2011 twenty-four merry days of Perl Feed

Munge All the Filenames!

rename - 2011-12-18

A Shameful Admission

It has come to pass many times that I've needed to rename a bunch of files. I had a bunch of strategies for doing this, and I never picked one and stuck to using it. I certainly never build a reusable tool for it, which meant that every time I did it, I screwed things up in a new and exciting way.

Sometimes a one-liner in shell would do it:

for f in *.txt; { mv "$f" "$f.old" }

Or maybe:

for f in *.txt; { BASE=$(basename $f .txt); mv "$f" "$BASE.yaml" }

Other times, I'd resort to Perl:

use autodie qw(rename);
my @files = grep { -f } <*>;
for my $file (@files) {
  my $new = $file =~ s/^([0-9]+)/sprintf '%4u', $1/e;
  rename $file => $new;

Actually, I am relieved to note that I can't even easily show what I usually did, which was just deplorable. I'd run Vim and do something like :r !find . -type f to get a list of files, then do some :v/.../d to delete files I didn't care to rename, then maybe put it in a __DATA__ section to iterate over or maybe use a block yank-and-put to build up a file that looked like this:

file1.pdf         "File 01.pdf"
file2.pdf "File 02.pdf"
file10.pdf "File 10.pdf"
file18.pdf "File 18.pdf"

Then I could just :%s/^/mv / to put a mv at the front of every line and pipe it through sh! Seriously, I did this. But I've gotten better. Now I use rename.

rename '$_ = ucfirst; s/([0-9]+)/sprintf "%02u", $1/e;' *.pdf

I love writing one-liners, but I always get them wrong. For example, that one above has a bug. With lots of one-liners, I can just run them over and over until I get it right. When renaming files, though, you don't want to actually do it until you have it right – so there's the -n switch to do a dry run:

  ~$ rename -n '$_ = ucfirst; s/([0-9]+)/sprintf "%02u", $1/e;' *.pdf
  rename file1.pdf File01.pdf
  rename file10.pdf File10.pdf
  rename file18.pdf File18.pdf
  rename file2.pdf File02.pdf

Oops. I need to add whitespace.

  ~$ rename -n '$_ = ucfirst; s/ *([0-9]+)/sprintf " %02u", $1/e;' *.pdf
  rename file1.pdf File 01.pdf
  rename file10.pdf File 10.pdf
  rename file18.pdf File 18.pdf
  rename file2.pdf File 02.pdf

Great! Drop the -n and our files get renamed!

If you need to do something more complicated – like keep a counter to number your files, you can write a whole subroutine in a "real" program:

use strict;
use File::Rename;
my @files = @ARGV || die "no files given!";

my $i = 1;
my $width = length scalar @files;
  sub { s/^/sprintf '%0*u - ', $width, $i++/e },

See Also

Gravatar Image This article contributed by: Ricardo Signes <rjbs@cpan.org>