sed question

[ed91270]

Hi,

Easy question hopefully. I have a file with the contents

foo
foo

bar
bar
bar

foo
foo

And I want to add a line 'a pointless line' after the last line containing the string bar. How would I do this with sed ?

I managed :

sed ' /bar/ a\
a pointless line' < ./foobar.txt

which prints :

foo
foo

bar
a pointless line
bar
a pointless line
bar
a pointless line

foo
foo


So not quite the desired effect as I only wanted it to print:

foo
foo

bar
bar
bar
a pointless line

foo
foo

[secondmouse]

Suppose your_file is

foo
foo

bar
bar
bar

foo
FOO

I change the last line to FOO so you know in the result file which line is first which is last, just for identification purpose, it has nothing to do with the commands that I am gonna use.

tac your_file | sed '0,/bar/s#bar#pointless line\nbar#' | tac > new_file

It would be nice if sed supports searching from the end of the file, in that case the above command can be written as

sed '$,/bar/s#bar#bar\npointless line#' your_file > new_file

A bit of explanation:
tac: the opposite of cat, that is, prints file content from the end to the beginning.
When file is reverse-printed, search from line 0 to the first occurrence of bar and replace that with two lines pointless line\nbar, this in effect inserts pointless line above the first occurrence of bar, then another reverse print gets what you want.

[ghostdog74]

$ more file
foo
foo

bar
bar
bar

foo
foo
bar

$ awk '{a[++d]=$0}/bar/{s=FNR}END{for(i=1;i<=d;i++){if (i==s){print a[i],"\nword"}else{print a[i]}}}' file
foo
foo

bar
bar
bar

foo
foo
bar
word

[ed91270]

Suppose your_file is

foo
foo

bar
bar
bar

foo
FOO

I change the last line to FOO so you know in the result file which line is first which is last, just for identification purpose, it has nothing to do with the commands that I am gonna use.

tac your_file | sed '0,/bar/s#bar#pointless line\nbar#' | tac > new_file

It would be nice if sed supports searching from the end of the file, in that case the above command can be written as

sed '$,/bar/s#bar#bar\npointless line#' your_file > new_file

A bit of explanation:
tac: the opposite of cat, that is, prints file content from the end to the beginning.
When file is reverse-printed, search from line 0 to the first occurrence of bar and replace that with two lines pointless line\nbar, this in effect inserts pointless line above the first occurrence of bar, then another reverse print gets what you want.

I agree, I couldn't work out how to get sed to search from the end of a file, but anyway, your solution is clever and simple! It never occured to me to use tac. Thanks very much!

Ed

[ed91270]

$ awk '{a[++d]=$0}/bar/{s=FNR}END{for(i=1;i<=d;i++){if (i==s){print a[i],"\nword"}else{print a[i]}}}' file

Also an impressive answer that verges on the dark side of awk

[secondmouse]

I agree, I couldn't work out how to get sed to search from the end of a file, but anyway, your solution is clever and simple! It never occured to me to use tac. Thanks very much!

Ed

I am glad it helps. You are welcome.