find elapse time with date rollover

I have a array with a bunch of times formatted as HH:MM:SS as follows:

11:58:00

11:59:00

00:00:02

00:02:04

I'm trying to find the elapse time in seconds from begining time. The time data I have starts a new day (11pm to 12am). All my attemps to use etime or subtract datenum result in returning a negative time when trying to subtract 00:00:02 from 11:58:00. I havn't figured out how to make matlabe know the 00:00:02 is actually a new date, since there is no actual date defined in my data just time.

I would like to get the following result, in seconds from start:

0

60

62

182

Answer 1

[Peter Perkins]

Get away from using datenum. Can't say this strongly enough.

>> tstr=["11:58:00"; "11:59:00"; "00:00:02"; "00:02:04"];
>> tod = duration(tstr)
tod = 
  4×1 duration array
   11:58:00
   11:59:00
   00:00:02
   00:02:04
>> newDay = [false;diff(tod)<0]
newDay =
  4×1 logical array
   0
   0
   1
   0

Don't know if you klnow what day the time start on. If you cross a Daylight Saving Time shift, it will matter. Use datetime, not datenum.

>> day0 = datetime(2021,7,26);
>> t = day0 + caldays(cumsum(newDay)) + tod
t = 
  4×1 datetime array
   26-Jul-2021 11:58:00
   26-Jul-2021 11:59:00
   27-Jul-2021 00:00:02
   27-Jul-2021 00:02:04
>> elapsed = seconds(t - t(1))
elapsed =
           0
          60
       43322
       43444

My answer is different than yours. Guess what? Your time stamps have no way of indicating am or pm. That would seem to be a problem. If you can, use 24-hour time stamps. If not, then how do you know if it's 11 am or pm? about the only thing you can do is assume you will have timestamps in both the first and second half of every day, and my newDay variable will indicate both noon and midnight crossings.

[Chunru]

It seems that here is no foolproof answer to the question where date time info is ambiguous. Just make the best guess.

Answer 2

[Chunru]

tstr=[
"11:58:00"
"11:59:00"
"00:00:02"
"00:02:04"];
t = datenum(tstr, 'HH:MM:SS')

t = 4×1

1.0e+05 * 7.3816 7.3816 7.3816 7.3816

d = diff(t)*24*3600;
% check for negative difference
d(d<0) = d(d<0)+12*3600;
d = [0; d]

d = 4×1

0 60.0000 62.0000 122.0000