How to do unidirectional many-to-many relationships with already existing registers?

Question

I'm new to Spring/JPA and I'm trying to create a relationship pretty much like this post from Vlad, but with one difference. My Tags already exist in another table.

So, if I do just like Vlad does in its post, creating a post, adding some tags to it and then persisting it, everything works as expected. I get a register on Post, two on Tag and two on PostTag.

Post newPost = new Post("Title");
newPost.addTag(new Tag("TagName"));
newPost.addTag(new Tag("TagName2"));
this.postRepository.save(newPost);

But, if I try to create a tag and save it before creating a post, I get an error.

Tag tag = new Tag("TagAlreadyCreated");
this.tagRepository.save(tag);
Post newPost = new Post("Title");
newPost.addTag(tag);
this.postRepository.save(newPost);
// Error: detached entity passed to persist: com.***.***.Tag

I get it that I don't want to create the Tag if it already exists and that the detached message means my Tag already has an ID, so I tried to change the CascadeType to MERGE, but then I don't get a register created on PostTag. Code for the classes:

Post

@Entity(name = "Post")
@Table(name = "post")
public class Post {

    @Id
    @GeneratedValue
    private Long id;

    private String title;

    @OneToMany(
            mappedBy = "post",
            cascade = CascadeType.ALL,
            orphanRemoval = true
    )
    private List<PostTag> tags = new ArrayList<>();

    public Post() {
    }

    public Post(String title) {
        this.title = title;
    }

    public void addTag(Tag tag) {
        PostTag postTag = new PostTag(this, tag);
        tags.add(postTag);
    }

    public void removeTag(Tag tag) {
        for (Iterator<PostTag> iterator = tags.iterator();
             iterator.hasNext(); ) {
            PostTag postTag = iterator.next();

            if (postTag.getPost().equals(this) &&
                    postTag.getTag().equals(tag)) {
                iterator.remove();
                postTag.setPost(null);
                postTag.setTag(null);
            }
        }
    }

    @Override
    public boolean equals(Object o) {
        if (this == o) return true;

        if (o == null || getClass() != o.getClass())
            return false;

        Post post = (Post) o;
        return Objects.equals(title, post.title);
    }

    @Override
    public int hashCode() {
        return Objects.hash(title);
    }

    public Long getId() {
        return id;
    }
}

Tag

@Entity(name = "Tag")
@Table(name = "tag")
@NaturalIdCache
@Cache(usage = CacheConcurrencyStrategy.READ_WRITE)
public class Tag {

    @Id
    @GeneratedValue
    private Long id;

    public Long getId() {
        return id;
    }

    @NaturalId
    private String name;

    public Tag() {
    }

    public Tag(String name) {
        this.name = name;
    }

    @Override
    public boolean equals(Object o) {
        if (this == o) return true;

        if (o == null || getClass() != o.getClass())
            return false;

        Tag tag = (Tag) o;
        return Objects.equals(name, tag.name);
    }

    @Override
    public int hashCode() {
        return Objects.hash(name);
    }
}

PostTag

@Entity(name = "PostTag")
@Table(name = "post_tag")
public class PostTag {

    @EmbeddedId
    private PostTagId id;

    @ManyToOne(fetch = FetchType.LAZY)
    @MapsId("postId")
    private Post post;

    @ManyToOne(fetch = FetchType.LAZY)
    @MapsId("tagId")
    private Tag tag;

    @Column(name = "created_on")
    private Date createdOn = new Date();

    private PostTag() {}

    public void setPost(Post post) {
        this.post = post;
    }

    public void setTag(Tag tag) {
        this.tag = tag;
    }

    public PostTag(Post post, Tag tag) {
        this.post = post;
        this.tag = tag;
        this.id = new PostTagId(post.getId(), tag.getId());
    }

    //Getters and setters omitted for brevity

    @Override
    public boolean equals(Object o) {
        if (this == o) return true;

        if (o == null || getClass() != o.getClass())
            return false;

        PostTag that = (PostTag) o;
        return Objects.equals(post, that.post) &&
                Objects.equals(tag, that.tag);
    }

    public Post getPost() {
        return post;
    }

    public Tag getTag() {
        return tag;
    }

    @Override
    public int hashCode() {
        return Objects.hash(post, tag);
    }
}

PostTagId

@Embeddable
public class PostTagId
        implements Serializable {

    @Column(name = "post_id")
    private Long postId;

    @Column(name = "tag_id")
    private Long tagId;

    private PostTagId() {}

    public PostTagId(
            Long postId,
            Long tagId) {
        this.postId = postId;
        this.tagId = tagId;
    }

    //Getters omitted for brevity

    @Override
    public boolean equals(Object o) {
        if (this == o) return true;

        if (o == null || getClass() != o.getClass())
            return false;

        PostTagId that = (PostTagId) o;
        return Objects.equals(postId, that.postId) &&
                Objects.equals(tagId, that.tagId);
    }

    @Override
    public int hashCode() {
        return Objects.hash(postId, tagId);
    }
}

Answer 1

The spring-data-jpa is a layer on top of JPA. Each entity has its own repository and you have to deal with that. I've seen that tutorial mentioned above and it's for JPA and it's also setting ID's to null which seems off a bit and probably the cause of your error. I didn't look that close. For dealing with the issue in spring-data-jpa you need a separate repository for the link table.

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

    @OneToMany(mappedBy = "post", cascade = CascadeType.ALL, orphanRemoval = true)
    private List<PostTag> tags;

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

    @OneToMany(mappedBy = "tag", cascade = CascadeType.ALL, orphanRemoval = true)
    private List<PostTag> posts;

@Entity
public class PostTag {
    @EmbeddedId
    private PostTagId id = new PostTagId();

    @ManyToOne(fetch = FetchType.LAZY)
    @MapsId("postId")
    private Post post;

    @ManyToOne(fetch = FetchType.LAZY)
    @MapsId("tagId")
    private Tag tag;

    public PostTag() {}
    public PostTag(Post post, Tag tag) {
        this.post = post;
        this.tag = tag;
    }

@SuppressWarnings("serial")
@Embeddable
public class PostTagId implements Serializable {
    private Long postId;
    private Long tagId;
    @Override
    public boolean equals(Object o) {
        if (this == o)
            return true;
        if (o == null || getClass() != o.getClass())
            return false;
        PostTagId that = (PostTagId) o;
        return Objects.equals(postId, that.postId) && Objects.equals(tagId, that.tagId);
    }
    @Override
    public int hashCode() {
        return Objects.hash(postId, tagId);
    }

And to use it, as show above:

@Transactional
private void update() {
    System.out.println("Step 1");
    Tag tag1 = new Tag();
    Post post1 = new Post();
    PostTag p1t1 = new PostTag(post1, tag1);
    tagRepo.save(tag1);
    postRepo.save(post1);
    postTagRepo.save(p1t1);

    System.out.println("Step 2");
    Tag tag2 = new Tag();
    Post post2 = new Post();
    PostTag p2t2 = new PostTag(post2, tag2);
    postRepo.save(post2);
    tagRepo.save(tag2);
    postTagRepo.save(p2t2);

    System.out.println("Step 3");
    tag2 = tagRepo.getOneWithPosts(2L);
    tag2.getPosts().add(new PostTag(post1, tag2));
    tagRepo.save(tag2);

    System.out.println("Step 4 -- better");
    PostTag p2t1 = new PostTag(post2, tag1);
    postTagRepo.save(p2t1);
}

Note there are few changes. I don't explicitly set the PostTagId id's. These are handled by the persistence layer (hibernate in this case).

Note also that you can update PostTag entries either explicity with its own repo or by adding and removing them from the list since CascadeType.ALL is set, as shown. The problem with using the CascadeType.ALL for spring-data-jpa is that even though you prefetch the join table entities spring-data-jpa will do it again anyway. Trying to update the relationship through the CascadeType.ALL for new entities is problematic.

Without the CascadeType neither the posts or tags lists (which should be Sets) are the owners of the relationship so adding to them wouldn't accomplish anything in terms of persistence and would be for query results only.

When reading the PostTag relationships you need to specifically fetch them since you don't have FetchType.EAGER. The problem with FetchType.EAGER is the overhead if you don't want the joins and also if you put it on both Tag and Post then you will create a recursive fetch that gets all Tags and Posts for any query.

@Query("select t from Tag t left outer join fetch t.posts tps left outer join fetch tps.post where t.id = :id")
Tag getOneWithPosts(@Param("id") Long id);

Finally, always check the logs. Note that creating an association requires spring-data-jpa (and I think JPA) to read the existing table to see if the relationship is new or updated. This happens whether you create and save a PostTag yourself or even if you prefetched the list. JPA has a separate merge and I think you can use that more efficiently.

create table post (id bigint generated by default as identity, primary key (id))
create table post_tag (post_id bigint not null, tag_id bigint not null, primary key (post_id, tag_id))
create table tag (id bigint generated by default as identity, primary key (id))
alter table post_tag add constraint FKc2auetuvsec0k566l0eyvr9cs foreign key (post_id) references post
alter table post_tag add constraint FKac1wdchd2pnur3fl225obmlg0 foreign key (tag_id) references tag

Step 1
insert into tag (id) values (null)
insert into post (id) values (null)
select posttag0_.post_id as post_id1_1_0_, posttag0_.tag_id as tag_id2_1_0_ from post_tag posttag0_ where posttag0_.post_id=? and posttag0_.tag_id=?
insert into post_tag (post_id, tag_id) values (?, ?)

Step 2
insert into post (id) values (null)
insert into tag (id) values (null)
select posttag0_.post_id as post_id1_1_0_, posttag0_.tag_id as tag_id2_1_0_ from post_tag posttag0_ where posttag0_.post_id=? and posttag0_.tag_id=?
insert into post_tag (post_id, tag_id) values (?, ?)

Step 3
select tag0_.id as id1_2_0_, posts1_.post_id as post_id1_1_1_, posts1_.tag_id as tag_id2_1_1_, post2_.id as id1_0_2_, posts1_.tag_id as tag_id2_1_0__, posts1_.post_id as post_id1_1_0__ from tag tag0_ left outer join post_tag posts1_ on tag0_.id=posts1_.tag_id left outer join post post2_ on posts1_.post_id=post2_.id where tag0_.id=?
select tag0_.id as id1_2_1_, posts1_.tag_id as tag_id2_1_3_, posts1_.post_id as post_id1_1_3_, posts1_.post_id as post_id1_1_0_, posts1_.tag_id as tag_id2_1_0_ from tag tag0_ left outer join post_tag posts1_ on tag0_.id=posts1_.tag_id where tag0_.id=?
select posttag0_.post_id as post_id1_1_0_, posttag0_.tag_id as tag_id2_1_0_ from post_tag posttag0_ where posttag0_.post_id=? and posttag0_.tag_id=?
insert into post_tag (post_id, tag_id) values (?, ?)

Step 4 -- better
select posttag0_.post_id as post_id1_1_0_, posttag0_.tag_id as tag_id2_1_0_ from post_tag posttag0_ where posttag0_.post_id=? and posttag0_.tag_id=?
insert into post_tag (post_id, tag_id) values (?, ?)

Answer 2

I think I found the answer.

First think to know is that if you have open-in-view set to true, Spring will keep the JPA/Hibnernate Session open for the whole lifespan of the request. This means that your code (which is let's say inside a service method) should work without problems. Of course this is not very efficient (see here why): https://vladmihalcea.com/the-open-session-in-view-anti-pattern/

Now, if you set open-in-view to false, it seems that Spring opens a new Session for each repository call. So a new session for fetching the Tags and then a new session for saving the Post. That is why the Tags entities are detached when saving the Post.

To solve this, you need to annotate your service call with @Transactional. Spring will try to reuse the same session inside a transaction so the entities will not become detached. See here for example: How does @Transactional influence current session in Hibernate?

The final thing to know which is crucial is that the service methods must be public! If you have any other visibility on your method the @Transactional is ignored, no errors are thrown: https://docs.spring.io/spring/docs/current/spring-framework-reference/data-access.html#transaction-declarative-annotations (see Method visibility and @