Hello, I am very new to shell scripting and have a directory path such as:
/usr/dev/blah/Arch/release/812-1235-P05/files/list and I want to output:

812-1235-P05

I think using sed with a regex like [0-9]-[0-9]-[0-9]?[a-z] would be the way to go, but I am having much trouble getting it to work. Any suggestions? Thank you.

you need to add some quantifiers and limiters in there
which will look for anything followed by \( exactly 3 digits followed by a - followed by exactly 4 digits, followed by a - followed by a combination of 3 letters and/or digits \) followed by anything and replace it all with the part above in \( .. \)

Thank you so much for your reply. Looks like it should work, I am piping
/usr/dev/blah/Arch/release/812-1235-P05/files/list which is from a previous grep command into sed so like and it just displays the whole path again, it doesn't strip out everything around the 812-1235-P05. Am I doing something completely wrong? I must be.

Thanks again!

That is strange, if I pipe that string directly into sed on my machine I get the required result. Can you run the grep without the sed and show us the exact output of the grep (including all spacing and 'wierd' chars).

my sed version:

chris@druid: ~ $ sed --version
GNU sed version 4.1.5

I got it, my directory path also has a timestamp at the end of it and I didn't enclose it with quotes and it used some <>'s Newbie mistake I guess. Thank you much for all your help...certainly made my day easier.