Annotation of libwww/Robot/src/robot.sh, revision 1.1
1.1 ! frystyk 1: #!/bin/sh
! 2: if [ $# -lt 3 ]
! 3: then
! 4: echo
! 5: echo "Usage: $0 RootURI ImageRootURI LogPrefix [ flags ]"
! 6: echo
! 7: echo "where"
! 8: echo " RootURI is the URI prefix for links, for example http://www.w3.org"
! 9: echo " ImageRootURI is the URI prefix for inlined images, for example http://www.w3.org"
! 10: echo " LogPrefix is the prefix for log files, for example w3c"
! 11: echo " flags are any additional command line flags, for example -vp"
! 12: echo
! 13: echo "See"
! 14: echo " http://www.w3.org/Robot/"
! 15: echo
! 16: echo "for more information"
! 17: exit 1
! 18: fi
! 19:
! 20: ROOT=$1
! 21: IMGROOT=$2
! 22: LOG=$3
! 23: FLAGS=$4
! 24:
! 25: ROBOT=./webbot
! 26:
! 27: ${ROBOT} ${FLAGS} -ss -q -n -depth 1 \
! 28: -exclude "/ArchiveBrowser/|/History/|/member/|/team/" \
! 29: -check "\.gz$|\.Z$|\.zip$|\.ZIP$|\.tar$|\.ps$|\.doc$|\.pdf$|\.xplot$|\.java$|\.c$|\.h$|\.txt$|\.ppt$|\.gif$|\.GIF$|\.tiff$|\.png$|\.PNG$|\.jpeg$|\.jpg$|\.JPE$" \
! 30: -prefix ${ROOT} \
! 31: -img -imgprefix ${IMGROOT} \
! 32: -l ${LOG}-log-clf.txt \
! 33: -alt ${LOG}-log-alt.txt \
! 34: -hit ${LOG}-log-hit.txt \
! 35: -lm ${LOG}-log-lastmodified.txt \
! 36: -title ${LOG}-log-title.txt \
! 37: -referer ${LOG}-log-referer.txt \
! 38: -negotiated ${LOG}-log-negotiated.txt \
! 39: -404 ${LOG}-log-notfound.txt \
! 40: -reject ${LOG}-log-reject.txt \
! 41: -format ${LOG}-log-format.txt \
! 42: -charset ${LOG}-log-charset.txt \
! 43: ${ROOT}
Webmaster