解题思路：
1、1-N排名，重复的值排名相同，将salary去重后倒序排列，然后row_number得出不重复salary的1-N的排名；
2、原表再与结果集left join

select a.emp_no,a.salary,b.t_rank
from salaries a
left join (select salary,
row_number() over(order by salary desc) as t_rank
from salaries
group by salary) b
on a.salary=b.salary
order by b.t_rank,a.emp_no;