The Repository Pattern

The repository pattern

A repository makes persistence look like a collection of domain objects, hiding the storage technology:

public interface ExpenseRepository {
    Expense save(Expense e);            // insert or update
    Optional<Expense> findById(String id);
    List<Expense> findAll();
    List<Expense> findByCategory(String category);
    boolean deleteById(String id);
}

Service code depends on the interface. Production binds a JDBC implementation; tests bind an in-memory one. That is dependency inversion applied to data:

final class InMemoryExpenseRepository implements ExpenseRepository {
    private final Map<String, Expense> store = new LinkedHashMap<>();
    public Expense save(Expense e) { store.put(e.id(), e); return e; }
    public Optional<Expense> findById(String id) { return Optional.ofNullable(store.get(id)); }
    ...
}

Benefits: services unit-test with zero infrastructure; the storage swap is one constructor argument; queries are named domain operations, not SQL strings leaking everywhere.