+ # this is craptacular, but leading space is hacked off by unmime.
+ # Save it.
+ my $leading_space = '';
+ $leading_space = $1 if $string =~ s/^(\ +)//;
+ # we must do this to switch off the utf8 flag before calling decode_mimewords
+ $string = encode_utf8($string);
+ my @mime_words = MIME::Words::decode_mimewords($string);
+ my $tmp = $leading_space .
+ join('',
+ (map {
+ if (@{$_} > 1) {
+ convert_to_utf8(${$_}[0],${$_}[1]);
+ } else {
+ decode_utf8(${$_}[0]);
+ }
+ } @mime_words)
+ );
+ return $tmp;
+}
+
+=head2 encode_rfc1522
+
+ encode_rfc1522('Dön Armströng <don@donarmstrong.com>')
+
+Encodes headers according to the RFC1522 standard by calling
+MIME::Words::encode_mimeword on distinct words as appropriate.
+
+=cut
+
+# We cannot use MIME::Words::encode_mimewords because that function
+# does not handle spaces properly at all.
+
+sub encode_rfc1522 {
+ my ($rawstr) = @_;
+
+ # handle being passed undef properly
+ return undef if not defined $rawstr;
+
+ # convert to octets if we are given a string in perl's internal
+ # encoding
+ $rawstr= encode_utf8($rawstr) if is_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. 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 = '';
+ for my $word (@words) {
+ if ($word !~ m#[\x00-\x1F\x7F-\xFF]#o and $word ne ' ') {
+ $string = $word.$string;
+ $previous_word_encoded=0;
+ }
+ elsif ($word =~ /^[\s\n]$/) {
+ $string = $word.$string;
+ $previous_word_encoded = 0 if $word eq "\n";
+ }
+ else {
+ my $encoded = MIME::Words::encode_mimeword($word, 'q', 'UTF-8');
+ # RFC 1522 mandates that segments be at most 76 characters
+ # long. If that's the case, we split the word up into 10
+ # character pieces and encode it. We must use the Encode
+ # magic here to avoid breaking on bit boundaries here.
+ if (length $encoded > 75) {
+ # Turn utf8 into the internal perl representation
+ # so . is a character, not a byte.
+ my $tempstr = is_utf8($word)?$word:decode_utf8($word,Encode::FB_DEFAULT);
+ my @encoded;
+ # Strip it into 10 character long segments, and encode
+ # the segments
+ # XXX It's possible that these segments are > 76 characters
+ while ($tempstr =~ s/(.{1,10})$//) {
+ # turn the character back into the utf8 representation.
+ my $tempword = encode_utf8($1);
+ # It may actually be better to eventually use
+ # the base64 encoding here, but I'm not sure
+ # if that's as widely supported as quoted
+ # printable.
+ unshift @encoded, MIME::Words::encode_mimeword($tempword,'q','UTF-8');
+ }
+ $encoded = join(" ",@encoded);
+ # If the previous word was encoded, we must
+ # include a trailing _ that gets encoded as a
+ # space.
+ $encoded =~ s/\?\=$/_\?\=/ if $previous_word_encoded;
+ $string = $encoded.$string;
+ }
+ else {
+ # If the previous word was encoded, we must
+ # include a trailing _ that gets encoded as a
+ # space.
+ $encoded =~ s/\?\=$/_\?\=/ if $previous_word_encoded;
+ $string = $encoded.$string;
+ }
+ $previous_word_encoded = 1;
+ }
+ }
+ return $string;