Oracle interval arithmetic for aggregating rows - sql

Question

I'm writing a query to group rows by their timestamps into aggregation blocks. The query parameters include the start time start of the first aggregation block and a positive integer period which is the length of each and every aggregation block in minutes. Given a row's timestamp row_stamp, which is later than start, I want to calculate its block_start such that
block_start <= row_stamp < block_start + period minutes, and
block_start = start + (N * period minutes) where N is a nonnegative integer
It's easy enough to find row_stamp - start, which appears to be an interval. I figure I'll want block_start to be either start + (period minutes) * FLOOR((row_stamp - start) / (period minutes)) or row_stamp - MOD(row_stamp - start, period minutes). I know neither of these is exact syntax, but I think you see the algorithms I'm going for. Unfortunately, it looks like neither FLOOR nor MOD works well with an interval as its first parameter. What's the recommended way to turn an interval into a number of minutes and then, after the math, turn it back to an interval again?
My apologies for not mentioning earlier that I'm looking for the number of rows in each aggregation block. Edited to add an example.
CREATE TABLE "DEV_JKNIGHT" ( "ROW_STAMP" TIMESTAMP (6) NOT NULL );
insert into dev_jknight values (TIMESTAMP '2022-06-27 14:27:00');
insert into dev_jknight values (TIMESTAMP '2022-06-27 14:32:00');
insert into dev_jknight values (TIMESTAMP '2022-06-27 14:33:00');
insert into dev_jknight values (TIMESTAMP '2022-06-27 15:01:00');
insert into dev_jknight values (TIMESTAMP '2022-06-27 16:32:00');
Suppose the query parameters are a start time of '2022-06-27 14:15:00' and a period of 15 minutes. Then the aggregation blocks are as follows. The first begins at the specified start time and lasts for the period number of minutes. The next block starts immediately after the first one ends and lasts the same length.
14:15:00 - 14:30:00, June 27
14:30:00 - 14:45:00, June 27
14:45:00 - 15:00:00, June 27
15:00:00 - 15:15:00, June 27
and so on
If I run the query with those parameters, then I'm looking for these four rows of output.
Block start = '2022-06-27 14:15:00', count = "1"
Block start = '2022-06-27 14:30:00', count = "2"
Block start = '2022-06-27 15:00:00', count = "1"
Block start = '2022-06-27 16:30:00', count = "1"
The first row of output indicates that there is one, and only one, dev_jknight row in the aggregation block which starts at 14:15:00 -- namely, the row at 14:27:00.
The second row of output indicates that there are two dev_jknight rows in the aggregation block which starts at 14:30:00 -- the rows at 14:32 and 14:33.
Because there are zero dev_jknight rows in the third aggregation block (14:45 - 15:00), there is no output row for it.
The last two rows of output indicate that there is one dev_jknight rows in the aggregation block which starts at 15:00 and one in the block which starts at 16:30.
Thank you for your patience while I clarified this.

Answer 1

Turn an interval into a number of minutes?
Use EXTRACT:
SELECT EXTRACT(DAY FROM value) * 24 * 60
+ EXTRACT(HOUR FROM value) * 60
+ EXTRACT(MINUTE FROM value)
+ EXTRACT(SECOND FROM value) / 60 AS minutes
FROM (
SELECT INTERVAL '1 12:34:56.789' DAY TO SECOND AS value
FROM DUAL
);
Which outputs:
MINUTES
2194.946483333333333333333333333333333333
and then turn it back to an interval again?
Multiply a 1 minute interval by the number of minutes:
SELECT INTERVAL '1' MINUTE * 2194.94648333333 AS interval_value
FROM DUAL;
or, use the NUMTODSINTERVAL function:
SELECT NUMTODSINTERVAL(2194.94648333333, 'MINUTE') AS interval_value FROM DUAL;
INTERVAL_VALUE
+000000001 12:34:56.789000000
db<>fiddle here