SQL Pivot table returning NULL for non-existent child table values - sql

Question

I have a typical RDMS setup where records in a main table can have optional records in a related table via a M2M join. I'm trying to PIVOT this data but in cases where there is no relation I want to return a default value. The join I have below is returning NULL.
select *
from
(
SELECT s.Biz_Name, la.Name AS Association, ISNULL(i.Location, 'Default') as Location
FROM dbo.ShopAssociations sa
INNER JOIN dbo.LookupAssociations la
ON sa.AssociationID = la.AssociationID
RIGHT JOIN dbo.Basic_Shop_Info s
ON sa.ShopID = s.ShopID
INNER JOIN dbo.Images i
ON la.ImageID = i.ImageID
) DataTable
PIVOT
(
min(Location)
for association in
([OnCall],[OCGuy],[ASCLogo],[ASC_OtherSt],[ASE],[AASP],[AASP_PA],
[ASE_BlueSeal],[AAA],[AAA-B],[ASA],[ATRA],[ICAR],[CAA],[ACDelco],
[Cert],[ASC],[BBB],[Goodyear],[Limos],[RVs],[Bosch],[NARSA],
[DiscTire],[BigO],[Tires],[Firestone],[ASCCA],[JustTires],[ASE_Blue])
) PivotTable
The output looks like this:
BizName OnCall OCGuy ASCLogo ASC_OtherSt ASE ...
"Wonderful Biz" somevalue somevalue NULL somevalue NULL
What I am trying to achieve is if a child record doesn't exist in INNER JOIN from Basic_Shop_Info to ShopAssociations that we get "Default" instead of NULL. I've tried ISNULL(), Coalesce() and even a CASE statement, all with the same results.

Answer 1

Based on your comment it sounds like you found a solution. I am only answering this to provide a suggestion based on the fact you are pivoting so many columns and they are all hard-coded. You can use dynamic SQL for a PIVOT and your query would look something like this:
DECLARE #cols AS NVARCHAR(MAX),
#query AS NVARCHAR(MAX),
#colsPivot AS NVARCHAR(MAX)
select #cols = STUFF((SELECT distinct ',' + QUOTENAME(Name)
from dbo.LookupAssociations
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
select #colsPivot = STUFF((SELECT distinct ', IsNull(' + QUOTENAME(Name) +', ''Default'')'
from dbo.LookupAssociations
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
set #query = 'SELECT Bizname, ' + #colsPivot + ' from
(
SELECT s.Biz_Name, la.Name AS Association, ISNULL(i.Location, ''Default'') as Location
FROM dbo.ShopAssociations sa
INNER JOIN dbo.LookupAssociations la
ON sa.AssociationID = la.AssociationID
RIGHT JOIN dbo.Basic_Shop_Info s
ON sa.ShopID = s.ShopID
INNER JOIN dbo.Images i
ON la.ImageID = i.ImageID
) x
pivot
(
min(Location)
for association in (' + #cols + ')
) p
'
execute(#query)
The value #colsPivot is adding the IsNull() around each of you columns so you can put in place the Default value. But this should provide the same result as your original query where everything was hard-coded.
This will get the list of columns at run-time so then you do not have to hard-code anything and it will accept new values without having to change the query.

Answer 2

I got this:
DECLARE #cols AS NVARCHAR(MAX),
#query AS NVARCHAR(MAX)
select #cols = STUFF((SELECT ',' + QUOTENAME(c.col+cast(rn as varchar(10)))
from
(
select row_number() over(partition by person_nbr
order by person_nbr,first_name, last_name, medication_name) rn
from TA_PIVOT
) d
cross apply
(
select 'diag' col, 1 sort
) c
group by col, rn, sort
order by rn, sort
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
set #query = 'SELECT person_nbr, first_name, last_name,medication_name,' + #cols + '
from
(
select person_nbr,first_name,last_name,medication_name,
col+cast(rn as varchar(10)) col,
value
from
(
-- when you perform an unpivot the datatypes have to be the same.
-- you might have to cast the datatypes in this query
select person_nbr,first_name,last_name, medication_name, cast(icd_code_id as varchar(500)) diag,
row_number() over(partition by person_nbr order by person_nbr, first_name, last_name,medication_name) rn
from ta_pivot
) src
unpivot
(
value
for col in (diag)
) unpiv
) d
pivot
(
max(value)
for col in (' + #cols + ')
) p '
execute(#query);