Oracle SQL 开发人员 - 循环遍历行值

Question

I have a table below where I know the the Start and End Positions in the Position Column. The table is order by Start Time in DESC order.

ID	PREV End Time	Start Time	End Time	Position
1	1/1/2022 10:00	1/1/2022 10:30	1/1/2022 11:30	Start
1	1/1/2022 12:30	1/1/2022 13:30	1/1/2022 14:30	null
1	1/1/2022 15:30	1/1/2022 16:30	1/1/2022 17:30	End
1	1/1/2022 18:30	1/1/2022 19:30	1/1/2022 20:30	Start
1	1/1/2022 21:30	1/1/2022 22:30	1/1/2022 23:30	null
1	1/2/2022 0:30	1/2/2022 1:30	1/2/2022 2:30	null
1	1/2/2022 3:30	1/2/2022 4:30	1/2/2022 5:30	End

I want to be able to make two new columns where I can identify Trip 1 and 2 distinctly. And I also want to number each leg for each trip. Below, is the table I desire.

ID	PREV End Time	Start Time	End Time	Position	LEG	Trip
1	1/1/2022 10:00	1/1/2022 10:30	1/1/2022 11:30	Start	1	1
1	1/1/2022 12:30	1/1/2022 13:30	1/1/2022 14:30	null	2	1
1	1/1/2022 15:30	1/1/2022 16:30	1/1/2022 17:30	End	3	1
1	1/1/2022 18:30	1/1/2022 19:30	1/1/2022 20:30	Start	1	2
1	1/1/2022 21:30	1/1/2022 22:30	1/1/2022 23:30	null	2	2
1	1/2/2022 0:30	1/2/2022 1:30	1/2/2022 2:30	null	3	2
1	1/2/2022 3:30	1/2/2022 4:30	1/2/2022 5:30	End	4	2

Sometimes the data, isn't as perfect as the first row starting as the Start position. Sometimes it starts in the middle of a trip. How can I exclude? And sometimes the end, ends in the middle of a trip. How can I exclude?

Please let me know if you have any questions. Thank you for taking the time to read.

Answer 1

From Oracle 12, you can use MATCH_RECOGNIZE to do row-by-row processing:

SELECT ID,
       PREV_End_Time,
       Start_Time,
       End_Time,
       Position,
       Leg,
       Trip
FROM   table_name
MATCH_RECOGNIZE(
  PARTITION BY ID
  ORDER     BY start_time
  MEASURES
    MATCH_NUMBER() AS trip,
    COUNT(*) AS leg
  ALL ROWS PER MATCH
  PATTERN (start_row intermediate_row* end_row)
  DEFINE
    start_row        AS position = 'Start',
    intermediate_row AS position IS NULL,
    end_row          AS position = 'End'
)

Which, for the sample data:

CREATE TABLE table_name (ID, PREV_End_Time, Start_Time, End_Time, Position) AS
SELECT 1, DATE '2022-01-01' + INTERVAL '10:00' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '10:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '11:30' HOUR TO MINUTE, 'Start' FROM DUAL UNION ALL
SELECT 1, DATE '2022-01-01' + INTERVAL '12:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '13:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '14:30' HOUR TO MINUTE, NULL    FROM DUAL UNION ALL
SELECT 1, DATE '2022-01-01' + INTERVAL '15:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '16:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '17:30' HOUR TO MINUTE, 'End'   FROM DUAL UNION ALL
SELECT 1, DATE '2022-01-01' + INTERVAL '18:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '19:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '20:30' HOUR TO MINUTE, 'Start' FROM DUAL UNION ALL
SELECT 1, DATE '2022-01-01' + INTERVAL '21:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '22:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '23:30' HOUR TO MINUTE, NULL    FROM DUAL UNION ALL
SELECT 1, DATE '2022-01-02' + INTERVAL '00:30' HOUR TO MINUTE, DATE '2022-01-02' + INTERVAL '01:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '02:30' HOUR TO MINUTE, NULL    FROM DUAL UNION ALL
SELECT 1, DATE '2022-01-02' + INTERVAL '03:30' HOUR TO MINUTE, DATE '2022-01-02' + INTERVAL '04:30' HOUR TO MINUTE, DATE '2022-01-01' + INTERVAL '05:30' HOUR TO MINUTE, 'End'   FROM DUAL;

Outputs:

ID	PREV_END_TIME	START_TIME	END_TIME	POSITION	LEG	TRIP
1	2022-01-01 10:00:00	2022-01-01 10:30:00	2022-01-01 11:30:00	Start	1	1
1	2022-01-01 12:30:00	2022-01-01 13:30:00	2022-01-01 14:30:00	null	2	1
1	2022-01-01 15:30:00	2022-01-01 16:30:00	2022-01-01 17:30:00	End	3	1
1	2022-01-01 18:30:00	2022-01-01 19:30:00	2022-01-01 20:30:00	Start	1	2
1	2022-01-01 21:30:00	2022-01-01 22:30:00	2022-01-01 23:30:00	null	2	2
1	2022-01-02 00:30:00	2022-01-02 01:30:00	2022-01-01 02:30:00	null	3	2
1	2022-01-02 03:30:00	2022-01-02 04:30:00	2022-01-01 05:30:00	End	4	2

db<>fiddle here