Replace Nth linewise search result

Question

I want to replace all excluding first result

I have txt file:

AAA
BBB
CCC
AAA
BBB
CCC
AAA
BBB
CCC

I want to get this:

AAA
BBB <-- stay same
CCC
AAA
XXX <-- 2nd find replaced
CCC
AAA
XXX <-- 3rd and nth find replaced
CCC

I looking something similar to this, but for whole lines, not for words in lines

sed -i 's/AAA/XXX/2' ./test01

I've edited out the answer from question... to indicate solved status, please accept the answer that helped you by clicking the tick button.. — Sundeep, Apr 09 '18 at 14:14
Obciously you want to replace `BBB` wtih `XXX` and not `AAA` — George Vasiliou, Apr 14 '18 at 22:57

score 4 · Answer 1 · answered Apr 09 '18 at 13:52

4

Use branching:

sed -e'/BBB/ {ba};b; :a {n;s/BBB/XXX/;ba}'

I.e. on the first BBB, we branch to :a, otherwise b without parameter starts processing of the next line.

Under :a, we read in a new line, replace all BBB by XXX and branch to a again.

answered Apr 09 '18 at 13:52

choroba

231,213
25
204
289

score 0 · Answer 2 · answered Apr 09 '18 at 13:55

0

Following awk may also help you on same.

awk '{++a[$0];$0=$0=="AAA"&&(a[$0]==2||a[$0]==3)?$0 ORS "XXX":$0} 1' Input_file

answered Apr 09 '18 at 13:55

RavinderSingh13

130,504
14
57
93

This isn't working it's second and third BBB line which should be replaced. – Claes Wikner Apr 10 '18 at 20:31

score 0 · Answer 3 · answered Apr 09 '18 at 14:19

$ # replace all occurrences greater than s
$ # use gsub instead of sub to replace all occurrences in line
$ # whole line: awk -v s=1 '$0=="AAA" && ++c>s{$0="XXX"} 1' ip.txt
$ awk -v s=1 '/AAA/ && ++c>s{sub(/AAA/, "XXX")} 1' ip.txt 
AAA
BBB
CCC
XXX
BBB
CCC
XXX
BBB
CCC

$ # replace exactly when occurrence == s
$ awk -v s=2 '/AAA/ && ++c==s{sub(/AAA/, "XXX")} 1' ip.txt 
AAA
BBB
CCC
XXX
BBB
CCC
AAA
BBB
CCC

Further reading: Printing with sed or awk a line following a matching pattern

score 0 · Answer 4 · answered Apr 14 '18 at 22:39

0

awk  '/BBB/{c++;if(c >=2)sub(/BBB/,"XXX")}1' file

AAA
BBB
CCC
AAA
XXX
CCC
AAA
XXX
CCC

answered Apr 14 '18 at 22:39

Claes Wikner

1,457
1
9
8

score 0 · Answer 5 · answered Apr 14 '18 at 23:08

As soon as your file does not contain null chars (\0) you can fool sed to consider the whole file as a big string by intstructing sed to separate records using null char \0 instead of the default \n with sed option -z:

$ sed -z 's/BBB/XXX/2g' file66
AAA
BBB
CCC
AAA
XXX
CCC
AAA
XXX
CCC

/2g at the end means from second match and then globally.

You can combine -i with -z without problem.

Replace Nth linewise search result

5 Answers5