You can't have a space between the variable= and the value to be assigned to that variable. The syntax you used, sets st to an empty string and includes it in the environment of the command that follows. But that won't get rid of the Thu Jan 29 or the CST 2015

Note that your original post didn't say anything about the string started at appearing in the input (and was not in your sample input).

Assuming you're working in an English locale, you could try something like:

Code:

awk '
match($0, /[[:digit:]]{1,2}:[[:digit:]]{2}:[[:digit:]]{2} *([AP]M){0,1}/) {
	print substr($0, RSTART, RLENGTH)
}' file

The ERE can be simplified if you only have 24-hour format times.

With the following contents in your input file:

Code:

Nothing found on this line
 a.sh start time is Fri Jan  9 17:17:33 CST 2015
a.sh end time is Fri Jan  9 17:47:33 CST 2015
Single digit hour Fri Jan  9  5:50:59 PM CST 2015
Double digit hour Sat Jan 10 11:22:33 AM CST 2015
No space single Fri Jan  9  5:50:59PM CST 2015
No space double Sat Jan 10 11:22:33AM CST 2015

it produces the output:

Code:

17:17:33 
17:47:33 
5:50:59 PM
11:22:33 AM
5:50:59PM
11:22:33AM

Fairly simple changes would allow you to find more than one timestamp on a line.

If you want to try this on a Solaris/SunOS system, change awk to /usr/xpg4/bin/awk, /usr/xpg6/bin/awk, or nawk