Question

I need to create a table (possibly a view) that relates two tables, one table of configs with a results table. The problem is that it’s necessary transform a simple row in the results table in multiple rows of the configs table.

I will give you an example of what I need:

Configs Table: primary key = (id_file,cpattr)

id_file     cpattr      type
------------------------------------
f01         dim01       merchant_id
f01         dim02       card_number
f01         dim03       trans_code
f02         dim01       card_number
f02         dim02       amount

Results Table

id_file dim01   dim02   dim03
-------------------------------------------------------
f01      01      88      015
f02      99     0.78    null

And I want to get a table(View) with the following output:

id_file type            data
--------------------------------------------------
f01     merchant_id     01
f01     card_number     88
f01     trans_code      015
f02     card_numer      99
f02     amount          0.78 

I’m not seeing a way of do that without using dynamic sql, but I’m trying to avoid it. Hope you can help me.

Thanks in advance.

Answer 1

Since it appears you only have the three dim columns in your results table, the easiest thing would be a case statement:

select 
    c.id_file
    , [type]
    , case c.cpattr
        when 'dim01' then r.dim01
        when 'dim02' then r.dim02
        when 'dim03' then r.dim03
        else ''
    end [data]
from 
    configs c
    left join results r
        on c.id_file = r.id_file

Edited to add:

Since there are an indeterminate number of dim columns, a slightly better approach might be:

select
    c.id_file, c.type, o.data
from 
    _configs c
    left join 
    (
        select * from _results
        unpivot
        (
            [data] for cpattr in (dim01, dim02, dim03, ...)
        ) as [output]
    ) as [o]
        on c.id_file = o.id_file
        and c.cpattr = o.cpattr

You still have to identify all the dim columns affected, but you only have to do it once, not twice. This will also throw an error if any of the dim columns are removed or renamed, although it won’t populate from new dim columns. Not sure of any way to do that.