenarios, Page Objects, step definitions, Cucumber profiles, and Allure configuration. The README explains the choice of a separate browser context per scenario and sequential execution for scenarios that touch one shared account.

The README also notes intermittent timing issues around cart merging after sign-in and sign-out. No current, independently verified pass rate is presented here.

Technical structure

Execution context

The README states that scenarios touching the shared account run sequentially with one worker, while the remaining tests run separately in parallel. It also documents npm run report:generate for Allure reporting. Account details are expected through .env; this portfolio did not run the suite.

Sources

The path to evidence

These links lead to public repository files. The presence of code does not mean the tests currently pass.

Open GitHub repository ↗
Technical knowledge

Page Object Model with Playwright: separating scenarios from page details ↗

Verification status

What do we not know yet?

The project tests were not run as part of this portfolio work. No current, dated, shareable report was added; behavior descriptions are statements from the repository README.

Contact

Let's talk about
test strategy.

Write to me about projects or how I approach testing.

Contact details ↗