Spring Data JPA 分页 HHH000104

Question

我得到了这个存储库代码：

@Query(value = "select distinct r from Reference r " +
        "inner join fetch r.persons " +
        "left outer join fetch r.categories " +
        "left outer join fetch r.keywords " +
        "left outer join fetch r.parentReferences",
        countQuery = "select count(distinct r.id) from Reference r " +
                "inner join r.persons " +
                "left outer join r.categories " +
                "left outer join r.keywords " +
                "left outer join r.parentReferences")
Page<Reference> findsAllRelevantEntries(Pageable pageable);

当我对该查询运行测试时，我收到了 Hibernate 警告：
HHH000104: firstResult/maxResults specified with collection fetch; applying in memory!

@Test
void testFindAllRelevantAsync() throws ExecutionException, InterruptedException {
    CompletableFuture<Page<Reference>> all = referenceService.findAllRelevantAsync(PageRequest.of(1, 20));
    CompletableFuture.allOf(all).join();
    assertThat(all.get()).isNotNull();
    assertThat(all.get()).isNotEmpty();
}

存储库代码封装在此处未显示的服务方法中。它（服务方法）只是将来自服务的调用编组到存储库并返回。

此外，生成的 sql 查询不会生成 limit 子句。虽然它确实触发了两个查询。

一个用于count，另一个用于获取所有记录。
因此它会获取所有记录并在内存中应用分页。
这会导致查询执行速度非常慢。

我怎样才能使分页与这个查询一起工作？

编辑

我知道这里经常被建议作为解决方案： How can I avoid the Warning "firstResult/maxResults specified with collection fetch; applying in memory!" when using Hibernate?

有没有办法使用 Spring Data JPA 实现分页？ 我不想硬连线EntityManager，我也不想 从BasicTransformerAdapter扩展代码

Answer 1

您可以使用基于两个查询方法的通用/可重用方法。

一个用于检索实体的IDs 的 SQL 查询和一个带有 IN 谓词的第二个查询，包括来自第二个查询的 IDs。

实现自定义 Spring Data JPA Executor：

@NoRepositoryBean
public interface AsimioJpaSpecificationExecutor<E, ID extends Serializable> extends JpaSpecificationExecutor<E> {

  Page<ID> findEntityIds(Pageable pageable);
}


public class AsimioSimpleJpaRepository<E, ID extends Serializable> extends SimpleJpaRepository<E, ID>
        implements AsimioJpaSpecificationExecutor<E, ID> {

  private final EntityManager entityManager;
  private final JpaEntityInformation<E, ID> entityInformation;

  public AsimioSimpleJpaRepository(JpaEntityInformation<E, ID> entityInformation, EntityManager entityManager) {
    super(entityInformation, entityManager);
    this.entityManager = entityManager;
    this.entityInformation = entityInformation;
  }

  @Override
  public Page<ID> findEntityIds(Pageable pageable) {
    CriteriaBuilder criteriaBuilder = this.entityManager.getCriteriaBuilder();
    CriteriaQuery<ID> criteriaQuery = criteriaBuilder.createQuery(this.entityInformation.getIdType());
    Root<E> root = criteriaQuery.from(this.getDomainClass());

    // Get the entities ID only
    criteriaQuery.select((Path<ID>) root.get(this.entityInformation.getIdAttribute()));

    // Update Sorting
    Sort sort = pageable.isPaged() ? pageable.getSort() : Sort.unsorted();
    if (sort.isSorted()) {
      criteriaQuery.orderBy(toOrders(sort, root, criteriaBuilder));
    }

    TypedQuery<ID> typedQuery = this.entityManager.createQuery(criteriaQuery);

    // Update Pagination attributes
    if (pageable.isPaged()) {
      typedQuery.setFirstResult((int) pageable.getOffset());
      typedQuery.setMaxResults(pageable.getPageSize());
    }

    return PageableExecutionUtils.getPage(typedQuery.getResultList(), pageable,
      () -> executeCountQuery(this.getCountQuery(null, this.getDomainClass())));
  }

  protected static long executeCountQuery(TypedQuery<Long> query) {
    Assert.notNull(query, "TypedQuery must not be null!");

    List<Long> totals = query.getResultList();
    long total = 0L;

    for (Long element : totals) {
      total += element == null ? 0 : element;
    }

    return total;
  }
}

您可以在https://tech.asimio.net/2021/05/19/Fixing-Hibernate-HHH000104-firstResult-maxResults-warning-using-Spring-Data-JPA.html阅读更多内容

Answer 2

我自己找到了解决方法。基于此：
How can I avoid the Warning "firstResult/maxResults specified with collection fetch; applying in memory!" when using Hibernate?

首先：通过分页获取ID：

@Query(value = "select distinct r.id from Reference r " +
        "inner join r.persons " +
        "left outer join r.categories " +
        "left outer join r.keywords " +
        "left outer join r.parentReferences " +
        "order by r.id",
        countQuery = "select count(distinct r.id) from Reference r " +
                "inner join r.persons " +
                "left outer join r.categories " +
                "left outer join r.keywords " +
                "left outer join r.parentReferences " +
                "order by r.id")
Page<UUID> findsAllRelevantEntriesIds(Pageable pageable);

第二：使用 Id 进行in 查询

@Query(value = "select distinct r from Reference r " +
        "inner join fetch r.persons " +
        "left outer join fetch r.categories " +
        "left outer join fetch r.keywords " +
        "left outer join fetch r.parentReferences " +
        "where r.id in ?1 " +
        "order by r.id",
        countQuery = "select count(distinct r.id) from Reference r " +
                "inner join r.persons " +
                "left outer join r.categories " +
                "left outer join r.keywords " +
                "left outer join r.parentReferences ")
@QueryHints(value = {@QueryHint(name = "hibernate.query.passDistinctThrough", value = "false")},
        forCounting = false)
List<Reference> findsAllRelevantEntriesByIds(UUID[] ids);

注意： 我得到的是List<Reference 而不是Pageable，所以你必须像这样自行构建Pageable：

private Page<Reference> processResults(Pageable pageable, Page<UUID> result) {
    List<Reference> references = referenceRepository.findsAllRelevantEntriesByIds(result.toList().toArray(new UUID[0]));
    return new PageImpl<>(references, pageable, references.size());
}

这看起来不太好，做了两个语句，但它使用limit 进行查询，所以只获取需要的记录。

Answer 3

先获取 id 然后进行主查询的方法有效，但效率不高。我认为这是Blaze-Persistence 的完美用例。

Blaze-Persistence 是基于 JPA 的查询构建器，它支持 JPA 模型之上的许多高级 DBMS 功能。它附带的分页支持可以处理您可能遇到的所有问题。

它还集成了 Spring Data，因此您可以像现在一样使用相同的代码，只需添加依赖项并进行设置：https://persistence.blazebit.com/documentation/entity-view/manual/en_US/index.html#spring-data-setup

Blaze-Persistence 有许多不同的分页策略可供您配置。默认策略是将 id 查询内联到主查询中。像这样的：

select r 
from Reference r 
inner join r.persons
left join fetch r.categories 
left join fetch r.keywords
left join fetch r.parentReferences 
where r.id IN (
  select r2.id
  from Reference r2
  inner join r2.persons
  order by ...
  limit ...
)
order by ...