remove lines based on file input pattern using sed

Question

I have been trying to solve a simple sed line deletion problem. Looked here and there. It didn't solve my problem. My problem could simply be achieved by using sed -i'{/^1\|^2\|^3/d;}' infile.txt which deletes lines beginning with 1,2 and 3 from the infile.txt.

But what I want instead is to take the starting matching patterns from a file than manually feeding into the stream editor.

E.g: deletePattern

1 
3
2

infile.txt

1 Line here
2 Line here
3 Line here 
4 Line here

Desired output

4 Line here

Thank you in advance,

You could read deletepat.txt into a string, and then try that string as the content of a character class in sed. Something like while read -r ch; do a+=$ch; done <deletepat.txt. Then sed -i "{/^[$a]/d;}" infile.txt (note: you will need to test) — David C. Rankin
– David C. Rankin, Commented Jun 29, 2015 at 6:49

anubhava · Accepted Answer · 2015-06-30 06:56:53Z

1

This grep should work:

grep -Fvf deletePattern infile.txt
4 Line here

But this will skip a line if patterns in deletePattern are found anywhere in the 2nd file.

More accurate results can be achieved by using this awk command:

awk 'FILENAME == ARGV[1] && FNR==NR{a[$1];next} !($1 in a)' deletePattern infile.txt
4 Line here

edited Jun 30, 2015 at 6:56

answered Jun 29, 2015 at 6:40

anubhava

790k67 gold badges603 silver badges671 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

deepseefan Over a year ago

the grep is an excellent one. awk works but a glitch if the pattern file is empty, it produces empty file. Just try touch deletePattern and use awk. Thank you.

anubhava Over a year ago

Thanks for your point regarding awk command. I've corrected that as well.

David C. Rankin · Accepted Answer · 2015-06-29 07:11:14Z

1

Putting together a quick command substitution combined with a character class will allow a relatively short oneliner:

$ sed -e "/^[$( while read -r ch; do a+=$ch; done <pattern.txt; echo "$a" )]/d" infile.txt
4 Line here

Of course, change the -e to -i for actual in-place substitution.

answered Jun 29, 2015 at 7:11

David C. Rankin

85.1k6 gold badges67 silver badges95 bronze badges

2 Comments

deepseefan Over a year ago

It works perfect except when the pattern is an empty file. touch pattern.txt and running sed produces _ unterminated address regex_ error . Thank you.

David C. Rankin Over a year ago

Glad I could help. And... you got me... I never thought to test it on a blank file. All the different answers just go to show there is always more than one way to skin a cat in shell computing. Pick your favorite cat!

Sato Katsura · Accepted Answer · 2015-06-29 07:13:44Z

1

With GNU sed (for -f -):

sed 's!^[0-9][0-9]*$!/^&[^0-9]/d!' deletePattern | sed -f - infile.txt

The first sed transforms deletePattern into a sed script, then the second sed applies this script.

answered Jun 29, 2015 at 7:13

Sato Katsura

3,10417 silver badges22 bronze badges

Comments

Maroun · Accepted Answer · 2015-06-29 06:42:22Z

0

Try this:

sed '/^[123]/ d' infile.txt

answered Jun 29, 2015 at 6:42

Maroun

96.3k30 gold badges195 silver badges249 bronze badges

Collectives™ on Stack Overflow

remove lines based on file input pattern using sed

4 Answers 4

2 Comments

2 Comments

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

2 Comments

2 Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related