如何正确保存关联实体？

Question

假设我们有以下三个域模型实体：Company、Departament 和 Employee。

@Getter @Setter @NoArgsConstrutor
public class Employee {
    private String name;

    @ManyToOne(fetch = FetchType.LAZY)
    @JoinColumn(name = "department_id", nullable = false, insertable = false, updatable = false)
    private Department department;

    @JoinColumn(name = "department_id", nullable = false)
    private int department_id;
}

@Getter @Setter @NoArgsConstrutor
public class Department {
    private String name;

    @ManyToOne(fetch = FetchType.LAZY)
    @JoinColumn(name = "company_id", nullable = false, insertable = false, updatable = false)
    private Company company;

    @JoinColumn(name = "company_id", nullable = false)
    private int company_id;

    @OneToMany(mappedBy = "department")
    private List<Employee> employees;
}

@Getter @Setter @NoArgsConstrutor
private class Company {
    private String name;

    @OneToMany(mappedBy = "company")
    private List<Department> departments;
}

对于每个实体，我们都有 Repositories 扩展 JpaRepository、Services 和 Controllers。在每个Service 中我们@Autowire 各自的Repository，在每个实体Controller 中我们调用来自实体Service 的方法。

我的问题如下：我无法保存整个Company，因为Departments 需要Company ID，而Employees 需要Deparment ID。因此，首先，在我的CompanyService 中保存然后清除departments 列表，执行saveAndFlush 为我的company 分配一个ID。我将接收到的 ID 分配给先前保存的 departments 列表的每个实体中的每个 company_id，然后将该列表附加到 company 并执行另一个 saveAndFlush，我再为 @987654347 执行一次@列表。

@RestController
public class CompanyController {
    @Autowire
    private CompanyService companyService;

    @PostMapping("/companies")
    public Company createCompany(@RequestBody Company newCompany) {
        return companyService.createCompany(newCompany);
    }
}

@Service
public class CompanyService {
    @Autowire
    private CompanyRepository companyRepository;

    public Company createCompany(Company company) {
        List<Department> departments = new ArrayList<>(company.getDepartments());
        company.getDepartments().clear();

        companyRepository.saveAndFlush(company);

        int company_id = company.getId();

        departments.forEach (department ->
            department.setCompany_id(company_id);
        );

        //here I save a copy of the previously saved departments, because I still need the employees
        company.getDepartments().addAll(departments.stream().map(department -> department.clone(department)).collect(Collectors.toList()));
        company.getDepartments().forEach(department -> department.getEmployees().clear());

        companyRepository.saveAndFlush(company);

        //here I assign each employee it's corresponding department ID
        for (int i = 0; i < company.getDepartments().size(); i++) {
            Department departmentInSavedCompany = company.getDepartments().get(i);
            Department departmentWhichStillHasEmployees = departments.get(i);

            departmentWhichStillHasEmployees.setId(departmentInSavedCompany.getId());
            departmentWhichStillHasEmployees.getEmployees().forEach(employee -> employee.setDepartment_id(departmentInSavedCompany.getId()));
        }

        company.getDepartments.clear();
        company.getDepartments.addAll(departments);

        return companyRepository.saveAndFlush(company);
    }
}

@Repository
public interface CompanyRepository extends JpaRepository<Company, Integer> {

}

我目前不喜欢这个实现，也不觉得它很好。哪种方法适合这种情况？

Answer 1

使用 JPA 时，不要使用 ID，而是使用对象引用。

在您的情况下，这意味着删除重复引用的 id 属性。

为了获得正确的 ID 实体，请使用 JpaRepository.getOne。如果实体已经在一级缓存中，它将返回实体，或者只是包装 id 的代理，因此它不会命中数据库。

这允许您组装您的对象图，并从没有引用其他实体的实体开始，一次性将其持久化。

如果您认为实体是同一个聚合的一部分，您也可以考虑配置级联，即它们应该一起加载和持久化。