Hey friends,

I want to write a script that will descend through a parent directories subdirectories and move each specified file to another directory. I have .mp3 files gallore but they are mixed into different directories and it would be annoying to have to go through them one by one... So that's why I need this script... anyone?

I've gotten help on that sort of things a number of times. Just today, I' linked them all to this thread http://www.linuxquestions.org/questi...d.php?t=469411

Not exactly what you're looking for, but it might get you started.

How about:

find -xdev -iname '*mp3' -exec mv {} /home/you/some/dir \;

Lyle.

It's a very simple application of -exec but it's best not to use it.
Piping find's output to xargs gets the same job done in a more efficient and safer way. (Though not much of each of those in this situation - I'm being a bit of a facist here ;-) )

Try

Each of the other find commands posted so far has a problem! OK, its actually the mv command that has a problem...

Many Linux users are used to the -i option to mv always being there because of an alias many distributions (notably RedHat) put in the .bashrc file ... but when used in find as has been shown here, the -i option will not be aliased in... as a result, duplicated filenames will be lost, with only the last "found" file remaining.

The mv -i option prevents "clobbering" like-named files by interactively prompting you to confirm any overwrite. Being an "old-timer," I get annoyed by the -i (especially on cp)and comment out all of the aliases in my .bashrc -- but that's because I'm an old-timer!

Maybe this would be no loss for music files (because you're not likely to keep them all named music.mp3 or anything), but if there are other files (backup.tgz or something) then you'll clobber them! (I know that both my Eric Claption "Layla.mp3" files are really very different! -- One is the original, the other is "unplugged" & I'd hate to lose either one!)

I'm not so sure about the benefit (at least in this instance) of xargs over -exec -- but in either case above, the -i option added to the mv command might very well prevent the loss of some valuable music!

For any newbies, this results in the command:

NOTE: I also added a place for find to look (other than CWD): the ~ is your home directory. And I placed quotes around the {} in mv just in case there are spaces or other wierdnesses in my file/directory names.

Peace, Love, and Unix!

Yes, very good. I've been hit by such clobbering before *shudder*
As you say, xargs is really not necessary at all but my mother told me there was something wrong with me if I didn't write with my right hand and use xargs ;-)
You could do some godawful intermediate stuff with sort and uniq but the above code works best - don't want to spend as much time debugging as it would take to do it by hand.
-p for xargs does the same as -i for mv but you'd have a list of about 100 filenames that you'd have to manually check for uniqueness... Jesus I'm rambling with poor grammar/clarity

Go with it4soho!

Like many of you, I like the simple one-line solutions. But sometimes these result in scripts that are either too simple (like the one which clobbers like-named files), or too complex, or sometimes just too hard to use correctly. For instance, what if you do a "find / ...." ? You'll get lots of error messages about trying to move a file over itself.

In cases like this, you're better off not trying to write a clever one-liner. Here's a script that only took a few minutes to write....

-- It is written in ksh, but it ought to work
in bash and most modern Bourne shells as well.

-- It automatically renames the files to
file_2.mp3, file_3.mp3, etc. in the
event of a naming conflict.

-- It keeps a log of where the files came from
and what they were ultimately named.

-- It handles both mp3 and MP3 extensions. It doesn't
do mixed case extensions (i.e. .Mp3 or .mP3),
though it could.

-- It will be OK even if the target directory is
underneath the search directory (e.g. "fromTree=/")


#!/usr/bin/ksh
#################3
# Quick script to move all MP3s on a system to a single directory.

fromTree=/fromDir
targDir=/toDir
logFile=$targDir/xfer.log

echo >> $logFile
echo ========== `date` =========== >> $logFile

find $fromTree \( -name "*.mp3" -o -name "*.MP3" \) -print | while true; do
read mp3File || break

## Remove .mp3 and .MP3 extensions
## This is inefficient, but it's good enough for quick-and-dirty script
base=`basename $mp3File .mp3`
base=`basename $base .MP3`

## Where shall we put the file
target=$targDir/$base.mp3

## Make sure we are not moving the file over itself!
[ $target = $mp3File ] && continue

## If there is a name conflict, append a number
cnt=1
while [ -f $target ]; do #
## Use "expr" since some shells don't have $(( ))
cnt=`expr $cnt + 1`
target=$targDir/${base}_$cnt.mp3
done

echo "$mp3File $target" >> $logFile # There's a <TAB> in that echo
mv $mp3File $target

done ## end of "find | while true; read.... "

what's the difference of this two in terms of efficiency and safety?

I had no idea my post would create so much controversy. That's great, I learned something. I guess I thought the filenames would all be different.

Lyle.

I'd just like to thank you all for your valuable input and confirm that it worked like a charm.