@JoinColumnsOrFormulas實現附屬表的多字段查詢

/**
 * 產品分類
 * @author simm
 */
@Entity
@Table
@Data
@NoArgsConstructor
@Where(clause = "is_deleted = 0")
public class ProductCategory extends FullAuditedAggregateRoot {

    @Id
    @GeneratedValue(strategy = GenerationType.IDENTITY)
    private Long id;

    /**
     * 圖標
     */
    @OneToOne(cascade = CascadeType.REMOVE,fetch = FetchType.LAZY)
 @JoinColumnsOrFormulas(value = { @JoinColumnOrFormula(column = @JoinColumn(name="id",referencedColumnName = "refId",insertable = false,updatable = false)), @JoinColumnOrFormula(formula = @JoinFormula(value = "1",referencedColumnName = "refModule")), @JoinColumnOrFormula(formula = @JoinFormula(value = "1",referencedColumnName = "refType")), @JoinColumnOrFormula(formula = @JoinFormula(value = "1",referencedColumnName = "isAck")),
 }) private FileRecord icon;
}

 @JoinColumnOrFormula 的 column 屬性，用於設置主表與關聯表的主外鍵關聯關系，formula允許寫表達式（做多字段的常量查詢）。上述的JPA設置方式，相關與sql語句的關聯查詢，類似於

select file_record.* from product_category as a
inner join file_record as b on a.id = b.ref_id and b.ref_module=1 and b.ref_type=1 and is_ack=1

事實上，JPA的查詢並不會因為我們設置了@JoinColumn而生成一個關聯查詢語句獲取實體，它是分兩次發起請求。

-- 1、先查詢主表 ----
select * from product_category where id = 3;
-- 2、再查詢明細表 -------
select * from file_record where ref_id=3 and ref_module=1 and ref_type=1 and is_ack=1;

當然，這樣的做法對框架來說，取數是最方便的，不管與附屬表是OneToOne 還是 ManyToOne還是ManyToMany的關系，主實體的查詢永遠是一次檢索就能正確加載。不讀視圖就沒有主表信息重復的問題。對JPA來講，@JoinColumn的聲明對於附屬表的級聯更新就有了依據，數據就可以實現級聯保存。