simple pattern match with awk, sed
 

I have a line with the following pattern:

some text (x, y)

where x y are the numbers
I am looking for a simple awk or sed or bash script that would print out just x and y

thanks

depends how strict the formatting is but this should work...

$ echo "some text (123, 456)" | sed -e "s/^.*(\([0-9]*\),\ \([0-9]*\)).*$/\1 \2/"
123 456

Can anyone tell *me* though, why it doesn't work if any of the *'s are replaced with +'s? * = 0 or more, + = 1 or more, so i'm clearly missing something, and have done for a long long time...

IIRC, a bare + is not special to sed, you need to escape it.

If you want sed to accept + you need to use the -r switch.


Cheers,
Tink

running the command above = simply running echo. Am i missing something?

Apparently... it works fine here. What does your exact string
actually look like?


Cheers,
Tink

*blink*

-r extended regex

Where the frell did that come from? I never noticed that before. I'm using Slack 10.2 here (too lazy to upgrade) and I just noticed I have it here, probably been there for years. Gah, what sneaks past my myopia, gets shot down by my attention deficit disorder.

no, i tried exactly the same line
prints 'some text (123, 456)'

ok ,fixed it.... space missing.. thanx a lot

Btw, depending on your input, you might want to use no printing mode, so lines that don't match will be ignored. Give sed the -n option and whack a p after the final delimiter,

If all your input is in the format you specified it shouldn't matter, but if it has extraneous stuff you're not interested in, it will be printed out by default without the -n option.

This also makes it easy to see if your pattern is working, because sed gives you no output at all if it doesn't match, rather than simply echoing the input.

cool.. thanx!