#!/bin/bash
#
# squidGuard blacklist update script v1.3 (incomplete!)
# http://www.bn-paf.de/filter/ - filter@bn-paf.de
#
# Todo (somebody wants to help?):
# - notify if:
#  downloaded files are broken
#  squidGuard goes into emergency mode for some reason

SQGDIR=/usr/share/squidGuard-1.1.4
SQGLOG=/var/log/squidGuard/log
CONFIG=/etc/squid/squidGuard/squidGuard.conf

# Use "proxy" for Debian, otherwise usually "squid"
USER=nobody

# Copy blacklists locally for testing purposes
# --- not implemented yet ---
DEVEL=0

# Workaround for some installations having problems with diffs
TEXT=0



if test -z $1;
  then echo; echo Please specifiy what to update: all, diffs or custom; echo; exit;
fi



custom-update ()
{ echo Compiling custom text files...
  #/usr/local/squidGuard/bin/squidGuard -C  all -c $CONFIG 2>$SQGLOG/custom.log
  /usr/bin/squidGuard -C  bad/domains -c $CONFIG 2>$SQGLOG/custom.log             
  /usr/bin/squidGuard -C bad/urls -c $CONFIG 2>>$SQGLOG/custom.log      
  /usr/bin/squidGuard -C good/domains -c $CONFIG 2>>$SQGLOG/custom.log     
  /usr/bin/squidGuard -C good/urls    -c $CONFIG 2>>$SQGLOG/custom.log; }

if test "$1" = custom; then custom-update; /usr/sbin/squid -k reconfigure; exit; fi



dircreate  () { if test ! -d $1; then mkdir  $1; fi; }
filecreate () { if test ! -f $1; then echo > $1; fi; }

dircreate  $SQGDIR
dircreate  $SQGDIR/db
dircreate  $SQGDIR/db/custom
dircreate  $SQGDIR/db/custom/bad
filecreate $SQGDIR/db/custom/bad/domains
filecreate $SQGDIR/db/custom/bad/urls
dircreate  $SQGDIR/db/custom/good
filecreate $SQGDIR/db/custom/good/domains
filecreate $SQGDIR/db/custom/good/urls
dircreate  $SQGDIR/tmp
dircreate  $SQGLOG
filecreate $SQGLOG/blocked.log



rm -rf $SQGDIR/tmp/*
cd $SQGDIR/tmp

echo Downloading blacklists...

if test "$1" = all -o ! -f ../blacklists.tar.gz; then
  wget -nv ftp://ftp.ost.eltele.no/pub/www/proxy/squidGuard/contrib/blacklists.tar.gz
  #cp $SQGDIR/devel/blacklists.tar.gz .
  mv blacklists.tar.gz ..
fi
tar xfz ../blacklists.tar.gz

wget -nv http://www.bn-paf.de/filter/de-blacklists.tar.gz
#cp $SQGDIR/devel/de-blacklists.tar.gz .
tar xfz de-blacklists.tar.gz

wget -nv ftp://ftp.univ-tlse1.fr/pub/reseau/cache/squidguard_contrib/adult.tar.gz
wget -nv ftp://ftp.univ-tlse1.fr/pub/reseau/cache/squidguard_contrib/publicite.tar.gz
wget -nv ftp://ftp.univ-tlse1.fr/pub/reseau/cache/squidguard_contrib/redirector.tar.gz
wget -nv ftp://ftp.univ-tlse1.fr/pub/reseau/cache/squidguard_contrib/warez.tar.gz
#cp $SQGDIR/devel/adult.tar.gz .
#cp $SQGDIR/devel/publicite.tar.gz .
#cp $SQGDIR/devel/redirector.tar.gz .
#cp $SQGDIR/devel/warez.tar.gz .
mkdir fr-blacklists
   cd fr-blacklists
tar xfz ../adult.tar.gz
tar xfz ../publicite.tar.gz
tar xfz ../redirector.tar.gz
tar xfz ../warez.tar.gz
cd ..
mv fr-blacklists/adult fr-blacklists/porn
mv fr-blacklists/publicite fr-blacklists/ads
mv fr-blacklists/redirector fr-blacklists/hacking

wget -nv http://squidguard.mesd.k12.or.us/mesd-blacklists.tgz
#cp $SQGDIR/devel/mesd-blacklists.tgz .
tar xfz mesd-blacklists.tgz




if ! test "$TEXT" = 1; then

echo Compiling diffs...

diffgen ()
    { if test -f $1;
        then sed 's/^/+/' $1 >> $SQGDIR/db/$filter/$2.diff; fi
      if test -f $1.exclude;
        then sed 's/^/-/' $1.exclude >> $SQGDIR/db/$filter/$2.diff; fi; }

diffadd ()
    { cat $source-blacklists/$filter/$1.diff>>$SQGDIR/db/$filter/$1.diff; }

for filter in ads aggressive audio-video drugs gambling hacking porn violence warez;
do
  # Creates empty domain/url files to avoid emergency mode when compilling
  # custom lists later.
  dircreate $SQGDIR/db/$filter
  mv blacklists/$filter/*.db $SQGDIR/db/$filter/  
  filecreate $SQGDIR/db/$filter/domains; filecreate $SQGDIR/db/$filter/urls;
  for source in de fr mesd
  do
    if test -f $source-blacklists/$filter/domains.diff;
      then diffadd domains;
      else diffgen $source-blacklists/$filter/domains domains;
    fi
    if test -f $source-blacklists/$filter/urls.diff;
      then diffadd urls;
      else diffgen $source-blacklists/$filter/urls urls;
    fi
  done
done

else
# Much slower than using diffs, creates plain text urls/domains
# Thanks Wolfgang for this part!
echo Creating text files...
for filter in ads aggressive audio-video drugs gambling hacking porn violence warez;
do
  if test ! -d $SQGDIR/db/$filter; then mkdir $SQGDIR/db/$filter; fi
  for addrtype in domains urls;
  do
# We filter possible <LF>s using tr (comm is not working with that) and
# comment lines which are confusing to read after sorting
    cat blacklists/$filter/$addrtype > temp.txt
    for source in de fr mesd
    do
      if test -f $source-blacklists/$filter/$addrtype;
      then cat $source-blacklists/$filter/$addrtype >> temp.txt; fi
    done
    comm -2 -3 <(tr -d '\r' < temp.txt | grep -v "^#" | sort) <(tr -d '\r' < de-blacklists/$filter/$addrtype.exclude | grep -v "^#" | sort) > $SQGDIR/db/$filter/$addrtype
  done
done
rm temp.txt

fi



echo Securing files...
chownmod () { chown $1 $2
              chmod ug+rw,o-rw $2
	      chown -R $1 $2/*
              chmod -R ug+rw,o-rw $2/*; }
chownmod $USER:$USER $SQGDIR/db
chownmod  root:root  $SQGDIR/tmp
chownmod $USER:$USER $SQGLOG



if ! test "$TEXT" = 1; then
  if ! test "$1" = custom; then
    echo Applying diffs...
    /usr/bin/squidGuard -u -c $CONFIG 2>$SQGLOG/update.log
    # Mandrake RPM: /usr/squidGuard/log/squidGuard.log will be used instead of
    # update.log if directory exists
  fi
  custom-update
else
  echo Compiling text files...
  /usr/bin/squidGuard -C all -c $CONFIG 2>$SQGLOG/update.log
fi


echo Reconfiguring Squid...; /usr/sbin/squid -k reconfigure