Comments on: Bash Script for Replace Multiple Words in a Multiple File

By: Masim "Vavai" Sugianto

Masim "Vavai" Sugianto — Fri, 15 Jan 2010 04:38:26 +0000

@Pascal,

Thank you. I’m not really familiar with Perl but I would love to try your code.

By: Pascal Bleser

Pascal Bleser — Thu, 07 Jan 2010 15:56:19 +0000

As Diego already said, it is a lot easier to use the “-i” in-place editing flag, because it does it properly (locking, really temporary/random filename, …). If not, it is a lot safer to use “mktemp” instead of a fixed filename.

But be aware that those “words” are actually regular expressions, and that you cannot use e.g. “/” in the word to be replaced or the replacement.

I would rather resort to Perl, because it has the ability to escape complete strings in regular expressions. Oh, and Perl also has the -i flag

perl -pi -e ‘s/\Qword1\E/\Qreplacement1\E/g;s/\Qword2\E/\Qreplacement2\E/g’ file1 file2 file3

By: Vavai

Vavai — Thu, 07 Jan 2010 12:04:19 +0000

Hi Diego,

Thanks for the suggestion.

By: Diego E. "Flameeyes" Pettenò

Diego E. "Flameeyes" Pettenò — Thu, 07 Jan 2010 10:52:58 +0000

On modern GNU sed (that is, from at least the last five years) – and you’re already assuming GNU sed when you use multiple -e parameters – you can use the -i parameter to do “inline replace”.

This way you don’t have to copy the file to a temporary version and then back again for the replacement, nor you really need to grep for the string: just use sed -i -e ‘…’ -e ‘…’ /list/of/files/*.foo