CODEWITHSUNDEEP
sqlsql-server
irshan syed
irshan syed

Reputation: 89

Pivot rows to columns with id not repeated

I have some data coming in from an external source that I need to pivot into the right columns. Problem is that the id and one more column are not repeated for all the rows. Example data: 

Rownumber ID     Type             Category Value
1         100   1-Purchase Order  Address  Cedar Rd
2                                 City     Beachwood
3                                 State    Ohio
4                                 Zip      44122
5         300   1-Purchase Order  Address  Rockside Rd
6                                 City     Independence
7                                 State    Ohio
8                                 Zip      44131
9         200   1-Purchase Order  Address  Rockside Rd
10                                City     Independence
11                                State    Ohio
12                                Zip      44131

Desired output is 

Rownumber  ID   Type              Address      City        State  Zip
1         100  1-Purchase Order  Cedar Rd      Beachwood   Ohio   44122
2         300   1-Purchase Order Rockside Rd  Independence Ohio   44122
3         200   1-Purchase Order Rockside Rd  Independence Ohio   44122

Please note the ID appears in random order, because of which this code did not work:

    select row_number() over (order by min(id)) as rownumber,
           new_id as id,
           max(type) as type,
           max(case when category = 'Address' then value end) as address,
           max(case when category = 'City' then value end) as city,
           max(case when category = 'State' then value end) as state,
           max(case when category = 'Zip' then value end) as zip
    from (select t.*,
                 coalesce(id, max(id) over (order by rownumber)) as new_id
          from t
         ) t
    group by new_id;

Upvotes: 3

Views: 44

Answers (1)

Gordon Linoff
Gordon Linoff

Reputation: 1270191

You can easily fix this code because you don't really need the ids. You just need a way to define the groups. So, a cumulative count should do the same thing:

select row_number() over (order by min(rownumber)) as rownumber,
       max(id) as id,
       max(type) as type,
       max(case when category = 'Address' then value end) as address,
       max(case when category = 'City' then value end) as city,
       max(case when category = 'State' then value end) as state,
       max(case when category = 'Zip' then value end) as zip
from (select t.*,
             count(id) over (order by rownumber) as grp
      from t
     ) t
group by grp;

This assigns all NULL values to the immediately preceding non-null id grouping.

Here is a db<>fiddle.

Upvotes: 5

Related Questions