#!/bin/bash
#
# $Header: /CVSROOT/public/scylla-charybdis/md5backup/bin/check.sh,v 1.5 2005/03/02 23:32:11 tino Exp $
#
# Run this script in the directory (atop out/)
# to verify if the backed up files are still valid.
#
# This script is designed to be interrupted anywhere
# at any time and restart work from where it left.
# So if a "check.list" exists, it is rerun.
# To be sure everything is OK run it with argument -f
#
# $Log: check.sh,v $
# Revision 1.5  2005/03/02 23:32:11  tino
# first version for multi filestore
#
# Revision 1.4  2004/10/05 02:03:17  tino
# You can give the path to the script now
#
# Revision 1.3  2004/04/24 01:10:46  tino
# Intermediate version, I start to implement linked file store now
#
# Revision 1.2  2004/01/07 12:05:09  tino
# Scripts improved, some spelling corrections, and CVS headers added.
# md5backup.c now has a sanity check which slows down things factor 10.

rm -f check.fail.tmp

if [ ! -s "check.list" ] || { [ .-f = ".$1" ] && shift; }
then
	echo "part1:a: building file list"
	> check.list.tmp
	[ -f check.list.old ] || >> check.list.old
	count=-`wc -l < check.list.old | tr -d ' '`
	tail -f check.list.tmp 2>&1 |
	while read a
	do
		let count=count+1
		echo -n "$count 
"
	done &
	bgpid="$!"
	find "${1:-out}"/* -type f -name '*.md5' -print >> check.list.tmp
	kill $bgpid
	echo "part1:b: sorting file list"
	sort check.list.tmp > check.list || exit 1
	sync
	> check.list.tmp
	rerun=false
else
	echo "part1: reusing file list"
	rerun=:
fi

# As this is designed to be stopped anywhere we have to sort first
if [ ! -s check.ok ]
then
	fullrun=:
	echo "part2: creating check.ok"
	> check.ok
else
	fullrun=false
	echo "part2:a: sorting check.ok (delete this for full new check)"
	sort check.ok > check.ok.tmp || exit 1
	if ! cmp -s check.ok check.ok.tmp
	then
		echo "part2:b: oklist changed"
		sync
		# Sanity check.  I really don't trust.
		sort check.ok | cmp - check.ok.tmp || exit 1
		mv -f check.ok.tmp check.ok
	fi
fi

echo "part3:a: locating new files"
> check.fail.tmp
comm -13 check.ok check.list > check.todo
count=`wc -l < check.todo`

echo "part3:b: checking $count files"
while read a
do
	let count=count-1
	echo -n "$count 
"
	case "$a" in
	*.md5)		processor="cat";;
# Those two might show up in future
	*.md5.gz)	processor="gzcat";;
	*.md5.bz2)	processor="bzcat";;
# Ignore unknown stuff
	*)	echo "don't know hot to handle $a" >&2; continue;;
	esac
	sum="`$processor "$a" | md5sum`"
	if [ ".${sum// *}" = ".`basename "${a//.*}"`" ]
	then
		echo "$a" >> check.ok
	else
		echo "$a" >> check.fail.tmp
		echo "failed: $a" >&2
	fi
done < check.todo

echo "part4: syncing filesystems"
sync

[ -f check.fail ] && mv -f check.fail check.fail.old
if [ -s check.fail.tmp ]
then
	echo "part4:b: `wc -l < check.fail.tmp` mismatches found"
	echo "MD5 mismatches found, see file check.fail!" >&2
	mv -f check.fail.tmp check.fail
	echo -n ""
	ret=2
else
	if $rerun
	then
		echo "part4:b: please rerun to be safe everything is ok"
		echo "OK but rerun" >&2
		ret=3
	elif $fullrun
	then
		echo "part4:b: MD5 sums ok (full run)"
		ret=0
	else
		echo "part4:b: MD5 sums ok (incremental run)"
		ret=0
	fi
fi
sync
mv -f check.list check.list.old
sync