It's tricky, but it does deal with subdirectories.

Assume, that NAS was set to /usr4/nas, and the first file returned by find was /usr4/nas/2010/pic1.jpg. The result of this line of the script:

Code:

  STR=${file##$NAS/}

sets STR to 2010/pic1.jpg. The meaning of ${variable##pattern} is to remove the longest occurance of pattern from the front of the contents of variable. The variable file would have been /usr4/nas/2010/pic1.jpg and the value of $NAS would have been /usr4/nas, so the result is everything after /usr4/nas/, or 2010/pic1.jpg which includes the subdirectory.

Assuming XBMC was set to /xbmc, and taking it further

Code:

if [ -f $XBMC/$STR ];

would test to see if /xbmc/2010/pic1.jpg existed -- properly assuming the directory structure from the source filesystem exists on the target filesystem. Likewise, the copy command assumes the same directory structure on the target filesystem. (It does assume that the directory structure exists on the target.)

There are two things that I would tweek:

First, use find piped to while rather than assuming that the output of the find command can be handled as a list to for.

Code:

find $NAS -type f | while read file
do
 ....
done

Secondly, assuming Kshell or Bash, use [[ ]] to supply the test expression rather than the single bracket. The double bracketed expressions are more efficient because the evaluation is done by the shell and not by an external command (test).

Code:

if [[ ! -f $XBMC/$STR ]]

In addition, if the script should make the directories on the target that don't exist, the following needs to be added:

Code:

targetd=$XBMC/${STR%/*}   # get the target dir by chopping off the basename
if [[ ! -d $targetd ]]
then
    mkdir -p $targetd
fi

This should go immediately before the copy.