Parsing directories in a shell script

[javaguy]

I have a bunch of JPGs arranged in directories that describe when the pictures where taken, i.e. there's a directory called "2002", one called "2003" and so on. Under each there are month directories, "2003 June", "2003 July" and so on, and under each of these there are further directories called batch1, batch2 and so forth.

 

Some of the batch directories, however, are empty, and I've been thinking about a script to clean them up. My first step was to write a script to identify the empty directories. Here's my first attempt.

 

#!/bin/sh

cd /home/javaguy/Documents/Photos
DIRS=$(ls -d 2*)

for DIR in $DIRS
do
echo "Changing to directory $DIR"
cd "$DIR"
MONTHS=$(ls -d *)
for MONTH in $MONTHS
do
	echo "Checking $MONTH..."
	if [[ -d '$MONTH' ]]
	then
		echo "Changing to month directory $MONTH"
		cd "$MONTH"
		BATCHES=$(ls -d batch*)
		for BATCH in $BATCHES
		do
			if [[ ! `ls $BATCH` ]]
			then
				echo "$BATCH is empty."
			fi
		done
		cd ..
	fi
done
cd ..
done

 

Here's the problem with this. When it lists all the directories under "2002", it sees directories called "2002 January" and such. I'd like it to check to see if "2002 January" is a directory and, on finding that it is, to cd into it and further check all its subdirectories. But what it does instead is first check to see if "2002" is a directory and then if "January" is a directory (neither exists at that level). How can I make it recognize that "2002 January" is a single directory name?

[neddie]

Have you tried something like this?

find . -type d -empty

[javaguy]

Woo hoo! :)

 

Much simpler. Thanks!

[iphitus]

Just another tip, try not to use ls for file listings as a users settings may mess up the script.

 

By default, my ls shows detailed information, and this tends to screw up any scripts that use ls. find is much better.

 

James