masses/rule-qa/corpus-nightly

#!/bin/sh

# settings are located in $HOME/.corpus

echo "Using corpus-nightly settings from $HOME/.corpus at"
date
. /etc/profile
. $HOME/.corpus

# use $PERL from env if unset
[ "x${PERL:-}" = x ] && PERL=perl
export PERL

set -x

renice -n +10 -p $$

set -e          # after renice, which typically fails

# ensure we're running after 0900 UTC; complex due to daylight savings.
# If we're running in the 08xx UTC hour-space, sleep for 1 hour; when
# we wake, we will be in 09xx.
(
  TIME="%e,%U,%S"; TZ=UTC; export TIME TZ
  if date | egrep '^... ... .. 08:'; then
    date; echo "sleeping for 1 hour to compensate for DST"
    sleep 3600 ; date 
  fi
)

# day of week
net=""
if date | egrep '^Sat' || [ "$1" = "net" ]; then net="net-" ; fi

# enter tree
if [ -d $tree ] ; then 
  cd $tree
else
  echo "$tree does not exist yet. you need to check it out first" 1>&2
  exit 2
fi

# find current revision
rm -f weekly-versions.txt nightly-versions.txt
if ! wget --tries=120 --wait=30 \
    http://rsync.spamassassin.org/weekly-versions.txt \
    http://rsync.spamassassin.org/nightly-versions.txt
then
	echo "wget failed" 1>&2
	exit 0
fi

if [ -n "$net" ]; then
	revision=$(tail -1 weekly-versions.txt|awk '{print $2}')
else
	revision=$(tail -1 nightly-versions.txt|awk '{print $2}')
fi

# update
# TODO: replace all this with rsync, it's much safer ;)
set +e
retry=0
while true; do
        rm -rf rules    # causing trouble for some reason
        rm -rf rulesrc/sandbox

	if svn update -r $revision < /dev/null && \
            svn update -r $revision rulesrc < /dev/null
        then
		break;
	fi

        # if that failed, get aggressive about getting a full
        # checkout, regardless of what may have been here already;
        # delete most of the checkout to do so, and just use "svn co"
        # instead of "svn up".

        svnpath=`svn info .| grep -i ^URL: | cut -d ' ' -f 2`

        rm -rf rules lib build rulesrc spamd spamc
        svn cleanup < /dev/null
        svn co --non-interactive $svnpath .

        # just 10 times; our SVN repo shouldn't be *that* unreliable!
	if [ $retry -eq 20 ]; then
		echo "svn update/co failed" 1>&2
		exit 1
	fi

	retry=$(( $retry + 1 ))

        # longer and longer between retries
	sleep `expr 20 '*' $retry`
done
set -e

# once we have everything checked out, run "part 2"
exec bash $tree/masses/rule-qa/corpus-nightly.post-svn