Bootstrap

分组查询取每组最新的数据(order by 和group by使用问题)

1.第一种方法: 

2:第二种方法: 

表结构
CREATE TABLE cy_biz_message
(
 id                INT AUTO_INCREMENT
 COMMENT '主键'
   PRIMARY KEY,
 send_id           INT                                 NOT NULL
 COMMENT '发送者',
 receive_id        INT                                 NOT NULL
 COMMENT '接收者',
 content           VARCHAR(500)                        NULL
 COMMENT '信息内容',
 `read`            TINYINT(1) DEFAULT '0'              NOT NULL
 COMMENT '信息已读',
 type              INT DEFAULT '2'                     NOT NULL
 COMMENT '类型(1:系统信息;2:普通信息)',
 sort              INT                                 NULL
 COMMENT '排序',
 deleted           CHAR DEFAULT '0'                    NOT NULL
 COMMENT '是否删除 0否 1是',
 enabled           CHAR DEFAULT '1'                    NOT NULL
 COMMENT '是否启用: 0否, 1是',
 creation_date     TIMESTAMP DEFAULT CURRENT_TIMESTAMP NOT NULL
 COMMENT '创建时间',
 created_by        INT                                 NULL
 COMMENT '创建人',
 last_updated_date TIMESTAMP DEFAULT CURRENT_TIMESTAMP NOT NULL
 COMMENT '最后修改时间',
 last_updated_by   INT                                 NULL
 COMMENT '最后修改人',
 login_count       INT DEFAULT '0'                     NULL
)
 COMMENT '消息记录'
 ENGINE = InnoDB;

CREATE INDEX cy_biz_message_send_id_index
 ON cy_biz_message (send_id);

CREATE INDEX cy_biz_message_receive_id_index
 ON cy_biz_message (receive_id);
错误写法:
SELECT * FROM
  (SELECT a.id, a.send_id, b.fullname AS sendName, a.receive_id, c.fullname AS receiveName, a.content, a.creation_date
FROM cy_biz_message a
  LEFT JOIN sys_user b ON a.send_id = b.id
  LEFT JOIN sys_user c ON a.receive_id = c.id
WHERE a.receive_id = 1025
ORDER BY a.creation_date DESC) AS t
GROUP BY t.send_id ORDER BY t.creation_date DESC;
正确写法:
SELECT * FROM
  (SELECT a.id, a.send_id, b.fullname AS sendName, a.receive_id, c.fullname AS receiveName, a.content, a.creation_date
FROM cy_biz_message a
  LEFT JOIN sys_user b ON a.send_id = b.id
  LEFT JOIN sys_user c ON a.receive_id = c.id
WHERE a.receive_id = 1025
ORDER BY a.creation_date DESC LIMIT 10000) AS t
GROUP BY t.send_id ORDER BY t.creation_date DESC;
原理分析:
我们这里使用了临时表排序,继而对其结果进行分组,结果显示失败,加了limit后结果正确,原因是因为临时表(派生表derived table)中使用order by且使其生效,必须满足三个条件:
1:外部查询禁止分组或者聚合
2:外部查询未指定having,HAVING, order by
3:外部查询将派生表或者视图作为from句中唯一指定源
不满足这三个条件,order by会被忽略。
一旦外部表使用了group by,那么临时表(派生表 derived table)将不会执行filesort操作(即order by 会被忽略)。之后我使用了limit可以使其生效,原因是因为要使派生表order by生效,派生表可以通过使用group by、limit、having、distinct等等使其生效(方法有好多,详情可看文档https://dev.mysql.com/doc/refman/5.7/en/derived-table-optimization.html)
(如有发现文章错误,请在评论处指出,谢谢。)

 单表也可以这样:

SELECT * FROM  (SELECT * FROM banners ORDER BY id DESC LIMIT 1000)  AS  t GROUP BY t.title

 Tp写法:

$subQuery = Db::table('log log')
              ->where('log.operation_id',$operationId)
              ->limit(10000000000) //不加这一行代码就有可能出现不是最新的代码
              ->buildSql();//构建查询语句


$res =  Db::table($subQuery.'a')
             ->join('user u','a.uid=u.id')
             ->field('a.*,u.username,u.logo')
             ->order('a.add_time','desc')
             ->group('uid')
             ->limit(($page - 1) * $page_size , $page_size)
             ->select();

 

悦读

道可道,非常道;名可名,非常名。 无名,天地之始,有名,万物之母。 故常无欲,以观其妙,常有欲,以观其徼。 此两者,同出而异名,同谓之玄,玄之又玄,众妙之门。

;