SQL server - Pivot only on a part of the result

谁说我不能喝 提交于 2019-12-12 02:37:51

问题


I need to put columns in rows only for the last columns of my query's results.

The query is :

select 
     dpp.CODE_PORTEFEUILLE,
     dpr.SOCIETE_GESTION,
     lgp.CODE_GERANT,
     lgp.DEVISE_GERANT,
     lgp.CODE_INTERVENANT
from 
     GP3DBA.DESCRIPTIF_PORTEFEUILLE dpp  
          join GP3DBA.DESCRIPTIF_PORT_REPORTING dpr on dpp.CODE_PORTEFEUILLE = dpr.CODE_PORTEFEUILLE  
          join GP3DBA.LIEN_GERANT_PORTEFEUILLE lgp  on dpp.CODE_PORTEFEUILLE = lgp.CODE_PORTEFEUILLE 

RESULT:

DESIRED RESULT:

I have tried to do it myself, but without any satisfying result.


回答1:


EDIT

I don't see an easy way to do this. You actual result is in #tmpResult in this example.

select CODE_PORTEFEUILLE, SOCIETE_GESTION, count(1) as 'number'
into #columnNumber
from #tmpResult
group by CODE_PORTEFEUILLE, SOCIETE_GESTION

declare @columnMaxNumber int;

SELECT @columnMaxNumber = MAX(number) 
FROM #columnNumber

declare @tableSql = 'CREATE TABLE #finalResult( 
    CODE_PORTEFEUILLE int not null, 
    SOCIETE_GESTION varchar(10) not null'

declare @cnt int = 0;
WHILE @cnt < @columnMaxNumber 
BEGIN
   set @tableSql = @tableSql + ', CODE_GERANT' + @cnt + ' varchar(2)' 
   set @tableSql = @tableSql + ', DEVISE_GERANT' + @cnt + ' varchar(3)'
   set @tableSql = @tableSql + ', CODE_INTERVENANT' + @cnt + ' varchar(3)'
   SET @cnt = @cnt + 1;
END

set @tableSql = @tableSql + ')'

exec @tableSql 

then you just have to inject data in this table

declare cursor_portefeuille CURSOR FOR
select * from #tmpResult /*order by fields you want to order by*/

declare @code_portefeuille int, @societe_gestion varchar(3), @code_gerant varchar(2), @device_gerant varchar(3), @code_intervenant varchar(3), @gerant_index int = 0, @last_portefeuille int = 0, @last_societe varchar(3) = ''


OPEN cursor_portefeuille 

FETCH NEXT FROM cursor_portefeuille 
INTO @code_portefeuille, @societe_gestion, @code_gerant, @device_gerant, @code_intervenant

WHILE @@FETCH_STATUS = 0  
BEGIN  
    if(@last_portefeuille = @code_portefeuille and @last_societe = @societe_gestion)
    begin
        set @gerant_index = @gerant_index + 1
    end
    else
    begin
        set @last_portefeuille = @code_portefeuille
        set @last_societe = @societe_gestion
        set @gerant_index = 0
    end
    if exists(select 1 from #finalResult where CODE_PORTEFEUILLE = @code_portefeuille and SOCIETE_GESTION = @societe_gestion)
    begin
        declare @sqlQuery varchar(512) = 'update #finalResult
        set CODE_GERANT' + @gerant_index + ' = @code_gerant,
        DEVISE_GERANT' + @gerant_index + ' = @device_gerant,
        CODE_INTERVENANT' + @gerant_index + ' = @code_intervenant
        WHERE CODE_PORTEFEUILLE = @code_portefeuille and SOCIETE_GESTION = @societe_gestion'
        exec @sqlQuery;
    end
    else
    begin
         insert into #finalResult(CODE_PORTEFEUILLE, SOCIETE_GESTION, CODE_GERANT0, DEVISE_GERANT0, CODE_INTERVENANT0)
         values(@code_portefeuille, @societe_gestion, @code_gerant, @device_gerant, @code_intervenant)
    end
    FETCH NEXT FROM cursor_portefeuille 
    INTO @code_portefeuille, @societe_gestion, @code_gerant, @device_gerant, @code_intervenant
END
CLOSE cursor_portefeuille ;  
DEALLOCATE cursor_portefeuille ; 

SELECT * from #finalResult

/*Don't forget to drop temp tables */

Didn't try it, I have no actual case at reach but I know the logic works. But you should avoid doing this... It runs fairly slowly



来源:https://stackoverflow.com/questions/39390972/sql-server-pivot-only-on-a-part-of-the-result

易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!