Develop Reference

Convert HQL query in the form of "select foo from Foo foo, Bar bar where ..." to Criteria query

In our (ancient) project, we use loads of select queries with HQL on Hibernate version 3.5.6-Final. Because Hibernate will do an auto-commit, because it doesn't know they are select queries, I'm in the process of rewriting all HQL select queries to Hibernate Criteria queries so it won't do in-between commits when we don't want it yet. This is pretty straight-forward in most cases, but I'm currently looking at a query like this, and I'm not sure how to transform it:
Query query = session.createQuery("select municapilityStreet"
+ " from MunicapilityStreet munStreet, Street street"
+ " where munStreet.id = street.MunicapilityStreet.id"
+ " and street.id = :streetId");
query.setParameter(":streetId", streetId);
MunicapilityStreet result = (MunicapilityStreet) query.uniqueResult();
return result;
Here what I have thus far in the process of transforming it:
Criteria criteria = session.createCriteria(MunicapilityStreet.class);
// No idea what to set here to only get the "municapilityStreet" as result
criteria.setProjection(??);
// I'm not sure if this is correct. With a criteria on a single table it would have been simply "Id".
// Both tables have the column-name Id, and I'm not sure how to differentiate between them.
criteria.add(Restrictions.eq("Street.Id", streetId));
MunicapilityStreet result = (MunicapilityStreet) criteria.uniqueResult();
return result;
Maybe I should create a different question for each, but converting the above HQL query to a Criteria has three points I'm not sure about how to do:
How to do a select with multiple tables (the from MunicapilityStreet munStreet, Street street part)?
How to have a projection to only return a single table of the two (the select municapilityStreet part)?
How to have an equal-Restriction on a column name of one table, even though both tables have the same column-name (the and street.id = :streetId part)?

I do oppose the rewrite approach, I hope I'm not impolite doing so.
Hibernate allows to control commits (autocommit is by default off), and what you're experiencing are Entitymanager-flushes, they are auto by default and can be disabled too. And, finally, I think, there is no difference if you're running HQL or criteria queries, same machinery underneath.

Criteria API limit results in subquery

I'm trying to write a query similar to
select * from Table a
where a.parent_id in
(select b.id from Table b
where b.state_cd = ?
and rownum < 100)
using the Criteria API. I can achieve the query without the rownum limitation on the subquery fine using similar code to https://stackoverflow.com/a/4668015/597419 but I cannot seem to figure out how to appose a limit on the Subquery

In Hibernate, you can add the actual SQL restriction, but it is worth noting this will be Oracle-specific. If you switched over to PostgreSQL, this would break and you'd need LIMIT 100 instead.
DetachedCriteria criteria = DetachedCriteria.forClass(Domain.class)
.add(Restrictions.sqlRestriction("rownum < 100"));
In the JPA API, the short answer is that you can't... In your question you proposed using the Criteria API (along with a SubQuery). However it is not until you actually call EntityManager.createQuery(criteriaQuery) that you'll get a TypedQuery where you can specify the maxResult value.
That said, you could break it into 2 queries, the first where you get the inner-select results (max 100) and then a 2nd Criteria where you take the resulting list in an in():
// inner query
CriteriaBuilder cb = em.getCriteriaBuilder();
CriteriaQuery<YourClass> innerCriteriaQuery = cb.createQuery(YourClass.class);
Root<YourClass> yourClass = innerCriteriaQuery.from(YourClass.class);
innerCriteriaQuery.select(yourClass).where(
cb.equal(yourClass.get(YourClass_.stateCode), someStateValue));
// list of 100 parent ids
List<YourClass> list = em.createQuery(innerCriteriaQuery).setMaxResults(100).getResultList();
// outer query
CriteriaQuery<YourClass> criteriaQuery = cb.createQuery(YourClass.class);
Root<YourClass> yourClass = criteriaQuery.from(YourClass.class);
criteriaQuery.select(yourClass).where(
cb.in(yourClass.get(YourClass_.parentId)).value(list);
return em.createQuery(criteriaQuery).getResultList();

There is no JPA Criteria solution for this. You could make use of a custom SQL function that runs during SQL query generation time. All JPA providers support something like that in one way or another.
If you don't want to implement that yourself or even want a proper API for constructing such queries, I can only recommend you the library I implemented called Blaze-Persistence.
Here is the documentation showcasing the limit/offset use case with subqueries: https://persistence.blazebit.com/documentation/core/manual/en_US/index.html#pagination
Your query could look like this with the query builder API:
criteriaBuilderFactory.create(entityManager, SomeEntity.class)
.where("id").in()
.select("subEntity.id")
.from(SomeEntity.class, "subEntity")
.where("subEntity.state").eq(someValue)
.orderByAsc("subEntity.id")
.setMaxResults(100)
.end()
It essentially boils down to using the LIMIT SQL function that is registered by Blaze-Persistence. So when you bootstrap Blaze-Persistence with your EntityManagerFactory, you should even be able to use it like this
entityManager.createQuery(
"select * from SomeEntity where id IN(LIMIT((" +
" select id " +
" from SomeEntity subEntity " +
" where subEntity.state = :someParam " +
" order by subEntity.id asc" +
"),1)) "
)
or something like
criteriaQuery.where(
cb.in(yourClass.get(YourClass_.parentId)).value(cb.function("LIMIT", subquery));
If you are using EclipseLink the calling convention of such functions looks like OPERATOR('LIMIT', ...).

Hibernate Criteria filter Entity where ManyToMany relation contains multiple objects

I need help with Hibernate Criteria API.
I have a class Job that contain a list of Skill in ManyToMany relationship.
I need to select jobs based on a skill list given as parameter.
I've tried with Restriction.in("skill.id",ids) but this gives me wrong list.If i've selected 2 skills in search form i want the jobs that contain BOTH of them,so i need somehow to implement AND clause.
I've tried with Conjunction:
Conjunction and = Restrictions.conjunction();
for(Skill skill:job.getSkills()){
and.add(Restrictions.eq("skill.id",skill.getId()));
}
And this:
Criteria crit = criteria.createCriteria("skills",JoinType.LEFT_OUTER_JOIN);
for(Skill skill:job.getSkills()){
crit.add(Restrictions.eq("id", skill.getId()));
}
but it creates same alias for skill and it gives me no result.
sql is and (skill1_.ID=? and skill1_.ID=?)
Can anyone help me with this ?thanks
UPDATE:
HQL Query will be like:
select a from Job a where exists (select skill1.id from Skill skill1 join skill1.jobs r where r.id=a.id and skill1.id=1) and exists (select skill2.id from Skill skill2 join skill2.jobs r where r.id=a.id and skill2.id=4)
I need Criteria based on this.

for(Skill skill:job.getSkills()){
DetachedCriteria subquery = DetachedCriteria.forClass(Skill.class,"skill");
subquery.add(Restrictions.eq("id",skill.getId()));
subquery.setProjection(Projections.property("id"));
subquery.createAlias("jobs", "job");
subquery.add(Restrictions.eqProperty("job.id", "Job.id"));
criteria.add(Subqueries.exists(subquery));
}
I managed to solve it.now it works perfect.

Kundera for Cassandra - Deleting record by row key

I'm trying to delete specific record from database by row key. But when I try to execute this query:
Query query = em.createQuery(
"DELETE FROM User u WHERE u.userId = :u");
query.setParameter("u", userID).executeUpdate();
I got this exception: "Condition = is not suported for query on row key!".
Is there any workaround, or I missing something?

What you can do as a workaround is:
Find using:
User u = em.find(User.class, userId)
and then,
em.delete(u);

Also,
http://groups.google.com/group/kundera-discuss/subscribe
can give you quick and better support to discuss issues around Kundera.
-Vivek

A possible approach to perform such a delete (in one command using Kundera, version 2.2 at the moment) is to use "native query", for example:
EntityManagerFactory emf = Persistence.createEntityManagerFactory("cassandra_pu");
EntityManager em = emf.createEntityManager();
// "table" is the table name (case sensitive) you name your table in Cassandra
String query = "delete from table where key = 'keyValue'";
// "TablePersistencyEntity" is the Kundera Persistency Entity (Class) for the "table"
Query q = em.createNativeQuery(query, TablePersistencyEntity.class);
q.getResultList();
em.close();

Get record with max id, using Hibernate Criteria

Using Hibernate's Criteria API, I want to select the record within a table with the maximum value for a given column.
I tried to use Projections, creating an alias for max(colunName), then using it in restrictions.eq(), but it keeps telling me "invalid number".
What's the correct way to do that with Hibernate?

You can use a DetachedCriteria to express a subquery, something like this:
DetachedCriteria maxId = DetachedCriteria.forClass(Foo.class)
.setProjection( Projections.max("id") );
session.createCriteria(Foo.class)
.add( Property.forName("id").eq(maxId) )
.list();
References
Hibernate Core Reference Guide
15.8. Detached queries and subqueries

I found that using addOrder and setMaxResults together worked for me.
Criteria c = session.createCriteria(Thingy.class);
c.addOrder(Order.desc("id"));
c.setMaxResults(1);
return (Thingy)c.uniqueResult();
Using the MySQL dialect, this generates a SQL prepared statement about like this (snipping out some of the fields):
select this_.id ... from Thingy this_ order by this_.id desc limit ?
I am not sure if this solution would be effective for dialects other than MySQL.

Use
addOrder(Order.desc("id"))
and fetch just the first result :)

HQL:
from Person where person.id = (select max(id) from Person)
Untested. Your database needs to understand subselects in the where clause.
Too lazy to find out if/how such a subselect can be expressed with the criteria api. Of course, you could do two queries: First fetch the max id, then the entity with that id.

The cleaner solution would also be :
DetachedCriteria criteria = DetachedCriteria.forClass(Foo.class).setProjection(Projections.max("id"));
Foo fooObj =(Foo) criteria.getExecutableCriteria(getCurrentSession()).list().get(0);

Date maxDateFromDB = null;
Session session = (Session) entityManager.getDelegate();
//Register is and Entity and assume maxDateFromDB is a column.
//Status is another entity with Enum Applied.
//Code is the Parameter for One to One Relation between Register and Profile entity.
Criteria criteria = session.createCriteria(Register.class).setProjection(Projections.max("maxDateFromDB") )
.add(Restrictions.eq("status.id", Status.Name.APPLIED.instance().getId()));
if(code != null && code > 0) {
criteria.add(Restrictions.eq("profile.id", code));
}
List<Date> list = criteria.list();
if(!CollectionUtils.isEmpty(list)){
maxDateFromDB = list.get(0);
}

To do it entirely with Detached Criteria (because I like to construct the detached criteria without a session)
DetachedCriteria maxQuery = DetachedCriteria.forClass(Foo.class)
.setProjection( Projections.max("id") );
DetachedCriteria recordQuery = DetachedCriteria.forClass(Foo.class)
.add(Property.forName("id").eq(maxId) );

For the max() function in hibernate:
criteria.setProjection(Projections.max("e.encounterId"));