path: root/localwebstats

#!/bin/bash
# /etc/cron.daily/localstats: Webstats maintenance script
# Written by Jonas Smedegaard <dr@jones.dk>

# halt on errors
#set -e

function usage()    {
    echo "Usage: `basename $0` update|prelogrotate|postlogrotate|reset <website> [<website>...]"
    echo "  If no website is given, all are attempted"
    echo "  Tip: Automagically runs when symlinked to /etc/cron.{daily,weekly,monthly}/"
    exit 1
}

# automagically configure when run from cron dirs
case `dirname $0` in
    /etc/cron.daily)
    stamp=ignore
    ;;
    /etc/cron.weekly)
    stamp=update
    ;;
    /etc/cron.monthly)
    stamp=ignore
    ;;
    *)
    stamp=$1
    shift || usage
    ;;
esac

case "$stamp" in
    update|prelogrotate|postlogrotate|reset|ignore)
    ;;
    *)
    usage
    ;;
esac

if [ "$stamp" == "ignore" ]; then
    if [ $DEBUG ]; then
        echo "Asked to ignore - exiting silently..."
    fi
    exit 0
fi

WEBROOT=/var/www
LOGROOT=/var/log/apache
WEBALIZER_OPTIONS="-q"

function webdir()   { echo $1/VIRTUAL/$2; }
function webdirs()  { echo `find /var/log/apache -type d -print`; }
#function webdirs() { for dir in $(find ${LOGROOT} -type d -print); do echo $dir; done; }
#function logdir()  { echo ${LOGROOT}/${WEBSITE}; }
#function logfiles()    { $(ls -r $LOGDIR/*-access*.gz) $(ls -r $LOGDIR/access*.??.gz) $(ls -r $LOGDIR/access*.?.gz) $(ls -r $LOGDIR/access*.?); }
function logfiles() { ls $1/$2/????.??.??; }
function logfilesgz()   { ls $1/$2/????.??.??.gz; }
function host()     { cat $1/VIRTUAL/$2/hostname || exit 1; }
function domain()   { cat $1/VIRTUAL/$2/domainname || exit 1; }

# The above can be overridden in /etc/local/www

. /etc/local/www || exit 1

# variables too boring to be configurable
JDRESOLVE_DB="/var/cache/jdresolve/hosts.db"
#JDRESOLVE_OPTIONS="-r -t 5 --database=${JDRESOLVE_DB} --dbfirst --expiredb=48"
JDRESOLVE_OPTIONS="-r --database=${JDRESOLVE_DB} --dbfirst --expiredb=48"
JDRESOLVE_MERGE_OPTIONS="--mergedb --database=${JDRESOLVE_DB}"

test -d ${WEBROOT} -a -d ${LOGROOT} || exit 1

WEBSITES=$@
if [ "$WEBSITES" == "" ]; then
    WEBSITES=$(webdirs)
fi

for WEBSITE in $WEBSITES; do
    WEBDIR=$(webdir $WEBROOT ${WEBSITE}) || exit 1

    test -d ${WEBDIR} || exit 1

    HOST=$(host ${WEBROOT} ${WEBSITE}) || exit 1
    DOMAIN=$(domain ${WEBROOT} ${WEBSITE}) || exit 1
    LOGFILES=$(logfiles ${LOGROOT} ${WEBSITE}) || exit 1
    LOGFILESGZ=$(logfilesgz ${LOGROOT} ${WEBSITE}) || exit 1

    if [ -x /usr/bin/jdresolve ]; then
        [ ${DEBUG} ] && echo "Checking access to ${JDRESOLVE_DB}"
        touch ${JDRESOLVE_DB} || exit 1
    fi

    WEBALIZER_OPTIONS="-c /etc/webalizer.conf -c $WEBDIR/webalizer.conf -o $WEBDIR/stats/webalizer -f"
    ANALOG_OPTIONS="-G +g/etc/analog.conf +g$WEBDIR/analog.conf -a"
    AWSTATS_OPTIONS="-update"
    FQDN_ESC=`echo "$HOST.$DOMAIN" | sed -e 's/\./\\\./g'` # needed for awstats config

    if [ ${DEBUG} ]; then
        echo "Making stats for ${WEBSITE} in ${WEBDIR}:"
        JDRESOLVE_OPTIONS="${JDRESOLVE_OPTIONS} -p"
        ANALOG_OPTIONS="${ANALOG_OPTIONS} +q"
        WEBALIZER_OPTIONS="${WEBALIZER_OPTIONS} -T"
        AWSTATS_OPTIONS="${AWSTATS_OPTIONS} -showsteps"
    else
        JDRESOLVE_OPTIONS="${JDRESOLVE_DB} -n"
        ANALOG_OPTIONS="${ANALOG_OPTIONS} -q"
        WEBALIZER_OPTIONS="${WEBALIZER_OPTIONS} -Q"
        AWSTATS_OPTIONS="${AWSTATS_OPTIONS}"
    fi

    if [ $stamp == "reset" ]; then
        rm -rf $WEBDIR/stats
        mkdir $WEBDIR/stats
    fi

    LOGDATA="${WEBDIR}/stats/rawlog.txt"
    if [ -x /usr/bin/jdresolve ]; then
        zcat ${LOGFILESGZ} | jdresolve ${JDRESOLVE_OPTIONS} - > ${LOGDATA}
        cat ${LOGFILES} | jdresolve ${JDRESOLVE_OPTIONS} - >> ${LOGDATA}
        jdresolve ${JDRESOLVE_MERGE_OPTIONS} ${LOGDATA}
    elif [ -x /usr/sbin/logresolve ]; then
        zcat ${LOGFILESGZ} | logresolve > ${LOGDATA}
        cat ${LOGFILES} | logresolve >> ${LOGDATA}
    else
        zcat ${LOGFILESGZ} > ${LOGDATA}
        cat ${LOGFILES} >> ${LOGDATA}
    fi

    if [ -x /usr/bin/analog ]; then
        if [ $stamp == "reset" ]; then
            test ! -f $WEBDIR/analog.conf && echo "\
HOSTNAME    $HOST.$DOMAIN
"\
            >$WEBDIR/analog.conf
            mkdir ${WEBDIR}/stats/analog
            cat ${LOGDATA} | analog ${ANALOG_OPTIONS} -C"LOGFILE none" -C"LOGFILE stdin" > $WEBDIR/stats/analog/index.html
        fi
#       cat ${LOGDATA} | analog ${ANALOG_OPTIONS} -C"LOGFILE none" -C"LOGFILE stdin" > $WEBDIR/stats/analog/index.html
    fi
    if [ -x /usr/bin/webalizer ]; then

        if [ $stamp == "reset" ]; then
            test ! -f $WEBDIR/webalizer.conf && echo "\
HostName    $HOST.$DOMAIN
HideSite    *$DOMAIN
HideReferrer    $DOMAIN/
"\
            > $WEBDIR/webalizer.conf
            mkdir $WEBDIR/stats/webalizer
            cat ${LOGDATA} | webalizer $WEBALIZER_OPTIONS -N 0 -
        fi
#       cat ${LOGDATA} | webalizer $WEBALIZER_OPTIONS -N 0 -
    fi
    if [ -x /usr/lib/cgi-bin/awstats.pl ]; then
        if [ $stamp == "reset" ]; then
            test ! -f /etc/awstats/awstats.$WEBSITE.conf && echo "\
LogFile=\"/var/www/VIRTUAL/$WEBSITE/stats/rawlog.txt\"
LogFormat=4
DNSLookup=0
DirData=\"/var/cache/awstats\"
DirIcons=\"/awstats-icon\"
SiteDomain=\"$FQDN_ESC\"
HostAliases=\"$FQDN_ESC localhost 127\.0\.0\.1\"
Lang=\"dk\"
DirLang=\"/usr/share/awstats/lang\"
DefaultFile=\"index.html\"
ShowFlagLinks=0
AllowToUpdateStatsFromBrowser=0
ShowLinksOnUrl=0
DirCgi=\"http://cgi.jones.dk/cgi-bin\"
DirIcons=\"http://stats.jones.dk/awstats-icon\"
"\
            >/etc/awstats/awstats.$WEBSITE.conf
            mkdir $WEBDIR/stats/awstats
            /usr/lib/cgi-bin/awstats.pl -config=${WEBSITE} $AWSTATS_OPTIONS -output > $WEBDIR/stats/awstats/index.html
        fi
        /usr/lib/cgi-bin/awstats.pl -config=${WEBSITE} $AWSTATS_OPTIONS -output > $WEBDIR/stats/awstats/index.html
    fi
    test -L $WEBDIR/www/stats || ln -s $WEBDIR/stats $WEBDIR/www/stats
    test -L $WEBDIR/stats/stats && rm $WEBDIR/stats/stats
done