#!/bin/sh

# 50 * * * * /home/jm/ftp/spamassassin/build/gen_freqs_from_nightly_mc

outdir=/home/jm/spamassassin.taint.org/qa/freqs
cd /home/jm/ftp/spamassassin/masses

(
echo "STARTING AT"; date

# you need to create this file with RSYNC_USERNAME and RSYNC_PASSWORD
# settings for your account in it.

. nightly_mc_passwords
export RSYNC_PASSWORD

mkdir all-results >/dev/null 2>&1
rsync -vrz $RSYNC_USERNAME@belphegore.hughes-family.org::corpus/. all-results


# now grep out the ones built today, and put copies into "fresh".
# copy both today and y'day into "mixed"
today=`date +"%b %e"`   # "Oct 17"
yday=`date +"%b %e" --date='yesterday'`   # "Oct 16"

rm -rf fresh; mkdir fresh >/dev/null 2>&1
rm -rf mixed; mkdir mixed >/dev/null 2>&1
rm -rf freshstamps; mkdir freshstamps >/dev/null 2>&1
rm -rf mixedstamps; mkdir mixedstamps >/dev/null 2>&1
for f in all-results/*.log ; do
  if grep "mass-check results from" $f | egrep "$today" > dates ; then
    cp $f fresh
    cp $f mixed
    sort dates | tail -1 > tmp/x
    cat tmp/x > freshstamps/`basename $f`
    cat tmp/x > mixedstamps/`basename $f`
  elif grep "mass-check results from" $f | egrep "$yday" > dates ; then
    cp $f mixed
    sort dates | tail -1 > mixedstamps/`basename $f`
  fi
done



cat fresh/nonspam-*.log > nonspam.log
cat fresh/spam-*.log > spam.log
make clean ; make freqs
echo "
Last update: `date`
Nonspam results used: `echo fresh/nonspam*`
Spam results used: `echo fresh/spam*`

`cat freqs`

Timestamps from mass-check output used:

`egrep . freshstamps/*`
" > $outdir/DETAILS.today
cp freqs $outdir/freqs.today



cat mixed/nonspam-*.log > nonspam.log
cat mixed/spam-*.log > spam.log
make clean ; make freqs
echo "
Last update: `date`
Nonspam results used: `echo mixed/nonspam*`
Spam results used: `echo mixed/spam*`

`cat freqs`

Timestamps from mass-check output used:

`egrep . mixedstamps/*`
" > $outdir/DETAILS.mixed
cp freqs $outdir/freqs.mixed



echo $outdir; ls -l $outdir
echo "FINISHED AT"; date

) > nightly.log 2>&1