Replace leading whitespace with sed (or similar)

Question

I'd like to replace the leading space or tab characters on each line of a file with a like-number of other characters (let's use _ for simplicity).

E.g.

foo bar
 foo bar
  line 3

becomes

foo bar
_foo bar
__line 3

Note that non-leading whitepsace is not affected, otherwise this would be easy! I suspect it's still easy, but I'm missing the trick.

Answer 1

This will work for you:

sed ':a;s/^\([[:space:]]*\)[[:space:]]/\1_/;ta' file

Example

$ sed ':a;s/^\([[:space:]]*\)[[:space:]]/\1_/;ta' <<<$'foo bar\n foo bar\n\t\tline 3\n _ x'
foo bar
_foo bar
__line 3
__ x

Answer 2

Are you O.K. with a Perl one-liner? You can write:

perl -pe 's/^([ \t]+)/"_" x length $1/e'

(either piping in your file on standard-input, or specifying a filename at the end of the cmomand).

Edited to add: William Pursell, in a comment above, asks whether you want "a tab get replaced with a single '_', or enough to fill out a tabstop". The above command would replace a tab with a single underscore. If you want enough to fill out a tabstop, the simplest approach is to use the expand utility, which converts tabs to spaces, before passing the input to Perl:

expand -i | perl -pe 's/^([ \t]+)/"_" x length $1/e'

(either piping in your file on standard-input, or specifying a filename at the end of the expand part, just before the | character).