Question

I have a view like this one

id     name     characteristic     value     isList
1      cube         sides           6          0
1      cube         color          blue        0
2    triangle       sides           3          0
3    hexagon        (null)        (null)     (null)
4    rectangle      weight          15         0

I need to select all ids and names and retrieve some characteristics and the respective value. for instance, I want to retrieve all figures (ids 1, 2, 3 and 4) and the characteristics sides and color (if are available, if not, only id and name are filled up; the others are null).

I tried

select *
from shapes_view
where (id = 1 or id = 2 or id = 3 or id = 4) and (characteristic like 'sides' or characteristic like 'color')

but it, obviously, retrieves the ids 1 and 2 but not 3 and 4.

My guess is that I need some sort of subquery to do this but when I try to join this view with itself, I get a long list of combinations that aren’t anywhere near of what I need.

What i intend to get is something like

id     name     characteristic     value     isList
1      cube         sides           6          0
1      cube         color          blue        0
2    triangle       sides           3          0
3    hexagon        (null)        (null)     (null)
4    rectangle      (null)        (null)     (null)

I know I can select all values and exclude what I do not what in the java side, but it doesn’t sound very correct…

Can anyone help me on this one?
Best regards

Answer 1

You can use a self join– actually, a couple of self joins

select all_ids.id,
       all_ids.name,
       s.value sides,
       c.value color
  from shapes_view all_ids
       left outer join (select *
                          from shapes_view
                         where characteristic = 'sides') s
               on( s.id = all_ids.id )
       left outer join (select *
                          from shapes_view
                         where characteristic = 'color') c
               on( c.id = all_ids.id )

Or you can pivot the data

select id,
       name,
       max( case when characteristic = 'sides'
                 then value
                 else null
              end) sides,
       max( case when characteristic = 'color'
                 then value
                 else null
              end) color
  from shapes_view
 group by id, name

The complexity of the queries required to extract data for N different attributes is one of the reasons that this sort of very generic entity-attribute-value data model is generally frowned upon.