#!/bin/sh
#######################################################################
#
# badaba (BAckup DAta BAckups)
# ----------------------------
#
# Abstract:
# ---------
# This SHell-compatible script can be used for backing up large amounts
# of data onto CD-Recordables for example.
#
# Description:
# ------------
# It slices the data which is to be backed up into parts which fit on
# a medium size you can specify (e.g. 650 MB for CD-R)
# The data is backuped using the GNU tar archiver. TAR supports natively
# the multi-volume option, so slicing is done  with TAR.
# For each of the tar-archives, some more information is generated, e.g.
# the date, a CONTENTS file and md5sums of each backuped file.
# These files are converted into an ISO 9660 file system image file, with
# support for JOLIET (windows) and ROCKRIDGE (UNIX) extensions.
#
# Logging:
# --------
# All output to standard error is logged to the file called "LOG" in the
# backup-directory
#
# Disk space usage:
# -----------------
# While doing the backup-jobs, badaba needs at least free disk space
# for double the amount of data you have specified for one slice.
# Example: if you make slices with 650 MB, you will need approx.
#          1300 MB of free disk space.
# This is a result of the fact that mkisofs can not read data from
# standard input AND cannot append files to an existing image.
# We first have to create our tar files, md5 sums and then generate
# the image, after which we can delete the files.
# If you have a good and usable(!) idea on how to improve this, send
# me an email.
#
# BUGS:
# -----
# - md5sum : The md5 sums are generated using a file containing
#            filenames. This is a problem, for example when the filename
#            contains whitespaces.
#
# - exlude : when specifying files/dirs to be excluded from the backup,
#            you have to omit the leading "/" (root directory).
#            this is because removing it in the script for use with the
#            tar --exclude option is difficult or would need a special
#            filter/program like sed. (I want to avoid sed)
#
# Author:
# -------
# Paul Kremer (paulkremer NOSPAM[AT]gmx dot net)
#
# License:
# --------
# BSD artistic license
#
# Changelog:
# ----------
# 0.1 initial release (april 2001)
# 0.11 small bugfix in an "if file-exists" clause (parentheses) (june 2001)
# 0.12 small bugfixes
#      added PATH env check
#      the directory for the ISO is now deleted upon successful mkisofs
#      EXCLUDE can now be empty without enerating an error
#
# Dependencies:
# -------------
# md5sum, head, tar, date, du, cut, mkisofs, cdrecord-clone
#
# Trivia:
# -------
# The script is called badaba (backup data backups), because we first
# make a backup to a tar file and additionnally backup this file to a
# different medium. ;-)
#
#######################################################################

#######################################################################
# CONFIG SECTION
# what to backup
SOURCE="/home/pkremer/bin/"
# temporary dir for backup
TEMP_DIR="/home/pkremer/tmp/"
# slices are x MB
SLICE_SIZE="1"
# string for backup-name (tar-files and CD-TITLE) max. 128 CHARS
BACKUP_NAME="APRIL_2001"
# default nice level
NICE="10"
# EXCLUDE is a space separated list of dirs or files to exclude in backup
#  no leading/ending slashes "/" !!!!!
EXCLUDE="home/pkremer/bin/todo"
# cdrecord command:
CDRECORD="cdrecord -eject -data -v -isosize speed=4 dev=0,6,0"
# END CONFIG SECTION
#######################################################################
#######################################################################
function make_iso() {
   echo "   creating ISO9660 JOLIET ROCKRIDGE file system image ($1.iso)..."
   echo "COMMAND: mkisofs -quiet -A $1 -J -pad -r -o $1.iso $1" >> LOG
   mkisofs -quiet -A $1 -J -pad -r -o $1.iso $1 2>> LOG
   if [ $? != 0 ]; then
      echo "Error creating ISO image of $1. Exit."
      exit 1
   fi
   # still have to remove the directory
   rm -rf $1
}
#######################################################################
function count_file () {
# this function counts the archives, returns current number
# all output will be captured in the CONTENTS files
	TARFILE="$1"
	COUNT_F="$2"
#	echo "TARFILE $TARFILE"
#	echo "COUNT_F $COUNT_F"
	let COUNT=`cat $COUNT_F`-1
	if [ "$3" == "--last" ]; then
		let COUNT=$COUNT+1
	fi
	echo "Created archive $TARFILE-$COUNT"
	mkdir $TARFILE-$COUNT
	mv $TARFILE $TARFILE-$COUNT/$TARFILE.tar
	# make a small README
	echo "   generating some info..."
	echo `date` > $TARFILE-$COUNT/DATE
	echo "This is archive number #$COUNT." > $TARFILE-$COUNT/README
	echo "Use tar with the --multi-volume option for extracting." >> $TARFILE-$COUNT/README
	echo "The file md5sum.txt contains the MD5 sums for each" >> $TARFILE-$COUNT/README
	echo "individual file in the archive." >> $TARFILE-$COUNT/README
	# end make a small README
	echo "COMMAND: tar --list --file $TARFILE-$COUNT/$TARFILE.tar > $TARFILE-$COUNT/CONTENTS-$COUNT" >> LOG
	tar --list --file $TARFILE-$COUNT/$TARFILE.tar > $TARFILE-$COUNT/CONTENTS-$COUNT 2>> LOG
	cat $TARFILE-$COUNT/CONTENTS-$COUNT >> CONTENTS 2>> LOG
	# now get md5 sums for each file in this slice and save
	echo "   generating MD5 sums..."
	FNAME="1"
	echo "COMMAND: md5sum /\$FNAME >> $TARFILE-$COUNT/md5sum.txt" >> LOG
	while [ "$FNAME" != "" ]
	do
		read FNAME
		if [ -f "/$FNAME" ]; then
			echo "md5sum /$FNAME" >> LOG
			md5sum "/$FNAME" >> $TARFILE-$COUNT/md5sum.txt 2>>LOG
		fi
	done < $TARFILE-$COUNT/CONTENTS-$COUNT
	# done md5 sums
	# make iso image
	make_iso $TARFILE-$COUNT
	# done make iso image
	echo -n "  shall I burn the disc now? [N] "
	read continue_
	if [ "$continue_" != "Y" ]; then
		exit 0
	elif [ "$continue_" == "Y" ]; then
		echo "$CDRECORD $TARFILE-$COUNT.iso"
		$CDRECORD $TARFILE-$COUNT.iso
	fi
	exit 0
} # end count_file
#######################################################################

function check_env() {
   IS_PATH_GOOD=`which badaba`
   if [ "$IS_PATH_GOOD" == "" ];
   then
      echo "'badaba' must be in your PATH: $PATH"
      exit 1
   fi
   IS_TAR_GOOD=`which tar`
   if [ "$IS_TAR_GOOD" == "" ];
   then
      echo "'tar' can't be found in your PATH: $PATH"
      exit 1
   fi
# add checks for md5sum, head, date, du, cut, mkisofs, cdrecord
}

#######################################################################
if [ "$1" == "--count" ]; then
  count_file $2 $3 $4
fi
#######################################################################
CURDIR=$PWD
check_env
# go to TEMP_DIR and clean it
echo -n "I will clean \"$TEMP_DIR\" completely! Sure? [N] "
read var
if [ "$var" != "Y" ]; then
   exit 1
fi
cd $TEMP_DIR
rm -rf *
# done with clearing target directory
#######################################################################
echo "Gathering info..."
# generate correct exclude string for tar and putting into FILE EXCLUDE
# and calcualte size of excluded files
if [ "$EXCLUDE" != "" ];
then
for entry in $EXCLUDE
do
      echo "/$entry" >> EXCLUDE
      EXCLUDES="$EXCLUDES --exclude /$entry"
       if [ -e "/$entry" ]; then
           S=`nice -n $NICE du --summarize --total "/$entry" | head -n 1 | cut -f 1 -d "	"`
       fi
       let EXCLUDED_SIZE=$EXCLUDED_SIZE+$S
   done
else
   EXCLUDES=""
   let EXCLUDED_SIZE=0
fi
# done generate correct exclude string for tar
echo "START: `date`" > TIMING
# get size and generate some stats before starting
TOTALSIZE=`nice -n $NICE du --summarize --total $SOURCE | head -n 1 | cut -f 1 -d "	"`
let TOTALSIZE=$TOTALSIZE-$EXCLUDED_SIZE
let TOTALSIZE=$TOTALSIZE/1024

echo "----------------------- STATISTICS: ----------------------------"
echo "  (sizes may be wrong: location of excluded files not checked)"
echo "----------------------------------------------------------------"
echo "SOURCE:	$SOURCE"
echo "SIZE:	$TOTALSIZE MB"
let SLICE_NUM=$TOTALSIZE/$SLICE_SIZE+1
echo "SLICES:	$SLICE_NUM slices with $SLICE_SIZE MB (approx.)"
echo "----------------------------------------------------------------"
# done get size and generate some stats before starting
# find a temporary filename for counting the archives
prefix=temp
suffix=`eval date +%s`
COUNT_FILE=$prefix.$suffix
while [ -e $COUNT_FILE ]; do
    suffix=`eval date +%s`
    COUNT_FILE=$prefix.$suffix
done
# done temporary filename for counting the archives
let SIZE="$SLICE_SIZE*1024"
# now create the tar files
echo "Creating tar archives..."
echo "COMMAND: nice -n $NICE tar --create $EXCLUDES --one-file-system --tape-length $SIZE --multi-volume --new-volume-script=\"$0 --count $BACKUP_NAME $COUNT_FILE\" --volno-file=$COUNT_FILE --file $BACKUP_NAME $SOURCE" >> LOG
nice -n $NICE tar --create $EXCLUDES --one-file-system --tape-length $SIZE --multi-volume --new-volume-script="$0 --count $BACKUP_NAME $COUNT_FILE" --volno-file=$COUNT_FILE --file $BACKUP_NAME $SOURCE 2>> LOG
echo "COMMAND: $0 --count $BACKUP_NAME $COUNT_FILE --last" >> LOG
$0 --count $BACKUP_NAME $COUNT_FILE --last 2>> LOG
# done create tar files
# get the number of tar files created
let COUNT=`cat $COUNT_FILE`
# remove count-file
rm $COUNT_FILE
echo "Created $COUNT archives."

# create iso-image files for each subdirectory and burn

echo "END: `date`" >> TIMING
ls -lh

cd $CURDIR