03Usr
03Usr

Reputation: 3435

Conditional selecting rows based on a column value

ID  RANGE_ID            START_DATE    END_DATE    BAND_TYPE           FLAG_LINE
3     1               01/03/2013    31/03/2013          R                   1
4     1               01/03/2013    31/03/2013          R                   0
5     2               01/03/2013    31/03/2013          R                   1
6     2               01/03/2013    31/03/2013          R                   0
7     3               01/03/2013    31/03/2013          R                   0
8     3               01/03/2013    31/03/2013          N                   0

From this table, for each RANGE_ID, I need to select rows using the following conditions: If there are rows with identical columns (apart from the ID field) then only select the row which has FLAG_LINE = 1, if there are identical rows but none of them contain a FLAG_LINE=1 column then select all of them, based on this the query should return the following results:

ID  RANGE_ID          START_DATE    END_DATE      BAND_TYPE           FLAG_LINE
3     1               01/03/2013    31/03/2013          R                   1
5     2               01/03/2013    31/03/2013          R                   1
7     3               01/03/2013    31/03/2013          R                   0
8     3               01/03/2013    31/03/2013          N                   0

I tried doing it in chunks: i.e run something similar for each RANGE:

begin
  for x in ( select count(*) cnt
               from dual 
              where exists (
                select 1 FROM myTable 
                WHERE RANGE_ID = 1 AND FLAG_LINE = 1) )
  loop
        if ( x.cnt = 1 ) 
        then
           dbms_output.put_line('flag line exists');
           --insert the line with FLAG_LINE = 1 into temp table for this range
        else 
           dbms_output.put_line('does not exist');
           --insert the lines into temp table for this range
        end if;
  end loop;
end;

using this method for each RANGE I populate a temp table and return the results at the end, but this is not quite flexible, is there another way that this can be achieved?

Thanks

Upvotes: 2

Views: 25709

Answers (3)

Amit Singh
Amit Singh

Reputation: 8109

try like this...

Select * from tablename where flag=1 
union 
(Select * from tablename a where  (Select count(*) from tablename b 
where a.Range_id=b.RANGE_ID  and b.flag=1)<1)

SQL FIDDLE Demo

Upvotes: 3

mtwaddell
mtwaddell

Reputation: 189

Here's another way to get the results you're wanting

SELECT *
  FROM info
 WHERE flag_line = 1
UNION
SELECT *
  FROM info i1
 WHERE NOT EXISTS (SELECT NULL
                     FROM info i2
                    WHERE i2.range_id = i1.range_id
                      AND i2.flag_line = 1
                   )

Upvotes: 0

Noel
Noel

Reputation: 10525

Try this.

select * from myTable
 where flag_line = 1
       or
       (range_id, start_date, end_date, band_type) in (
                  select range_id, start_date, end_date, band_type
                    from myTable
                group by range_id, start_date, end_date, band_type
                  having max(flag_line) = 0)

Upvotes: 1

Related Questions