Modifying display in oracle

Question

Dear all;
I have an interesting problem
I have the following code
create table table_A
      ID varchar2(200),
      qty number(6)
insert into table_a
  (id, qty)
values
  ('A', 10);
insert into table_a
  (id, qty)
values
  ('B', 6);
create table table_B
      ID varchar2(200),
      colour varchar2(200)
insert into table_b
  (id, colour)
values
  ('A', 'Red');
insert into table_b
  (id, colour)
values
  ('B', 'Black');
create table table_info
       ID varchar2(200),
       addr varchar2(1000)
insert into table_info
  (id, addr)
values
  ('A', '2034 James Street');
insert into table_info
  (id, addr)
values
  ('B', '2044 Main Street');and I am using a select similar to do this
select g.id, g.qty, h.colour from table_A g, table_B h
where g.id = h.idwhich gives me the following display below
ID     QTY    COLOUR
A       10      Red
B         6      Blacknow I would like to modify the display so that I can get this instead
ID       Addr
A        '2034 James Street'
ID       QTY       Colour
A          10         Red
ID      Addr
B       '2044 Main Street'
ID      QTY        Colour
B         6           BlackAll help is appreciated. Thank you.

Answer 1

Hi,
Here's one way:
WITH     union_data  AS
     SELECT    g1.id
     ,       TO_CHAR (g1.qty)     AS column_2
     ,       h1.colour
     ,       g1.id               AS sort_id
     ,       2               AS sort_line
     FROM        table_A      g1
     ,       table_B      h1
     WHERE       g1.id            = h1.id
         UNION ALL
     SELECT       'ID'               AS id
     ,       'QTY'               AS column_2
     ,       'Colour'          AS colour
     ,       g2.id               AS sort_id
     ,       1               AS sort_line
     FROM        table_A      g2
     ,       table_B      h2
     WHERE       g2.id            = h2.id
         UNION ALL
        SELECT    t1.id
     ,       t1.addr          AS column_2
     ,       NULL               AS colour
     ,       t1.id               AS sort_id
     ,       0               AS sort_line
     FROM       table_info   t1
         UNION ALL
        SELECT    'ID'               AS id
     ,       'Addr'          AS column_2
     ,       NULL               AS colour
     ,       t2.id               AS sort_id
     ,       -1               AS sort_line
     FROM       table_info   t2
SELECT       id, column_2, colour
FROM       union_data
ORDER BY  sort_id
,            sort_line
;Here's how it works:
You already know how to get the output lines that contain the values of the qty and colour.
You can generate a header line for each of those lines, that has the labels 'ID', 'QTY' and 'COLOUR', and do a UNION to combine those result sets.
SELECT    g.id
,       TO_CHAR (g.qty)     AS qty
,       h.colour
FROM        table_A      g
,       table_B      h
WHERE       g.id            = h.id
    UNION ALL
SELECT       'ID'
,       'QTY'
,       'Colour'
FROM        table_A      g
,       table_B      h
WHERE       g.id            = h.id
;Output:
ID         QTY        COLOUR
A          10         Red
B          6          Black
ID         QTY        Colour
ID         QTY        ColourThe problem with the output above is that the header lines should be interleaved with the data lines, first a header, then the data that goes with it. To do that, we can add some extra columns to each branch of the UNION that are just for sorting:
SELECT    g1.id
,       TO_CHAR (g1.qty)     AS qty
,       h1.colour
,       g1.id               AS sort_id
,       1               AS sort_line
FROM        table_A      g1
,       table_B      h1
WHERE       g1.id            = h1.id
    UNION ALL
SELECT       'ID'               AS id
,       'QTY'               AS qty
,       'Colour'          AS colour
,       g2.id               AS sort_id
,       2               AS sort_line
FROM        table_A      g2
,       table_B      h2
WHERE       g2.id            = h2.id
ORDER BY  sort_id
,            sort_line
;Output:
ID         QTY        COLOUR     SORT_ID     SORT_LINE
A          10         Red        A                   1
ID         QTY        Colour     A                   2
B          6          Black      B                   1
ID         QTY        Colour     B                   2That solved one problem, but created another one. We don't want to see the sort_id and sort_line columns; we just want to use them in the ORDER BY clause. However, in a UNION you cn't ORDER BY anything that's not in the SELECT clause. Depending on your front end, you might be able to hide those columns (in SQL*Plus, for instance, you can say "COLUMN sort_id NOPRINT"), but here's how you can get around the problem in pure SQL, by doing the UNION in a separate sub-query from the ORDER BY:
WITH     union_data  AS
     SELECT    g1.id
     ,       TO_CHAR (g1.qty)     AS qty
     ,       h1.colour
     ,       g1.id               AS sort_id
     ,       2               AS sort_line
     FROM        table_A      g1
     ,       table_B      h1
     WHERE       g1.id            = h1.id
         UNION ALL
     SELECT       'ID'               AS id
     ,       'QTY'               AS qty
     ,       'Colour'          AS colour
     ,       g2.id               AS sort_id
     ,       1               AS sort_line
     FROM        table_A      g2
     ,       table_B      h2
     WHERE       g2.id            = h2.id
SELECT       id, qty, colour
FROM       union_data
ORDER BY  sort_id
,            sort_line
;Output:
ID         QTY        COLOUR
ID         QTY        Colour
A          10         Red
ID         QTY        Colour
B          6          BlackThat's perfect for the data from table_a and table_b. To get the data from table_info as well, just add 2 more branches to the UNION, as shown at the beginning of this message.
It's a little inefficient to do 2 separate queries, one to get the data, and another to get the header, when those 2 queries use the same tables and the same WHERE clause. An alternative is to cross-join the table to another table (or, more likely, a result set functioning as a table), that has the values n=1 and n=2, and use CASE expressions to display either the data or its header. I'll leave that as an exercise for the reader.
Edited by: Frank Kulash on Jan 18, 2011 1:02 PM
Explanation added.