compare_script.sh

### Usage: compare_script.sh directory1 directory2

# This script can be used with 2 different functionalities, list_same or list_different
#functionality=list_same
functionality=list_different
# list_different will include files that exist only in the first directory. For
# that reason you will probably want to run it as:
# compare_script.sh dir1 dir2; compare_script.sh dir2 dir1

# gnu stat is very different that BSD stat, on a Mac you should install
# the gnu coreutils
stat=$(which gstat || which stat)

### Define the functions we'll be needing

same_size(){
    if [[ -f $2 ]] && [[ $($stat --printf "%s" $1) == $( $stat --printf "%s" $2 ) ]]; then
        return 0
    else
        return 1
    fi
}

list_same(){
    same_size $1 $2 && echo $1
}

list_different(){
    same_size $1 $2 || echo $1
}

path_complement(){
    # Usage: path_complement directory_1 directory_2 path_to_file_in_directory_1
    # returns path_to_file_in_directory_1 path_to_file_in_directory_2
    echo $3 $2/${3#${1}/}
}

# This is called by the find in main() below
callback(){
    path=$($functionality $(path_complement $2 $3 $4))
    if [[ -n "$path" ]]; then
        # to return the whole path do:
        echo $path
        # to remove the dir1 portion from the path do:
        #echo ${path#${2}/}
    fi
}

main(){
    # remove trailing slashes if they are present
    dir1=${1%/}
    dir2=${2%/}
    # remove double slashes if they are present
    dir1=${dir1/\/\//\/}
    dir2=${dir2/\/\//\/}

    find $dir1 -type f -exec $0 --callback $dir1 $dir2 {} \; 
}

### Procedure

# The find command in main() cannot -exec functions, not files. For that reason
# we must give this script way to call itself as a callback.
if [[ $1 == "--callback" ]]; then
    callback $*
else
    main $*
fi
exit 0

### BEGIN UNIT TEST CODE ###
echo Creating client directories in /tmp/ for testing...
echo Removing /tmp/client*
rm -rf /tmp/client*
echo Create a dir structure to prove that it works on deep dirs
mkdir -p /tmp/client01/images/landscape/thumbnails
echo Create 10 files in each dir
for ((i=1; i <= 10; i++)); do find /tmp/client01 > /tmp/client01/images/landscape/file$(printf '%02d' $i).ext; done
for ((i=1; i <= 10; i++)); do find /tmp/client01 > /tmp/client01/images/landscape/asset$(printf '%02d' $i).ext; done
for ((i=1; i <= 10; i++)); do find /tmp/client01 > /tmp/client01/images/landscape/thumbnails/thumbnail$(printf '%02d' $i).ext; done
echo Copy the images dir over to a movies dir
cp -r /tmp/client01/images /tmp/client01/movies
echo Remove a file from each of the images dirs
find /tmp/client01/images -name '*5*' -exec rm {} \;
echo Remove a different file from each of the movies dirs
find /tmp/client01/movies -name '*6*' -exec rm {} \;
echo Put some alternate data in to create non-matching filesizes
for f in $(find /tmp/client01/images/ -name '*3*'); do ls -l > $f; done
for f in $(find /tmp/client01/movies/ -name '*4*'); do ls -l > $f; done
echo Create 10 /tmp/clients "(they'll all be the same, but you get the point)"
for ((i=2; i <= 10; i++)); do cp -r /tmp/client01 /tmp/client$(printf '%02d' $i); done
echo Run the script on all client directories
echo; echo;
find /tmp/ -maxdepth 1 -type d -name 'client*' -exec ./compare_script.sh {}/images {}/movies \;