In sed documentation, we read that "sed works by making only one pass over the input(s)".

I would like to write a sed script (to be run like this : sed -f my_script.sed), that performs several operations including on the operations that were performed earlier in the script.

Is it possible to perform a second (or n-th) pass on the edited stream content ?

e.g. :

  • initial stream content :


  • stream content after first pass :
/titi/ i (
/titi/ a )
$ cat test.txt
  • after another pass :

I would like the output to be

$ cat test.txt

This is just an example to show that i'd like to perform stream editions that have effects on previous commands inside a single .sed script.


1 Answer 1


sed only gets one pass over the input stream. This is specifically stated in the man page (at least, on Fedora 33, anyway).

Whenever I have to do something like this, I pipe the output of sed into a second instance of sed to do the subsequent edits.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .