SQL - Ugly combination of GROUP BY and COALESCE

2019-02-24 18:14发布

I have a table with data similar to the following:

[ID], [State], [foo], [DateCreated], [DateUpdated]

The longer I work on this, the uglier my SQL is getting, which tells me I'm probably doing something wrong. What I want is a unique list of each State so long as foo is always the same for that State (if foo is not the same for all records in that State, I don't want that State at all). Also, I want to COALESCE DateCreated and DateUpdated and want the maximum value for that State.

So given this data:

[ID], [State], [foo], [DateCreated], [DateUpdated]
1,  MA, data1,  05/29/2012, 06/02/2012
2,  MA, data1,  05/29/2012, 06/03/2012
3,  RI, data2,  05/29/2012, NULL
4,  RI, data3,  05/29/2012, NULL
5,  NH, data4,  05/29/2012, NULL
6,  NH, data4,  05/29/2012, 06/05/2012

I'd like only these results:

[State], [foo], [LastUpdated]
MA, data1,  06/03/2012
NH, data4,  06/05/2012

What's the most elegant way to get what I'm after?

5条回答
孤傲高冷的网名
2楼-- · 2019-02-24 18:24

Not as elegant, but for you poor SQL 2000 souls:

SELECT T1.State, T2.Foo, T1.LastUpdated
FROM (
    SELECT State, MAX([ID]) AS [ID], 
        MAX(COALESCE(DateUpdated, DateCreated)) AS LastUpdated
    FROM YourTable
    GROUP BY State
    HAVING COUNT(DISTINCT Foo) = 1
) T1 
INNER JOIN YourTable T2 ON T1.State = T2.State AND T1.[ID] = T2.[ID]
查看更多
Lonely孤独者°
3楼-- · 2019-02-24 18:27

Assuming you are using SQL Server 2005 or >

Try this:

WITH Data AS
(
    SELECT  *,
        COALESCE([DateCreated], [DateUpdated]) AS LastUpdated,
        ROW_NUMBER() OVER(PARTITION BY State ORDER BY COALESCE([DateCreated], [DateUpdated]) DESC) Position
      FROM <YOUR-TABLE> a
     WHERE NOT EXISTS
     (
        SELECT  1 
            FROM    <YOUR-TABLE> b
         WHERE  a.State = b.State
            AND a.foo <> b.foo
     )
)
SELECT State, foo, LastUpdated
  FROM Data
 WHERE Positon = 1
查看更多
霸刀☆藐视天下
4楼-- · 2019-02-24 18:34

Another one:

http://sqlfiddle.com/#!6/fd219/1

SELECT
  t.State,
  MAX(t.foo),
  MAX( COALESCE( t.DateUpdated, t.DateCreated ))
FROM t
GROUP BY t.State
HAVING COUNT(DISTINCT t.foo) = 1;
查看更多
等我变得足够好
5楼-- · 2019-02-24 18:43

A simple Group by with nested queries should suffice:

Select State, coalesce(max_created,max_updated) from (
  Select State, min(foo) as min_foo, max(foo) as max_foo, 
    max(DateCreated) as max_created,
    max(DateUpdated) as max_updated
  From Data
  Group by State)
 Where min_foo = max_foo
查看更多
Bombasti
6楼-- · 2019-02-24 18:45

Try this:

select state_name,foo,max(dateUpdated) from state where state_name in (select state_name from state group by state_name having count(distinct foo)=1) group by state_name,foo;
查看更多
登录 后发表回答