#!/bin/bash
dir1=$1
dir2=$2
work_dir=$3
mail_to=$4
mail_from=$5
hour_to_mail_regardless=$6
cache_results=$7
#hour_to_mail_regardless=""

start_date=$(date)

[ "$mail_from" == "" ] && mail_from="root@moparisthebest.com"
[ "$hour_to_mail_regardless" == "" ] && hour_to_mail_regardless="$(date +%k)"
( [ "$cache_results" != "0" ] && [ "$cache_results" != "1" ] ) && cache_results="0"
#( [ "$cache_results" != false ] && [ "$cache_results" != true ] ) && cache_results=false

mkdir -p $work_dir

list_files(){
	find -L . -not -type d -not -type l | egrep -v "^./rdiff-backup-data" | egrep -v '.bash_history$' | egrep -v 'logs/lighttpd/.*\.log$' | egrep -v 'cache/(data|lang)_.*\.php$' | egrep -v "worldmap/.*\.png$" | egrep -v "(crc_file|log)$" | egrep -v "moparscape\.org/timestamp$" | egrep -v "moparisthebest\.com/downloads/current_simba/.*(\.7z|\.exe|-linux)$" | egrep -v "moparisthebest\.com/rs/gamepack.*\.(jar|properties)$" | egrep -v "moparisthebest\.com/smf/attachments/([0-9]+_[0-9a-f]{40}|post_tmp_[0-9]+_[0-9])$"
}

print_report(){
	[ -s "$2" ] && (echo -e "$1"; cat "$2"; echo; echo -en "$3") >> $work_dir/report
}

# copy over cached files if requested, and they exist
if [ "$cache_results" == "1" ]
then
	[ -f "$dir1/dir1.files" ] && cp "$dir1/dir1.files" "$work_dir/dir1.files"
	[ -f "$dir1/dir1.md5" ] && cp "$dir1/dir1.md5" "$work_dir/dir1.md5"
#	[ -f "$dir1/files_to_checksum" ] && cp "$dir1/files_to_checksum" "$work_dir/files_to_checksum"
fi

cd $dir1
# list files if cache_results is false or dir1.files doesn't exist
( [ "$cache_results" == "0" ] || [ ! -f "$work_dir/dir1.files" ] ) && list_files > $work_dir/dir1.files
# if cache_results is true, and a cached file list doesn't exist, copy it
( [ "$cache_results" == "1" ] && [ ! -f "$dir1/dir1.files" ] ) && cp "$work_dir/dir1.files" "$dir1/dir1.files"


cd $dir2
list_files > $work_dir/dir2.files

diff -Naurw $work_dir/dir1.files $work_dir/dir2.files > $work_dir/files.diff

egrep "^\+" $work_dir/files.diff | tail -n+2 | sed "s/^\+//" > $work_dir/files_added
egrep "^-"  $work_dir/files.diff | tail -n+2 | sed "s/^-//" > $work_dir/files_removed
cat $work_dir/dir1.files $work_dir/dir2.files | sort | uniq -d > $work_dir/files_to_checksum
cat $work_dir/dir1.files $work_dir/dir2.files | sort | uniq -u > $work_dir/files_added_or_removed

# if cache is enabled, and files_to_checksum is different, refresh dir1.md5sum
#( [ "$cache_results" == "1" ] && [ "$(md5sum "$dir1/files_to_checksum" | cut -f1 -d ' ')" != "$(md5sum "$work_dir/files_to_checksum" | cut -f1 -d ' ')" ] ) && ( rm "$work_dir/dir1.md5" "$dir1/dir1.md5"; cp "$work_dir/files_to_checksum" "$dir1/files_to_checksum" )

temp_ifs=$IFS
IFS="
"

if ( [ "$cache_results" == "0" ] || [ ! -f "$work_dir/dir1.md5" ] )
then
	rm -f $work_dir/dir1.md5
	cd $dir1
	for f in $(cat $work_dir/files_to_checksum)
	do
		md5sum "$f" >> $work_dir/dir1.md5
	done
fi
( [ "$cache_results" == "1" ] && [ ! -f "$dir1/dir1.md5" ] ) && cp "$work_dir/dir1.md5" "$dir1/dir1.md5"

cd $dir2
cat $work_dir/dir1.md5 | md5sum -c | egrep -v ": OK$" | sed "s/: FAILED$//" | sed "/: FAILED open or read$/d"  > $work_dir/files_changed

rm -f $work_dir/files_changed.diff
dir1_sed=$(echo $dir1 | sed -e 's/\(\.\|\/\|\*\|\[\|\]\|\\\)/\\&/g')
dir2_sed=$(echo $dir2 | sed -e 's/\(\.\|\/\|\*\|\[\|\]\|\\\)/\\&/g')
for f in $(cat $work_dir/files_changed | sed "s/\.\///")
do
	tmp=$(diff -Naurw "$dir1/$f" "$dir2/$f")
	# just use sed on the first 2 lines, so we know we won't mess up anything else
	echo "$tmp" | head -n2 | sed "s/^--- $dir1_sed/--- a/" | sed "s/^+++ $dir2_sed/+++ b/" >> $work_dir/files_changed.diff
	# copy from lines 3 on down
	echo "$tmp" | tail -n+3 >> $work_dir/files_changed.diff
done
IFS=$temp_ifs

rm -f $work_dir/report
print_report "Files added:\n" "$work_dir/files_added"
print_report "Files removed:\n" "$work_dir/files_removed"
print_report "Files changed:\n" "$work_dir/files_changed" "Diff file: $work_dir/files_changed.diff\n\n"

[ ! -s "$work_dir/report" ] && attach="no" && echo -e "No changes!\n" >> $work_dir/report

(echo "Started:"; echo $start_date; echo "Finished:"; date; uptime; echo; echo ) >> $work_dir/report

[ "$mail_to" != "" ] && ( [ "$attach" != "no" ] || [ "$(date +%k)" == "$hour_to_mail_regardless" ] ) && ( echo -e "To: $mail_to\nFrom: $mail_from\nSubject: Diff Report\n"; cat $work_dir/report $work_dir/files_changed.diff; echo; [ "$attach" != "no" ] && uuencode $work_dir/report report.txt; [ -s "$work_dir/files_changed.diff" ] && uuencode $work_dir/files_changed.diff files_changed.diff.txt ) | sendmail -f $mail_from $mail_to

if [ $? != 0 ]
then
	# if the last command (sendmail) failed (maybe diff file too big?), send a shorter email saying so, without the diff file or attachment
	[ "$mail_to" != "" ] && ( [ "$attach" != "no" ] || [ "$(date +%k)" == "$hour_to_mail_regardless" ] ) && ( echo -e "To: $mail_to\nFrom: $mail_from\nSubject: Diff Report\n"; cat $work_dir/report; echo "Diff file too big!"; echo; ) | sendmail -f $mail_from $mail_to
	# if the last command (sendmail) failed (maybe report too big?), send a short-and-sweet email saying so, without anything else
	[ $? != 0 ] && [ "$mail_to" != "" ] && ( [ "$attach" != "no" ] || [ "$(date +%k)" == "$hour_to_mail_regardless" ] ) && ( echo -e "To: $mail_to\nFrom: $mail_from\nSubject: Diff Report\n"; echo "Changes too large to mail!"; echo; ) | sendmail -f $mail_from $mail_to
fi


exit 0;