#!/bin/bash
# $Header: /CVSROOT/public/scylla-charybdis/md5backup/bin/compare.sh,v 1.4 2006/10/03 21:16:33 tino Exp $
#
# Simple compare script.
#
# THIS CAN BE SEEN AS A BASE FOR A RESTORE SCRIPT!
# Note that there will never be restore program.
# You always will be able to restore using the shell only.
#
# Read the log in reverse and compare all the files found
# and report the differences.
#
# $Log: compare.sh,v $
# Revision 1.4  2006/10/03 21:16:33  tino
# See Changelog, commit for dist
#
# Revision 1.3  2005/10/10 23:19:13  tino
# typo corrected (md5check -> md5sum)
#
# Revision 1.2  2005/03/05 23:28:39  tino
# support for multi file store added
#
# Revision 1.1  2004/10/05 02:02:48  tino
# added
#

# Change "compare" to "restore" to have a protoype restore method
# But see below, there is a second safety belt below (hunt for
# restore).  If you don't know how, better don't try ;)
MODE=compare

host="`hostname -f`"
[ -z "$host" ] && host="`hostname`"
[ -z "$host" ] && host="`cat /etc/hostname`"
[ -z "$host" ] && host="HOSTNAME"

if [ 0 = $# ]
then
	echo "Usage: `basename "$0"` /backup/md5backup/log/$host [outpath] [grep]
	Compare files in the file store with their originals
	and report differences to stderr.
	The path can be '' for default.
	If you get a lot of errors, try running this as root.
	Usually you don't need this, you can also do:
	md5sum --check /backup/md5backup/md5/$host
	cd /backup/md5backup && bin/check.sh" >&2
	exit 1
fi

SRC="$1"

[ -z "$SRC" ] && SRC="/backup/md5backup/log/$host"

DIR="`dirname "$SRC"`"
DIR="`dirname "$DIR"`/out"
OUT="${2:-$DIR}"
GREP="$3"

if [ ! -d "$OUT" ]
then
	echo "Required paths not found: $OUT
	Give correct file store path as second argument"
	exit 1
fi

# SET DUMMY TO NOTHING TO RESTORE (I need it to be able to test!)
DUMMY=echo
restore()
{
if [ ! -f "$1" ]
then
	echo "missing $1 ($2)" >&2
	inc miss
	return
fi
[ -f "$2" ] && inc diff
inc clean

if ! $DUMMY cp --backup=t "$@"
then
	inc del
	$DUMMY mkdir -p "`dirname "$2"`" &&
	$DUMMY cp --backup=t "$@"
fi &&
echo "restored $2" >&2 ||
echo "failed $2" >&2
}

CLR="`tput el`"

tac "$SRC" |

# old: YYYY-MM-DD HH:MM:SS md5 MD5SUM nr PATH
# new: YYYY-MM-DD HH:MM:SS md5 nnnKB MD5SUM nr PATH
# new: YYYY-MM-DD HH:MM:SS known md5 MD5SUM nr PATH
# In V0.4.x this format will change again.
awk -v GREP="$GREP" '
$3=="md5" || $3=="known" {

sub(/^[^ ]* [^ ]* [^ ]* /,"");
if (length($1)!=32)
  sub(/^[^ ]* /,"");

md5=$1
nr=$2

sub(/^[^ ]* [^ ]* /,"");
if (had[$0]++)
  next

if (GREP=="" || $0 ~ GREP)
  print substr(md5,1,2) "/" substr(md5,3,2) "/" md5 "." nr ".md5 " $0
}' |
{
warn=false
nr=0
diff=0
del=0
miss=0
clean=0
max=69

inc()
{
eval test -z \"\${$1//9/}\" && let max--
let $1++
}

compare()
{
if [ ! -f "$2" ]
then
	if [ -f "$1" ]
	then
		echo -n "$CLR"
		echo "deleted $2" >&2
		inc del
	else
		# yes, reported to stdout
		echo "clean $2$CLR"
		inc clean
	fi
elif [ ! -f "$1" ]
then
	echo "missing $1$CLR"
	echo "Advice! Make a full backup! Remove dbm/`basename "$SRC"`"
	warn=:
	echo "missing $1" >&2
	inc miss
else
	echo -n "$CLR"
	echo "diff $2" >&2
	inc diff
fi
}

while read -r md5 file
do
	inc nr
	if [ -n "${file:max}" ]
	then
		echo -n "$nr-$clean+$diff*$del?$miss ${file:0-max:max}
"
	else
		echo -n "$nr-$clean+$diff*$del?$miss $file$CLR
"
	fi
	for dir in "$OUT" "$OUT"[0-9]
	do
		src="$dir/$md5"
		[ -f "$src" ] && break
	done
	cmp -s "$src" "$file" ||
	$MODE "$src" "$file"
done

if $warn
then
	echo "Serious inconsistencies detected! Hunt for 'missing' files"
	echo "It looks like you need a full backup.  For this do:"
	echo "Remove database dbm/`basename "$SRC"`"
	echo "Then run md5backup again."
	echo "Serious inconsistencies detected! Hunt for 'missing' files" >&2
fi
}