What GNU/Linux command-line tool would I use for performing a search and replace on a file?

Question

What GNU/Linux command-line tool would I use for performing a search and replace on a file?

Can the search text, and replacement, be specified in a regex format?

Answer 1

sed, the stream editor, and yes, it uses regex.

Answer 2

Perl was invented for this:

perl -pi -e 's/foo/bar/g;' *.txt

Any normal s/// pattern in those single quotes. You can keep a backup with something like this:

perl -pi.bak -e 's/foo/bar/g;' *.txt

Or pipeline:

cat file.txt | perl -ne 's/foo/bar/g;' | less

But that's really more sed's job.

Answer 3

sed 's/a.*b/xyz/g;' old_file > new_file

GNU sed (which you probably have) is even more versatile:

sed -r --in-place 's/a(.*)b/x\1y/g;' your_file

Here is a brief explanation of those options:

-i[SUFFIX], --in-place[=SUFFIX] edit files in place (makes backup if extension supplied)

-r, --regexp-extended use extended regular expressions in the script.

The FreeBSD version also supports these options. The NetBSD and OpenBSD versions support only -r.

If you want to learn more about sed, Cori has suggested this tutorial.

Answer 4

Consider Ruby as an alternative to Perl. It stole most of Perl's one-liner commandline args (-i, -p, -l, -e, -n) and auto-sets $_ for you like Perl does and has plenty of regex goodness. Additionally Ruby's syntax may be more comfortable and easier to read or write than Perl's or sed's. (Or not, depending on your tastes.)

ruby -pi.bak -e '$_.gsub!(/foo|bar/){|x| x.upcase}' *.txt

vs.

perl -pi.bak -e 's/(foo|bar)/\U\1/g' *.txt

In many cases when dealing with one-liners, performance isn't enough of an issue to care whether you use lightweight sed or heavyweight Perl or heaveier-weight Ruby. Use whatever is easiest to write.