How do I put a crosstab on a form?

Question

I need to set up a matrix-type canvas on a form where each cell will become a record on a table with the row value and column value forming the key. The row and column values will be retrieved from another table.
Any ideas? I don't even know where to start.

Answer 1

for example I have a table:
emps( empno number, start_time date, end_time date );
And I'd like to have a form that looks like:
EMPNO 8am 9am 10am 11am 12pm 1pm 2pm
1 - x x x x - x ....
2 x x x x - x x ....
that shows empno = 1 works from 9am-12pm, is off for 1 hour (and
so on...)
and I said...
Well, once we create a view, this is easy. The view isn't easy
to come up with perhaps but once there -- its trivial to use.
We'll start with an extra table:
[email protected]> create table hours
2 as
3 select rownum-1 hour
4 from all_objects
5 where rownum <= 24;
Table created.
That has a row for every hour in the day...
Here is my emp table. I have a column DAY that is for
illustrative purposes -- you might not need it (you could also
substitute TRUNC(start_date) where I reference DAY in the view
below). I made it a column so I could index it and query it
easily...
[email protected]> create table emp
2 ( empno number,
3 day date,
4 start_time date,
5 end_time date
6 )
7 /
Table created.
[email protected]> create index emp_idx on emp(empno);
Index created.
[email protected]> create index start_idx on emp(day);
Index created.
[email protected]> alter session set
nls_date_format='dd-mon-yyyy hham';
Session altered.
[email protected]>
[email protected]> insert into emp values
2 ( 1, '07-jul-2000', '07-jul-2000 9am', '07-jul-2000 12pm'
1 row created.
[email protected]> insert into emp values
2 ( 1, '07-jul-2000', '07-jul-2000 1pm', '07-jul-2000 6pm' );
1 row created.
[email protected]>
[email protected]> insert into emp values
2 ( 2, '07-jul-2000', '07-jul-2000 8am', '07-jul-2000 11am'
1 row created.
[email protected]> insert into emp values
2 ( 2, '07-jul-2000', '07-jul-2000 12pm','07-jul-2000 5pm' );
1 row created.
[email protected]>
[email protected]> insert into emp
2 select empno, day-1,
3 start_time-(1+1/24), end_time-(1+1/24)
4 from emp;
4 rows created.
[email protected]>
[email protected]> select * from emp;
EMPNO DAY START_TIME END_TIME
1 07-jul-2000 12am 07-jul-2000 09am 07-jul-2000 12pm
1 07-jul-2000 12am 07-jul-2000 01pm 07-jul-2000 06pm
2 07-jul-2000 12am 07-jul-2000 08am 07-jul-2000 11am
2 07-jul-2000 12am 07-jul-2000 12pm 07-jul-2000 05pm
1 06-jul-2000 12am 06-jul-2000 08am 06-jul-2000 11am
1 06-jul-2000 12am 06-jul-2000 12pm 06-jul-2000 05pm
2 06-jul-2000 12am 06-jul-2000 07am 06-jul-2000 10am
2 06-jul-2000 12am 06-jul-2000 11am 06-jul-2000 04pm
8 rows selected.
So that is the data we have to work with. 2 days worth for 2
emps. Hours are different by day...
[email protected]> alter session set
nls_date_format='dd-mon-rr';
Session altered.
Here comes the view:
[email protected]> create or replace view empV
2 as
3 select empno, day,
4 max( decode( hour, 0, flag, null ) ) "0",
5 max( decode( hour, 1, flag, null ) ) "1",
6 max( decode( hour, 2, flag, null ) ) "2",
7 max( decode( hour, 3, flag, null ) ) "3",
8 max( decode( hour, 4, flag, null ) ) "4",
9 max( decode( hour, 5, flag, null ) ) "5",
10 max( decode( hour, 6, flag, null ) ) "6",
11 max( decode( hour, 7, flag, null ) ) "7",
12 max( decode( hour, 8, flag, null ) ) "8",
13 max( decode( hour, 9, flag, null ) ) "9",
14 max( decode( hour, 10, flag, null ) ) "10",
15 max( decode( hour, 11, flag, null ) ) "11",
16 max( decode( hour, 12, flag, null ) ) "12",
17 max( decode( hour, 13, flag, null ) ) "13",
18 max( decode( hour, 14, flag, null ) ) "14",
19 max( decode( hour, 15, flag, null ) ) "15",
20 max( decode( hour, 16, flag, null ) ) "16",
21 max( decode( hour, 17, flag, null ) ) "17",
22 max( decode( hour, 18, flag, null ) ) "18",
23 max( decode( hour, 19, flag, null ) ) "19",
24 max( decode( hour, 20, flag, n ull ) ) "20",
25 max( decode( hour, 21, flag, null ) ) "21",
26 max( decode( hour, 22, flag, null ) ) "22",
27 max( decode( hour, 23, flag, null ) ) "23"
28 from ( select empno, day, hour,
29 decode( sign(
start_time-(trunc(start_time)+
30 hour/24)), 1, NULL,
31 decode( sign(end_time-(trunc(end_time)+
32 hour/24)), 1, 'X', NULL ) ) flag
33 from ( select * from hours ),
34 emp
35 )
36 group by empno, day
37 /
View created.
The view works by:
o "from (select * from hours ), emp" creates 24 rows for each
row in the emp table we are interested in. Each of these rows
has an hour associated with it now. So, where we had 8 rows,
we'll get 8x24 rows (less if we use a predicate on the view as
see below)...
o the decode in the from list says:
if ( start_time - HOUR ) > 0
then
-- start time is bigger then hour, we did not work in that
-- hour
return NULL
else
if ( end_time - HOUR ) > 0
then
-- end_time happens AFTER that hour hence we WORKED in
-- that hour
return 'X'
else
return NULL
end if
end if
o we now have a set such that for each row in emp we have 24
rows with the numbers 0..23 and a flag that says whether we
worked in that hour or not...
o we can use a simple decode and group by to pivot on empno and
day...
Now lets see how it works and how it performs
[email protected]> select * from empV;
EMPNO DAY 0 1 2 3 4 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2
1 06-jul-00 X X X X X X X X
1 07-jul-00 X X X X X X X X
2 06-jul-00 X X X X X X X X
2 07-jul-00 X X X X X X X X
That shows empno=1 works from 8am-11am (upto not including
11am) and 1pm-5pm on 6/jul (easier to read with a wider screen
The following shows that we can efficiently query this as
well:
[email protected]> set autotrace on explain
[email protected]> select * from empV where empno = 1;
EMPNO DAY 0 1 2 3 4 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2
1 06-jul-00 X X X X X X X X
1 07-jul-00 X X X X X X X X
Execution Plan
0 SELECT STATEMENT Optimizer=CHOOSE
1 0 VIEW OF 'EMPV'
2 1 SORT (GROUP BY)
3 2 NESTED LOOPS
4 3 TABLE ACCESS (FULL) OF 'HOURS'
5 3 TABLE ACCESS (BY INDEX ROWID) OF 'EMP'
6 5 INDEX (RANGE SCAN) OF 'EMP_IDX' (NON-UNIQUE)
Even though the view looks too complex to merge predicates
into -- it can. It used the index on emp to find empno 1 and
then did the work. The following shows the same can be true of
the other indexed columns:
[email protected]> select * from empV where day = to_date(
'07-jul-00' );
EMPNO DAY 0 1 2 3 4 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2
1 07-jul-00 X X X X X X X X
2 07-jul-00 X X X X X X X X
Execution Plan
0 SELECT STATEMENT Optimizer=CHOOSE
1 0 VIEW OF 'EMPV'
2 1 SORT (GROUP BY)
3 2 NESTED LOOPS
4 3 TABLE ACCESS (FULL) OF 'HOURS'
5 3 TABLE ACCESS (BY INDEX ROWID) OF 'EMP'
6 5 INDEX (RANGE SCAN) OF 'START_IDX'
(NON-UNIQUE)
null