[SOLVED] How to mv while keeping matching sub-directories

MisterBark · 07-19-2012, 09:14 PM

Hi!

My problem is kinda ridiculous but I can't find a simple solution!

I would like to make :

Code:

mv */*foo* /directory/

so that the files are moved to /directory/*/*foo* and not /directory/*foo*
(so it would preserve the matching paths)

It is important to make a REAL move, and not a cp+rm because it's on the same disk and it's huge.

THANKS A LOT!

chrism01 · 07-20-2012, 12:41 AM

You could create a test area with same dir file layout and a few small files to test, but I don't think(?) mv is that smart ie multi-levels of wildcards.
I'd use a script with find cmd, and test first eg in a test area or use 'echo' to echo out what it would have done instead of doing the actual mv.

MisterBark · 07-20-2012, 08:49 AM

ok, I just wrote this program :

Code:

#!/usr/bin/perl
use strict;


my @src = $ARGV[ 0 .. ($#ARGV-1) ];

my $dest = $ARGV[$#ARGV];
$dest =~ s!/*$!/!;

foreach(@src){
	
	my $path = $_;
	$path =~ s!/[^/]+$!!;
	
	print("$_  >  $dest$path/  ...\n");
	
	unless( -d "$dest$path" ){
		mkdir("$dest$path");
	}
	
	system("mv '$_' '$dest$path/'");
	
}

print("DONE MOVING $#ARGV FILES.\n");
exit;

Didn't test it yet but it should work

lleb · 07-25-2012, 09:13 AM

if you have the storage space you could

rsync -aviS source destination

then just rm -rf /foo

that script looks interesting too. sadly i know nothing about perl yet.

MisterBark · 07-25-2012, 09:30 AM

Thanks

I didn't know rsync, that looks great (but not for what I wanted of course)
So it's some kind of MySQL replication concept, but for files?

But the thing is, I was trying to find a way to make a real MOVE, not copy + rm.
Copy + rm is very long for these files, and makes a problem of confidentiality since I'm gonna have to shred everything at the end...
But my program worked

Perl worth it lleb, it's an awesome language. There is nothing you cannot do in perl.

jschiwal · 07-25-2012, 09:36 AM

You could use `find' to locate the files and create a bash script to move them.

Code:

find dir??/*foo* -type f -printf "mv \"%p\" \"directory/%p\"\n" >movescript.sh

>cat movescript.sh
mv "dir10/1foo10/file5" "directory/dir10/1foo10/file5"
mv "dir10/1foo10/file3" "directory/dir10/1foo10/file3"
mv "dir10/1foo10/file1" "directory/dir10/1foo10/file1"
mv "dir10/1foo6/file5" "directory/dir10/1foo6/file5"
...

If the destination directory is on the same filesystem, the files directory entry is moved. If the destination is on a different filesystem, then the `mv' command does a copy & delete. If confidentiality is important, you could change the one liner adding && shred <pathname> to the end.

Code:

find dir??/*foo* -type f -printf "cp \"%p\" \"directory/%p\" && shred \"%p\"\n" >movescript.sh

lleb · 07-25-2012, 09:38 AM

the nice thing about rsync (remote Synchronize) is that it is extremely light on the system, does compression during the move as well as encryption for security.

sadly the rm is very heavy on the system, but not as heavy as mv.

i just have not had time to learn perl. wanted to learn C first. im only a new programmer as in less then 4-6 months with BASH. my next goal is to learn C, then work on perl.

MisterBark · 07-25-2012, 09:52 AM

Quote:

Originally Posted by lleb

sadly the rm is very heavy on the system, but not as heavy as mv.

The dilemma was not rm VS mv but cp+rm VS mv

lleb · 07-25-2012, 11:34 AM

got ya.