6 use vars qw($VERSION @EXPORT_OK);
11 @EXPORT_OK = qw(parse decode_rfc1522 encode_rfc1522 convert_to_utf8 create_mime_message);
18 use MIME::WordDecoder qw();
19 use Encode qw(decode encode encode_utf8 decode_utf8 is_utf8);
28 my $type = $entity->effective_type;
29 if ($type eq 'text/plain' or
30 ($type =~ m#text/# and $type ne 'text/html') or
31 $type eq 'application/pgp') {
32 return $entity->bodyhandle;
33 } elsif ($type eq 'multipart/alternative') {
34 # RFC 2046 says we should use the last part we recognize.
35 for my $part (reverse $entity->parts) {
36 my $ret = getmailbody($part);
40 # For other multipart types, we just pretend they're
41 # multipart/mixed and run through in order.
42 for my $part ($entity->parts) {
43 my $ret = getmailbody($part);
52 # header and decoded body respectively
53 my (@headerlines, @bodylines);
55 my $parser = new MIME::Parser;
56 mkdir "mime.tmp.$$", 0777;
57 $parser->output_under("mime.tmp.$$");
58 my $entity = eval { $parser->parse_data($_[0]) };
60 if ($entity and $entity->head->tags) {
61 @headerlines = @{$entity->head->header};
64 my $entity_body = getmailbody($entity);
65 @bodylines = $entity_body ? $entity_body->as_lines() : ();
68 # Legacy pre-MIME code, kept around in case MIME::Parser fails.
69 my @msg = split /\n/, $_[0];
72 for ($i = 0; $i <= $#msg; ++$i) {
75 while ($msg[$i + 1] =~ /^\s/) {
77 $_ .= "\n" . $msg[$i];
79 push @headerlines, $_;
82 @bodylines = @msg[$i .. $#msg];
85 rmtree "mime.tmp.$$", 0, 1;
88 shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
90 # Strip off RFC2440-style PGP clearsigning.
91 if (@bodylines and $bodylines[0] =~ /^-----BEGIN PGP SIGNED/) {
92 shift @bodylines while @bodylines and length $bodylines[0];
93 shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
94 for my $findsig (0 .. $#bodylines) {
95 if ($bodylines[$findsig] =~ /^-----BEGIN PGP SIGNATURE/) {
96 $#bodylines = $findsig - 1;
100 map { s/^- // } @bodylines;
103 return { header => [@headerlines], body => [@bodylines]};
106 =head2 create_mime_message
108 create_mime_message([To=>'don@debian.org'],$body,[$attach1, $attach2]);
110 Creates a MIME encoded message with headers given by the first
111 argument, and a message given by the second.
113 Optional attachments can be specified in the third arrayref argument.
115 Headers are passed directly to MIME::Entity::build, the message is the
118 Each of the elements of the attachment arrayref is attached as an
119 rfc822 message if it is a scalar or an arrayref; otherwise if it is a
120 hashref, the contents are passed as an argument to
125 sub create_mime_message{
126 my ($headers,$body,$attachments) = @_;
127 $attachments = [] if not defined $attachments;
129 die "The first argument to create_mime_message must be an arrayref" unless ref($headers) eq 'ARRAY';
130 die "The third argument to create_mime_message must be an arrayref" unless ref($attachments) eq 'ARRAY';
133 my $msg = MIME::Entity->build(@{$headers},
137 # Attach the attachments
138 for my $attachment (@{$attachments}) {
139 if (ref($attachment) eq 'HASH') {
140 $msg->attach(%{$attachment});
143 # This is *craptacular*, but because various MTAs
144 # (sendmail and exim4, at least) appear to eat From
145 # lines in message/rfc822 attachments, we need to make
146 # sure the From line is the first thing in the
147 # attachement, not the second, so if it gets eaten, the
148 # headers don't collide into the body.
149 if (ref($attachment) eq 'ARRAY' and $attachment->[1] =~ /^From /) {
150 # make a copy so that we don't screw up anything
151 # that is expecting this arrayref to stay constant
152 $attachment = [@{$attachment}];
153 # remove the from line
154 splice @$attachment, 1, 1;
156 elsif (not ref($attachment)) {
157 # It's a scalar; remove the from line
158 $attachment =~ s/^(Received:[^\n]+\n)(From [^\n]+\n)/$1/s;
160 $msg->attach(Type => 'message/rfc822',
166 return $msg->as_string;
172 sub convert_to_utf8 {
173 my ($data, $charset) = @_;
174 # raw data just gets returned (that's the charset WordDecorder
175 # uses when it doesn't know what to do)
176 return $data if $charset eq 'raw' or is_utf8($data,1);
179 # this encode/decode madness is to make sure that the data
180 # really is valid utf8 and that the is_utf8 flag is off.
181 $result = encode("utf8",decode($charset,$data))
184 warn "Unable to decode charset; '$charset' and '$data': $@";
191 =head2 decode_rfc1522
193 decode_rfc1522('=?iso-8859-1?Q?D=F6n_Armstr=F3ng?= <don@donarmstrong.com>')
195 Turn RFC-1522 names into the UTF-8 equivalent.
200 # Set up the default RFC1522 decoder, which turns all charsets that
201 # are supported into the appropriate UTF-8 charset.
202 MIME::WordDecoder->default(new MIME::WordDecoder(
203 ['*' => \&convert_to_utf8,
207 sub decode_rfc1522 ($)
211 # unmime calls the default MIME::WordDecoder handler set up at
212 # initialization time.
213 return MIME::WordDecoder::unmime($string);
216 =head2 encode_rfc1522
218 encode_rfc1522('Dön Armströng <don@donarmstrong.com>')
220 Encodes headers according to the RFC1522 standard by calling
221 MIME::Words::encode_mimeword on distinct words as appropriate.
225 # We cannot use MIME::Words::encode_mimewords because that function
226 # does not handle spaces properly at all.
228 sub encode_rfc1522 ($) {
231 # We process words in reverse so we can preserve spacing between
232 # encoded words. This regex splits on word|nonword boundaries and
233 # nonword|nonword boundaries.
234 my @words = reverse split /(?:(?<=[\s\n])|(?=[\s\n]))/m, $rawstr;
236 my $previous_word_encoded = 0;
238 for my $word (@words) {
239 if ($word !~ m#[\x00-\x1F\x7F-\xFF]#o and $word ne ' ') {
240 $string = $word.$string;
241 $previous_word_encoded=0;
243 elsif ($word =~ /^[\s\n]$/) {
244 $string = $word.$string;
245 $previous_word_encoded = 0 if $word eq "\n";
248 my $encoded = MIME::Words::encode_mimeword($word, 'q', 'UTF-8');
249 # RFC 1522 mandates that segments be at most 76 characters
250 # long. If that's the case, we split the word up into 10
251 # character pieces and encode it. We must use the Encode
252 # magic here to avoid breaking on bit boundaries here.
253 if (length $encoded > 75) {
254 # Turn utf8 into the internal perl representation
255 # so . is a character, not a byte.
256 my $tempstr = decode_utf8($word,Encode::FB_DEFAULT);
258 # Strip it into 10 character long segments, and encode
260 # XXX It's possible that these segments are > 76 characters
261 while ($tempstr =~ s/(.{1,10})$//) {
262 # turn the character back into the utf8 representation.
263 my $tempword = encode_utf8($1);
264 # It may actually be better to eventually use
265 # the base64 encoding here, but I'm not sure
266 # if that's as widely supported as quoted
268 unshift @encoded, MIME::Words::encode_mimeword($tempword,'q','UTF-8');
270 $encoded = join(" ",@encoded);
271 # If the previous word was encoded, we must
272 # include a trailing _ that gets encoded as a
274 $encoded =~ s/\?\=$/_\?\=/ if $previous_word_encoded;
275 $string = $encoded.$string;
278 # If the previous word was encoded, we must
279 # include a trailing _ that gets encoded as a
281 $encoded =~ s/\?\=$/_\?\=/ if $previous_word_encoded;
282 $string = $encoded.$string;
284 $previous_word_encoded = 1;