Timestamp to_char subquery - oracle

Question

I have the following query but I need a particular format (HH24,MI,SS) for the results that I get in column "elapsed". Can I do that with a subquery?
SELECT
trunc(dstamp) "DATE",
to_char(dstamp, 'HH24:MI:SS') "Time"
, user_id
, dstamp - lag(dstamp) OVER (
PARTITION BY user_id ORDER BY dstamp
) AS elapsed
, COUNT(CODE) "Lines_Picked"
, ROUND(sum(update_qty / cast(substr(sku_id, instr(sku_id, '-', 1, 1) + 1
, instr(sku_id, '-', 1, 2) - 1 - instr(sku_id, '-', 1, 1)) as integer))) "CASES_PICKED"
FROM v_inventory_transaction
WHERE client_id = 'USKIDS2CA'
AND code = 'Pick'
AND list_id IS NOT NULL
AND STATION_ID LIKE 'R%'
AND reference_id NOT LIKE '%-FK%'
AND trunc(dstamp) = to_date('03/23/2022','mm/dd/yyyy')
GROUP BY
user_id,
dstamp
, cast(substr(sku_id, instr(sku_id, '-', 1, 1) + 1
, instr(sku_id, '-', 1, 2) - 1 - instr(sku_id, '-', 1, 1)) as integer)

Answer 1

What is "timestamp" from the title? Which datatype is it? DATE or TIMESTAMP?
If it is TIMESTAMP, extract helps a lot:
SQL> with temp (user_id, dstamp) as
2 (select 1, to_timestamp('27.03.2022 08:08', 'dd.mm.yyyy hh24:mi') from dual union all
3 select 1, to_timestamp('27.03.2022 05:30', 'dd.mm.yyyy hh24:mi') from dual union all
4 select 1, to_timestamp('27.03.2022 02:28', 'dd.mm.yyyy hh24:mi') from dual
5 )
6 select user_id,
7 dstamp,
8 --
9 to_char(extract (hour from dstamp - lag(dstamp) over (partition by user_id order by dstamp)), 'FM00') ||':'||
10 to_char(extract (minute from dstamp - lag(dstamp) over (partition by user_id order by dstamp)), 'FM00') ||':'||
11 to_char(extract (second from dstamp - lag(dstamp) over (partition by user_id order by dstamp)), 'FM00') result
12 from temp
13 order by user_id, dstamp;
USER_ID DSTAMP RESULT
---------- ----------------------------------- -----------
1 27-MAR-22 02.28.00.000000000 AM ::
1 27-MAR-22 05.30.00.000000000 AM 03:02:00
1 27-MAR-22 08.08.00.000000000 AM 02:38:00
SQL>
If it is DATE, then ELAPSED currently represents number of days between dstamp and its previous row's value per user_id. As it seems that that number is always less than 1 day (as you want result in hh24:mi:ss format), a simple option is to apply combination of
to_char(to_date(elapsed, 'sssss'), 'hh24:mi:ss')
and get the result. Step-by-step (so that you'd see which result you get for each step):
SQL> with temp (user_id, dstamp) as
2 (select 1, to_date('27.03.2022 08:08', 'dd.mm.yyyy hh24:mi') from dual union all
3 select 1, to_date('27.03.2022 05:30', 'dd.mm.yyyy hh24:mi') from dual union all
4 select 1, to_date('27.03.2022 02:28', 'dd.mm.yyyy hh24:mi') from dual
5 )
6 select user_id,
7 dstamp,
8 --
9 dstamp - lag(dstamp) over (partition by user_id order by dstamp) as elapsed_days,
10 --
11 round((dstamp - lag(dstamp) over (partition by user_id order by dstamp)) *
12 24 * 60 * 60) elapsed_seconds,
13 --
14 to_char(to_date(round((dstamp - lag(dstamp) over (partition by user_id
15 order by dstamp)) *
16 24 * 60 * 60), 'sssss'),
17 'hh24:mi:ss') result
18 from temp
19 order by user_id, dstamp;
USER_ID DSTAMP ELAPSED_DAYS ELAPSED_SECONDS RESULT
---------- ------------------- ------------ --------------- --------
1 27.03.2022 02:28:00
1 27.03.2022 05:30:00 .126388889 10920 03:02:00
1 27.03.2022 08:08:00 .109722222 9480 02:38:00
SQL>
However, if elapsed can be larger than 1 day so that it exceeds number of seconds in a day (that's 24 * 60 * 60 = 86400), that won't work so you'll have to actually calculate the result. However, query then becomes really ugly so you'd rather use another CTE (or a subquery) to first extract elapsed and work with it, instead of using dstamp - lag(...) all the time.
Therefore, I hope it is the timestamp after all.