SQL Server: Kombinieren COUNT (*) für verschiedene Tabellen auf einmal

Question

Ich habe ein paar Abfragen, die ich in ONE Abfrage kombinieren möchte, um nicht auf den Server mehrere Male aufrufen müssen.

Ein Beispiel für die Abfragen Ich verwende:

SELECT COUNT(*) AS mailCount1 
FROM [WebContact].[dbo].[memberEmails] 
WHERE contactdatetime > '01/01/06' 
    AND contactdatetime < '02/01/06' 

SELECT COUNT(*) AS mailCount2 
FROM [WebContact].[dbo].[otherEmails] 
WHERE contactdatetime > '01/01/06' 
    AND contactdatetime < '02/01/06' 

SELECT COUNT(*) AS mailCount3 
FROM [WebContact].[dbo].[memberEmails] 
WHERE contactdatetime > '02/01/06' 
    AND contactdatetime < '03/01/06' 

SELECT COUNT(*) AS mailCount4 
FROM [WebContact].[dbo].[otherEmails] 
WHERE contactdatetime > '02/01/06' 
    AND contactdatetime < '03/01/06' 

etc etc... 

So wie die obigen Beispiele, was nur, dass Änderungen sind:

• Die FROM (memberEmails & otherEmails)

• Die > & < Monate (01/01/06, 02/01/06 | 02/01/06, 03/01/06 | etc ...)

Ist dies mit einer einzigen Abfrage möglich?

Answer 1

Verwenden Sie zuerst group by und nur zwei Abfragen verwenden:

select year(contactdatetime) as yyyy, month(contactdatetime) as mm, count(*) 
from WebContact].[dbo].[memberEmails] 
group by year(contactdatetime), month(contactdatetime); 

und:

select year(contactdatetime) as yyyy, month(contactdatetime) as mm, count(*) 
from WebContact].[dbo].[otherEmails] 
group by year(contactdatetime), month(contactdatetime); 

Dann, wenn Sie möchten, können Sie diese in einer einzigen Abfrage kombinieren:

select coalesce(me.yyyy, oe.yyyy) as yyyy, coalesce(me.mm, oe.mm) as mm, 
     coalesce(me.cnt, 0) as memberemailcnt, 
     coalesce(oe.cnt, 0) as otheremailcnt 
from (select year(contactdatetime) as yyyy, month(contactdatetime) as mm, count(*) as cnt 
     from WebContact].[dbo].[memberEmails] 
     group by year(contactdatetime), month(contactdatetime) 
    ) me full outer join 
    (select year(contactdatetime) as yyyy, month(contactdatetime) as mm, count(*) as cnt 
     from WebContact].[dbo].[otherEmails] 
     group by year(contactdatetime), month(contactdatetime) 
    ) oe 
    on me.yyyy = oe.yyyy and me.mm = oe.mm; 

Eine full outer join ist nicht erforderlich, wenn beide Tabellen Daten für alle Monate enthalten.

Answer 2

Ich habe die Syntax oder die Leistung nicht, aber man kann etwas tun,

 WITH cte (
     countvalue 
     ,description 
     ) 
    AS (
     SELECT COUNT(*) 
      ,'mailCount1' 
     FROM [WebContact].[dbo].[memberEmails] 
     WHERE contactdatetime > '01/01/06' 
      AND contactdatetime < '02/01/06' 

     UNION ALL 

     SELECT COUNT(*) 
      ,'mailCount2' 
     FROM [WebContact].[dbo].[otherEmails] 
     WHERE contactdatetime > '01/01/06' 
      AND contactdatetime < '02/01/06' 

     UNION ALL 

     SELECT COUNT(*) 
      ,'mailCount3' 
     FROM [WebContact].[dbo].[memberEmails] 
     WHERE contactdatetime > '02/01/06' 
      AND contactdatetime < '03/01/06' 

     UNION ALL 

     SELECT COUNT(*) 
      ,'mailCount4' 
     FROM [WebContact].[dbo].[otherEmails] 
     WHERE contactdatetime > '02/01/06' 
      AND contactdatetime < '03/01/06' 
     ) 

    SELECT mailCount1 
    ,mailCount2 
    ,mailCount3 
    ,mailCount4 
FROM (
    SELECT countvalue 
     ,description 
    FROM cte 
    ) d 
pivot(max(countvalue) FOR description IN (mailCount1, mailCount2, mailCount3, mailCount4)) piv; 

this helps ..

Answer 3

declare @emailCount table(tablename varchar(20), year int, month int, qty int) 

insert into @emailCount 
select 'memberEmails', year(contactdatetime), month(contactdatetime), count(*) 
from [WebContact].[dbo].[memberEmails] 
group by year(contactdatetime), month(contactdatetime) 

insert into @emailCount 
select 'otherEmails',year(contactdatetime), month(contactdatetime), count(*) 
from [WebContact].[dbo].[otherEmails] 
group by year(contactdatetime), month(contactdatetime) 

select tablename, year, month, qty from @emailCount 

WHERE Klausel hinzufügen, wenn Datumsbereiche beschränken benötigt. (Edit - vereinfacht, Jahr() und Monat() Funktionen zu verwenden.)