Re: Count values that match elements in csv string - Mailing list pgsql-sql

sqlnewbie2015 wrote
> Hello All- 
> I am having difficulty limiting my query to only count items which match
> any part of a csv string table.  My current query runs but it seems to
> only count one of the elements listed in the csv and stops rather then
> count all the activities that match with the csv.
>  
> select (sum (kept)/60) from 
>                                      (select distinct sa.staff_id,
>                                                           
> sa.service_date, 
>                                                            sa.client_id,
>                                                           
> sa.actual_duration as kept
>                                       from rpt_scheduled_activities as sa
>                                       inner join
> rpt_staff_performance_target as spt
>                                       on sa.staff_id = spt.staff_id 
>                                       where (sa.status = 'Kept' 
>                                                 and sa.service_date
> between '01-nov-2014' and '30-nov-2014' 
>                                                 and sa.activity_name in 
>                                                                      
> (select regexp_split_to_table(spt.activity,',') 
>                                                                       
> from rpt_staff_performance_target as spt))
>                                                 ) as p
> 
>  Details: I need to calculate how many hours our staff spends seeing
> clients. Each staff has different appointments that can count toward this.
> The specified appointments for each staff are listed as comma separated
> values. My existing query calculates the appointment hours for each staff
> in a given time period. However, I need help limiting my query to only
> include specified activities for each staff. My current where clause uses
> IN to compare the appointment (i.e. activity) listed in the staff's
> schedule with what is listed an an approved appointment type (i.e.
> performance target activity). 
> 
> Thank you for your consideration and for any tips you can provide!
> 
> Cara

How about something considerably easier to read...

FROM ... sa
JOIN ... spt USING (staff_id)
WHERE sa.activity_name = ANY( regexp_split_to_array( spt.activity, ',' ) )
AND [status, service-name]

You haven't explained or shown why you believe DISTINCT is necessary so I
question its presence.

It would be a lot easier if you either show schema or, even better, provide
a query of the form

WITH sa AS ( ... ), spt AS ( ... )
SELECT ...
FROM sa JOIN spt USING (staff_id)

and provide some sample data inside the WITH/CTE constructs so that someone
can simply copy/paste your example and run it without having to create
tables.

The lack of any example data makes suggestions difficult to give here.  The
ANY(array) expression is much cleaner and from what I can tell the presence
of the IN (subquery) with its own FROM clause is pointless if not outright
problematic.

David J.




--
View this message in context:
http://postgresql.nabble.com/Count-values-that-match-elements-in-csv-string-tp5833121p5833136.html
Sent from the PostgreSQL - sql mailing list archive at Nabble.com.