learning sed, having trouble extracting text between to values

Question

I have a file with text in it that looks like this:

ssid="sdfsdf" #psk="sdfsfsdf" psk=zzzs93j03r093ur0fjfs39uj }

I'm learning sed and want to use it to extract just the string that starts with zzz.

I though I could just grab everything between " psk=" and " }" but this does not seem to work:

sed 's_ psk=\(.*\) }_\1_' /tmp/myfile

I am also curious why it doesn't work. psk appears in the text twice but I though searching for " psk" would distinguish it from "#psk"

score 1 · Accepted Answer · edited May 23 '17 at 10:29

1

With GNU sed:

sed 's/.* psk=\([^ ]*\).*/\1/' file

Output:

zzzs93j03r093ur0fjfs39uj

edited May 23 '17 at 10:29

Community

answered May 29 '16 at 21:25

Cyrus

Add two `.*`: `sed 's_.* psk=$.*$ .*}_\1_' /tmp/myfile` – Cyrus May 29 '16 at 21:28
1

Ah because I was only matching some of the text before and after and what I need to do was match between everything_before_substring( )everything_after_substring right? – red888 May 29 '16 at 21:32
you were trying to match a `' '` before the psk not a `'#'` as is present? Or I'm misreading.. – Stefan Hegny May 29 '16 at 21:38
@red888: yes, you want to replace **complete line** by part in round brackets. – Cyrus May 30 '16 at 06:00

score 0 · Answer 2 · answered May 29 '16 at 21:45

Your command are working just fine, It replaces everything from psk=...$ with zzzs..., what you want to add is .* in the front:

% sed 's_.* psk=\(.*\) }_\1_'
zzzs93j03r093ur0fjfs39uj

You can change your regex a little to handle cases where psk is not at the end:

sed 's/.*[^#]psk=\([[:alnum:]]*\).*/\1/'

2 Answers2