AZANIR/qa-data-factory

QA Data Factory

Purpose

Generate realistic, consistent, and reproducible test data for automated and manual testing. Supports fixtures (static JSON/YAML), factory patterns (dynamic generation with overrides), faker-based seeds (realistic random data), and builder patterns (fluent API for complex objects). Produces fixtures, factory classes, seed scripts, and data cleanup utilities.

Trigger Phrases

• "Generate test data for [entity type]"
• "Create fixtures for [Users/Products/Orders]"
• "Factory pattern for test data"
• "Faker-based seed data"
• "Test data builder for [complex object]"
• "Realistic test data with locale [en_US/de_DE]"
• "Seed script for test database"
• "Data cleanup utilities for tests"

Data Generation Approaches

| Approach | Use Case | Output | |----------|----------|--------| | Fixtures | Static, version-controlled data; predictable | JSON, YAML, CSV files | | Factory pattern | Dynamic generation with overrides; per-test customization | Factory classes (TypeScript/Python) | | Faker-based seeds | Realistic but random; locale-aware | Seed scripts, inline generation | | Builder pattern | Complex nested objects; fluent API | Builder classes |

See references/factory-patterns.md for implementation details.

TypeScript Tools

| Tool | Purpose | |------|---------| | @faker-js/faker | Realistic random data (names, emails, addresses, dates) | | fishery | Factory pattern with sequences and traits | | factory.ts | TypeScript factories with associations | | Test data builders | Fluent API for complex objects |

Python Tools

| Tool | Purpose | |------|---------| | Faker | Realistic random data; locale support | | factory_boy | Factory pattern with sequences, subfactories | | pytest-factoryboy | Pytest integration; fixtures from factories | | model_bakery | Django model factories; minimal boilerplate |

See references/faker-guide.md for usage patterns.

Data Types Supported

| Type | Examples | Locale-aware | |------|----------|--------------| | Users | name, email, username, password hash | Yes (names, formats) | | Products | SKU, title, price, category | Yes (currency, formats) | | Orders | order ID, items, totals, status | Yes (dates, currency) | | Addresses | street, city, postal code, country | Yes | | Payment info | card numbers (test), expiry, CVV | Test card patterns | | Dates | birthdate, created_at, expiry | Yes (formats) | | Emails | valid format, domain | Yes | | Names | first, last, full | Yes (locale-specific) |

Output Artifacts

1. Test data fixtures — Static JSON/YAML/CSV for predictable scenarios
2. Factory classes — TypeScript or Python factories with overrides
3. Seed scripts — Scripts to populate test DB with faker data
4. Data cleanup utilities — Teardown scripts, truncate/reset helpers

Workflow

1. Identify data needs — Entity types, required fields, relationships
2. Choose approach — Fixtures vs factory vs faker based on use case
3. Select tools — Match project stack (TS/Python, framework)
4. Generate artifacts — Fixtures, factories, seed scripts
5. Add cleanup — Teardown for isolation between test runs

Scope

Can do (autonomous):

• Generate fixtures, factories, seed scripts for specified entities
• Use Faker/faker-js for realistic locale-aware data
• Produce TypeScript (fishery, factory.ts) and Python (factory_boy, Faker) code
• Create data cleanup utilities
• Reference factory-patterns and faker-guide

Cannot do (requires confirmation):

• Add new dependencies without approval
• Modify production schemas or migrations
• Generate data for proprietary/restricted formats

Will not do (out of scope):

• Execute seed scripts or modify databases directly
• Generate production data or PII
• Bypass data validation or security constraints

References

| Topic | File | |-------|------| | Factory pattern implementations (TS, Python) | references/factory-patterns.md | | Faker library usage for realistic data | references/faker-guide.md |

Quality Checklist

• [ ] Generated data matches entity schema/contract
• [ ] Locale specified when names/addresses/formats matter
• [ ] No real PII or production-like secrets
• [ ] Fixtures are deterministic; faker seeds use configurable seed for reproducibility
• [ ] Cleanup utilities included for test isolation
• [ ] Factory overrides documented for common scenarios

Troubleshooting

| Symptom | Likely Cause | Fix | |---------|--------------|-----| | Data conflicts between tests | Shared state; no cleanup | Add teardown; use unique IDs/emails per test | | Unrealistic data | Default faker; wrong locale | Set locale; use appropriate faker providers | | Factory too verbose | Over-specification | Use sensible defaults; override only when needed | | Seed non-reproducible | Random seed not set | Use faker.seed(123) or equivalent | | Schema mismatch | Generated fields don't match API/DB | Align with OpenAPI/DB schema; validate output |