Question

我正在使用Spring JPA和Hibernate构建REST API。我正在搜索2天，但没有找到解决此问题的任何解决方案。

在某些查询中，我有多个JOIN FETCH子句。当我执行查询时，我在结果集中多次包含父对象，实际上与父亲生孩子的次数一样多。
示例：

    @Query("SELECT DISTINCT p AS post," +
        " <some subquery>" +
        "FROM Post p  JOIN FETCH p.user u LEFT JOIN FETCH p.linkedUsers INNER JOIN FETCH p.track track  " 
        "WHERE ( (<some condition>) OR p.user = :me) " +
        "ORDER BY p.created DESC")
List<Object []> getData(@Param("me")User me,
                                 Pageable pageable);

例如，如果我有一个具有2个链接用户的帖子，则我的帖子将在我的结果集中至少出现2次。如果我不执行JOIN FETCH或其他JOIN，则数据会延迟加载。但这对我来说不是目标，因为它会导致性能下降。

编辑：所以我的问题是，如何执行一个查询，其中所有数据都被提取并且所有符合指定条件的帖子在我的结果集中只有一次。

编辑：

示例对象为：

 [{
    "id": 1767,
    "track": {
        "id": 1766,
        "title": "VVS",
        ...
        "streams": [
            {
                "id": 1764,
                 ....
            },
            {
                "id": 1765,
                  ...

            }
        ],
        "isrc": "DEQ021801393"
    },
    "user": {
        "id": 999998,
        "username": "My username",
         ....

    },
    "created": "2018-08-21T22:18:56.451Z",
     ...
    "linked_users": []
},
<this object another time, beacause two stream objects in track>
<many other objects two times>
...
]

编辑：

事实证明，子查询与“区别”冲突。如果我从查询中删除它们，则会得到不同的帖子。如果我编辑本机sql查询并将“ distinct”更改为“ distinct on”，则它可以正常工作。但是我认为休眠中不存在“与众不同”子句。那么有什么好主意怎么办？

我将不胜感激：）

Answer 1

在jpa 2.1中，它支持实体图以解决“重复和N + 1问题”

@NamedEntityGraph(name = "graph.Order.items", 
               attributeNodes = @NamedAttributeNode(value = "items", subgraph = "items"), 
subgraphs = @NamedSubgraph(name = "items", attributeNodes = @NamedAttributeNode("product")))

使用实体图加载数据：

Map hints = new HashMap();
hints.put("javax.persistence.fetchgraph", graph);

return this.em.find(Order.class, orderId, hints);

更多细节：

https://www.thoughts-on-java.org/jpa-21-entity-graph-part-1-named-entity/

Answer 2

如果有人对我如何解决问题感兴趣：

在忽略了这个问题几周后，我找到了解决方案。我很确定不能通过使用HQL来解决它，因此我通过使用postgreSQLs json_build_object方法编写了本机查询。所有数据均在数据库级别生成，并作为JSON对象列表获取。这样的优点是查询的性能要好得多。

查询结构：

DECLARE @DB SYSNAME = 'adventureworks'
SELECT 'Collation' as Property, DATABASEPROPERTYEX (@DB, 'Collation') as Value UNION
SELECT 'ComparisonStyle' as Property, DATABASEPROPERTYEX (@DB, 'ComparisonStyle') as Value UNION
SELECT 'Edition' as Property, DATABASEPROPERTYEX (@DB, 'Edition') as Value UNION
SELECT 'IsAnsiNullDefault' as Property, DATABASEPROPERTYEX (@DB, 'IsAnsiNullDefault') as Value UNION
SELECT 'IsAnsiNullsEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsAnsiNullsEnabled') as Value UNION
SELECT 'IsAnsiPaddingEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsAnsiPaddingEnabled') as Value UNION
SELECT 'IsAnsiWarningsEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsAnsiWarningsEnabled') as Value UNION
SELECT 'IsArithmeticAbortEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsArithmeticAbortEnabled') as Value UNION
SELECT 'IsAutoClose' as Property, DATABASEPROPERTYEX (@DB, 'IsAutoClose') as Value UNION
SELECT 'IsAutoCreateStatistics' as Property, DATABASEPROPERTYEX (@DB, 'IsAutoCreateStatistics') as Value UNION
SELECT 'IsAutoCreateStatisticsIncremental' as Property, DATABASEPROPERTYEX (@DB, 'IsAutoCreateStatisticsIncremental') as Value UNION
SELECT 'IsAutoShrink' as Property, DATABASEPROPERTYEX (@DB, 'IsAutoShrink') as Value UNION
SELECT 'IsAutoUpdateStatistics' as Property, DATABASEPROPERTYEX (@DB, 'IsAutoUpdateStatistics') as Value UNION
SELECT 'IsClone' as Property, DATABASEPROPERTYEX (@DB, 'IsClone') as Value UNION
SELECT 'IsCloseCursorsOnCommitEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsCloseCursorsOnCommitEnabled') as Value UNION
SELECT 'IsFulltextEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsFulltextEnabled') as Value UNION
SELECT 'IsInStandBy' as Property, DATABASEPROPERTYEX (@DB, 'IsInStandBy') as Value UNION
SELECT 'IsLocalCursorsDefault' as Property, DATABASEPROPERTYEX (@DB, 'IsLocalCursorsDefault') as Value UNION
SELECT 'IsMemoryOptimizedElevateToSnapshotEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsMemoryOptimizedElevateToSnapshotEnabled') as Value UNION
SELECT 'IsMergePublished' as Property, DATABASEPROPERTYEX (@DB, 'IsMergePublished') as Value UNION
SELECT 'IsNullConcat' as Property, DATABASEPROPERTYEX (@DB, 'IsNullConcat') as Value UNION
SELECT 'IsNumericRoundAbortEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsNumericRoundAbortEnabled') as Value UNION
SELECT 'IsParameterizationForced' as Property, DATABASEPROPERTYEX (@DB, 'IsParameterizationForced') as Value UNION
SELECT 'IsQuotedIdentifiersEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsQuotedIdentifiersEnabled') as Value UNION
SELECT 'IsPublished' as Property, DATABASEPROPERTYEX (@DB, 'IsPublished') as Value UNION
SELECT 'IsRecursiveTriggersEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsRecursiveTriggersEnabled') as Value UNION
SELECT 'IsSubscribed' as Property, DATABASEPROPERTYEX (@DB, 'IsSubscribed') as Value UNION
SELECT 'IsSyncWithBackup' as Property, DATABASEPROPERTYEX (@DB, 'IsSyncWithBackup') as Value UNION
SELECT 'IsTornPageDetectionEnabled' as Property, DATABASEPROPERTYEX (@DB, 'IsTornPageDetectionEnabled') as Value UNION
SELECT 'IsVerifiedClone' as Property, DATABASEPROPERTYEX (@DB, 'IsVerifiedClone') as Value UNION
SELECT 'IsXTPSupported' as Property, DATABASEPROPERTYEX (@DB, 'IsXTPSupported') as Value UNION
SELECT 'LastGoodCheckDbTime' as Property, DATABASEPROPERTYEX (@DB, 'LastGoodCheckDbTime') as Value UNION
SELECT 'LCID' as Property, DATABASEPROPERTYEX (@DB, 'LCID') as Value UNION
SELECT 'MaxSizeInBytes' as Property, DATABASEPROPERTYEX (@DB, 'MaxSizeInBytes') as Value UNION
SELECT 'Recovery' as Property, DATABASEPROPERTYEX (@DB, 'Recovery') as Value UNION
SELECT 'ServiceObjective' as Property, DATABASEPROPERTYEX (@DB, 'ServiceObjective') as Value UNION
SELECT 'ServiceObjectiveId' as Property, DATABASEPROPERTYEX (@DB, 'ServiceObjectiveId') as Value UNION
SELECT 'SQLSortOrder' as Property, DATABASEPROPERTYEX (@DB, 'SQLSortOrder') as Value UNION
SELECT 'Status' as Property, DATABASEPROPERTYEX (@DB, 'Status') as Value UNION
SELECT 'Updateability' as Property, DATABASEPROPERTYEX (@DB, 'Updateability') as Value UNION
SELECT 'UserAccess' as Property, DATABASEPROPERTYEX (@DB, 'UserAccess') as Value UNION
SELECT 'Version' as Property, DATABASEPROPERTYEX (@DB, 'Version') as Value

从SpringBoot执行查询：

SELECT DISTINCT ON(post.created)json(json_build_object('id',...
'user', (SELECT json_build_object('id',... 
            FROM users WHERE users.id = post.user_id)
'track', (SELECT json_build_object('id',....
     'streams',ARRAY(SELECT json_build_object('id'...
FROM
<APPORXIMATELY Same JOINS AS BEFORE>
WHERE
<CONDITIONS>

要完成这项工作，需要以下依赖项（取决于您的休眠版本）：

 List<JsonNode> decoratedPosts = (List<JsonNode>) entityManager
                .createNativeQuery(NativeQueries.getPostsByUser)
                .setParameter("userId", user.getId())
                .setParameter("me", requestor.getId())
                .setParameter("limit", pageable.getPageSize())
                .setParameter("offset", pageable.getOffset())
                .unwrap(org.hibernate.query.NativeQuery.class)
                .addScalar("json", JsonNodeBinaryType.INSTANCE).getResultList();

休眠方言需要扩展。

    <dependency>
        <groupId>com.vladmihalcea</groupId>
        <artifactId>hibernate-types-52</artifactId>
        <version>2.3.4</version>
    </dependency>

休眠JOIN FETCH-对象在结果集中多次出现

2 个答案: