1 # This module is part of debbugs, and is released
2 # under the terms of the GPL version 2, or any later
3 # version at your option.
4 # See the file README and COPYING for more information.
6 # [Other people have contributed to this file; their copyrights should
8 # Copyright 2006 by Don Armstrong <don@donarmstrong.com>.
11 package Debbugs::MIME;
15 Debbugs::MIME -- Mime handling routines for debbugs
19 use Debbugs::MIME qw(parse decode_rfc1522);
33 use base qw(Exporter);
34 use vars qw($DEBUG $VERSION @EXPORT_OK %EXPORT_TAGS @EXPORT);
38 $DEBUG = 0 unless defined $DEBUG;
42 %EXPORT_TAGS = (mime => [qw(parse create_mime_message getmailbody)],
43 rfc1522 => [qw(decode_rfc1522 encode_rfc1522)],
46 Exporter::export_ok_tags(keys %EXPORT_TAGS);
47 $EXPORT_TAGS{all} = [@EXPORT_OK];
54 use POSIX qw(strftime);
55 use List::MoreUtils qw(apply);
58 use Debbugs::UTF8 qw(convert_to_utf8 encode_utf8_safely);
61 use MIME::WordDecoder qw();
62 use Encode qw(decode encode encode_utf8 decode_utf8 is_utf8);
70 my $type = $entity->effective_type;
71 if ($type eq 'text/plain' or
72 ($type =~ m#text/?# and $type ne 'text/html') or
73 $type eq 'application/pgp') {
75 } elsif ($type eq 'multipart/alternative') {
76 # RFC 2046 says we should use the last part we recognize.
77 for my $part (reverse $entity->parts) {
78 my $ret = getmailbody($part);
82 # For other multipart types, we just pretend they're
83 # multipart/mixed and run through in order.
84 for my $part ($entity->parts) {
85 my $ret = getmailbody($part);
94 # header and decoded body respectively
95 my (@headerlines, @bodylines);
97 my $parser = MIME::Parser->new();
98 my $tempdir = File::Temp::tempdir();
99 $parser->output_under($tempdir);
100 my $entity = eval { $parser->parse_data($_[0]) };
102 if ($entity and $entity->head->tags) {
103 @headerlines = @{$entity->head->header};
106 my $entity_body = getmailbody($entity);
107 my $entity_body_handle;
109 if (defined $entity_body) {
110 $entity_body_handle = $entity_body->bodyhandle();
111 $charset = $entity_body->head()->mime_attr('content-type.charset');
113 @bodylines = $entity_body_handle ? $entity_body_handle->as_lines() : ();
114 @bodylines = map {convert_to_utf8($_,$charset)} @bodylines;
117 # Legacy pre-MIME code, kept around in case MIME::Parser fails.
118 my @msg = split /\n/, $_[0];
121 for ($i = 0; $i <= $#msg; ++$i) {
124 while ($msg[$i + 1] =~ /^\s/) {
126 $_ .= "\n" . $msg[$i];
128 push @headerlines, $_;
131 @bodylines = @msg[$i .. $#msg];
134 rmtree $tempdir, 0, 1;
136 # Remove blank lines.
137 shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
139 # Strip off RFC2440-style PGP clearsigning.
140 if (@bodylines and $bodylines[0] =~ /^-----BEGIN PGP SIGNED/) {
141 shift @bodylines while @bodylines and
142 length $bodylines[0] and
143 # we currently don't strip \r; handle this for the
144 # time being, though eventually it should be stripped
145 # too, I think. [See #565981]
146 $bodylines[0] ne "\r";
147 shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
148 for my $findsig (0 .. $#bodylines) {
149 if ($bodylines[$findsig] =~ /^-----BEGIN PGP SIGNATURE/) {
150 $#bodylines = $findsig - 1;
154 map { s/^- // } @bodylines;
157 return { header => [@headerlines], body => [@bodylines]};
160 =head2 create_mime_message
162 create_mime_message([To=>'don@debian.org'],$body,[$attach1, $attach2],$include_date);
164 Creates a MIME encoded message with headers given by the first
165 argument, and a message given by the second.
167 Optional attachments can be specified in the third arrayref argument.
169 Whether to include the date in the header is the final argument; it
170 defaults to true, setting the Date header if one is not already
173 Headers are passed directly to MIME::Entity::build, the message is the
176 Each of the elements of the attachment arrayref is attached as an
177 rfc822 message if it is a scalar or an arrayref; otherwise if it is a
178 hashref, the contents are passed as an argument to
183 sub create_mime_message{
184 my ($headers,$body,$attachments,$include_date) = @_;
185 $attachments = [] if not defined $attachments;
186 $include_date = 1 if not defined $include_date;
188 die "The first argument to create_mime_message must be an arrayref" unless ref($headers) eq 'ARRAY';
189 die "The third argument to create_mime_message must be an arrayref" unless ref($attachments) eq 'ARRAY';
192 my %headers = apply {lc($_)} @{$headers};
193 if (not exists $headers{date}) {
196 strftime("%a, %d %b %Y %H:%M:%S +0000",gmtime)
202 # MIME::Entity is stupid, and doesn't rfc1522 encode its headers, so we do it for it.
203 my $msg = MIME::Entity->build('Content-Type' => 'text/plain; charset=utf-8',
204 'Encoding' => 'quoted-printable',
205 (map{encode_rfc1522(encode_utf8_safely($_))} @{$headers}),
206 Data => encode_utf8_safely($body),
209 # Attach the attachments
210 for my $attachment (@{$attachments}) {
211 if (ref($attachment) eq 'HASH') {
212 $msg->attach(%{$attachment});
215 # This is *craptacular*, but because various MTAs
216 # (sendmail and exim4, at least) appear to eat From
217 # lines in message/rfc822 attachments, we need eat
218 # the entire From line ourselves so the MTA doesn't
219 # leave \n detrius around.
220 if (ref($attachment) eq 'ARRAY' and $attachment->[1] =~ /^From /) {
221 # make a copy so that we don't screw up anything
222 # that is expecting this arrayref to stay constant
223 $attachment = [@{$attachment}];
224 # remove the from line
225 splice @$attachment, 1, 1;
227 elsif (not ref($attachment)) {
228 # It's a scalar; remove the from line
229 $attachment =~ s/^(Received:[^\n]+\n)(From [^\n]+\n)/$1/s;
231 $msg->attach(Type => 'message/rfc822',
237 return $msg->as_string;
243 =head2 decode_rfc1522
245 decode_rfc1522('=?iso-8859-1?Q?D=F6n_Armstr=F3ng?= <don@donarmstrong.com>')
247 Turn RFC-1522 names into the UTF-8 equivalent.
252 # Set up the default RFC1522 decoder, which turns all charsets that
253 # are supported into the appropriate UTF-8 charset.
254 MIME::WordDecoder->default(new MIME::WordDecoder(
255 ['*' => \&convert_to_utf8,
262 # this is craptacular, but leading space is hacked off by unmime.
264 my $leading_space = '';
265 $leading_space = $1 if $string =~ s/^(\s+)//;
266 # unmime calls the default MIME::WordDecoder handler set up at
267 # initialization time.
268 return $leading_space . MIME::WordDecoder::unmime($string);
271 =head2 encode_rfc1522
273 encode_rfc1522('Dön Armströng <don@donarmstrong.com>')
275 Encodes headers according to the RFC1522 standard by calling
276 MIME::Words::encode_mimeword on distinct words as appropriate.
280 # We cannot use MIME::Words::encode_mimewords because that function
281 # does not handle spaces properly at all.
286 # handle being passed undef properly
287 return undef if not defined $rawstr;
288 # convert to octets if is_utf8 is set
289 $rawstr= encode_utf8_safely($rawstr);
290 # We process words in reverse so we can preserve spacing between
291 # encoded words. This regex splits on word|nonword boundaries and
292 # nonword|nonword boundaries. We also consider parenthesis and "
293 # to be nonwords to avoid escaping them in comments in violation
295 my @words = reverse split /(?:(?<=[\s\n\)\(\"])|(?=[\s\n\)\(\"]))/m, $rawstr;
297 my $previous_word_encoded = 0;
299 for my $word (@words) {
300 if ($word !~ m#[\x00-\x1F\x7F-\xFF]#o and $word ne ' ') {
301 $string = $word.$string;
302 $previous_word_encoded=0;
304 elsif ($word =~ /^[\s\n]$/) {
305 $string = $word.$string;
306 $previous_word_encoded = 0 if $word eq "\n";
309 my $encoded = MIME::Words::encode_mimeword($word, 'q', 'UTF-8');
310 # RFC 1522 mandates that segments be at most 76 characters
311 # long. If that's the case, we split the word up into 10
312 # character pieces and encode it. We must use the Encode
313 # magic here to avoid breaking on bit boundaries here.
314 if (length $encoded > 75) {
315 # Turn utf8 into the internal perl representation
316 # so . is a character, not a byte.
317 my $tempstr = is_utf8($word)?$word:decode_utf8($word,Encode::FB_DEFAULT);
319 # Strip it into 10 character long segments, and encode
321 # XXX It's possible that these segments are > 76 characters
322 while ($tempstr =~ s/(.{1,10})$//) {
323 # turn the character back into the utf8 representation.
324 my $tempword = encode_utf8($1);
325 # It may actually be better to eventually use
326 # the base64 encoding here, but I'm not sure
327 # if that's as widely supported as quoted
329 unshift @encoded, MIME::Words::encode_mimeword($tempword,'q','UTF-8');
331 $encoded = join(" ",@encoded);
332 # If the previous word was encoded, we must
333 # include a trailing _ that gets encoded as a
335 $encoded =~ s/\?\=$/_\?\=/ if $previous_word_encoded;
336 $string = $encoded.$string;
339 # If the previous word was encoded, we must
340 # include a trailing _ that gets encoded as a
342 $encoded =~ s/\?\=$/_\?\=/ if $previous_word_encoded;
343 $string = $encoded.$string;
345 $previous_word_encoded = 1;