同一查询中的DISTINCT计数和总和

Question

this is my table

id , tiket_id , agency , problem
 1  , 123     , 1      ,  x
 2  , 123     , 1      ,  y
 3  , 124     , 2      ,  z

i have a query that gives me the stats of the table based on the agency ,basically couple of counts and sums

 SELECT  COUNT(*) as total , 
 SUM(CASE WHEN `agancy` = '1' THEN 1 ELSE 0 END) as ag_1
 SUM(CASE WHEN `agancy` = '2' THEN 1 ELSE 0 END) as ag_2

but i want rows with the same ticket_id to be counted only once

for example the first two row have the same ticket id and they are actually one ticket .

i can do this for the total count

 SELECT  COUNT(distinct(ticket_id)) as total , 

i'm not sure how to do this for the SUM ? i know this doesn't works

 SUM(  DISTINCT ticket_id CASE WHEN `agancy` = '1' THEN 1 ELSE 0 END) as ag_1

Answer 1

Well, if you need only one result row:

select sum(cnt) as total, SUM(case when agency = 1 then 1 else 0 end) as ag1, SUM(case when agency = 2 then 1 else 0 end) as ag2 from (     
    SELECT COUNT(*) as cnt, agency
    from foo
    group by tiket_id,agency
) b

http://www.sqlfiddle.com/#!2/cde84/1

Note that this assumes that two different agencies will not share a ticket_id.

Edit. Well, I reread your post, and I'm not sure if you want the TOTAL number of rows in the table or just the unique ticket ids. If you just want the unique ticket ids then you can do:

select COUNT(*) as total , SUM(case when agency = 1 then 1 else 0 end) as ag1, SUM(case when agency = 2 then 1 else 0 end) as ag2 from (        
    SELECT COUNT(*) as cnt, agency
    from foo
    group by tiket_id,agency
) b;

See: http://www.sqlfiddle.com/#!2/cde84/5

Answer 2

I would say use group by

SELECT  COUNT(*) as total , 
SUM(CASE WHEN `agancy` = '1' THEN 1 ELSE 0 END) as ag_1
SUM(CASE WHEN `agancy` = '2' THEN 1 ELSE 0 END) as ag_2
from mytable
group by tiket_id