#!/bin/sh
#
#       @(#)procmail_mindspring 1999/07/14 Anne Bennett
#
# Download Mindspring's list of bad senders for blocking spam,
# reformat it for our site, and compare it to what
# we've got.  Report if different, and prepare new file.
#
# Last modified: 2000/02/18 Anne Bennett
#                  (removed entries with no "@" sign or ".")

PATH=/local/paths:/usr/bin:/sbin:/usr/sbin
export PATH


thisprog=procmail_mindspring
logger="logger -p mail.info ${thisprog}:"

url="http://cgi.mindspring.com/cgi-bin/spamlist.pl"
cur_rc=/local/etc/procmail/spam/tag-mindspring
new_rc=/local/etc/procmail/spam/NEW.tag-mindspring
mailto="anne"

tmp_new=/tmp/${thisprog}.$$.tmp_new
tmp_old=/tmp/${thisprog}.$$.tmp_old
tmp_diff=/tmp/${thisprog}.$$.tmp_diff
tmp_raw=/tmp/${thisprog}.$$.tmp_raw

all_tmp="${tmp_new} ${tmp_old} ${tmp_diff} ${tmp_raw}"


# Start.
#
${logger} starting.


# Download Mindspring material; note the date and time.
#
lynx -dump -source ${url} > ${tmp_raw}
down_date=`date '+%Y/%m/%d %H:%M:%S %Z'`

# If we got nothing, stop now.
#
if [ ! -s ${tmp_raw} ]
then
  echo "Empty or failed download" | \
    mailx -s "New Mindspring procmail data FAILURE" ${mailto}
  rm -f ${all_tmp}
  ${logger} end of run, failed.
  exit
fi

# Format the material.
#
touch ${tmp_new}

# (Pick out envelope senders.)
< ${tmp_raw} sed -n '/<h3>Envelope Senders<\/h3>/,/<\/pre>/p' \
   | sed -e '/^</d'  \
         -e '/^[        ]*$/d' \
         -e 's%\.%\\.%g' \
         -e 's%\[%\\[%g' \
         -e 's%\]%\\]%g' \
   | sort -u \
   | egrep '@|\.' \
   | gawk '{printf(":0 Hf\n* ^Received: from\\<%s\\>\n| formail -b -f -A \"$trash_header siteban per-mindspring envelope sender\"\n\n",$1)}' \
   >> ${tmp_new}

# (Pick out header senders.)
< ${tmp_raw} sed -n '/<h3>Header Senders<\/h3>/,/<\/pre>/p' \
   | sed -e '/^</d'  \
         -e '/^[        ]*$/d' \
         -e 's%\.%\\.%g' \
         -e 's%\[%\\[%g' \
         -e 's%\]%\\]%g' \
   | sort -u \
   | egrep '@|\.' \
   | gawk '{printf(":0 Hf\n* ^From:.*\\<%s\\>\n| formail -b -f -A \"$trash_header siteban per-mindspring header sender\"\n\n",$1)}' \
   >> ${tmp_new}

# If the result is empty, report the problem and stop.
#
if [ ! -s ${tmp_new} ]
then
  echo "Empty after formatting" | \
    mailx -s "New Mindspring procmail data FAILURE" ${mailto}
  rm -f ${all_tmp}
  ${logger} end of run, failed.
  exit
fi


# Isolate our current MindSpring-derived material.
#
cat ${cur_rc} \
   | sed -n '/^# ========== MindSpring-derived material below ==========$/,$p' \
   | tail +2 \
   >> ${tmp_old}


# Are they different?  If not, clean up and exit quietly.
#
diff ${tmp_old} ${tmp_new} > ${tmp_diff}
if [ ! -s ${tmp_diff} ]; then
   # No changes.
   ${logger} no change.
   rm -f ${all_tmp}
   ${logger} end of run.
   exit
fi


# There were changes!  Prepare a new file and notify maintainer.
#
${logger} changes found.
cat ${cur_rc} \
   | sed -n '1,/^# ========== MindSpring-derived material below ==========$/p' \
   | sed -e "s%^# Last download on .*%# Last download on ${down_date}%" \
   > ${new_rc}
cat ${tmp_new} >> ${new_rc}

${logger} notifying maintainer.
( echo ""
  echo "New Mindspring procmail data is in ${new_rc}"
  echo ""
  echo "Differences follow:"
  echo ""
  echo "===================================="
  echo ""
  cat ${tmp_diff}
) | mailx -s "New Mindspring procmail data ready" ${mailto}


# Clean up and leave
rm -f ${all_tmp}
${logger} end of run.
exit