#!/bin/sh -eu
# digest-checker -- See if a digest has any bad message separators etc.
# Copyright       : http://www.fsf.org/copyleft/gpl.html
# Author          : Dan Jacobson -- http://jidanni.org/comp/mail/
# Created On      : Mon May 31 01:28:34 2004
# Last Modified By: 
# Last Modified On: Sat Jan 15 14:44:53 2011
# Update Count    : 83

# Pipe a potential e.g. RISKS digest through this program to find bad
# separators etc. before the the issue hits the newsstands.  Of recent
# RISKS issues at the time of this writing, 23.34, 23.39 had separator
# discrepancies. The RISKS moderator enjoyed this program.

# A bad separator could result in readers missing an article,
# especially if it is tacked at the back of a boring article, as it
# often won't appear in their reader program's table of contents.

t=/tmp/risks.$$ exitval=0
cat > $t
w=-mwarnings=FATAL,all
echo -n "`basename $0`: checking: "
perl $w -nwe 's/^Subject: // && print && exit' $t
echo Probable separators:
perl $w -nwe 'next if /^-{70}$/;#ought to check this too though
  print if /-{20}/' $t|sort|tee $t.1|uniq -c|cat -e
echo
tr \\n @ <$t| perl $w -plwe 's/\@\@-{30}\@\@/\n_OK_SEPARATOR_\n/g'|
perl $w -nwe 'print if/^_OK_SEPARATOR_$/'>$t.2
set `wc -l $t.[12]`
probable=$1 ok=$3
echo $probable probable, and $ok OK separators.

# Also check for rarely intentional shameful trailing whitespace,
# visible to me thanks to emacs' show-trailing-whitespace, and now
# letting you know too.
# Indeed, cleaning up trailing whitespace will often also cure one's invalid
# separators, if any.

#whitespace_lines=`perl $w -nlwe 'print $. if /\s$/' $t`

if test $probable -ne $ok
    then
    echo "********** $probable PROBABLE != $ok OK !!!"
    echo OK separators are '\n\n-{30}\n\n', no more, no less.
    exitval=11
fi

set +e
perl $w -nlwe '
BEGIN { $w = $l = ""; $maxchars = 80; $e = 0 };
$w .= " $." if /\s$/;
$l .= " $." if ( length > $maxchars );
END {
    if ( length $w ) { print "\nTrailing whitespace on lines:$w"; $e++ }
    if ( length $l ) {
        print "\nMore than $maxchars chars on lines:$l";
        $e++;
    }    #Checking long lines too.
         #Of course we often thus produce more long lines :-)
    exit $e;
}
' $t
exitval=`expr $exitval + $?`
set -e
rm $t $t.1 $t.2
exit $exitval

# I didn't check if there were other programs that did the same.
# Could be extended to check for other bad things.

:||<<EOF
Using this splitdigest.config:
Subject: Risks Digest %d.%d
	End of RISKS-FORUM Digest %d.%d
	risks.%d.%d
	70
	30
wasn't as picky as my program above. I use gnus, which is picky,
whereas splitdigest split the messages that would have stuck together in gnus.
EOF