CODEWITHSUNDEEP
sed
Sean Nguyen
Sean Nguyen

Reputation: 13128

how to use result from a pipe in the next sed command?

I want to use sed to do this. I have 2 files:

keys.txt:
host1
host2

test.txt
host1 abc
host2 cdf
host3 abaasdf

I want to use sed to remove any lines in test.txt that contains the keyword in keys.txt. So the result of test.txt should be

host3 abaasdf

Can somebody show me how to do that with sed?

Thanks,

Upvotes: 2

Views: 301

Answers (3)

potong
potong

Reputation: 58430

This might work for you (GNU sed):

sed 's|.*|/^&\\>/d|' keys.txt | sed -f - test.txt

Upvotes: 0

glenn jackman
glenn jackman

Reputation: 246817

fgrep -v -f is the best solution. Here are a couple of alternatives:

A combination of comm and join

comm -13 <(join keys.txt test.txt) test.txt

or awk

awk 'NR==FNR {key[$1]; next} $1 in key {next} 1' keys.txt test.txt

Upvotes: 2

Zsolt Botykai
Zsolt Botykai

Reputation: 51613

I'd recommend using grep for this (especially fgrep since there are no regexps involved), so

fgrep -v -f keys.txt test.txt

does it fine. With sed quickly this works:

sed -i.ORIGKEYS.txt ^-e 's_^_/_' -e 's_$_/d_' keys.txt
sed -f keys.txt test.txt

(This modifies the original keys.txt in place - with backup - to a sourceable sed script.)

Upvotes: 2

Related Questions