Extract one word after a specific word on the same line [duplicate]

Question

How can I extract a word that comes after a specific word in Linux (csh)? More precisely, I have a file which has a single line which looks like this:

[some useless data] --pe_cnt 100 --rd_cnt 1000 [some more data]

I want to extract the number 100 which is after the --pe_cnt word. I cannot use sed as that works only if you want to extract an entire line. Maybe I can use awk?

Also, I have multiple files that have different values instead of 100 so I need something that extracts the value but doesn't depend on the value.

Qulllon

"awk" is one possibility. For example, echo 5 4 | awk '{ print $2 }' should display "4". – Feoffee

Answer 1

With awk:

awk '{for(i=1;i<=NF;i++) if ($i=="--pe_cnt") print $(i+1)}' inputFile

Basically loop over each word of the line. When you find the first you are looking for, grab the next word and print it.

With grep:

grep -oP "(?<=--pe_cnt )[^ ]+" inputFile

Legendre

You need to add a space at the end of the lookbehind pattern in grep. – Somatology

Yep sorry didn't test it. Thank you! :) – Legendre

N.B. grep on Mac OSX 10.8 doesn't have -P. – Desberg

JS웃: Thanks a lot for the quick reply. The first command worked perfectly! The second command did not give any output though. – Qulllon

@Qulllon Sorry I had missed a space in the look behind which was corrected by ctn. It should work now. – Legendre

Works great! But can you explain the grep regexp, in particular what (?<=--pe_cnt ) does? – Keeping

Nevermind, it's a perl regexp using look around assertions, as I found here. – Keeping

The sed answer is more efficient and more idiomatic than this answer, and should probably be the most upvoted, if not the accepted answer. – Norahnorbert

your answer helped me a lot. Thank you – Blindworm

Answer 2

You can use sed. Just make a group of want you want to match and replace the whole line with the group:

sed -n 's/^.*pe_cnt\s\+\([0-9]\+\).*$/\1/p' file

Somatology

There was an error in the pattern previously but it's fixed now. – Somatology

Answer 3

If there is a single-space character between --pe_cnt and 100, you may be able to use lookahead and lookbehind assertions

grep -oP '(?<=--pe_cnt\s)\d+(?=\s+--rd_cnt)'

Iconolatry

Probably don't even need the --rd_cnt, it doesn't necessarily matter what the following options are – Desberg

Ok. Thanks for the tip. I'm a newbie to Linux so these tips help a lot. – Qulllon

Answer 4

This may be the shortest version:

grep -oP 'pe_cnt \K[^ ]+' file

Egon