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;
17 Debbugs::MIME -- Mime handling routines for debbugs
21 use Debbugs::MIME qw(parse decode_rfc1522);
35 use Exporter qw(import);
36 use vars qw($DEBUG $VERSION @EXPORT_OK %EXPORT_TAGS @EXPORT);
40 $DEBUG = 0 unless defined $DEBUG;
44 %EXPORT_TAGS = (mime => [qw(parse create_mime_message getmailbody)],
45 rfc1522 => [qw(decode_rfc1522 encode_rfc1522 handle_escaped_commas)],
48 Exporter::export_ok_tags(keys %EXPORT_TAGS);
49 $EXPORT_TAGS{all} = [@EXPORT_OK];
52 use File::Path qw(remove_tree);
53 use File::Temp qw(tempdir);
55 use Mail::Message::Field;
57 use POSIX qw(strftime);
58 use List::AllUtils qw(apply);
61 use Debbugs::UTF8 qw(convert_to_utf8);
63 # for decode_rfc1522 and encode_rfc1522
64 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 = tempdir(CLEANUP => 1);
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 # assume us-ascii unless charset is set; probably bad, but we
122 # really shouldn't get to this point anyway
123 my $charset = 'us-ascii';
124 for ($i = 0; $i <= $#msg; ++$i) {
127 while ($msg[$i + 1] =~ /^\s/) {
129 $_ .= "\n" . $msg[$i];
131 if (/charset=\"([^\"]+)\"/) {
134 push @headerlines, $_;
136 @bodylines = map {convert_to_utf8($_,$charset)} @msg[$i .. $#msg];
139 remove_tree($tempdir,{verbose => 0, safe => 1});
141 # Remove blank lines.
142 shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
144 # Strip off RFC2440-style PGP clearsigning.
145 if (@bodylines and $bodylines[0] =~ /^-----BEGIN PGP SIGNED/) {
146 shift @bodylines while @bodylines and
147 length $bodylines[0] and
148 # we currently don't strip \r; handle this for the
149 # time being, though eventually it should be stripped
150 # too, I think. [See #565981]
151 $bodylines[0] ne "\r";
152 shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
153 for my $findsig (0 .. $#bodylines) {
154 if ($bodylines[$findsig] =~ /^-----BEGIN PGP SIGNATURE/) {
155 $#bodylines = $findsig - 1;
159 map { s/^- // } @bodylines;
162 return { header => [@headerlines], body => [@bodylines]};
165 =head2 create_mime_message
167 create_mime_message([To=>'don@debian.org'],$body,[$attach1, $attach2],$include_date);
169 Creates a MIME encoded message with headers given by the first
170 argument, and a message given by the second.
172 Optional attachments can be specified in the third arrayref argument.
174 Whether to include the date in the header is the final argument; it
175 defaults to true, setting the Date header if one is not already
178 Headers are passed directly to MIME::Entity::build, the message is the
181 Each of the elements of the attachment arrayref is attached as an
182 rfc822 message if it is a scalar or an arrayref; otherwise if it is a
183 hashref, the contents are passed as an argument to
188 sub create_mime_message{
189 my ($headers,$body,$attachments,$include_date) = @_;
190 $attachments = [] if not defined $attachments;
191 $include_date = 1 if not defined $include_date;
193 die "The first argument to create_mime_message must be an arrayref" unless ref($headers) eq 'ARRAY';
194 die "The third argument to create_mime_message must be an arrayref" unless ref($attachments) eq 'ARRAY';
197 my %headers = apply {defined $_ ? lc($_) : ''} @{$headers};
198 if (not exists $headers{date}) {
201 strftime("%a, %d %b %Y %H:%M:%S +0000",gmtime)
207 # MIME::Entity is stupid, and doesn't rfc1522 encode its headers, so we do it for it.
208 my $msg = MIME::Entity->build('Content-Type' => 'text/plain; charset=utf-8',
209 'Encoding' => 'quoted-printable',
210 (map{encode_rfc1522(encode_utf8(defined $_ ? $_:''))} @{$headers}),
211 Data => encode_utf8($body),
214 # Attach the attachments
215 for my $attachment (@{$attachments}) {
216 if (ref($attachment) eq 'HASH') {
217 $msg->attach(%{$attachment});
220 # This is *craptacular*, but because various MTAs
221 # (sendmail and exim4, at least) appear to eat From
222 # lines in message/rfc822 attachments, we need eat
223 # the entire From line ourselves so the MTA doesn't
224 # leave \n detrius around.
225 if (ref($attachment) eq 'ARRAY' and $attachment->[1] =~ /^From /) {
226 # make a copy so that we don't screw up anything
227 # that is expecting this arrayref to stay constant
228 $attachment = [@{$attachment}];
229 # remove the from line
230 splice @$attachment, 1, 1;
232 elsif (not ref($attachment)) {
233 # It's a scalar; remove the from line
234 $attachment =~ s/^(Received:[^\n]+\n)(From [^\n]+\n)/$1/s;
236 $msg->attach(Type => 'message/rfc822',
242 return $msg->as_string;
248 =head2 decode_rfc1522
250 decode_rfc1522('=?iso-8859-1?Q?D=F6n_Armstr=F3ng?= <don@donarmstrong.com>')
252 Turn RFC-1522 names into the UTF-8 equivalent.
259 # this is craptacular, but leading space is hacked off by unmime.
261 my $leading_space = '';
262 $leading_space = $1 if $string =~ s/^(\ +)//;
263 # we must do this to switch off the utf8 flag before calling decode_mimewords
264 $string = encode_utf8($string);
265 my @mime_words = MIME::Words::decode_mimewords($string);
266 my $tmp = $leading_space .
270 convert_to_utf8(${$_}[0],${$_}[1]);
272 decode_utf8(${$_}[0]);
279 =head2 encode_rfc1522
281 encode_rfc1522('Dön Armströng <don@donarmstrong.com>')
283 Encodes headers according to the RFC1522 standard by calling
284 MIME::Words::encode_mimeword on distinct words as appropriate.
288 # We cannot use MIME::Words::encode_mimewords because that function
289 # does not handle spaces properly at all.
294 # handle being passed undef properly
295 return undef if not defined $rawstr;
297 # convert to octets if we are given a string in perl's internal
299 $rawstr= encode_utf8($rawstr) if is_utf8($rawstr);
300 # We process words in reverse so we can preserve spacing between
301 # encoded words. This regex splits on word|nonword boundaries and
302 # nonword|nonword boundaries. We also consider parenthesis and "
303 # to be nonwords to avoid escaping them in comments in violation
305 my @words = reverse split /(?:(?<=[\s\n\)\(\"])|(?=[\s\n\)\(\"]))/m, $rawstr;
307 my $previous_word_encoded = 0;
309 for my $word (@words) {
310 if ($word !~ m#[\x00-\x1F\x7F-\xFF]#o and $word ne ' ') {
311 $string = $word.$string;
312 $previous_word_encoded=0;
314 elsif ($word =~ /^[\s\n]$/) {
315 $string = $word.$string;
316 $previous_word_encoded = 0 if $word eq "\n";
319 my $encoded = MIME::Words::encode_mimeword($word, 'q', 'UTF-8');
320 # RFC 1522 mandates that segments be at most 76 characters
321 # long. If that's the case, we split the word up into 10
322 # character pieces and encode it. We must use the Encode
323 # magic here to avoid breaking on bit boundaries here.
324 if (length $encoded > 75) {
325 # Turn utf8 into the internal perl representation
326 # so . is a character, not a byte.
327 my $tempstr = is_utf8($word)?$word:decode_utf8($word,Encode::FB_DEFAULT);
329 # Strip it into 10 character long segments, and encode
331 # XXX It's possible that these segments are > 76 characters
332 while ($tempstr =~ s/(.{1,10})$//) {
333 # turn the character back into the utf8 representation.
334 my $tempword = encode_utf8($1);
335 # It may actually be better to eventually use
336 # the base64 encoding here, but I'm not sure
337 # if that's as widely supported as quoted
339 unshift @encoded, MIME::Words::encode_mimeword($tempword,'q','UTF-8');
341 $encoded = join(" ",@encoded);
342 # If the previous word was encoded, we must
343 # include a trailing _ that gets encoded as a
345 $encoded =~ s/\?\=$/_\?\=/ if $previous_word_encoded;
346 $string = $encoded.$string;
349 # If the previous word was encoded, we must
350 # include a trailing _ that gets encoded as a
352 $encoded =~ s/\?\=$/_\?\=/ if $previous_word_encoded;
353 $string = $encoded.$string;
355 $previous_word_encoded = 1;
363 $header = handle_escaped_commas('','From: ')
365 Handle commas in addresses which have been RFC1522 escaped and now need to be
366 quoted to avoid parsing as a record separator.
370 sub handle_escaped_commas {
371 my ($modified_hdr, $orig_hdr) = @_;
373 my $field = Mail::Message::Field->new($orig_hdr);
374 # if the header isn't structured, it can't contain an address
375 if (not $field->isStructured()) {
378 if ($field->name() !~ m/^(?:to|from|reply-to)$/) {
381 my @addresses = $field->addresses();
382 if (not @addresses) {
385 my @return_addresses;
386 for my $address (@addresses) {
387 $address->phrase(decode_rfc1522($address->phrase()));
388 push @return_addresses, $address->format();
390 return join(', ',@return_addresses)