How do I use select in loop continue when statement? - sql

Question

I want to continue a oracle sql loop when I find more then one result in a query - so my simplified code:
declare
cursor foo_cursor select * from foo_table;
foo foo_cursor%ROWTYPE;
begin
open foo_cursor;
loop
fetch foo_cursor into foo;
exit when foo_cursor%NOTFOUND;
continue when ( -- the next query has entries or an entry,
-- but how do I do this?
select count(*) from bar_table where bar_column=foo.foo_column
group by bar_column having count(1)>1;
)
insert into uninterresting_table (some_column) VALUES
(foo.foo_column);
end loop;
close foo_cursor;
end;

Answer 1

It looks like you only want to act on the records in FOO_TABLE when at least two related records exists in BAR_TABLE. You can alter the definition of foo_cursor to take that requirement into account as shown below. That way you don't need to iteratively check for the existence of a record in BAR_TABLE each time.
declare
cursor foo_cursor is
select *
from foo_table foo
where exists (select 1 from bar_table bar
where bar.bar_column = foo.foo_column
having count(*) > 1);
foo foo_cursor%ROWTYPE;
begin
open foo_cursor;
loop
fetch foo_cursor into foo;
exit when foo_cursor%NOTFOUND;
insert into uninteresting_table (some_column) VALUES
(foo.foo_column);
end loop;
close foo_cursor;
end;
/
On the other hand if you are looking to skip records in FOO_TABLE that already have two or more records in BAR_TABLE, you can just invert the existence check and all else would be the same:
declare
cursor foo_cursor is
select *
from foo_table foo
where NOT exists (select 1 from bar_table bar
where bar.bar_column = foo.foo_column
having count(*) > 1);
foo foo_cursor%ROWTYPE;
begin
open foo_cursor;
loop
fetch foo_cursor into foo;
exit when foo_cursor%NOTFOUND;
insert into uninteresting_table (some_column) VALUES
(foo.foo_column);
end loop;
close foo_cursor;
end;
/
If you want to process all records in FOO_TABLE but do additional actions when two or more records exist in BAR_TABLE, you can still do that with a change to your foo_cursor:
declare
cursor foo_cursor is
select foo.*
, case when exists (select 1 from bar_table bar
where bar.bar_column = foo.foo_column
having count(*) > 1)
then 'Y'
else 'N'
end has_two_or_more
from foo_table foo;
foo foo_cursor%ROWTYPE;
begin
open foo_cursor;
loop
fetch foo_cursor into foo;
exit when foo_cursor%NOTFOUND;
continue when foo.has_two_or_more = 'Y';
insert into uninteresting_table (some_column) VALUES
(foo.foo_column);
end loop;
close foo_cursor;
end;
/

Answer 2

if the [select has ... one and more entries] i want go into the next iteration else do more stuff in this iteration.
If there are no records in bar_table you want to do some more processing otherwise you want to skip the processing. There's a way to do that: goto.
Oh yes :)
declare
cursor foo_cursor select * from foo_table;
foo foo_cursor%ROWTYPE;
n pls_integer;
begin
open foo_cursor;
loop
fetch foo_cursor into foo;
exit when foo_cursor%NOTFOUND;
select count(*) into n
from bar_table
where bar_column=foo.foo_column
group by bar_column having count(1)>1;
if n > 0 then
goto skip_point;
end if;
insert into uninterresting_table (some_column) VALUES
(foo.foo_column);
<< skip_point >>
end loop;
close foo_cursor;
end;
Obviously you could just put the whole skippable section into a branch of an IF .. ELSE statement, but where's the fun in that?

Answer 3

So thanks to #APC and #Sentinel -- I still made my version work, but your answers gave me the right dirrections:
declare
cursor foo_cursor select * from foo_table;
foo foo_cursor%ROWTYPE;
n pls_integer
begin
open foo_cursor;
loop
fetch foo_cursor into foo;
exit when foo_cursor%NOTFOUND;
begin
select count(1) into n from bar_table where bar_column=foo.foo_column
group by bar_column;
exception when NO_DATA_FOUND then continue;
end
continue when (n>1);
insert into uninterresting_table (some_column) VALUES
(foo.foo_column);
-- do some more stuff
end loop;
close foo_cursor;
end;
```