[SOLVED] [sed] Returns whole string instead of part

littlebigman · 07-03-2011, 10:49 AM

Hello

I'm not used to using "sed", but I need to use it to extract part of a string piped to it.

However, the "\1" mentioned in the tutorials I read doesn't extract the token between the two brackets and returns the whole input string:

Code:

/tmp# echo "before [myfile.txt] after" | sed 's/\[\(.+\)\]/\1/'                                                      before [myfile.txt] after

I expected simply "myfile.txt".

Does someone know why?

Thank you.

acid_kewpie · 07-03-2011, 10:58 AM

well yes you saying replace that part of the string with itself (minus the square brackets). You need to also match whatever is at the start and whatever is at the end.

/tmp# echo "before [myfile.txt] after" | sed -e 's/^.*\[$.*$\]*.$/\1/'

sycamorex · 07-03-2011, 10:58 AM

Try the following:

Code:

$ echo "before [myfile.txt] after" | sed 's/.*\[\(.*\)\].*/\1/'

acid_kewpie · 07-03-2011, 11:06 AM

Actually I was interpreting it wrong, it's not working as it's not matching. To that end I've always really struggled to remember / know why .+ never works here. IF the .+ was working, then the pattern would match and then be giving the output I anticipated:

# echo "before [myfile.txt] after" | sed 's/\[$.*$\]/\1/'
before myfile.txt after

littlebigman · 07-03-2011, 11:16 AM

Thanks guys. It's strange that sed doesn't seem to use "+" as usual, ie. "one or more characters".

Diantre · 07-03-2011, 11:44 AM

Quote:

Originally Posted by littlebigman

Thanks guys. It's strange that sed doesn't seem to use "+" as usual, ie. "one or more characters".

The + has to be escaped:

Code:

echo "before [myfile.txt] after" | sed 's/.*\[\(.\+\)\].*/\1/'

Returns myfile.txt.

acid_kewpie · 07-03-2011, 11:49 AM

It does, but that is mental... what's the flag to properly interpret the +?

Diantre · 07-03-2011, 11:53 AM

Quote:

Originally Posted by acid_kewpie

It does, but that is mental... what's the flag to properly interpret the +?

It's mental, I know. But it's in the sed manual. Check "info sed" section 3.3 "Overview of Regular Expression Syntax". Don't know (yet) if there's a flag to interpret + without escaping.

Diantre · 07-03-2011, 12:11 PM

Ok, I got it. It's extended regular expressions. From the sed info page:

Quote:

Appendix A Extended regular expressions
***************************************

The only difference between basic and extended regular expressions is in
the behavior of a few characters: `?', `+', parentheses, and braces
(`{}'). While basic regular expressions require these to be escaped if
you want them to behave as special characters, when using extended
regular expressions you must escape them if you want them _to match a
literal character_.

So the regex would be:

Code:

echo "before [myfile.txt] after" | sed -r 's/.*\[(.+)\].*/\1/'

Returns myfile.txt using an unescaped +.

littlebigman · 07-04-2011, 07:07 AM

Thanks for the tip on "-r".