data-seeding-fixtures-builder

Create a prisma/seed.ts script that clears existing records with deleteMany, then inserts data using factory functions. Register it in package.json as a db:seed script and run it with tsx.

Call faker.seed() with a fixed integer before generating data, such as faker.seed(12345). Every subsequent faker call produces the same values, making test data reproducible across runs.

Extend Playwright's base test with custom fixtures that create records via factory functions before the test and delete them afterward. This isolates each test with fresh, pre-seeded data.

Yes, truncate all tables except _prisma_migrations using TRUNCATE TABLE ... CASCADE, and temporarily disable foreign key checks with session_replication_role. This preserves migration history while clearing data.

Create the parent record first, then pass its id to child factories. For example, create an order with a userId, then create order items referencing both the order id and product ids.

Use minimal deterministic data for tests, small datasets for development debugging, and larger realistic volumes for staging. Branch on NODE_ENV in your main seed function to select the appropriate strategy.