my $entity = shift;
my $type = $entity->effective_type;
if ($type eq 'text/plain' or
- ($type =~ m#text/# and $type ne 'text/html') or
+ ($type =~ m#text/?# and $type ne 'text/html') or
$type eq 'application/pgp') {
return $entity->bodyhandle;
} elsif ($type eq 'multipart/alternative') {
# Strip off RFC2440-style PGP clearsigning.
if (@bodylines and $bodylines[0] =~ /^-----BEGIN PGP SIGNED/) {
- shift @bodylines while @bodylines and length $bodylines[0];
+ shift @bodylines while @bodylines and
+ length $bodylines[0] and
+ # we currently don't strip \r; handle this for the
+ # time being, though eventually it should be stripped
+ # too, I think. [See #565981]
+ $bodylines[0] ne "\r";
shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
for my $findsig (0 .. $#bodylines) {
if ($bodylines[$findsig] =~ /^-----BEGIN PGP SIGNATURE/) {
# handle being passed undef properly
return undef if not defined $rawstr;
+ if (is_utf8($rawstr)) {
+ $rawstr= encode_utf8($rawstr);
+ }
# We process words in reverse so we can preserve spacing between
# encoded words. This regex splits on word|nonword boundaries and
- # nonword|nonword boundaries.
- my @words = reverse split /(?:(?<=[\s\n])|(?=[\s\n]))/m, $rawstr;
+ # nonword|nonword boundaries. We also consider parenthesis and "
+ # to be nonwords to avoid escaping them in comments in violation
+ # of RFC1522
+ my @words = reverse split /(?:(?<=[\s\n\)\(\"])|(?=[\s\n\)\(\"]))/m, $rawstr;
my $previous_word_encoded = 0;
my $string = '';