harvest

harvest是一个下载html网页得机器人

#!/bin/sh
#
# /etc/init.d/harvest - start/stop the httpd-related daemons
#

# First, set up some useful variables.

HOME=/; export HOME
PATH=/bin:/usr/ucb:/usr/sbin:/etc:/sbin; export PATH

umask 022

# the paths and parameters for things we run.

R_GATHERD=/opt/harvest/gatherers/vsis-www/RunGatherd
R_BROKER=/opt/harvest/brokers/vsis-www/RunBroker
BROKER_OPTS=-nocol		# Don't redo collection on startup

### uncomment these if there seems to be a problem starting up.
# RUNLOG_G=/usr/tmp/www-run-log.gatherd
# RUNLOG_B=/usr/tmp/www-run-log.broker
RUNLOG_G=/dev/null
RUNLOG_B=/dev/null

case $1 in
'start')

   echo "Starting Harvest daemons:" #>/dev/console

   if [ -f ${R_GATHERD} ]; then
       echo " gatherd...\c"	#>/dev/console
#       /bin/su - wwwharv -c ${R_GATHERD} >${RUNLOG_G} 2>&1
       ${R_GATHERD} >${RUNLOG_G} 2>&1
   fi

   sleep 2

   if [ -f ${R_BROKER} ]; then
       echo " broker...\c"	#>/dev/console
#       /bin/su - wwwharv -c "${R_BROKER} ${BROKER_OPTS}" >${RUNLOG_B} 2>&1
       ${R_BROKER} ${BROKER_OPTS} >${RUNLOG_B} 2>&1
   fi

   echo " done!" #>/dev/console
   ;;
'stop')
   echo "Stopping Harvest daemons." #>/dev/console

	pid=`/bin/ps -x | /usr/bin/awk -- '($5~/^gatherd/) {print $1}'`
	if test "$pid"
	then
		kill $pid
	fi
	pid=`/bin/ps -x | /usr/bin/awk -- '($5~/^broker/) {print $1}'`
	if test "$pid"
	then
		kill $pid
	fi
	;;
*)
	echo "usage: /etc/init.d/harvest {start|stop}"
	;;
esac