Spring 数据 JPA 投影与多对一

Question

I have the following two entities:

@Entity
public class Author {
    @Id
    @GeneratedValue(strategy = GenerationType.IDENTITY)
    private Long id;

    private String userName;
    // Getters/setters
}

@Entity
public class Comment {

    @Id
    @GeneratedValue(strategy = GenerationType.IDENTITY)
    private Long id;

    @ManyToOne
    @JoinColumn(name = "author_id")
    private Author author;

    private String comment;
    // Getters/setters
}

I want to generate a JPQL query in a Spring Data JPA repository which gives me the authors with all their comments. My projection is

public interface AuthorProjection {
    String getUserName();
    List<String> getComments();
}

and my JPQL query is

public interface AuthorRepository extends JpaRepository<Author, Long> {

    @Query(value= "SELECT a.id as id, \n" +
            "    a.userName as userName,\n" +
            "    c.comment as comments\n" +
            "from Author a \n" +
            "left join Comment c on c.author = a")
    List<AuthorProjection> authorsWithComments();
}

public interface CommentRepository extends JpaRepository<Comment, Long> { 
}

This generates the following raw SQL query via Hibernate (Underlying DB is PostgreSQL):

select author0_.id as col_0_0_, author0_.user_name as col_1_0_,comment1_.comment as col_2_0_ from author author0_ left outer join comment comment1_ on (comment1_.author_id=author0_.id)

My test data consists of one Author with two Comments. Unfortunately, when I run this, I always get two authors (which is actually the same Author) with one comment each. This is of course how the DB returns each row (the raw SQL looks nice and how it should be). Is it possible for either Hibernate or Spring Data JPA (or whichever is responsible) to merge all comments into the list, or do I need to write Java code which merges the rows based on the author id?

If possible, I would like to have the DB perform all the data transformation already and have a nicely structured Projection as output from the query.

My test for demonstration is:

@DataJpaTest
@AutoConfigureTestDatabase(replace = AutoConfigureTestDatabase.Replace.NONE)
@Transactional(propagation = Propagation.NOT_SUPPORTED)
class JpatestApplicationTests {

    @Autowired
    private AuthorRepository authorRepository;

    @Autowired
    private CommentRepository commentRepository;

    @Test
    void contextLoads() {
        var author = new Author();
        author.setUserName("John Doe");
        var savedAuthor = authorRepository.save(author);

        var comment1 = new Comment();
        comment1.setComment("First comment").setAuthor(savedAuthor);
        commentRepository.save(comment1);

        var comment2 = new Comment();
        comment2.setComment("Second comment").setAuthor(savedAuthor);
        commentRepository.save(comment2);

        authorRepository.authorsWithComments().forEach(
                projection -> System.out.printf("%s, %s\n", projection.getUserName(), projection.getComments())
        );
    }
}

which gives the following output:

John Doe, [First comment]
John Doe, [Second comment]

And run directly in PostgreSQL, it gives:

 col_0_0_ | col_1_0_ |    col_2_0_
----------+----------+----------------
        1 | John Doe | First comment
        1 | John Doe | Second comment
(2 rows)

My expected Java output is:

John Doe, [First comment, Second comment]

Is this even possible with JPA?

Answer 1

Use relation in Author entity to fetch comments of author.

@OneToMany(mappedBy = "author")
private List<Comment> comments = new ArrayList<>();

Then query this way

@Query(value= "SELECT a.userName as userName, a.comments as comments from Author a")
List<AuthorProjection> authorsWithComments();

And projection like

public interface AuthorProjection {
    String getUserName();
    List<Comment> getComments();
}

Answer 2

You get all the comments by just query the authors. You need to add extra List in Author class.

@OneToMany(mappedBy = "author", fetch=FetchType.LAZY)
private List<Comment> comments = new ArrayList<>();

Then, you can use JAP Reposity

@Repository
public interface AuthorRepository  extends CrudRepository<Author, Long> {

}

CrudRepository already implement many useful API, such as findAll() , findById() , you don't need to write any sqls.