Playwright API Framework

A production-grade API test automation framework built with Python, pytest, and Playwright.

Tech Stack

• Python 3.11
• pytest — test runner
• requests — HTTP client
• Pydantic — response schema validation
• Faker — dynamic test data generation
• Allure — test reporting
• GitHub Actions — CI/CD

Architecture

playwright-api-framework/
├── core/
│   ├── client/         # HTTP client (requests wrapper with logging + error handling)
│   ├── config/         # Environment-based configuration
│   ├── data/           # Static test data constants
│   ├── factories/      # Dynamic test data generation (Faker)
│   ├── models/         # Pydantic response models for schema validation
│   ├── services/       # API service layer (one class per resource)
│   └── validators/     # Reusable assertion helpers
├── fixtures/           # pytest fixtures
├── tests/
│   └── api/            # API test suites
└── .github/workflows/  # CI/CD pipeline

Setup

Prerequisites: Python 3.11+, pip

git clone <repo-url>
cd playwright-api-framework
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Create a .env.dev file in the project root:

BASE_URL_API=https://reqres.in/api
BASE_URL_UI=https://reqres.in
API_KEY=your_api_key

Running Tests

Run all API tests (default: dev environment):

pytest -m api -v

Run against a specific environment:

ENV=staging pytest -m api -v
ENV=prod pytest -m api -v

Run with Allure reporting:

pytest -m api -v --alluredir=allure-results
allure serve allure-results

CI/CD

Tests run automatically on every push and pull request to main via GitHub Actions. The pipeline:

1. Spins up a clean Ubuntu environment
2. Installs all dependencies
3. Runs the API test suite against staging
4. Uploads Allure results as a downloadable artifact

Key Design Decisions

• Layered architecture — client, service, validator, and factory layers are separate concerns, making the framework easy to extend without modifying existing code
• Dynamic test data — Faker generates unique data per run, avoiding test pollution and data collisions
• Schema validation — Pydantic models validate full response structure (field names + types), not just individual keys
• Multi-environment support — switching environments requires no code changes, only an ENV variable
• Error handling — network failures (timeout, connection error) are caught and logged with context before re-raising