Oracle Partition By

/*
Parttion by 关键字是Oracle中分析性函数的一部分，
它和聚合函数不同的地方在于它能够返回一个分组中的多条记录，
而聚合函数一般只有一条反映统计值的结果。
*/

--示例查询出每个部门工资最低的员工编号【每个部门可能有多个最低的工资员工】  
create table tsaler(
userid varchar(10),
salary number,
deptid varchar(1)
);

insert into tsaler(userid, salary, deptid) values('1', '200', '1');
insert into tsaler(userid, salary, deptid) values('2', '2000', '1');
insert into tsaler(userid, salary, deptid) values('3', '200', '1');
insert into tsaler(userid, salary, deptid) values('4', '1000', '2');
insert into tsaler(userid, salary, deptid) values('5', '1000', '2');
insert into tsaler(userid, salary, deptid) values('6', '3000', '2');
commit;

--方法一
select tsaler.*
  from tsaler
 inner join (select min(salary) as salary, deptid
               from tsaler
              group by deptid) c
    on tsaler.salary = c.salary
   and tsaler.deptid = c.deptid;
   
--方法二
select *
  from tsaler
 inner join (select min(salary) as salary, deptid
               from tsaler
              group by deptid) c
 using (salary, deptid);
   
--方法三
--row_number() 顺序排序
select row_number() over(partition by deptid order by salary) my_rank ,deptid,USERID,salary from tsaler;
--rank() （跳跃排序，如果有两个第一级别时，接下来是第三级别）
select rank() over(partition by deptid order by salary) my_rank,deptid,USERID,salary from tsaler;
--dense_rank（）（连续排序，如果有两个第一级别时，接下来是第二级）
select dense_rank() over(partition by deptid order by salary) my_rank,deptid,USERID,salary from tsaler;

--①  
select *
 from (select       rank() over(partition by deptid order by salary) my_rank,
              deptid,
              USERID,
              salary
         from tsaler)
where my_rank = 1;
--②
select *
  from (select dense_rank() over(partition by deptid order by salary) my_rank,
               deptid,
               USERID,
               salary
          from tsaler)
 where my_rank = 1;