clean-architecture（干净的架构）

Clean Architecture, Hexagonal Architecture & DDD for Spring Boot

Overview

This skill provides comprehensive guidance for implementing Clean Architecture, Hexagonal Architecture (Ports & Adapters), and Domain-Driven Design tactical patterns in Java 21+ Spring Boot 3.5+ applications. It ensures clear separation of concerns, framework-independent domain logic, and highly testable codebases through proper layering and dependency management.

When to Use

• Architecting new Spring Boot applications with clear separation of concerns
• Refactoring tightly coupled code into testable, layered architectures
• Implementing domain logic independent of frameworks and infrastructure
• Designing ports and adapters for swappable implementations
• Applying Domain-Driven Design tactical patterns (entities, value objects, aggregates)
• Creating testable business logic without Spring context dependencies

Instructions

1. Understand the Core Concepts

Clean Architecture Layers (Dependency Rule)

Dependencies flow inward. Inner layers know nothing about outer layers.

Hexagonal Architecture (Ports & Adapters)

• Domain Core: Pure Java business logic, no framework dependencies
• Ports: Interfaces defining contracts (driven and driving)
• Adapters: Concrete implementations (JPA, REST, messaging)

Domain-Driven Design Tactical Patterns

• Entities: Objects with identity and lifecycle (e.g., Order, Customer)
• Value Objects: Immutable, defined by attributes (e.g., Money, Email)
• Aggregates: Consistency boundary with root entity
• Domain Events: Capture significant business occurrences
• Repositories: Persistence abstraction, implemented in infrastructure

2. Organize Package Structure

Follow this feature-based package organization:

com.example.order/
├── domain/
│   ├── model/              # Entities, value objects
│   ├── event/              # Domain events
│   ├── repository/         # Repository interfaces (ports)
│   └── exception/          # Domain exceptions
├── application/
│   ├── port/in/            # Driving ports (use case interfaces)
│   ├── port/out/           # Driven ports (external service interfaces)
│   ├── service/            # Application services
│   └── dto/                # Request/response DTOs
├── infrastructure/
│   ├── persistence/        # JPA entities, repository adapters
│   └── external/           # External service adapters
└── adapter/
    └── rest/               # REST controllers

3. Implement the Domain Layer (Framework-Free)

The domain layer must have zero dependencies on Spring or any framework.

• Use Java records for immutable value objects with built-in validation
• Place business logic in entities, not services (Rich Domain Model)
• Define repository interfaces (ports) in the domain layer
• Use strongly-typed IDs to prevent ID confusion
• Implement domain events for decoupling side effects
• Use factory methods for entity creation to enforce invariants

4. Implement the Application Layer

• Create use case interfaces (driving ports) in application/port/in/
• Create external service interfaces (driven ports) in application/port/out/
• Implement application services with @Service and @Transactional
• Use DTOs for request/response, separate from domain models
• Publish domain events after successful operations

5. Implement the Infrastructure Layer (Adapters)

• Create JPA entities in infrastructure/persistence/
• Implement repository adapters that map between domain and JPA entities
• Use MapStruct or manual mappers for domain-JPA conversion
• Configure conditional beans for swappable implementations
• Keep infrastructure concerns isolated from domain logic

6. Implement the Adapter Layer (REST)

• Create REST controllers in adapter/rest/
• Inject use case interfaces, not implementations
• Use Bean Validation on DTOs
• Return proper HTTP status codes and responses
• Handle exceptions with global exception handlers

7. Apply Best Practices

1. Dependency Rule: Domain has zero dependencies on Spring or other frameworks
2. Immutable Value Objects: Use Java records for value objects with built-in validation
3. Rich Domain Models: Place business logic in entities, not services
4. Repository Pattern: Domain defines interface, infrastructure implements
5. Domain Events: Decouple side effects from primary operations
6. Constructor Injection: Mandatory dependencies via final fields
7. DTO Mapping: Separate domain models from API contracts
8. Transaction Boundaries: Place @Transactional in application services
9. Factory Methods: Use Entity.create() for invariant enforcement during construction
10. Separate JPA Entities: Keep domain entities separate from JPA entities with mappers

8. Validate Architecture Compliance

After implementing each layer, verify the dependency rules are respected:

• Domain Layer Check: Run grep -r "@Service\|@Component\|@Autowired" domain/ to ensure zero Spring imports
• ArchUnit Test: Add dependency tests to verify no infrastructure imports in domain layer: noClasses().that().resideInPackage("..domain..").should().accessClassesThat().resideInAnyPackage("..spring..", "..infrastructure..");
• Entity Exposure Check: Verify JPA entities are never returned from domain services
• Transaction Check: Confirm @Transactional only on application layer services, never on domain

9. Write Tests

• Domain Tests: Pure unit tests without Spring context, fast execution
• Application Tests: Unit tests with mocked ports using Mockito
• Infrastructure Tests: Integration tests with @DataJpaTest and Testcontainers
• Adapter Tests: Controller tests with @WebMvcTest

Examples

Example 1: Domain Layer - Entity with Domain Events

// domain/model/Order.java
public class Order {
    private final OrderId id;
    private final List<OrderItem> items;
    private Money total;
    private OrderStatus status;
    private final List<DomainEvent> domainEvents = new ArrayList<>();

    private Order(OrderId id, List<OrderItem> items) {
        this.id = id;
        this.items = new ArrayList<>(items);
        this.status = OrderStatus.PENDING;
        calculateTotal();
    }

    public static Order create(List<OrderItem> items) {
        validateItems(items);
        Order order = new Order(OrderId.generate(), items);
        order.domainEvents.add(new OrderCreatedEvent(order.id, order.total));
        return order;
    }

    public void confirm() {
        if (status != OrderStatus.PENDING) {
            throw new DomainException("Only pending orders can be confirmed");
        }
        this.status = OrderStatus.CONFIRMED;
    }

    public List<DomainEvent> getDomainEvents() {
        return List.copyOf(domainEvents);
    }

    public void clearDomainEvents() {
        domainEvents.clear();
    }
}

Example 2: Domain Layer - Value Object with Validation

// domain/model/Money.java (Value Object)
public record Money(BigDecimal amount, Currency currency) {
    public Money {
        if (amount.compareTo(BigDecimal.ZERO) < 0) {
            throw new DomainException("Amount cannot be negative");
        }
    }

    public static Money zero() {
        return new Money(BigDecimal.ZERO, Currency.getInstance("EUR"));
    }

    public Money add(Money other) {
        if (!this.currency.equals(other.currency)) {
            throw new DomainException("Currency mismatch");
        }
        return new Money(this.amount.add(other.amount), this.currency);
    }
}

Example 3: Domain Layer - Repository Port

// domain/repository/OrderRepository.java (Port)
public interface OrderRepository {
    Order save(Order order);
    Optional<Order> findById(OrderId id);
}

Example 4: Application Layer - Use Case and Service

// application/port/in/CreateOrderUseCase.java
public interface CreateOrderUseCase {
    OrderResponse createOrder(CreateOrderRequest request);
}

// application/dto/CreateOrderRequest.java
public record CreateOrderRequest(
    @NotNull UUID customerId,
    @NotEmpty List<OrderItemRequest> items
) {}

// application/service/OrderService.java
@Service
@RequiredArgsConstructor
@Transactional
public class OrderService implements CreateOrderUseCase {
    private final OrderRepository orderRepository;
    private final PaymentGateway paymentGateway;
    private final DomainEventPublisher eventPublisher;

    @Override
    public OrderResponse createOrder(CreateOrderRequest request) {
        List<OrderItem> items = mapItems(request.items());
        Order order = Order.create(items);

        PaymentResult payment = paymentGateway.charge(order.getTotal());
        if (!payment.successful()) {
            throw new PaymentFailedException("Payment failed");
        }

        order.confirm();
        Order saved = orderRepository.save(order);
        publishEvents(order);

        return OrderMapper.toResponse(saved);
    }

    private void publishEvents(Order order) {
        order.getDomainEvents().forEach(eventPublisher::publish);
        order.clearDomainEvents();
    }
}

Example 5: Infrastructure Layer - JPA Entity and Adapter

// infrastructure/persistence/OrderJpaEntity.java
@Entity
@Table(name = "orders")
public class OrderJpaEntity {
    @Id
    private UUID id;
    @Enumerated(EnumType.STRING)
    private OrderStatus status;
    private BigDecimal totalAmount;
    @OneToMany(cascade = CascadeType.ALL, orphanRemoval = true)
    private List<OrderItemJpaEntity> items;
}

// infrastructure/persistence/OrderRepositoryAdapter.java
@Component
@RequiredArgsConstructor
public class OrderRepositoryAdapter implements OrderRepository {
    private final OrderJpaRepository jpaRepository;
    private final OrderJpaMapper mapper;

    @Override
    public Order save(Order order) {
        OrderJpaEntity entity = mapper.toEntity(order);
        return mapper.toDomain(jpaRepository.save(entity));
    }

    @Override
    public Optional<Order> findById(OrderId id) {
        return jpaRepository.findById(id.value()).map(mapper::toDomain);
    }
}

Example 6: Adapter Layer - REST Controller

// adapter/rest/OrderController.java
@RestController
@RequestMapping("/api/orders")
@RequiredArgsConstructor
public class OrderController {
    private final CreateOrderUseCase createOrderUseCase;

    @PostMapping
    public ResponseEntity<OrderResponse> createOrder(
            @Valid @RequestBody CreateOrderRequest request) {
        OrderResponse response = createOrderUseCase.createOrder(request);
        URI location = ServletUriComponentsBuilder
            .fromCurrentRequest()
            .path("/{id}")
            .buildAndExpand(response.id())
            .toUri();
        return ResponseEntity.created(location).body(response);
    }
}

Example 7: Domain Tests (No Spring Context)

class OrderTest {
    @Test
    void shouldCreateOrderWithValidItems() {
        List<OrderItem> items = List.of(
            new OrderItem(new ProductId(UUID.randomUUID()), 2, new Money("10.00", EUR))
        );

        Order order = Order.create(items);

        assertThat(order.getStatus()).isEqualTo(OrderStatus.PENDING);
        assertThat(order.getDomainEvents()).hasSize(1);
    }
}

Example 8: Application Tests (Unit with Mocks)

@ExtendWith(MockitoExtension.class)
class OrderServiceTest {
    @Mock OrderRepository orderRepository;
    @Mock PaymentGateway paymentGateway;
    @Mock DomainEventPublisher eventPublisher;

    @InjectMocks OrderService orderService;

    @Test
    void shouldCreateAndConfirmOrder() {
        when(paymentGateway.charge(any())).thenReturn(new PaymentResult(true, "tx-123"));
        when(orderRepository.save(any())).thenAnswer(i -> i.getArgument(0));

        OrderResponse response = orderService.createOrder(createRequest());

        assertThat(response.status()).isEqualTo(OrderStatus.CONFIRMED);
        verify(eventPublisher).publish(any(OrderCreatedEvent.class));
    }
}

Best Practices

• Domain purity: Keep the domain layer free of Spring annotations and framework imports — zero dependencies on outer layers
• Feature-based packages: Organize by business capability (order/, customer/) rather than technical role, with each feature containing all four layers
• Immutable value objects: Use Java records for value objects with built-in validation in compactors — immutable by design
• Rich domain models: Place business logic in entities and aggregates, not in application services — services orchestrate, entities encapsulate
• Always map: Separate JPA entities from domain models using MapStruct or manual mappers; never expose JPA entities outside infrastructure
• Domain events for decoupling: Use DomainEventPublisher to decouple cross-aggregate side effects instead of direct service calls
• Transaction boundaries in application layer: Place @Transactional only on application services, never on domain classes
• Factory methods for invariants: Use Entity.create(...) static methods to enforce invariants at construction time
• Enforce with ArchUnit: Add ArchUnit tests in the test suite to verify no Spring or infrastructure imports reach the domain layer
• Strongly-typed IDs: Use record OrderId(UUID value) instead of raw UUID to prevent ID confusion across aggregates

Constraints and Warnings

Critical Constraints

• Domain Layer Purity: Never add Spring annotations (@Entity, @Autowired, @Component) to domain classes
• Dependency Direction: Dependencies must only point inward (domain <- application <- infrastructure/adapter)
• Framework Isolation: All framework-specific code must stay in infrastructure and adapter layers

Common Pitfalls to Avoid

• Anemic Domain Model: Entities with only getters/setters, logic in services - place business logic in entities
• Framework Leakage: @Entity, @Autowired in domain layer - keep domain framework-free
• Lazy Loading Issues: Exposing JPA entities through domain model - use mappers to convert
• Circular Dependencies: Between domain aggregates - use IDs instead of direct references
• Missing Domain Events: Direct service calls instead of events for cross-aggregate communication
• Repository Misplacement: Defining repository interfaces in infrastructure - they belong in domain
• DTO Bypass: Exposing domain entities directly in API - always use DTOs for external contracts

Performance Considerations

• Separate JPA entities from domain models to avoid lazy loading issues
• Use read-only transactions for query operations
• Consider CQRS for complex read/write scenarios

References

• references/java-clean-architecture.md - Java-specific patterns (records, sealed classes, strongly-typed IDs)
• references/spring-boot-implementation.md - Spring Boot integration (DI patterns, JPA mapping, transaction management)