I have 2 classes, many-to-many relationship between them. Classes are Client and Role:
#Entity
#Table(name="iq_role")
public class Role extends IdEntity {
#ManyToMany(fetch = FetchType.LAZY, mappedBy = "roles")
private Set<Client> clients = new LinkedHashSet<>();
}
and
#Entity
#Table(name="iq_client")
public class Client extends DeletableEntity {
#ManyToMany(fetch = FetchType.LAZY)
#JoinTable(name = "iq_client_roles", joinColumns = #JoinColumn(name="iq_client", referencedColumnName = "id"), inverseJoinColumns = #JoinColumn(name="iq_role", referencedColumnName = "id"))
private Set<Role> roles = new HashSet<>();
}
when I select a specific client and join fetch his roles, this is the SQL (generated by Hibernate):
select
client.id,
role.id
from
iq_client client
left outer join
iq_client_roles client_roles
on client.id = client_roles.iq_client
left outer join
iq_role role
on client_roles.iq_role = role.id
where
client.id=1
in table iq_client I have record with ID 3, in table iq_role I have record with ID 1 and in table iq_client_roles I have record 3,1.
When I copy this generated SQL and run it from console, I get correct result, but hibernate is not able to map the result into objects obviously.
Any idea why?
Related
In a spring-boot app, I've got the following entity definition:
#Data
#Entity
#Table(name = "users")
public class User {
#Id
#Column(nullable = false, name = "username", length = 100)
private String username;
#JoinTable(name = "userrole",
joinColumns = { #JoinColumn(name = "username") },
inverseJoinColumns = { #JoinColumn(name = "role") }
)
#OneToMany(
cascade = CascadeType.ALL,
orphanRemoval = true
)
private List<Role> roles;`
I'm using Spring-data-jpa,Hibernate with H2 as the database.
The trouble is that spring-data-jpa, hibernate always generate/creates the join table (DDL) 'userrole' with a single column primary key. e.g. 'username'.
Hence, if records such as {'username', 'user_role'} and {'username', 'admin_role'} is inserted in the join table ('userrole'), the next insert fails with an error due to the 'duplicate' primary key.
I've tried using both columns in the above definition, as well as the following variation:
#OneToMany(
cascade = CascadeType.ALL,
orphanRemoval = true
)
#JoinColumns({
#JoinColumn(name = "username"),
#JoinColumn(name = "role") })
private List<Role> roles;`
But that they resulted in the same or worse problems, e.g. and in the latter, even table creation fails because only a single column is used as primary key for the jointable. Role is simply another table with 2 columns 'role' and 'description', basically a role catalog.
How do we specify to JPA that the #JoinTable should use both 'username' and 'role' columns as composite primary keys?
edit:
I tried using an independent table/entity as suggested, thanks #Kamil Bęben
#Data
#Entity
#Table(name = "users")
public class User {
#Id
#Column(nullable = false, name = "username", length = 100)
private String username;
#OneToMany(
fetch = FetchType.EAGER,
cascade = CascadeType.ALL,
mappedBy = "username",
orphanRemoval = true
)
#ElementCollection
private List<UserRole> roles;
UserRole is defined as such
#Data
#NoArgsConstructor
#Entity
#Table(name = "userrole")
public class UserRole {
#Id
#GeneratedValue(strategy = GenerationType.SEQUENCE, generator = "userrole_seq")
Long id;
#Column(nullable = false, name = "username", length = 100)
private String username;
#Column(nullable = false, name = "role", length = 50)
private String role;
the repository for that user-roles join table is defined as
#Repository
public interface UserRoleRepository extends CrudRepository<UserRole, Long> {
UserRole findByUsernameAndRole(String username, String role);
List<UserRole> findByUsername(String username);
List<UserRole> findByRole(String role);
}
Admittedly, ugly, but that it works. And that somehow, it seemed to use the correct findByUsername() method to retrieve the roles as is relevant to the user, probably related to the 'mappedBy' clause. 'black magic'! There's lots more that I'd still need to find my way around JPA, Spring, Spring-data
edit2:
further update:
the original #JoinTable works as well.
But that the relations need to be specified as #ManyToMany
#ManyToMany(
fetch = FetchType.EAGER,
cascade = CascadeType.MERGE
)
#JoinTable(name = "usersroles",
joinColumns = { #JoinColumn(name = "username") },
inverseJoinColumns = { #JoinColumn(name = "role") }
)
private List<Role> roles = new ArrayList<Role>();
This creates 2 column primary keys as expected for the 'users-roles' table
Thanks to #Roman
If Role only has two columns, eg user_id and role, the way to map this in jpa would be as following
#ElementCollection
#CollectionTable(name = "user_roles", joinColumns = #JoinColumn(name = "user_id"))
#Column(name = "role")
List<String> roles = new ArrayList<>();
Otherwise, jpa really requires each entity's identifier and join columns to be separate columns, so Role entity would have to have columns like id, user_id and role_name. Could look like this .:
class Role {
#Id
Long id;
#ManyToOne
#JoinColumn(name = "user_id", referencedColumnName = "id");
User user;
String roleName;
// Other fields
}
And in the User entity
#OneToMany(mappedBy = "user") // user is Field's name, not a column
List<Role> roles = new ArrayList<>();
Further reading
Given we have the following entities that form a many-to-many relationship:
#Entity
public class A {
#Id
private Long id;
private String name;
#ManyToMany(fetch = FetchType.LAZY, cascade = CascadeType.ALL)
#JoinTable(
name = "A_B",
joinColumns = #JoinColumn(name = "id_a"),
inverseJoinColumns = #JoinColumn(name = "id_b"))
private Set<B> listing;
}
#Entity
public class B {
#Id
private Long id;
}
I need to write a query that fetches B and applies some WHERE criteria on A side.
Since the relationsip is modeled from A entity's side it's very easy to write a query that joins these itsself:
new JPAQuery<>(entityManager)
.select(QB.b)
.from(QA.a)
.join(QA.a.listing,b)
.where(QA.a.name.eq("test"))
.fetch();
However since A_B table can be duplicated, this query can produce duplicate entries, which does not do for my scenario. So instead I need to start FROM B and JOIN A. And this is where I need help. I tried:
new JPAQuery<>(entityManager)
.select(QB.b)
.from(QB.b)
.join(QA.a).on(QA.a.listing.any().eq(QB.b))
.where(QA.a.name.eq("test"))
.fetch();
But that does not work as any() merely produces a subselect, instead of many to many join.
How do I write this query in Querydsl?
Short version
I have a basic setup where a user table is linked to a role table and a role table is linked to a right. These are both Many-to-Many relations. The roles are a dynamic entity and not of interest for the application (only for visual aspects). When I fetch a user I want to return the data in the user table including a list of the names of all rights.
To clarify, this is what I want the solution to do:
I managed to get the rights in my user object and return them, but it's inefficient due to the extra query calls hibernate makes after the original query was called.
Detailed version
Let me first give you some information on how to entities are linked and what the code looks like. My (simplified) database table structure looks like this:
User.java
#Entity
#Table(name = "user")
public class User {
#Id
#Column(name = "user_id", columnDefinition = "user_id")
private Long userId;
#Transient
private List<String> rights
#ManyToMany
#JoinTable(
name = "user_role",
joinColumns = #JoinColumn(name = "user_id", referencedColumnName = "user_id"),
inverseJoinColumns = #JoinColumn(name = "role_id", referencedColumnName = "role_id"))
#JsonIgnore
private List<Role> roles;
//Getters and setters
}
Role.java
#Entity
#Table(name = "role")
public class Role {
#Id
#Column(name = "role_id", columnDefinition = "role_id")
private Long roleId;
#ManyToMany
#JoinTable(
name = "user_role",
joinColumns = #JoinColumn(name = "role_id", referencedColumnName = "role_id"),
inverseJoinColumns = #JoinColumn(name = "user_id", referencedColumnName = "user_id"))
#JsonIgnore
private List<Employee> employees;
#ManyToMany
#JoinTable(
name = "role_right",
joinColumns = #JoinColumn(name = "role_id", referencedColumnName = "role_id"),
inverseJoinColumns = #JoinColumn(name = "right_id", referencedColumnName = "right_id"))
#JsonIgnore
private List<Right> rights;
//Getters and setters
}
Right.java
#Entity
#Table(name = "right")
public class Right {
#Id
#Column(name = "right_id", columnDefinition = "right_id")
private Long rightId;
#Column(name = "name", columnDefinition = "name")
private String name;
#ManyToMany
#JoinTable(
name = "role_right",
joinColumns = #JoinColumn(name = "right_id", referencedColumnName = "right_id"),
inverseJoinColumns = #JoinColumn(name = "role_id", referencedColumnName = "role_id"))
#JsonIgnore
private List<Role> roles;
//Getters and setters
}
It's important to know that I use the Java Specifications API to join the tables:
return (root, query, cb) -> {
query.distinct(true);
Join rolesJoin = root.join("roles", JoinType.LEFT);
Join rightsJoin = rolesJoin.join("rights", JoinType.LEFT);
return cb.conjunction();
};
This creates the correct query:
select <columns go here>
from employee user0_
left outer join user_role roles1_ on user0_.user_id=roles1_.user_id
left outer join role role2_ on roles1_.role_id=role2_.role_id
left outer join role_right rights3_ on role2_.role_id=rights3_.role_id
left outer join right right4_ on rights3_.right_id=right4_.right_id
Everything looked to good to me till now. But when I tried to fetch the names of all roles, there where more than two queries (count for page and the original one) being executed
//The original code uses lambda for this
for(Role role : user.getRoles()){
for(Right right: role.getRights()){
user.addRight(right.getName());
}
}
The extra query looks like:
select <column stuff>
from role_right rights0_
inner join right right1_ on rights0_.right_id=right1_.right_id
where rights0_.role_id=?
This makes the call very inefficient to me. In this case it's a single user, but with multiple users it adds up.
Is there a way to have a single query put the names of all rights in the user entity, without adding extra query executions?
Things I tried so far:
Using #SecondaryTable to directly define column from the Right table in my User entity. I could not get to first link the Role to the User and then use fields from the Role table to link the Right table. So in the end I would have to #SecondaryTable annotation on top of my User object and define columns of the Right object below.
Using #Formula in the User entity to insert a native call into the query. This did also not work as the annotation did not understand how to map everything into a list of rights.
There might be other options here, or I did something horribly wrong with implementing the ones above. But for now I don't which way to go in finding a solution for my problem. If someone could tell me, that would be great.
Thanks in advance,
Robin
You are using Root.join which does just the joining of tables for the purposes of the query; lazy associations in the loaded entities will still not be initialized.
As I see, your intention is to initialize the lazy collections as well. For that you have to use Root.fetch (defined in the interface method inherited from the FetchParent):
Create a fetch join to the specified collection-valued attribute using
the given join type.
However, your intention is not a good practice; do not join multiple collections in one query, otherwise the query result set will explode with full Cartesian product between the joined collections. Your result set contains <num of users> * <num of roles per user> * <num of rights per role> rows. So, each user data is repeated <num of roles per user> * <num of rights per role> times in the generated result set.
The approach I find to be the best and most straightforward is to specify batch size on lazy associations.
I have a many-to-many relationship between user and role entities. when I try to merge a role by adding new user objects to that, after merging i expect a new row in the joined table (in sql server database) to be inserted but nothing happens.
I guess the problem is the direction of the relationship which the owner is User entity now and should be switched to Role. But if I change it then my spring security wont work. Is there any other way to solve this? except changing many-to-many sides?
Thanks in advance.
User class
#ManyToMany(fetch = FetchType.EAGER, cascade = CascadeType.ALL)
#JoinTable(
name = "SEC_USER_ROLE",
joinColumns = {
#JoinColumn(name = "SEC_USER_ID", referencedColumnName = "SEC_USERS")},
inverseJoinColumns = {
#JoinColumn(name = "SEC_ROLE_ID", referencedColumnName = "SEC_ROLES")})
private List<Role> roles;
--
Role class
#ManyToMany(mappedBy = "roles", fetch = FetchType.EAGER, cascade = CascadeType.ALL)
private List<User> users;
And here is my merge function
#Override
#Transactional(readOnly = false, propagation = Propagation.REQUIRED)
public T merge(T o) {
o = em.merge(o);
em.flush();
return o;
}
I have a table that has two different many-to-many relations to two different tables. Let's say I have User <---> UserRole <--> Role and User <--> UserGroups <--> Groups. Since I am new to hibernate and database mapping I was wondering if having my User entity have attributes roles and groups in it, both with #ManytoMany annotations is good practice and acceptable?
i.e.:
#Entity(name = "User")
#Table(name = "USER")
public class User {
.... /* Obviously Id would go here and all other attributes */
#ManyToMany(cascade = CascadeType.ALL)
#JoinTable(name = "UserRole", joinColumns = { #JoinColumn(name="USER_ID") },
inverseJoinColumns = { #JoinColumn(name = "ROLE_ID") } )
private Set<Role> roles = new HashSet<Role>();
#ManyToMany(cascade = CascadeType.ALL)
#JoinTable(name = "UserGroup", joinColumns = { #JoinColumn(name="USER_ID") },
inverseJoinColumns = { #JoinColumn(name = "GROUP_ID") } )
private Set<Group> groups = new HashSet<Group>();
Sure; lots of types have multiple many-to-many.
I think minimizing many-to-many relationships is a good idea, but it's a natural structure, and is found all over the place.