Flatten rows in a join

[TysonLPrice]

I have a stored procedure that currently brings back one row per policy number. A new table has been introduced that needs to be joined to that stored procedure. That table currently has six entries per policy but that can increase or decrease. This is sort of what I am looking at.

Although it is much more complex I think this will illustrate the issue:

create table #Policies(PolicyNo int, SomeData Char(5))
insert into #Policies(policyno,somedata) values (1,'abcd')

create table #NewTable(PolicyNo int,NewData char(5),akey int)
insert into #NewTable(PolicyNo,NewData,akey) values(1, 'aaaa',1)
insert into #NewTable(PolicyNo,NewData,akey) values(1, 'bbbb',2)
insert into #NewTable(PolicyNo,NewData,akey) values(1, 'cccc',3)
insert into #NewTable(PolicyNo,NewData,akey) values(1, 'dddd',4)

Lets say this is the current stored procedure:

select * from #Policies p where policyno = 1

I need to join #NewTable to it:

select p.policyno,t.Newdata,t.akey from #Policies p
join #NewTable t on t.policyno = p.policyno
where p.policyno = 1

But I need the output to look like this:

PolicyNo NewData akey NewData akey NewData akey NewData akey

1 aaaa 1 bbbb 2 cccc 3 dddd 4

Any ideas assuming at this point I can't change the table structure. If the stored procdure always brought back one row I could flatten #NewTable first and join to that. The real issue is the stored procedure could bring back many policies and each one needs to be one row per policy with the #NewTable data flattened out per policy row.

 

[TysonLPrice]

Added:

#NewTable is dynamic so the number of entries is not fixed.

 

[simian336]

This kind of issue has come up several times in the past. Is there a maximum number of entries?

Is this for a report so that you could just concat them in a single field or do you need individual fields for each?

Simi

 

[wbodger]

Is this a place where you could use Pivot?

 

[TysonLPrice]

What I ended up doing just for expediency’s sake is:

In the select:

 ,'GR' + convert(Varchar,t1.pkRate_Barcodes) 'ac2'
,'GR' + convert(Varchar,t2.pkRate_Barcodes) 'ac26'
,'GR' + convert(Varchar,t3.pkRate_Barcodes) 'GrpRatCont'
,'GR' + convert(Varchar,t4.pkRate_Barcodes) 'EnrollQuest'
,'GR' + convert(Varchar,t5.pkRate_Barcodes) 'EnrollQuestInv'
,'GR' + convert(Varchar,t6.pkRate_Barcodes) 'U-153'

The join:

   join Rate_BarCodes t1 on gr.EstExperienceYear = t1.EstExperienceyear 
and ge.fkgroup = t1.fkgroup and t1.document_type_name = 'AC-2'
and e.policyno = t1.policyno

join Rate_BarCodes t2 on gr.EstExperienceYear = t2.EstExperienceyear	
and ge.fkgroup = t2.fkgroup and t2.document_type_name = 'AC-26'
and e.policyno = t2.policyno

join Rate_BarCodes t3 on gr.EstExperienceYear = t3.EstExperienceyear 
and ge.fkgroup = t3.fkgroup and t3.document_type_name = 'GrpRatCont'
and e.policyno = t3.policyno

join Rate_BarCodes t4 on gr.EstExperienceYear = t4.EstExperienceyear 
and ge.fkgroup = t4.fkgroup and t4.document_type_name = 'EnrollQuest'
and e.policyno = t4.policyno

join Rate_BarCodes t5 on gr.EstExperienceYear = t5.EstExperienceyear 
and ge.fkgroup = t5.fkgroup and t5.document_type_name = 'EnrollQuestInv'
and e.policyno = t5.policyno

join Rate_BarCodes t6 on gr.EstExperienceYear = t6.EstExperienceyear 
and ge.fkgroup = t6.fkgroup and t6.document_type_name = 'U-153'
and e.policyno = t6.policyno

I had left out part of the requirements. Each column name from the #NewTable had to have the column name in the select. That is becuase it needs to be exposed to Crystal reports.

I didn't check into cross tabs or pivot.


Thanks to those that replied.