如何在 SQL Server 中对 1:15 这种类型的值求和

Question

How to sum 1:15 this type of values in SQL server

I try this query:

select sum((SELECT REPLACE(
           CONVERT(VARCHAR(8), TotalTime, 108),
           ':',
           '.'
          ) where Associate='4587')
          )
from Tracking
where Associate='4587'

error showing:

Msg 130, Level 15, State 1, Line 45 Cannot perform an aggregate function on an expression containing an aggregate or a subquery.

Answer 1

You can try this using conditional aggregation

select sum(case when Associate='2614' then REPLACE(CONVERT(VARCHAR(8), TotalTime, 108), ':', '.' ) end) 
from Tracking where Associate='4587'

Answer 2

Well, you have chosen the wrong data type to store this value, and this can lead to to a lot of problems, starting with ill-formatted values (What's stopping you from having a value like '1:86' , or even worst - 'dyjst' in that column?), through having to perform needlessly complicated queries to get some of the results (just like this one), ambiguous values such as 13:2 - Is that 20 minutes or just 2?, and there are a bunch of other problems I didn't write because I think you should get the picture by now.

Surly the best way to solve this issue is to change the way you store the data. If this data represents an hour in a day, you should be using the Time data type. If it stores a duration that might be larger than 24 hours, you might want to simply calculate the number of minutes and store that in an int.

However, assuming that for some reason (and yes, some of them might be valid reasons) you can't change the way you store this data - you will need to write a query that will translate this data for you to the number of minutes it represents (This would also be the case if you where to store it in the Time data type - you can't do SUM on Time values), and then sum that value. If you want to display it back as the same format ( [h]h:mm ), you will need to translate the value from the sum back to this format.

I've put together a small demo that will show you one way to translate a string formatted as hh:mm to number of minutes and back - Once you get the number of minutes as an int, using sum is easy. Note : I didn't add any validation for the content of the strings!

So first, create and populate sample table ( Please , save us this step in your future questions):

DECLARE @Tracking AS TABLE
(
    TotalTime varchar(10),
    Associate varchar(10)
);
INSERT INTO @Tracking (TotalTime, Associate) VALUES
('1:15', '4587'),
('1:05', '4588'),
('22:33', '4589'),
('73:50', '4590'),
('120:85', '4591'); -- Note: this will translate back as 121:25

Then, create a common table expression to translate the value in the TotalTime column to the number of minutes it represents:

;WITH CTE AS
(
    SELECT  Associate, 
            TotalTime As TotalTime,
            CAST(LEFT(TotalTime, CHARINDEX(':', TotalTime) - 1) AS Int) * 60 +
            CAST(RIGHT(TotalTime, LEN(TotalTime) - CHARINDEX(':', TotalTime)) As int)
            As NumberOfMinutes
    FROM @Tracking
)

And a select statement to translate it back to a time string:

SELECT  Associate, TotalTime, NumberOfMinutes,
        CAST((NumberOfMinutes - (NumberOfMinutes % 60)) / 60 As varchar(3)) + ':' +
        RIGHT('00' + CAST(NumberOfMinutes % 60 As varchar(2)), 2) As TimeString
FROM CTE

Results:

Associate   TotalTime   NumberOfMinutes     TimeString
4587        1:15        75                  1:15
4588        1:05        65                  1:05
4589        22:33       1353                22:33
4590        73:50       4430                73:50
4591        120:85      7285                121:25

Answer 3

This is an example of how you can sum a time column. It returns a Days, Time and a datetime. Time will be relevant as long as the sum is less than 24 hours otherwise it should be combined with Days. The datetime will include the day part as days after 1900-01-01.

The query require valid data with data format hh:mi or h:mi:

DECLARE @ table(col int, TotalTime time)

INSERT @ values(1,'1:15'),(1,'0:30'),(1,'0:05'),(1,'23:30'),(2,'1:15'),(2,'1:30')
SELECT
  col,
  datediff(d, 0, cast(sum(cast(cast(totaltime as datetime) as float)) as datetime)) Days,
  cast(cast(sum(cast(cast(totaltime as datetime) as float)) as datetime) as time(0)) Time,
  cast(sum(cast(cast(totaltime as datetime) as float)) as smalldatetime) DT
FROM 
  @
GROUP BY col

Result:

col Days  Time     DT
1   1     01:20:00 1900-01-02 01:20:00
2   0     02:45:00 1900-01-01 02:45:00