Windows – Bulk copy files from CD, and rename disallowed filenames along the way

Updated:

We're under the assumption that "clip-2009-10-01 21;26;00.mov" is not the actual filename; one possibility is that the actual filename is "clip-2009-10-01 21:26:00.mov". However, we can't verify that under Windows.

We may not need to.

Failsafe Method:

Boot to a Linux LiveCD. Ubuntu 9.04 has good NTFS support, and Linux handles a lot more wonky-characters-in-filenames than Windows. The perl rename script may be included as the system's rename command.

This-Might-Actually-Work Batch Method (New Script!)

The DOS command DIR/X shows short filenames, if they exist on your system.

$ cmd
c:\test> dir /x
 Volume in drive E is NUVOL
 Volume Serial Number is 80D3-A96D

 Directory of e:\tor\test

10/04/2009  05:15 AM    <DIR>                       .
10/04/2009  05:15 AM    <DIR>                       ..
10/04/2009  05:11 AM                 0 CLIP-2~1.MOV clip-2009-10-01 21;26;00.mov
               1 File(s)              0 bytes
               2 Dir(s)   5,201,670,144 bytes free

If they do exist, the REN command will move them to a new name; the new name can be a new (valid) long filename.

c:\test> ren CLIP-2~1.MOV "clip-2009-10-01_21-26-00.mov"

That's how to fix one.

To batch process all of them, you need to 1) grab a listing of all the files you want to move; 2) run a short perl script to convert your listing into a batch file with the appropriate REN commands; and 3) run the resulting batch script.

c:\test> dir /x > mybrokenfiles.lst  
$ cat mybrokenfiles.lst | perl -lne 'next if not /MOV/; s/^.{1,39}//; s/^/ren /; s/ (\d\d);(\d\d);(\d\d)/_$1-$2-$3/; print' > fixmybrokenfiles.bat  
c:\test> fixmybrokenfiles.bat  

The perl commandline assumes a very particular input format, so if the DOS listing shows long filenames in something other than the "21;26;00.mov" format, it probably won't do exactly what you want. If you try it, double-check that the batch script looks right before running it.

If you are comfortable with perl (or sed/awk, python, whatever), you can script this yourself. But if DIR/X doesn't show the short filenames, your system has them disabled, and this solution won't help.

Original answer: not useful with what we know now, but if you copy this sort of file off of OSX again, you can use this BEFORE the copy as a preventative step.

I use the commandline a lot on both Windows and Linux systems. There's a handy perl script floating around the internet that allows batch file renames using standard perl regex's (google for rename.pl to find it).

Under Cygwin on windows, use this in the directory your files are located in to rename them:

$ ls
clip-2009-10-01 21;26;00.mov

$ rename.pl 'tr/ ;/_-/;' * 
$ ls
clip-2009-10-01_21-26-00.mov

Pretty sure my version came from the Perl Cookbook:

#!/usr/bin/perl -w
# rename - Larry's filename fixer
$op = shift or die "Usage: rename expr [files]\n";
chomp(@ARGV = <STDIN>) unless @ARGV;
for (@ARGV) {
    $was = $_;
    eval $op;
    die $@ if $@;
    rename($was,$_) unless $was eq $_;
}

If all your directories are in ~/foo, you can run this (assuming you want to rename everything that is in ~/foo):

cd ~/foo
for dir in *; do 
    t=`stat -c %y "$dir" | awk '{print $1"-"$2}' | 
       cut -d ":" -f 1,2 | sed 's/://'`
    mv "$dir" "$t"_"$dir";
done

For example:

$ ls -gG
total 20
drwxr-xr-x 2 4096 Jun 20 14:41 a
drwxr-xr-x 2 4096 Jun 21 14:40 b
drwxr-xr-x 2 4096 May 16 14:57 c
drwxr-xr-x 2 4096 Jun 21 14:33 d
drwxr-xr-x 2 4096 May  3 16:15 e
$ for dir in *; do 
    t=`stat -c %y "$dir" | awk '{print $1"-"$2}' | 
       cut -d ":" -f 1,2 | sed 's/://'`
    mv "$dir" "$t"_"$dir";
done
$ ls -gG
total 20
drwxr-xr-x 2 4096 May  3 16:15 2013-05-03-1615_e
drwxr-xr-x 2 4096 May 16 14:57 2013-05-16-1457_c
drwxr-xr-x 2 4096 Jun 20 14:41 2013-06-20-1441_a
drwxr-xr-x 2 4096 Jun 21 14:33 2013-06-21-1433_d
drwxr-xr-x 2 4096 Jun 21 14:40 2013-06-21-1440_b

The trick here is using stat to save the modification time in variable $t so we can use it as a name. If you want to preserve the modification dates of the directories and only change the name, do something like:

for dir in *; do 
    old=`mktemp` && touch -r "$dir" $old
    t=`stat -c %y "$dir" | awk '{print $1"-"$2}' | 
       cut -d ":" -f 1,2 | sed 's/://'`
    mv "$dir" "$t"_"$dir"; 
    touch -r $old "$t"_"$dir";
done