In SQL Server, I have a list of integers, datetimes, and st开发者_如何学JAVArings. For example,
number datetime string
6 2011-09-22 12:34:56 nameOne
6 2011-09-22 1:23:45 nameOne
6 2011-09-22 2:34:56 nameOne
5 2011-09-22 3:45:01 nameOne
5 2011-09-22 4:56:01 nameOne
5 2011-09-22 5:01:23 nameOne
7 2011-09-21 12:34:56 nameTwo
7 2011-09-21 1:23:45 nameTwo
7 2011-09-21 2:34:56 nameTwo
4 2011-09-21 3:45:01 nameTwo
4 2011-09-21 4:56:01 nameTwo
4 2011-09-21 5:01:23 nameTwo
I would to write a SQL statement that outputs only those rows whose number is a maximum for each string. In this example,
number datetime string
6 2011-09-22 12:34:56 nameOne
6 2011-09-22 1:23:45 nameOne
6 2011-09-22 2:34:56 nameOne
7 2011-09-21 12:34:56 nameTwo
7 2011-09-21 1:23:45 nameTwo
7 2011-09-21 2:34:56 nameTwo
I know that I could loop over each string in the string column, then get the maximum for that string, then select the rows matching that maximum. For example,
declare @max int
declare my_cursor cursor fast_forward for
select distinct string
from table
open my_cursor
fetch next from my_cursor into @string
while @@fetch_status = 0
begin
set @max = (select max(number) from table where string = @string)
select * from table where number = @max
fetch next from my_cursor into @string
end
close my_cursor
deallocate my_cursor
However, I am wondering if there is a way to accomplish this task WITHOUT using a loop (e.g., by using aggregate functions and grouping).
;WITH T as
(
SELECT *,
RANK() OVER (PARTITION BY string ORDER BY number DESC) RN
FROM YourTable
)
SELECT number,
datetime,
string
FROM T
WHERE RN=1;
WITH maxes AS (
SELECT string, MAX(number) AS max_number
FROM tbl
GROUP BY string
)
SELECT tbl.*
FROM tbl
INNER JOIN maxes
ON maxes.string = tbl.string AND maxes.max_number = tbl.number
精彩评论