Pagination with Hibernate Criteria and DISTINCT_ROOT_ENTITY

Question

I\'ve have already implemented pagination using the following code:

public Paginacao consultarPaginado(int pagina, Integer cidadeId) {

               


        

Answer 1

I'm not sure if I understand your question correctly, but if your query retrieves entities with toMany associations join-fetched, then pagination won't work as expected.

Indeed, suppose you have 4 A entities, each with 3 B entities, and suppose your query retrieves all the A entities with their Bs.

In that case, the SQL query will return 12 rows. If you use setMaxResults(7), it will retrieve (for example) three rows for A1 and its Bs, three rows for A2 and its Bs, and just 1 row for A3 and its first B.

And since you have used DISTINCT_ROOT_ENTITY, the criteria query will return only three entities: A1, A2, and A3 (which will have an incomplete set of Bs).

Answer 2

After a long time looking for a solution for my problem I managed to solve it. The problem that if you create a criteria or query that retrieves toMany associations using JOINS, and then you use setMaxResults and set the ResultTransformer to DISTINCT_ROOT_ENTITY the result won't be as you expected.

As JB Nizet said, suppose you have 4 A entities, each with 3 B entities, and suppose your query retrieves all the A entities with their Bs.

In that case, the SQL query will return 12 rows. If you use setMaxResults(7), it will retrieve (for example) three rows for A1 and its Bs, three rows for A2 and its Bs, and just 1 row for A3 and its first B.

And since you have used DISTINCT_ROOT_ENTITY, the criteria query will return only three entities: A1, A2, and A3 (which will have an incomplete set of Bs).

To solve this, you have to set the FETCH MODE for toMany (usually collections) relationships to SELECT or SUBSELECT, and you have basically 2 ways to achieve this:

The first way is to use @FetchMode(FetchMode.SUBSELECT) annotation on your attribute, and I don't like this approach because it causes every query to use SUBSELECT FETCH to retrieve the collection. But it will work.

The other way is to set fetch modes for relationships when you build your query. I prefer this way, because I can customize the query to my needs and I don't have to use SUBSELECTS to all queries. So, I've done this way:

public Paginacao<Anuncio> consultarPaginado(int pagina, Integer cidadeId) {

        Criteria criteria = this.sessionFactory.getCurrentSession().createCriteria(Anuncio.class);      
        criteria.add(Restrictions.eq("ativo", true));
        criteria.add(Restrictions.eq("statusLiberacao", AnunciosUtil.STATUS_ANUNCIO_LIBERADO));
        criteria.add(Restrictions.eq("statusVendaAnuncio", AnunciosUtil.STATUS_VENDA_ANUNCIO_DISPONIVEL));
        criteria.setFetchMode("imagens", FetchMode.SELECT);
        criteria.setFetchMode("pagamentos", FetchMode.SELECT);      

        if (cidadeId != null) {
            criteria.add(Restrictions.eq("cidade.id", cidadeId));
        }

        criteria.addOrder(Order.desc("dataPostagem"));
        criteria.setProjection(Projections.rowCount());

        Long count = (Long) criteria.uniqueResult();

        Paginacao<Anuncio> paginacao = new Paginacao<Anuncio>();
        int qtdPaginas = (count.intValue() / 7) + 1;

        paginacao.setQtdPaginas(qtdPaginas);

        criteria.setProjection(null);// reseta a criteria sem a projeção
        criteria.setResultTransformer(CriteriaSpecification.DISTINCT_ROOT_ENTITY);

        if (pagina > qtdPaginas) {
            pagina = qtdPaginas;
        }
        pagina = pagina - 1;
        criteria.setFirstResult(pagina * ConstantesGenericas.MAXIMO_OBJETOS_RETORNADOS);
        criteria.setMaxResults(ConstantesGenericas.MAXIMO_OBJETOS_RETORNADOS);

        paginacao.setRegistros(criteria.list());

        return paginacao;
    }

Hope that it helps anyone else. ;D

Answer 3

This was an issue for me, and it took some time to come up with a solution that works for all the scenarios I have.

What you want for each pagination page is 2 things, the total count of all results and your single page of results, but to do that you need to take 3 steps. 1) get the total count, 2) get the unique ids for your page, and 3) get the full data for the ids found in step 2. And you can do all that with a single criteria object:

1) get total count, using distinct ids (uniqueField = the name of your id in the entity class)

  Criteria criteria = session.createCriteria(YourEntity.class);
  Projection idCountProjection = Projections.countDistinct(uniqueField);
  criteria.setProjection(idCountProjection);
  //setup criteria, joins etc here
  int totalResultCount = ((Long)criteria.uniqueResult()).intValue();

2) reset projection and set start and length (you want the distinct ids)

  criteria.setProjection(Projections.distinct(Projections.property(uniqueField)));
  criteria.setFirstResult(start); 
  criteria.setMaxResults(length);
  List uniqueSubList = criteria.list();

3) reset projection and get distinct results that match ids

  criteria.setProjection(null);
  criteria.setFirstResult(0); criteria.setMaxResults(Integer.MAX_VALUE);
  criteria.add(Restrictions.in(uniqueField, uniqueSubList));
  criteria.setResultTransformer(CriteriaSpecification.DISTINCT_ROOT_ENTITY);
  List searchResults = criteria.list();
  //and now, however you want to return your results
  Map<String, Object> searchResultsMap = new HashMap<String, Object>();
  searchResultsMap.put("searchResults", searchResults);
  searchResultsMap.put("totalResultCount", totalResultCount);