zebractl.in

harvest是一个下载html网页得机器人

#!/bin/bash
#
# 20sep1999/kj   [e-mail]
#
# control script for zebra system
# type "zebractl help" for full usage information
#
# this script was stolen from apachectl, distributed with apache 1.3.9
#

PATH=@prefix@/bin:$PATH

# set base directory
# all data and directories are relative to this directory
BASEDIR=$PWD

CFG=$BASEDIR/zebra.cfg
DATADIR=$BASEDIR/objects-xml
ZEBRADIR=$BASEDIR/zebra
SRVLOG=$ZEBRADIR/log/zebrasrv.log
IDXLOG=$ZEBRADIR/log/zebraidx.log
PIDFILE=$ZEBRADIR/run/zebrasrv.pid

ZEBRAIDX="zebraidx -c $CFG -g harvest" # -v none"
ZEBRASRV="zebrasrv -c $CFG -w $BASEDIR -l $SRVLOG @:9999"

#######################################
# check status of server process      #
#######################################
function check_pid {
    # check for pidfile
    if [ -f $PIDFILE ]; then
	PID=`cat $PIDFILE`
	if [ "x$PID" != "x" ] && kill -0 $PID 2>/dev/null ; then
	    STATUS="zebrasrv (pid $PID) running"
	    RUNNING=1
	else
	    STATUS="zebrasrv (pid $PID?) not running"
	    RUNNING=0
	fi
    else
	STATUS="zebrasrv (no pid file) not running"
	RUNNING=0
    fi
}


# start work
ARGV="$@"
if [ "x$ARGV" = "x" ]; then
    ARGS="help"
fi

for ARG in $@ $ARGS
do
    case $ARG in
    analyze)
	$ZEBRAIDX -s update $DATADIR
	;;
    index)
	$ZEBRAIDX update $DATADIR 2>> $IDXLOG
	echo "Do 'zebractl commit' to register the updated index.";
	;;
    commit)
	$ZEBRAIDX commit 2>> $IDXLOG
	;;
    fastindex)
	$ZEBRAIDX -n update $DATADIR 2>> $IDXLOG
	;;
    fullindex)
	$ZEBRAIDX update $DATADIR 2>> $IDXLOG
	$ZEBRAIDX commit 2>> $IDXLOG
	;;
    start)
	check_pid
	if [ $RUNNING -eq 1 ]; then
	    echo "$0 $ARG: zebrasrv (pid $PID) already running"
	    continue
	fi
	$ZEBRASRV &
	# wait until zebrasrv has written its pidfile
	sleep 3
	check_pid
	if [ $RUNNING -eq 1 ]; then
	    echo "$0 $ARG: zebrasrv started"
	else
	    echo "$0 $ARG: zebrasrv could not be started"
	fi
	;;
    stop)
	check_pid
	if [ $RUNNING -eq 0 ]; then
	    echo "$0 $ARG: $STATUS"
	    continue
	fi
	if kill $PID ; then
	    echo "$0 $ARG: zebrasrv stopped"
	else
	    echo "$0 $ARG: zebrasrv could not be stopped"
	fi
	;;
    restart)
	check_pid
	if [ $RUNNING -eq 0 ]; then
	    echo "$0 $ARG: zebrasrv not running, trying to start"
	    $ZEBRASRV &
	    sleep 3
	    check_pid
	    if [ $RUNNING -eq 1 ]; then
		echo "$0 $ARG: zebrasrv started"
	    else
		echo "$0 $ARG: zebrasrv could not be started"
	    fi
	else
	    if kill $PID ; then
		$ZEBRASRV &
		sleep 3
		check_pid
		if [ $RUNNING -eq 1 ]; then
		    echo "$0 $ARG: zebrasrv restarted"
		else
		    echo "$0 $ARG: zebrasrv could not be restarted"
		fi
	    else
		echo "$0 $ARG: couldn't stop zebrasrv for restart"
	    fi
	fi
	;;
    *)
	echo
	echo "Usage: $0 <analyze|index|commit|fastindex|fullindex|start|stop|restart>"
	cat <<EOF

analyze   - don't make any changes, just show what will be done
index     - index changes for merging to main zebra index with commit
commit    - commit index changes
fastindex - update zebra index without shadowing
            (fast but database is not accessible until finished)
fullindex - update zebra index and commit changes
start     - start zebra server
stop      - stop zebra server
restart   - restart zebra server
help      - this screen

EOF
	;;
    esac
done