Shell Programming and Scripting

Hi,

I have a file containing timestamps (at micro-seconds granularity). It looks like the following:

06:49:42.383818
06:49:42.390190
06:49:42.392308
06:49:42.392712
06:49:42.393437
06:49:42.393960
06:49:42.402115

Now I need a sed/awk script to take the difference of two successive time values i.e 2nd value - 1st value, 3rd value - 2nd value , 4th - 3rd and so on....and print them in a file.

So if the input file contains 7 entries, the output file should have 6 entries.

Since the input file is quite big (large number of entries), I need sed/awk script to process it quickly.

Any help will be highly appreciated.

Thanks !

There's no check for a clock roll-over, but this might get you started:

Code:

 awk '
        {
                split( $1, a, ":" )
                t = a[1] * 3600;
                t += a[2] * 60;
                t += a[3];

                if( NR > 1 )
                        printf( "%.3f\n", t - last);

                last = t;
        }
'  <input-file

This is a one-liner version of agama's approach which may speed thing up a little bit.

Code:

awk -F: '{t=$1*3600+$2*60+$3} NR>1{printf( "%.6f\n", t-p)}{p=t}' infile