MySQL 实现row_number() over(partition by ) 分组排序功能
语法格式:
row_number() over(partition by 分组列 order by 排序列 desc)
row_number() over()分组排序功能:
在使用 row_number() over()函数时候,over()里头的分组以及排序的执行晚于 where 、group by,但不晚于 order by 的执行。
创建测试环境
在线数据库 http://sqlfiddle.com/
1. 创建表
-- 创建表
CREATE TABLE `a` (
`ID` INT(10) NULL DEFAULT NULL,
`class` INT(10) NULL DEFAULT NULL,
`score` INT(10) NULL DEFAULT NULL
)COLLATE='utf8_general_ci';
-- 插入数据
insert into a values (1,1,90);
insert into a values (2,1,70);
insert into a values (3,1,90);
insert into a values (4,1,80);
insert into a values (5,2,100);
insert into a values (6,2,80);
insert into a values (7,2,110);
insert into a values (8,2,80);
insert into a values (9,2,80);
insert into a values (10,2,60);
commit;
2. Oracle row_number() over(partition by) 分组排序功能
select *
from (select *,
row_number() over(partition by class order by score desc) rn
from a) b
where rn = 1
3. MySQL自定义实现row_number() over(partition by) 分组排序功能
select id, class, score, rank
from (select b.*,
-- 定义用户变量@rownum来记录数据的行号。通过赋值语句@rownum := @rownum+1来累加达到递增行号。
@rownum := @rownum + 1,
-- 如果当前分组编号和上一次分组编号相同,则@rank(对每一组的数据进行编号)值加1,否则表示为新的分组,从1开始
if(@pdept = b.class, @rank := @rank + 1, @rank := 1) as rank,
-- 定义变量@pdept用来保存上一次的分组id
@pdept := b.class
-- 这里的排序不确定是否需要,保险点还是加上吧
from (select * from a order by a.class, a.score desc) b,
-- 初始化自定义变量值
(select @rownum := 0, @pdept := null, @rank := 0) c
-- 该排序必须,否则结果会不对
order by b.class, b.score desc
) result
having rank < 2;
实例
Oracle
select app_xxzjbh, create_date, token_id, token_code, token_expire_time
from (select g.app_xxzjbh,
g.create_date,
t.token_id,
t.token_code,
t.token_expire_time,
(row_number()
over(partition by g.app_xxzjbh order by t.create_date desc)) rn
from RES_ESB_APP_TOKEN_GL g, RES_ESB_TOKEN t
where g.token_xxzjbh = t.token_id) e
where rn = 1
MySQL
select app_xxzjbh, token_id, token_code, token_expire_time, rank
from (select app_xxzjbh,
token_id,
token_code,
create_date,
token_expire_time,
@rownum := @rownum + 1,
if(@pdept = app_xxzjbh, @rank := @rank + 1, @rank := 1) as rank,
@pdept := app_xxzjbh
from (select g.app_xxzjbh,
g.create_date,
t.token_id,
t.token_code,
t.token_expire_time
from RES_ESB_APP_TOKEN_GL g, RES_ESB_TOKEN t
where g.token_xxzjbh = t.token_id) b,
(select @rownum := 0, @pdept := null, @rank := 0) c
order by b.app_xxzjbh, b.create_date desc
) result
having rank < 2;
参考