Is it possible to get several COUNT in one single SQL request? - sql

Question

I need help to write a simple procedure. Let me explain what I'm trying to do.
I have 3 tables
tJobOffer
tApplication
tApplicationStatus
I would like to create a procedure that return me a list of tJobOffer with the statistics of different status of this tJobOffer. tApplicationStatus is linked to tApplication that is linked to tJobOffer. An application can be CANDIDATE / ACCEPTED / REFUSED / IGNORED / ...
I created this query :
SELECT
[T].[JobOfferId],
[T].[JobOfferTitle],
COUNT([A].[ApplicationId]) AS [CandidateCount]
FROM [tJobOffer] AS [T]
LEFT JOIN [tApplication] AS [A]
INNER JOIN [tApplicationStatus] AS [S]
ON [S].[ApplicationStatusId] = [A].[ApplicationStatusId]
AND [S].[ApplicationStatusTechnicalName] = 'CANDIDATE'
ON [A].[JobOfferId] = [T].[JobOfferId]
GROUP BY
[T].[JobOfferId],
[T].[JobOfferTitle]
ORDER BY [T].[JobOfferTitle] ;
The result is
> 52ED7C67-21E1-49BB-A1F8-0601E6EED1EA Announce a 0
> F26B228D-0C81-4DA8-A287-F8F997CC1F9C Announce b 0
> 9DA60B23-F113-4C7F-9707-2B90C1556D5D Announce c 2
> 258E11A7-79C1-47B6-8C61-413AA54E2360 Announce d 0
> DA582383-5DF4-4E1D-837C-382371BDEF57 Announce e 1
The result is correct. I get my tJoboffers with statistic on status candidate. I have 2 candidates for Announce c and 1 candidate for announce e. If I change my string 'CANDIDATE' to 'ACCEPTED' or 'REFUSED' I can get the statistic on these status. Is it possible to get everything in one request?
Something like
> 52ED7C67-21E1-49BB-A1F8-0601E6EED1EA Announce a 0 0 2
> F26B228D-0C81-4DA8-A287-F8F997CC1F9C Announce b 0 0 1
> 9DA60B23-F113-4C7F-9707-2B90C1556D5D Announce c 2 0 0
> 258E11A7-79C1-47B6-8C61-413AA54E2360 Announce d 0 0 0
> DA582383-5DF4-4E1D-837C-382371BDEF57 Announce e 1 1 0

Answer 1

use SUM and CASE
SELECT
[T].[JobOfferId],
[T].[JobOfferTitle],
SUM(CASE WHEN [S].[ApplicationStatusTechnicalName] = 'CANDIDATE' THEN 1 ELSE 0 END) AS [CandidateCount],
SUM(CASE WHEN [S].[ApplicationStatusTechnicalName] = 'ACCEPTED' THEN 1 ELSE 0 END) AS [ACCEPTEDCount],
SUM(CASE WHEN [S].[ApplicationStatusTechnicalName] = 'REFUSED' THEN 1 ELSE 0 END) AS [REFUSEDCount]
FROM [tJobOffer] AS [T]
LEFT JOIN [tApplication] AS [A]
ON [A].[JobOfferId] = [T].[JobOfferId]
LEFT JOIN [tApplicationStatus] AS [S]
ON [S].[ApplicationStatusId] = [A].[ApplicationStatusId]
GROUP BY
[T].[JobOfferId],
[T].[JobOfferTitle]
ORDER BY [T].[JobOfferTitle] ;

Answer 2

Yes, it is. One way to do that is to use the PIVOT function. The other way to do this would be to use LEFT OUTER JOIN each time you need a count of items, something like that:
SELECT a.JobID, COUNT(b.JobID), COUNT(c.JobID)
FROM AllVacancies as a
LEFT OUTER JOIN
(SELECT JobID from AllVacancies WHERE ApplicationStatus = 'CANDIDATE') as b
ON a.JobID = b.JobID
LEFT OUTER JOIN
(SELECT JobID FROM AllVacancies WHERE ApplicationStatus = 'ACCEPTED') as c
ON a.JobID = cJobID
as many times as the categories that you need.

Answer 3

Yes you can carry as many counts as you want
try this
SELECT COUNT(1),COUNT(2) FROM demoTable;
this will give you the count of no of rows in column 1 and column two
usually this will result the same count unless you have any null values allowesd and existing in any of the column.
If any column has any null value then its count may differ , so basically the idea is to apply count on the primary Key column .
Select count(*) from demoTable ;
this line also results in count values but it applies for the complete table , so performance wise applying count on any particular column is better .
again on the accuracy issue this must be applied on the column with primary key or not null constraint .
moving further , you need not to restrain to a single table
SELECT COUNT(1),COUNT(2) FROM ( joins or any selection from any no of table);
just be aware of the no of columns existing in the selection set