两个非连续行之间的时间？ （MySQL的）

Question

I've looked here Calculate the time difference between of two rows .

I modified the code here:

SELECT A.Test_ID, B.TestDate - A.TestDate AS DaysBetweenTests FROM Exams A JOIN Exams B ON B.Test_ID = A.Test_ID WHERE (SELECT B.Test_ID FROM B.Exams ORDER BY Animal_ID, B.TestDate LIMIT 1 OFFSET 1)

This query isn't complete. I'm trying to calculate the days between TestDate with the next TestDate for the same Animal_ID . The problem here is that an Animal_ID TestDate is not directly consecutive with Test_ID . I can't use Test_ID +1 .

Sample Exams table:

Test_ID  |  TestDate  |  Animal_ID
1           2013-01-01   100
2           2007-06-18   162
3           2013-02-01   100
4           2013-04-16   100

The TestDate for a single Animal_ID is always increasing in order, thus a higher Test_ID number .

I'm having trouble with the subquery.

Answer 1

This query repeats the subquery that determines the next test date. You don't necessarily need to do that in production; it just helps to see the data when you're figuring things out.

SELECT T1.Animal_ID, 
       T1.TestDate,
       (SELECT min(TestDate)
        FROM Exams
        WHERE Animal_ID = T1.Animal_ID
          AND TestDate > T1.TestDate
        GROUP BY Animal_ID
        ) AS Next_TestDate, 
        datediff((SELECT min(TestDate)
                  FROM Exams
                  WHERE Animal_ID = T1.Animal_ID
                    AND TestDate > T1.TestDate
                  GROUP BY Animal_ID
                  ), T1.TestDate) AS Elapsed_Days
FROM Exams T1
ORDER BY Animal_ID, TestDate;

Answer 2

select
t.*,
coalesce(timestampdiff(day, @prevDate, if(@prevAnimal = Animal_ID, TestDate, @prevDate)), 0) as timedifference_in_days,
@prevAnimal := Animal_ID,
@prevDate := TestDate
from
t
, (select @prevDate := null, @prevAnimal := null) as var_init
order by Animal_ID, TestDate

• see it working live in an sqlfiddle

---

CREATE TABLE t
    (`Test_ID` int, `TestDate` date, `Animal_ID` int)
;

INSERT INTO t
    (`Test_ID`, `TestDate`, `Animal_ID`)
VALUES
    (1, '2013-01-01', 100),
    (2, '2007-06-18', 162),
    (3, '2013-02-01', 100),
    (4, '2013-04-16', 100)
;

Output:

| TEST_ID |                        TESTDATE | ANIMAL_ID | TIMEDIFFERENCE_IN_DAYS | @PREVANIMAL := ANIMAL_ID | @PREVDATE := TESTDATE |
|---------|---------------------------------|-----------|------------------------|--------------------------|-----------------------|
|       1 |  January, 01 2013 00:00:00+0000 |       100 |                      0 |                      100 |            2013-01-01 |
|       3 | February, 01 2013 00:00:00+0000 |       100 |                     31 |                      100 |            2013-02-01 |
|       4 |    April, 16 2013 00:00:00+0000 |       100 |                     74 |                      100 |            2013-04-16 |
|       2 |     June, 18 2007 00:00:00+0000 |       162 |                      0 |                      162 |            2007-06-18 |