#!/bin/bash
#Copyright 2003 William Stearns <wstearns@pobox.com>
#version 0.5.0
#Note! sed's undocumented '-r' switch turns on both '|' processing and the use of ( instead of \(

#In /etc/squid/squid.conf
#acl spammers url_regex "/etc/squid/sa-blacklist.domains"
#http_access deny all spammers


if [ -z "$1" ]; then
	echo I need a blacklist command line parameter, exiting.
	exit 1
fi

case $1 in
*master.cf)
	Blacklist="`echo "$1" | sed -e 's/\.master\.cf$//'`"

	if [ `grep -if /usr/src/sa-blacklist/sa-whitelist $1 | grep -v '[a-zA-Z0-9]@' | wc -l` -gt 0 ]; then
		grep -if /usr/src/sa-blacklist/sa-whitelist $1 | grep -v '[a-zA-Z0-9]@' 
		echo blacklist contains whitelist entries, please fix, exiting.
		exit 1
	fi
	if [ `cat $1 | sed -e 's/#.*//' | grep blacklist_from | sort | uniq -d | wc -l` -gt 0 ]; then
		cat $1 | sed -e 's/#.*//' | grep blacklist_from | sort | uniq -d
		echo are duplicates, please fix, exiting
		exit 1
	fi

	#This was the manual approach, now replaced by strip-host-from-domain
	# | sed -r \
	#    -e 's/.*\.([a-z0-9-]*\.(art.br|com\.ar|com\.br|com\.cn|co\.il|co\.jp|com\.np|com\.pk|ne\.jp|co\.tv|com\.tw|co\.uk|co\.zw|etc\.br|net\.il|net\.tw|org\.tw|org\.uk|pro\.br|rec\.br|tmp\.br))$/\1HANDLED/' \
	#    -e 's/[a-z0-9-]*\.(art.br|com\.ar|com\.br|com\.cn|co\.il|co\.jp|com\.np|com\.pk|ne\.jp|co\.tv|com\.tw|co\.uk|co\.zw|etc\.br|net\.il|net\.tw|org\.tw|org\.uk|pro\.br|rec\.br|tmp\.br)$/&HANDLED/' \
	#    -e 's/.*\.([^\.]*\.(ar|biz|br|bz|cc|cn|com|de|edu|fr|info|il|it|jp|lt|net|np|org|tc|tv|tw|uk|us|zw))$/\1/' \
	#    -e 's/HANDLED$//' \
	RawDomains="`cat "$1" \
	 | grep '^blacklist_from \*@' \
	 | tr A-Z a-z \
	 | sed \
	    -e 's/^blacklist_from \*@//' \
	    -e 's/^\*\.//' \
	    -e 's/^\*\.//' \
	    -e 's/^\*\.//' \
	 | strip-host-from-domain \
	 | grep -v '\*' \
	 | sort \
	 | uniq`"

	SerialHeader=`cat "$1" | grep '^#sa-blacklist: [0-9]*$'`

	#==== ALL domains, even ones no longer in whois ====
	echo "$RawDomains" | grep -v '^yahoo.com$'								 >$Blacklist.withdead.domains


	;;
*.withdead.domains)
	Blacklist="`echo "$1" | sed -e 's/\.withdead\.domains$//'`"

	RawDomains="`cat "$1" \
	 | tr A-Z a-z \
	 | sort \
	 | uniq`"

	SerialHeader="#sa-blacklist: `echo \"$1\" | sed -e 's/\.withdead\.domains$//' -e 's/.*\.\([0-9]*\)/\1/'`"

	;;
*)
	echo error, blacklist needs to be called something.master.cf or something.withdead.domains, exiting
	exit 1
esac


#Now we create a URL filter file from the domains file:
echo "$SerialHeader" | sed -e 's/#sa-blacklist: /#sa-blacklist.uri: /'						 >$Blacklist.withdead.uri.cf

#The sed script below converts "a.b.c" on stdin to "a.b.c a\.b\.c" on
#stdout, allowing a single sed to convert every line without needing to
#do a fork on each domain.

URICount=0
#(  echo drs.yahoo.com
#   echo rd.yahoo.com
#   echo srd.yahoo.com
#   cat $Blacklist.withdead.domains ) \
   cat $Blacklist.withdead.domains \
 | sed '{
h
s/\./\\\\./g
x
G
s/\
/ /
}' \
 | while read OneDomain OneQuotedDomain ; do
	URICount=$[ $URICount + 1 ]
	echo "describe WLS_URI_$URICount URI contains $OneDomain"
	echo "uri      WLS_URI_$URICount /\b${OneQuotedDomain}\b/i"
	echo "score    WLS_URI_$URICount 2.0"
done														>>$Blacklist.withdead.uri.cf

#==== LIVE domains, must be in whois ====
echo "$RawDomains" \
 | grep -v '^yahoo.com$' \
 | ( while read ADomain ; do
 	if [ ! -f "/usr/src/whois-out/${ADomain}-whois" -a ! -f "/usr/src/whois-out/notfound/${ADomain}-whois" ]; then	#Neither exists
		Neither=$[ $Neither + 1 ]
		echo $ADomain		
 	elif [ -f "/usr/src/whois-out/${ADomain}-whois" -a ! -f "/usr/src/whois-out/notfound/${ADomain}-whois" ]; then	#Live exists
		Live=$[ $Live + 1 ]
		echo $ADomain
 	elif [ ! -f "/usr/src/whois-out/${ADomain}-whois" -a -f "/usr/src/whois-out/notfound/${ADomain}-whois" ]; then	#Dead exists
		Dead=$[ $Dead + 1 ]
		#No write
 	elif [ -f "/usr/src/whois-out/${ADomain}-whois" -a -f "/usr/src/whois-out/notfound/${ADomain}-whois" ]; then	#Both exist
		if [ "/usr/src/whois-out/${ADomain}-whois" -nt "/usr/src/whois-out/notfound/${ADomain}-whois" ]; then
			BothLiveNewer=$[ $BothLiveNewer + 1 ]
			echo $ADomain
		else
			BothDeadNewer=$[ $BothDeadNewer + 1 ]
			#No write
		fi
	fi
 done
 echo "Neither $Neither"											 >$Blacklist.stats
 echo "Live $Live"												>>$Blacklist.stats
 echo "Dead $Dead"												>>$Blacklist.stats
 echo "BothLiveNewer $BothLiveNewer"										>>$Blacklist.stats
 echo "BothDeadNewer $BothDeadNewer"										>>$Blacklist.stats
 )														 >$Blacklist.domains
rm -f sa-blacklist.current.domains ; ln -sf $Blacklist.domains sa-blacklist.current.domains


echo '#### Start of Manual Blacklist ####'									 >$Blacklist.cf
echo "$SerialHeader"												>>$Blacklist.cf
echo '#This list provided by William Stearns <wstearns@pobox.com>, please send'					>>$Blacklist.cf
echo '#additions and corrections.'										>>$Blacklist.cf
echo '#The master copy is kept at:'										>>$Blacklist.cf
echo '#http://www.stearns.org/sa-blacklist/sa-blacklist.current.cf'						>>$Blacklist.cf
echo '#Some of the following may be trademarks, owned by their respective owners.'				>>$Blacklist.cf
echo '#'													>>$Blacklist.cf
cat "$Blacklist.domains" | grep -v '^[0-9\.]*$' | sed -e 's/^/blacklist_from *@/'				>>$Blacklist.cf
cat "$Blacklist.domains" | grep -v '^[0-9\.]*$' | sed -e 's/^/blacklist_from *@*./'				>>$Blacklist.cf
rm -f sa-blacklist.current ; ln -sf $Blacklist.cf sa-blacklist.current
rm -f sa-blacklist.current.cf ; ln -sf $Blacklist.cf sa-blacklist.current.cf


echo "$SerialHeader" | sed -e 's/#sa-blacklist: /#sa-blacklist.at-domains: /'					 >$Blacklist.at-domains
echo '#This file is designed to be used in Qmails "badmailfrom" file, found'					>>$Blacklist.at-domains
echo '#in the "control" directory.  Append these lines and restart qmail.'					>>$Blacklist.at-domains
echo '#The latest version of this list can be found at: '							>>$Blacklist.at-domains
echo '#http://www.stearns.org/sa-blacklist/'									>>$Blacklist.at-domains
cat "$Blacklist.domains" | sed -e 's/^/@/'									>>$Blacklist.at-domains
rm -f sa-blacklist.current.at-domains ; ln -sf $Blacklist.at-domains sa-blacklist.current.at-domains


echo "$SerialHeader" | sed -e 's/#sa-blacklist: /#sa-blacklist.sendmail-access: /'				 >$Blacklist.sendmail-access
echo '#This is designed to be used in Sendmails access file, usually found'					>>$Blacklist.sendmail-access
echo '#in /etc/mail/access .  Append these lines, run "make access.db",'					>>$Blacklist.sendmail-access
echo '#and restart sendmail.  The latest version of this file can be found'					>>$Blacklist.sendmail-access
echo '#at http://www.stearns.org/sa-blacklist/ .'								>>$Blacklist.sendmail-access
cat "$Blacklist.domains" | sed -e 's@$@		ERROR:550 "Blocked by http://www.stearns.org/sa-blacklist/"@'	>>$Blacklist.sendmail-access
rm -f sa-blacklist.current.sendmail-access ; ln -sf $Blacklist.sendmail-access sa-blacklist.current.sendmail-access


echo "$SerialHeader" | sed -e 's/#sa-blacklist: /#sa-blacklist.action: /'					 >$Blacklist.action
echo '#Designed for use with the Privoxy http proxy, see'							>>$Blacklist.action
echo '#http://www.privoxy.org/ .  This list can be found at'							>>$Blacklist.action
echo '#http://www.stearns.org/sa-blacklist/'									>>$Blacklist.action
echo '{+block}'													>>$Blacklist.action
cat "$Blacklist.domains" | sed -e 's@^@.@'									>>$Blacklist.action
rm -f sa-blacklist.current.action ; ln -sf $Blacklist.action sa-blacklist.current.action
rm -f sa-blacklist.current.actions ; ln -sf $Blacklist.action sa-blacklist.current.actions


echo "$SerialHeader" | sed -e 's/#sa-blacklist: /#sa-blacklist.reject: /'					 >$Blacklist.reject
echo '#Designed for use with the Postfix MTA, see'								>>$Blacklist.reject
echo '#http://www.postfix.org/ .  This list and install instructions can be'					>>$Blacklist.reject
echo '#found at http://www.stearns.org/sa-blacklist/'								>>$Blacklist.reject
cat "$Blacklist.domains" | sed -e 's@$@	REJECT@'								>>$Blacklist.reject
rm -f sa-blacklist.current.reject ; ln -sf $Blacklist.reject sa-blacklist.current.reject


#echo "$SerialHeader" | sed -e 's/#sa-blacklist: /;sa-blacklist.bind: /'					 >$Blacklist.bind
#Timestamp=`echo "$SerialHeader" | sed -e 's/#sa-blacklist: //'`
#echo ';Designed for use with the Bind dns server, see'								>>$Blacklist.bind
#echo ';http://www.isc.org/sw/bind/ .  This list and install instructions'					>>$Blacklist.bind
#echo ';can be found at http://www.stearns.org/sa-blacklist/'							>>$Blacklist.bind
#cat <<EOHEADER													>>$Blacklist.bind
#\$TTL	600
#@			IN	SOA	ns1.freeapp.net.	zone.surbl.org. (
#        		                $Timestamp 600 300 604800 600 )
#
#@			IN	NS	ns1.freeapp.net.
#@			IN	NS	dns1.littleredbat.net.
#@			IN	NS	a.rbl-auth.sr.sonic.net.
#@			IN	NS	b.rbl-auth.sr.sonic.net.
#
#@			IN	A	66.170.2.60
#
#test.surbl.org		IN	A	127.0.0.2	; permanent test point
#test.sc.surbl.org	IN	A	127.0.0.2	; permanent test point
#EOHEADER
#cat "$Blacklist.domains" | sed -e 's@$@	IN	A	127.0.0.2\
#		IN	TXT	"Message body contains spammer domain, see http://www.stearns.org/sa-blacklist"@'	>>$Blacklist.bind


#Now we create a URL filter file from the domains file:
echo "$SerialHeader" | sed -e 's/#sa-blacklist: /#sa-blacklist.uri: /'						 >$Blacklist.oneperrule.uri.cf
echo '#This is designed to be used with Spamassassin,'								>>$Blacklist.oneperrule.uri.cf
echo '#http://www.spamassassin.org .  Place it in /etc/mail/spamassassin/'					>>$Blacklist.oneperrule.uri.cf
echo '#and restart spamassassin.  The latest version of this list can be'					>>$Blacklist.oneperrule.uri.cf
echo '#found at http://www.stearns.org/sa-blacklist/ .'								>>$Blacklist.oneperrule.uri.cf
URICount=0
#(  echo drs.yahoo.com
#   echo rd.yahoo.com
#   echo srd.yahoo.com
#   cat $Blacklist.domains ) \
   cat $Blacklist.domains \
 | sed '{
h
s/\./\\\\./g
x
G
s/\
/ /
}' \
 | while read OneDomain OneQuotedDomain ; do
	URICount=$[ $URICount + 1 ]
	echo "describe WLS_URI_$URICount URI contains $OneDomain"
	echo "uri      WLS_URI_$URICount /\b${OneQuotedDomain}\b/i"
	echo "score    WLS_URI_$URICount 2.0"
done														>>$Blacklist.oneperrule.uri.cf



cat $Blacklist.oneperrule.uri.cf | dwinkler-convert-uri-list.pl							 >$Blacklist.uri.cf
rm -f sa-blacklist.current.uri.cf ; ln -sf $Blacklist.uri.cf sa-blacklist.current.uri.cf