Is there a simple way to save variables from sed/awk?

I'm struggling with the following:

Current File:

text1,text2,add(a,b),text3

Target File:

text1,text2,a b c,text3

where a and b vary.

Example:

text1,text2,add(11,12),text3
text1,text2,add(41,42),text3

would become:

text1,text2,11 12 c,text3
text1,text2,41 42 c,text3

I would like to do something like:

sed -i "s/add(*,*)/ a b c/g"

But I'm pretty sure the use of wildcards is completely wrong.

New to bash scripting, any advice is appreciated. Thanks.

Edit: To clarify, I want to save a and b into variables so that I can print them again.

asked Apr 23, 2020 at 7:58

help_me

12 bronze badges

Sorry but your question is unclear to me. What do you mean with $a and $b? Do you want to replace the content of add(a,b) with $a $b c where $a is the content of a shell variable named a (dito b)?
– kvantour
Commented Apr 23, 2020 at 8:00
Is this what you are after? $ echo 'text1,text2,add(foo,bar),text3' | sed 's/add($[^(),]\+$,$[^,()]\+$)/\1 \2 c/'
– kvantour
Commented Apr 23, 2020 at 8:08
Thank you!!! That's exactly what I was after
– help_me
Commented Apr 23, 2020 at 8:13
@kvantour Feel free to add an answer.
– Wiktor Stribiżew
Commented Apr 23, 2020 at 8:29
@WiktorStribiżew There is no need to make this an answer. There are enough questions out there regarding back referencing.
– kvantour
Commented Apr 23, 2020 at 8:54

| Show 1 more comment

Is there a simple way to save variables from sed/awk? [duplicate]