Spring 数据 JPA：有什么问题或如何更正此 jpql 查询

Question

实体关系是 - 一个学生属于一所大学，一所大学可以有多个学生。 所以在 Student --> College 之间存在 ManyToOne 关系，在 College --> Student 之间存在 OneToMany 关系。

实体如下。

@Entity
public class College {

    @Id
    @GeneratedValue
    private int collegeId;

    private String collegeName;

    @OneToMany(targetEntity = Student.class, mappedBy = "college") 
    private List<Student> students;

和

@Entity
public class Student {

    @Id
    @GeneratedValue
    private int studentId;

    private String studentName;

    @ManyToOne
    @JoinColumn(name = "collegeId_fk")
    private College college;

1) 我在 spring 数据 jpa 存储库中使用以下 jpql 查询。

@Query("SELECT c FROM College c LEFT JOIN FETCH c.students where c.collegeId IN (2)")
public List<College> findByCollegeIdsJPQL();

我希望返回一个包含单个大学实体的列表，因为 collegeId 是大学实体的主键，我只为IN提供一个 id。 但我得到的是所有具有相同主键 (collegeId=2) 的大学列表。 返回列表的大小等于学院的学生人数。

System.err.println("collegeRepo.findByCollegeIdsJPQL().size(): " + collegeRepo.findByCollegeIdsJPQL().size()); 
//output: collegeRepo.findByCollegeIdsJPQL().size(): 6

而对于

collegeRepo.findByCollegeIdsJPQL().forEach( System.err::println );

output：
大学 [collegeId=2, collegeName=college2]
大学 [collegeId=2, collegeName=college2]
大学 [collegeId=2, collegeName=college2]
大学 [collegeId=2, collegeName=college2]
大学 [collegeId=2, collegeName=college2]
大学 [collegeId=2, collegeName=college2]

2）我注意到的另一个问题，
第二次调用collegeRepo.findByCollegeIdsJPQL()会导致另一个 sql 查询，即数据库命中。 这不能从会话（第一级）缓存中提供。 我已经注释了调用方法

  @Transactional
  public void run(String... args) throws Exception {

output 日志 -

2022-02-20 15:10:59.140[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([student_1_12_0__] : [INTEGER]) - [6]
collegeRepo.findByCollegeIdsJPQL().size(): 6
[2m2022-02-20 15:10:59.153[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.s.t.i.TransactionInterceptor          [0;39m [2m:[0;39m No need to create transaction for [org.springframework.data.jpa.repository.support.SimpleJpaRepository.findByCollegeIdsJPQL]: This method is not transactional.
Hibernate: 
    select
        college0_.college_id as college_1_1_0_,
        students1_.student_id as student_1_12_1_,
        college0_.college_name as college_2_1_0_,
        students1_.college_id_fk as college_3_12_1_,
        students1_.student_name as student_2_12_1_,
        students1_.college_id_fk as college_3_12_0__,
        students1_.student_id as student_1_12_0__ 
    from
        college college0_ 
    left outer join
        student students1_ 
            on college0_.college_id=students1_.college_id_fk 
    where
        college0_.college_id in (
            2
        )
[2m2022-02-20 15:10:59.245[0;39m [32m INFO[0;39m [35m18416[0;39m [2m---[0;39m [2m[on(3)-127.0.0.1][0;39m [36mo.a.c.c.C.[Tomcat].[localhost].[/]      [0;39m [2m:[0;39m Initializing Spring DispatcherServlet 'dispatcherServlet'
[2m2022-02-20 15:10:59.246[0;39m [32m INFO[0;39m [35m18416[0;39m [2m---[0;39m [2m[on(3)-127.0.0.1][0;39m [36mo.s.web.servlet.DispatcherServlet       [0;39m [2m:[0;39m Initializing Servlet 'dispatcherServlet'
[2m2022-02-20 15:10:59.247[0;39m [32m INFO[0;39m [35m18416[0;39m [2m---[0;39m [2m[on(3)-127.0.0.1][0;39m [36mo.s.web.servlet.DispatcherServlet       [0;39m [2m:[0;39m Completed initialization in 1 ms
[2m2022-02-20 15:10:59.403[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_1_1_0_] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.404[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([student_1_12_1_] : [INTEGER]) - [1]
[2m2022-02-20 15:10:59.404[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_3_12_0__] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.404[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_1_1_0_] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.404[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([student_1_12_1_] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.404[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_3_12_0__] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.404[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_1_1_0_] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.405[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([student_1_12_1_] : [INTEGER]) - [3]
[2m2022-02-20 15:10:59.405[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_3_12_0__] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.405[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_1_1_0_] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.406[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([student_1_12_1_] : [INTEGER]) - [4]
[2m2022-02-20 15:10:59.406[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_3_12_0__] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.406[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_1_1_0_] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.406[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([student_1_12_1_] : [INTEGER]) - [5]
[2m2022-02-20 15:10:59.406[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_3_12_0__] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.406[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_1_1_0_] : [INTEGER]) - [2]
[2m2022-02-20 15:10:59.406[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([student_1_12_1_] : [INTEGER]) - [6]
[2m2022-02-20 15:10:59.406[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.h.type.descriptor.sql.BasicExtractor  [0;39m [2m:[0;39m extracted value ([college_3_12_0__] : [INTEGER]) - [2]
College [collegeId=2, collegeName=college2]
College [collegeId=2, collegeName=college2]
College [collegeId=2, collegeName=college2]
College [collegeId=2, collegeName=college2]
College [collegeId=2, collegeName=college2]
College [collegeId=2, collegeName=college2]
[2m2022-02-20 15:10:59.407[0;39m [32mTRACE[0;39m [35m18416[0;39m [2m---[0;39m [2m[           main][0;39m [36mo.s.t.i.TransactionInterceptor          [0;39m [2m:[0;39m Completing transaction for [com.demo.MyRunner.run]

Answer 1

(1) 因为LEFT JOIN将返回大学及其学生的组合。 因此，如果一所大学有 N 个学生，则将返回 N 条记录。 您必须添加DISTINCT以删除重复项：

@Query("SELECT distinct c FROM College c LEFT JOIN FETCH c.students where c.collegeId IN (2)")
public List<College> findByCollegeIdsJPQL();

但是，它会导致生成的 SQL 具有distinct关键字，这可能会对性能产生影响。 因此，自 Hibernate 5.2 以来，他们提供了一个名为hibernate.query.passDistinctThrough的查询提示，它可以配置不在生成的 SQL 中添加distinct关键字，而 Hibernate 将有助于删除重复项：

@Query("SELECT distinct c FROM College c LEFT JOIN FETCH c.students where c.collegeId IN (2)")
@QueryHints(@QueryHint(name = org.hibernate.jpa.QueryHints.HINT_PASS_DISTINCT_THROUGH, value = "false"))
public List<College> findByCollegeIdsJPQL();

有关详细信息，请参阅此博客文章。

(2) 正常。 您必须配置二级缓存和查询缓存以防止再次命中数据库。 一级缓存主要在同一事务中使用EntityManager#get()通过 ID 获取实体时起作用。 它在使用 JPQL 查询时不起作用。