Mybatis Generator通用Join的实现

通常，我们使用Mybatis实现join表关联的时候，一般都是通过在xml或注解里写自定义sql实现。

本文通过Mybatis Generator的插件功能新增一个JoinPlugin插件，只要在配置文件里加上该插件就可以使用。无其他第三方依赖。如下图：

该插件符合mbg plugin即插即用的特点，不影响生成的实体类，只对生成的Example文件做少量变动（新增一个内部类）。

首选我们看一下使用效果，如果符合你的要求，请关注支持。

/**
     * 简单join查询示例
     * select t0.user_id as t0_user_id,t0.user_name as t0_user_name,t0.login_account as t0_login_account,t0.login_password as t0_login_password,
     *  t0.user_sex as t0_user_sex,t0.user_email as t0_user_email,t0.user_mobile as t0_user_mobile,t0.user_avatar as t0_user_avatar,
     *  t0.user_company as t0_user_company,t0.user_dept as t0_user_dept,t0.is_del as t0_is_del,t0.is_admin as t0_is_admin,
     *  t0.system_type as t0_system_type,t0.last_login_time as t0_last_login_time,t0.create_time as t0_create_time,t0.create_user as t0_create_user,
     *  t0.update_time as t0_update_time,t0.update_user as t0_update_user,
     *  t2.role_id as t2_role_id,t2.role_name as t2_role_name
     * from auth_user as t0
     * inner join auth_user_role as t1 on t0.user_id = t1.user_id
     * left join auth_role as t2 on t2.role_id = t1.role_id
     * where ( ( t0.user_id is not null and ( t0.create_time is not null or t0.create_user is not null ) ) ) and ( t1.user_id <>999999 and t1.role_id >=1 )
     * order by t2.role_id desc,t0.create_time
     * limit  10,100
     */
    public void joinTable() {
        AuthUserExample authUserExample=new AuthUserExample();
        AuthUserExample.Criteria criteria=authUserExample.createCriteria();
        //where条件
        criteria.andUserIdIsNotNull();
        criteria.createOrCriteria().andCreateTimeIsNotNull().andCreateUserIsNotNull();

        //关联user_role
        AuthUserRoleExample urExample=new AuthUserRoleExample();
        //where中user_role表的条件
        urExample.createCriteria().andUserIdNotEqualTo(999999).andRoleIdGreaterThanOrEqualTo(1);
        criteria.createJoinCriteria()
        .innerJoinTable(urExample)
        //使用User表的user_id关联user_role表的user_id
        .on(a->a.getUserId(), a->a.equalTo(b->b.tableInfo.getUserId()));

        //关联role
        AuthRoleExample roleExample=new AuthRoleExample();
        criteria.createJoinCriteria()
        .leftJoinTable(roleExample)
        //使用role表的role_id关联user_role表的role_id
        .on(a->a.tableInfo.getRoleId(), a->a.equalTo(b->b.tableInfo.getRoleId()), urExample)
        //先按照role表的role_id降序
        .orderByDesc(a->a.tableInfo.getRoleId())
        //再按照user表的create_time升序
        .orderByFirstTable(a->a.getCreateTime())
        //跳过前10条
        .skip(10)
        //取100条记录
        .take(100)
        //只查询role表的role_id和role_name两个字段
        .select(a->new String[] {a.tableInfo.getRoleId(),a.tableInfo.getRoleName()});

        //执行查询
        List<Tuple> userList=userDao.selectJoinByExample(authUserExample);
        int totalItemCount=(int) userDao.countJoinByExample(authUserExample);
        System.out.println(userList.size()+",total:"+totalItemCount);
        for(Tuple tuple:userList) {
            AuthUser authUser=tuple.getObject(AuthUser.class);
            AuthUserRole authUserRole=tuple.getObject(AuthUserRole.class);
            AuthRole authRole=tuple.getObject(AuthRole.class);
            totalItemCount++;
        }
    }

控制台打印内容如下：

2020-04-23 23:09:16.326  INFO 10308 --- [nio-9000-exec-4] p.r.m.i.SqlStatementInterceptor          : ===>sql:select t0.user_id as t0_user_id,t0.user_name as t0_user_name,t0.login_account as t0_login_account,t0.login_password as t0_login_password,t0.user_sex as t0_user_sex,t0.user_email as t0_user_email,t0.user_mobile as t0_user_mobile,t0.user_avatar as t0_user_avatar,t0.user_company as t0_user_company,t0.user_dept as t0_user_dept,t0.is_del as t0_is_del,t0.is_admin as t0_is_admin,t0.system_type as t0_system_type,t0.last_login_time as t0_last_login_time,t0.create_time as t0_create_time,t0.create_user as t0_create_user,t0.update_time as t0_update_time,t0.update_user as t0_update_user,t0.`type` as t0_type,t0.title as t0_title,t0.bz as t0_bz,t2.role_id as t2_role_id,t2.role_name as t2_role_name
from auth_user as t0
inner join auth_user_role as t1 on t0.user_id = t1.user_id
left join auth_role as t2 on t2.role_id = t1.role_id
where ( ( t0.user_id is not null and ( t0.create_time is not null or t0.create_user is not null ) ) ) and ( t1.user_id <>999999 and t1.role_id >=1 )
order by t2.role_id desc,t0.create_time
limit  10,100
2020-04-23 23:09:16.329 DEBUG 10308 --- [nio-9000-exec-4] c.r.m.d.A.selectJoinByExample            : ==>  Preparing: select t0.user_id as t0_user_id,t0.user_name as t0_user_name,t0.login_account as t0_login_account,t0.login_password as t0_login_password,t0.user_sex as t0_user_sex,t0.user_email as t0_user_email,t0.user_mobile as t0_user_mobile,t0.user_avatar as t0_user_avatar,t0.user_company as t0_user_company,t0.user_dept as t0_user_dept,t0.is_del as t0_is_del,t0.is_admin as t0_is_admin,t0.system_type as t0_system_type,t0.last_login_time as t0_last_login_time,t0.create_time as t0_create_time,t0.create_user as t0_create_user,t0.update_time as t0_update_time,t0.update_user as t0_update_user,t0.`type` as t0_type,t0.title as t0_title,t0.bz as t0_bz,t2.role_id as t2_role_id,t2.role_name as t2_role_name from auth_user as t0 inner join auth_user_role as t1 on t0.user_id = t1.user_id left join auth_role as t2 on t2.role_id = t1.role_id WHERE ( ( t0.user_id is not null and ( t0.create_time is not null or t0.create_user is not null ) ) ) and ( t1.user_id <>? and t1.role_id >=? ) order by t2.role_id desc,t0.create_time limit 10,100
2020-04-23 23:09:16.330 DEBUG 10308 --- [nio-9000-exec-4] c.r.m.d.A.selectJoinByExample            : ==> Parameters: 999999(Integer), 1(Integer)
2020-04-23 23:09:16.361 DEBUG 10308 --- [nio-9000-exec-4] c.r.m.d.A.selectJoinByExample            : <==      Total: 100
2020-04-23 23:09:16.367  INFO 10308 --- [nio-9000-exec-4] p.r.m.i.SqlStatementInterceptor          : c.r.m.d.AuthUserMapper.selectJoinByExample:41ms
2020-04-23 23:09:16.371  INFO 10308 --- [nio-9000-exec-4] p.r.m.i.SqlStatementInterceptor          : ===>sql:select count(*)
from auth_user as t0
inner join auth_user_role as t1 on t0.user_id = t1.user_id
left join auth_role as t2 on t2.role_id = t1.role_id
where ( ( t0.user_id is not null and ( t0.create_time is not null or t0.create_user is not null ) ) ) and ( t1.user_id <>999999 and t1.role_id >=1 )
2020-04-23 23:09:16.375 DEBUG 10308 --- [nio-9000-exec-4] c.r.m.d.A.countJoinByExample             : ==>  Preparing: select count(*) from auth_user as t0 inner join auth_user_role as t1 on t0.user_id = t1.user_id left join auth_role as t2 on t2.role_id = t1.role_id WHERE ( ( t0.user_id is not null and ( t0.create_time is not null or t0.create_user is not null ) ) ) and ( t1.user_id <>? and t1.role_id >=? )
2020-04-23 23:09:16.376 DEBUG 10308 --- [nio-9000-exec-4] c.r.m.d.A.countJoinByExample             : ==> Parameters: 999999(Integer), 1(Integer)
2020-04-23 23:09:16.385 DEBUG 10308 --- [nio-9000-exec-4] c.r.m.d.A.countJoinByExample             : <==      Total: 1
2020-04-23 23:09:16.385  INFO 10308 --- [nio-9000-exec-4] p.r.m.i.SqlStatementInterceptor          : c.r.m.d.AuthUserMapper.countJoinByExample:14ms
100,total:5190

通过上面简单例子我们可以看到，生成的sql是完全符合预期的，并且不需要我们额外的手写sql了。

该插件特点：

1.可关联多表，on多个字段关联，on多种条件查询，例如:left join t_role as t1 on t0.role_id=t1.role_id and t0.sys_type=t1.role_type and t1.is_del=0 and t1.role_type='T'

2.可自定义表别名

3.sql中的所有表都可自定义要查询的字段，表字段多时，指定字段查询明显提高效率

4.分页、多表排序支持，函数式编程写法

5.支持in,exists等子查询

6.mybatis原生特性，无其他依赖

该插件原理：

由于mbg每个表都生成一个xml，在xml里自动组装生成的sql，并返回map对象。通过mybatis的拦截器拦截返回结果，将map转换成不同表的实体对象然后组装返回到一个类Tuple中。

本插件初衷：为了简化开发人员的数据库sql繁琐查询工作及改善mybatis写join关联时只能自定义的状况。开发过程中当表结构改变时，所有手写自定义的sql里的字段都要改一遍。

另外，mbg1.4.0的dynamic-sql不怎么好用，这么久了更新这样一个新版本很失望。