在 Mybatis (Mapper) 和 Java 中的属性 List 中分组结果

Question

I have a question about Mappers in Mybatis. I have two classes like this:

public class A{
  private String A1;
  private String A2;
  private List<B> listB;
  //getters and setters
  .
  .
  .
  public static class B {
   private String B1;
   private String B2;
   //getters and setters
   .
   .
   .

  }

}

Then I have a mapper class like this:

@Mapper
public interface ABMapper{

@Select("select b1,b2 from b where b.a1 = #{a1}")
public List<B> getBs(@Param("a1") String a1);

@Select ("select a1,a2 from a limit 100")
@Results({
@Result(property="a1", value = "a1"),
@Result(property="a2", value = "a2"),
@Result(property="listB", column="a1", many = @Many(select = "getBs"))
})
public List<A> getAs();
}

This works fine, but I know that when getAs() executes, getBs runs as many times as items have (limit 100 is an example). I wonder that if exists a way to run a query like select a.a1,a.a2,b.b1,b.b2 from aa inner join bb on a.a1 = b.a1 and then Mybatis (and Java) could group elements in List<A> and the attribute B is not empty. Perhaps, it necessary to use hash and equals in class A and B, but I don't know.

Thanks for your answers.

Answer 1

Mybatis can do that but only if you use xml mapping. The limitation of annotations in java makes it impossible to map associations with join:

You will notice that join mapping is not supported via the Annotations API. This is due to the limitation in Java Annotations that does not allow for circular references.

In this case the mapping might look like:

<resultMap id="bMap" type="B">
  <id property="b1" column="b1"/>
  <result property="b2" column="b2"/>
</resultMap>

<resultMap id="aMap" type="A">
    <id property="a1" column="a1"/>
    <result property="a2" column="a2"/>
    <collection property="listB" javaType="B" resultMap="bMap" columnPrefix="b_"/>
</resultMap>

<select id='getAs' resultMap='aMap'>
SELECT a.*, b.id B_id, b.b1 B_b1, b.b2 B_b2
FROM (
   select *
   from a
   LIMIT 100
  ) AS a
  LEFT JOIN AS b on a.a1 = b.a1
</select>

Some important notes:

1. both A and B should have some identifying field(s) configured with id element. The value in this field would be used to identify object and do what you call grouping. For table a this seems to be a1 (as you used it as a join field) and I used it in the example.
2. autoMapping="true" in resultMap might be useful if there are many fields to be mapped
3. you need to use left/right join to handle those records from table a that do not have anything in b .
4. in order to LIMIT work correctly with join you need to do it on the select that gets records from a and not on the join result otherwise you may get less than 100 records in the result if more than 100 records from b are joined.
5. it depends on the use case but usually if you use LIMIT you need to specify some order, otherwise records will be returned in unpredictable order.
6. In the older versions of mybatis there was a bug that required that column prefixes in the query should be in upper case (may be this is fixed now, I'm not sure).