Combine multiple results as columns, not rows

问题

I need to perform a number of counts on various tables in the database
and I would like to combine those counts into a single result.

Conider the following queries:

SELECT 100 As SomeCount
SELECT 200 As SomeOtherCount
SELECT 300 As YetAnotherCount

If I combine them using UNION, each of the results will be a row in the final result:

SELECT 100 As SomeCount
UNION
SELECT 200 As SomeOtherCount
UNION
SELECT 300 As YetAnotherCount

Output:

> SomeCount
> --------- 
> 100 
> 200 
> 300

What I want instead is

> SomeCount | SomeOtherCount | YetAnotherCount
> --------------------------------------------
>   100     |      200       |     300

The only other way I could think off to 'name' the results is using something like this:

SELECT 'SomeCount' As Name, 100 As Value
UNION ALL
SELECT 'SomeOtherCount', 200
UNION ALL
SELECT 'YetAnotherCount', 300

In which case the result looks like:

>     Name          |      Value
> ---------------------------------
> 'SomeCount'       |       100
> 'SomeOtherCount'  |       200
> 'YetAnotherCount' |       300

Is there a way to get the results I want, or is the last method the way to go?

I should mention the queries above are very simple in order to explain the core problem. In reality two queries that need to be combined might look like this:

Query 1:

SELECT Count(Id) As UndeliveredSms
FROM
(
 SELECT Id
 FROM IncomingSms
 WHERE Id NOT IN (SELECT IncomingSmsId
                  FROM DeliveryAttempt
                  WHERE Status = 'Delivered' OR Status = 'FailedPermanently')
 )

Query 2:

SELECT Count(Id) As UndeliveredEMail FROM
(
 SELECT Id
 FROM IncomingEMail
 WHERE Id NOT IN (SELECT IncomingEMailId
                  FROM DeliveryAttempt
                  WHERE Status = 'Delivered' OR Status = 'FailedPermanently')
)

Wrapping these in another SELECT statement does not work with SQlite.

Using the last method in the examples does work, and I might go with that solution unless it's a bad idea:

SELECT 'UndeliveredSms' As Name,  Count(Id) As Value
FROM
(
 SELECT Id
 FROM IncomingSms
 WHERE Id NOT IN (SELECT IncomingSmsId
                  FROM DeliveryAttempt
                  WHERE Status = 'Delivered' OR Status = 'FailedPermanently')
 )

UNION

SELECT 'UndeliveredEMail', Count(Id) FROM
(
 SELECT Id
 FROM IncomingEMail
 WHERE Id NOT IN (SELECT IncomingEMailId
                  FROM DeliveryAttempt
                  WHERE Status = 'Delivered' OR Status = 'FailedPermanently')
)

Which results in something like this:

>     Name           |     Value
> ---------------------------------
> UndeliveredEMail   |       82
> UndeliveredSms     |       0

And of course, in reality, there are a lot more things to count

回答1:

You should be able to use a CROSS JOIN between the queries:

SELECT *
FROM
(
  SELECT Count(Id) As UndeliveredSms
  FROM
  (
   SELECT Id
   FROM IncomingSms
   WHERE Id NOT IN (SELECT IncomingSmsId
                    FROM DeliveryAttempt
                    WHERE Status = 'Delivered' OR Status = 'FailedPermanently')
  )
)
CROSS JOIN
(
  SELECT Count(Id) As UndeliveredEMail FROM
  (
   SELECT Id
   FROM IncomingEMail
   WHERE Id NOT IN (SELECT IncomingEMailId
                    FROM DeliveryAttempt
                    WHERE Status = 'Delivered' OR Status = 'FailedPermanently')
  )
);

See SQL Fiddle with Demo

回答2:

I don't understand the use of the UNION in your queries, but perhaps I'm misunderstanding. Not in a subset union not in all = not in a subset...

Seems like it can be simplified to the following:

select *
from
(
  SELECT Count(Id) As UndeliveredSms
  FROM IncomingSms
  WHERE Id NOT IN (SELECT IncomingSmsId
                    FROM DeliveryAttempt
                    WHERE Status = 'Delivered' OR Status = 'FailedPermanently')
) t, 
(
  SELECT Count(Id) As UndeliveredEMail 
  FROM IncomingEMail
  WHERE Id NOT IN (SELECT IncomingEMailId
                    FROM DeliveryAttempt
                    WHERE Status = 'Delivered' OR Status = 'FailedPermanently')
  ) t2

来源：https://stackoverflow.com/questions/17667599/combine-multiple-results-as-columns-not-rows

标签

sql

sqlite