Best way to do nested case statement logic in SQL

2019-01-09 22:54发布

I'm currently writing an SQL Query, where a few of the columns returned need to be calculated depending on quite a lot of conditions.

I'm currently using nested case statements, but its getting messy. Is there a better (more organised and/or readable) way?

(I am using Microsoft SQL Server, 2005)


A simplified example:

SELECT
    col1,
    col2,
    col3,
    CASE
        WHEN condition 
        THEN
            CASE
                WHEN condition1 
                THEN
                    CASE 
                        WHEN condition2
                        THEN calculation1
                        ELSE calculation2
                    END
                ELSE
                    CASE 
                        WHEN condition2
                        THEN calculation3
                        ELSE calculation4
                    END
            END
        ELSE 
            CASE 
                WHEN condition1 
                THEN 
                    CASE
                        WHEN condition2 
                        THEN calculation5
                        ELSE calculation6
                    END
                ELSE
                    CASE
                        WHEN condition2 
                        THEN calculation7
                        ELSE calculation8
                    END
            END            
    END AS 'calculatedcol1',
    col4,
    col5 -- etc
FROM table

8条回答
ら.Afraid
2楼-- · 2019-01-09 23:28

You could try some sort of COALESCE trick, eg:

SELECT COALESCE(
  CASE WHEN condition1 THEN calculation1 ELSE NULL END,
  CASE WHEN condition2 THEN calculation2 ELSE NULL END,
  etc...
)
查看更多
小情绪 Triste *
3楼-- · 2019-01-09 23:29

a user-defined function may server better, at least to hide the logic - esp. if you need to do this in more than one query

查看更多
劳资没心,怎么记你
4楼-- · 2019-01-09 23:34

Here's a simple solution to the nested "Complex" case statment: --Nested Case Complex Expression

select  datediff(dd,Invdate,'2009/01/31')+1 as DaysOld, 
    case when datediff(dd,Invdate,'2009/01/31')+1 >150 then 6 else
        case when datediff(dd,Invdate,'2009/01/31')+1 >120 then 5 else 
            case when datediff(dd,Invdate,'2009/01/31')+1 >90 then 4 else 
                case when datediff(dd,Invdate,'2009/01/31')+1 >60 then 3 else 
                    case when datediff(dd,Invdate,'2009/01/31')+1 >30 then 2 else 
                        case when datediff(dd,Invdate,'2009/01/31')+1 >30 then 1 end 
                    end
                end
            end
        end
    end as Bucket
from rm20090131atb

Just make sure you have an end statement for every case statement

查看更多
Juvenile、少年°
5楼-- · 2019-01-09 23:35

I personally do it this way, keeping the embedded CASE expressions confined. I'd also put comments in to explain what is going on. If it is too complex, break it out into function.

SELECT
    col1,
    col2,
    col3,
    CASE WHEN condition THEN
      CASE WHEN condition1 THEN
        CASE WHEN condition2 THEN calculation1
        ELSE calculation2 END
      ELSE
        CASE WHEN condition2 THEN calculation3
        ELSE calculation4 END
      END
    ELSE CASE WHEN condition1 THEN 
      CASE WHEN condition2 THEN calculation5
      ELSE calculation6 END
    ELSE CASE WHEN condition2 THEN calculation7
         ELSE calculation8 END
    END AS 'calculatedcol1',
    col4,
    col5 -- etc
FROM table
查看更多
地球回转人心会变
6楼-- · 2019-01-09 23:46

Wrap all those cases into one.


SELECT
    col1,
    col2,
    col3,
    CASE
        WHEN condition1 THEN calculation1 
        WHEN condition2 THEN calculation2
        WHEN condition3 THEN calculation3
        WHEN condition4 THEN calculation4
        WHEN condition5 THEN calculation5
        ELSE NULL         
    END AS 'calculatedcol1',
    col4,
    col5 -- etc
FROM table
查看更多
Anthone
7楼-- · 2019-01-09 23:47

I went through this and found all the answers super cool, however wants to add to answer given by @deejers

    SELECT
    col1,
    col2,
    col3,
    CASE
        WHEN condition1 THEN calculation1 
        WHEN condition2 THEN calculation2
        WHEN condition3 THEN calculation3
        WHEN condition4 THEN calculation4
        WHEN condition5 THEN calculation5         
    END AS 'calculatedcol1',
    col4,
    col5 -- etc
FROM table

you can make ELSE optional as its not mandatory, it is very helpful in many scenarios.

查看更多
登录 后发表回答