#!/bin/bash -e

# Set "V" to "true" for verbose output.
V=false

# Set the following three variables to the username, hostname and
# directory in which you want to store the remote backup. If you are
# running the script automatically, e.g., in a cron job or in
# /etc/cron.daily, then you need to make sure that ssh has the ability
# to log in as this user on this host without a password, e.g., by
# creating an SSH key with no passphrase whose public key is in the
# .ssh/authorized_keys file for the user and whose private key is
# listed in root's .ssh/config on your VPS.
# 
# For additional security you may wish the remote user to run inside a
# chroot environment, but setting that up is complicated and beyond
# the scope of this posting.

REMOTE_USER=[remote-username]
REMOTE_HOST=[remote-hostname]
REMOTE_DIR=[remote-backup-directory]

# Search for "CONFIG:" below to find additional configuration you may
# wish to change.

D=/tmp/backup.$$
mkdir $D

# Find paths in all RPMs.  The find the ones that have changed.  Then
# find the ones that have *not* changed by removing the ones that have
# changed from the complete list.  Finally, remove directories from
# the list of unchanged paths, to arrive at a list of files and
# devices which are identical to what's in their RPMs.  These are the
# paths that are excluded from the backup.

$V && echo "Finding paths in all RPMs"
rpm -qla | fgrep -v '(contains no files)' | sort > $D/all-rpm-paths

$V && echo "Finding changed RPM paths"
rpm --verify --nomd5 -a | grep '^[\.S]' | grep -v '^missing' |
    awk '{print $NF}' | sort >| $D/changed-rpm-paths

# Need to do this because some config files are explicitly not
# verified by rpm.
$V && echo "Finding changed config files"
rpm -q --dump -a |
perl -ne 'split; next if (! $_[7]); print;' |
while read path size mtime md5sum mode owner group rest; do
    if [ -d $path ]; then
	:
    elif [ ! -e $path ]; then
	:
    elif [ "$(stat -c "%s %Y %U %G" $path)" != "$size $mtime $owner $group" ]; then
	echo $path
    elif [ 0$((echo 'obase=8; ibase=16;'; stat -c %f $path | tr a-z A-Z) | bc) != $mode ]; then
	echo $path
    elif [ $(md5sum $path | awk '{print $1}') != $md5sum ]; then
	echo $path
    fi
done |
sort -u -o $D/changed-rpm-paths $D/changed-rpm-paths -

$V && echo "Finding unchanged RPM paths"
comm -23 $D/all-rpm-paths $D/changed-rpm-paths > $D/unchanged-rpm-paths

$V && echo "Extracting non-directories from exclude path list"
while read file; do
    test -L "$file" -o \! -d "$file" && echo "$file"
done < $D/unchanged-rpm-paths > $D/excludes

$V && echo "Adding extra excludes"

# CONFIG: Add paths you don't wish to back up to this list.  These can
# be files or directories (in which case the entire contents of the
# directory are excluded).  NOTE: Do *not* use globbing patterns here.
# They won't work, because the command below uses --include-filelist
# and --exclude-filelist instead of --include-globbing-filelist and
# --exclude-globbing-filelist.  Globbing patterns slow down
# rdiff-backup *tremendously*.  See below for how to do globbing if
# you need to.

cat >> $D/excludes <<EOF
/tmp
/var/tmp
/var/spool
/var/run
/var/cache
/var/log/logwatch.log
/usr/src/redhat/RPMS
/var/lib/imap
/usr/lib/locale/locale-archive
/usr/libexec/webmin
/var/www/big8/dokuwiki/data/cache
/var/lib/php/session
/var/log
/d/mail-backup
/home/rhj/stmp/admin/tmp.rhj
/home/jik/.cpan
/home/andrea/.tmda
/home/jik/Mail/procmail.log
EOF

# CONFIG: Here's how to do globbing excludes -- do the globbing here,
# in the shell, rather than in the excludes file, and then add the
# matching paths to the excludes file.

ls /home/jik/Mail/bogospam* >> $D/excludes
ls /home/jik/Mail/notspam* >> $D/excludes
ls /home/*/.bogofilter/wordlist.db >> $D/excludes
ls /home/*/.bogofilter/wordlist.db.autobak.gz >> $D/excludes

ls /var/lib/news/history.* >> $D/excludes

find / -xdev -name '.#*' -print >> $D/excludes

$V && echo "Creating includes"

# CONFIG: You may wish to have some exceptions to the directories
# excluded above.  Edit the following as needed.  You can also use the
# globbing technique shown above for the includes file, if you need
# to.

cat >> $D/includes <<EOF
/var/log/rpmpkgs
/var/spool/at
/var/spool/cron
/var/spool/imap
EOF

if $V; then
    echo "Running backup"
    VFLAG=-v5
fi

rdiff-backup "$@" $VFLAG --exclude-other-filesystems \
    --include-filelist $D/includes \
    --exclude-filelist $D/excludes \
    --exclude-sockets \
    / $REMOTE_USER@$REMOTE_HOST::$REMOTE_DIR

$V && echo "Cleaning up"

rm -rf $D