Concatenate fields using connect by prior

Question

Having the following table my_tabe:

M01 |   1
M01 |   2
M02 |   1

I want to query over it in order to obtain:

M01 |   1,2
M02 |   1

I managed to get close using the following query:

with my_tabe as
(
    select 'M01' as scycle, '1' as sdate from dual union
    select 'M01' as scycle, '2' as sdate from dual union
    select 'M02' as scycle, '1' as sdate from dual
)
SELECT scycle, ltrim(sys_connect_by_path(sdate, ','), ',')
FROM
(
    select scycle, sdate, rownum rn
    from my_tabe
    order by 1 desc
)
START WITH rn = 1
CONNECT BY PRIOR rn = rn - 1

Yielding:

SCYCLE      |   RES
M02         |   1,2,1
M01         |   1,2

Which is wrong. It's seems I'm close, but I'm afraid I don't what's the next step...

Any tips?

Answer 1

You need to restrict your connect by to the same scycle value, and also count the number of matches and filter on that to avoid seeing intermediate results.

with my_tabe as
(
    select 'M01' as scycle, '1' as sdate from dual union
    select 'M01' as scycle, '2' as sdate from dual union
    select 'M02' as scycle, '1' as sdate from dual
)
select scycle, ltrim(sys_connect_by_path(sdate, ','), ',')
from
(
    select distinct sdate,
        scycle,
        count(1) over (partition by scycle) as cnt,
        row_number() over (partition by scycle order by sdate) as rn
    from my_tabe
)
where rn = cnt
start with rn = 1
connect by prior rn + 1 = rn
and prior scycle = scycle
/

SCYCLE LTRIM(SYS_CONNECT_BY_PATH(SDATE,','),',')
------ -----------------------------------------
M01    1,2
M02    1

If you're on 11g you can use the built-in LISTAGG function instead:

with my_tabe as
(
    select 'M01' as scycle, '1' as sdate from dual union
    select 'M01' as scycle, '2' as sdate from dual union
    select 'M02' as scycle, '1' as sdate from dual
)
select scycle, listagg (sdate, ',') 
within group (order by sdate) res
from my_tabe
group by scycle
/ 

Both approaches (and others) are shown here.