rtf.sum

harvest是一个下载html网页得机器人

#!/bin/sh

# RTF.sum
#
# D. Wessels 3/13/95

rbow_tmp=$TMPDIR/rbow$$

# Note, we have to do bizarre redirections because
# rbmaker prints a heading to stdout which confuses
# essence.  We could redirect everything from
# rbmaker to /dev/null, but we do want to see important
# errors.

# Ack.  This is because some Mac software seems to export RTF with CR's only
# and no LF's!!  So we replace a \r not followed by a \n with a \n.

rtf_tmp=$TMPDIR/rtf$$
perl -pe 's/\r([^\n])/\n$1/g;' < $1 > $rtf_tmp

# Convert the RTF into Rainbow compliant SGML
#
$HARVEST_HOME/lib/gatherer/rbmaker \
  -in $rtf_tmp \
  -out $rbow_tmp \
  -figdir  $TMPDIR \
  -datadir $HARVEST_HOME/lib/gatherer/rbmaker.dat \
  -tempdir $TMPDIR \
  2>&1 1>/dev/null  | grep -v '% completed' 1>&2


# cleanup garbage
#
rm -f graph???.eqn graf????.pct

# Run the SGML summarizer with the Rainbow DTD
#
SGML.sum Rainbow $rbow_tmp && rm -f $rbow_tmp $rtf_tmp