path: root/localwebsearch

#!/bin/bash
#
# /usr/local/sbin/localwebsearch
# Copyright 2001-2002 Jonas Smedegaard <dr@jones.dk>
#
# $Id: localwebsearch,v 1.5 2004-01-20 22:39:19 jonas Exp $
#
# Web search engine maintenance script
#

# halt on errors (NB! this is a bashism...)
set -e

function usage()    {
    echo "Usage: $(basename $0) initprep|init|update|ignore <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
}

function exit1()    {
    echo "Error: $1"
    echo "Exiting..."
    exit 1
}

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

case "$stamp" in
    initprep|init|update|ignore)
    ;;
    *)
    usage
    ;;
esac

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

WEBROOT=/var/www
LOGROOT=/var/log/apache

# Options: $1=LOGROOT, $2=WEBROOT, $3=WEBSITE
function searchdir()        { echo $2/VIRTUAL/search.$(dnsdomainname)/www/$3; }
function webdirs()      { find /etc/htdig -type f -name '*.conf' ! -name 'htdig.conf' -exec basename '{}' .conf \; ; }
#function host()            { cat $2/VIRTUAL/$3/hostname || exit1 "Unable to get hostname for virtual host."; }
#function domain()      { cat $2/VIRTUAL/$3/domainname || exit1 "Unable to get domainname for virtual host."; }
function htdig_cfg()        { echo /etc/htdig/$3.conf; }
function pre_init()     { true; }
function post_init()        { true; }
function pre_update()       { true; }
function post_update()      { true; }

# The above can be overridden
LOCALCONFIG=/etc/local/websearch.conf
. $LOCALCONFIG || exit1 "Unable to read local config file $LOCALCONFIG"

# variables and functions too boring to be configurable
HTDIG_BIN="/usr/bin/rundig"
HTDIG_REAL_BIN="/usr/local/sbin/localrundig" #TODO: Convince Debian maintainer to change official rundig

#function awstats_setlog()  { sed -e "s!^\(LogFile=\).*\$!\\1$2!" $1 > $1.tmp; mv $1.tmp $1; }

# Webiste/independent checks
test -d $WEBROOT || exit1 "Webroot \"$WEBROOT\" doesn't exist"
test -d $LOGROOT || exit1 "Logroot \"$LOGROOT\" doesn't exist"

# Index searches for websites from stdin or all default sites
WEBSITES=$@
if [ "$WEBSITES" = "" ]; then
    WEBSITES=$(webdirs $LOGROOT $WEBROOT $WEBSITE)
fi

for WEBSITE in $WEBSITES; do
    SEARCHDIR=$(searchdir $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve SEARCHDIR."
# FIXME test -d $SEARCHDIR/.. || exit1 "Directory above SEARCHDIR doesn't exist."

    HTDIG_CFG=$(htdig_cfg $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to resolve HTDIG_CFG."

    if [ $stamp = "init" ]; then
        [ $DEBUG ] && echo "Execute $WEBSITE PRE_INIT"
        pre_init $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing PRE_INIT."
    fi
    if [ $stamp = "update" ]; then
        [ $DEBUG ] && echo "Execute $WEBSITE PRE_UPDATE"
        pre_update $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing PRE_UPDATE."
    fi

#   HOST=$(host $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to get hostname for virtual host."
#   DOMAIN=$(domain $LOGROOT $WEBROOT $WEBSITE) || exit1 "Unable to get domainname for virtual host."

#   HTDIG_OPTIONS="-a" #TODO: This seems to always init currently
    HTDIG_OPTIONS=""
    [ "$WEBSITE" != "htdig" ] && HTDIG_OPTIONS="$HTDIG_OPTIONS -c $HTDIG_CFG"

    if [ $DEBUG ]; then
        echo "Indexing search for $WEBSITE in $SEARCHDIR:"
        HTDIG_OPTIONS="$HTDIG_OPTIONS -v -s"
    else
        HTDIG_OPTIONS="$HTDIG_OPTIONS"
    fi

    if [ "$stamp" = "init" ]; then
        HTDIG_OPTIONS="$HTDIG_OPTIONS -i"
    fi

    if [ "$stamp" = "initprep" ]; then
        [ $DEBUG ] && echo "Mark next run as a full-scale, and move on to next site..."
        touch /etc/htdig/full_refresh_$WEBSITE
        continue
    fi

# TODO
    if [ $stamp = "initXXX" ]; then
        [ $DEBUG ] && echo "$WEBSITE: Purge SEARCHDIR"
        rm -rf $SEARCHDIR
        mkdir $SEARCHDIR
        if [ -x $HTDIG_BIN ]; then
            [ $DEBUG ] && echo "$WEBSITE: Create/update htdig config"
            echo "\
# NB! This file is automatically generated. Do not edit directly!
# Instead, put additions/overrides in $HTDIG_CFG.local

HOSTNAME    $HOST.$DOMAIN
HOSTURL     http://$HOST.$DOMAIN/
BASEURL     http://$HOST.$DOMAIN
LANGUAGE    DANISH
"\
            > $HTDIG_CFG
            [ -s $HTDIG_CFG.local ] && cat $HTDIG_CFG.local >> $HTDIG_CFG
            mkdir $SEARCHDIR/htdig
        fi
    fi

#   if ! lockfile-create /var/run/localwebsearch_$WEBSITE; then
#       # Another htdig indexing cronjob is already running
#       [ $DEBUG ] && echo "Another $WEBSITE indexing is already running. Ignoring silently..."
#       continue
#   fi

#   lockfile-touch /var/run/localwebsearch_$WEBSITE &
#   # Save the PID of the lockfile-touch process
#   BADGER="$!"

    if [ $stamp = "init" -o $stamp = "update" ]; then
        if [ -x $HTDIG_BIN -a -x $HTDIG_REAL_BIN ]; then
            export TMPDIR=/tmp
            [ $DEBUG ] && echo "$WEBSITE: Update htdig search"
            if [ $stamp = "init" -o -f /etc/htdig/full_refresh_$WEBSITE ]; then
                $HTDIG_REAL_BIN $HTDIG_OPTIONS
                rm -f /etc/htdig/full_refresh_$WEBSITE
            else
                $HTDIG_REAL_BIN $HTDIG_OPTIONS
            fi
        fi
    fi

#   kill "${BADGER}"
#   lockfile-remove /var/run/localwebsearch_$WEBSITE

    if [ $stamp = "init" ]; then
        [ $DEBUG ] && echo "Execute $WEBSITE POST_INIT"
        post_init $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing POST_INIT."
    fi
    if [ $stamp = "update" ]; then
        [ $DEBUG ] && echo "Execute $WEBSITE POST_UPDATE"
        post_update $LOGROOT $WEBROOT $WEBSITE || exit1 "Error executing POST_UPDATE."
    fi

done