1 # Copyright (C) 2016-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # read-only utilities for Email::MIME
5 package PublicInbox::MsgIter;
9 our @EXPORT = qw(msg_iter msg_part_text);
10 use PublicInbox::MIME;
12 # Like Email::MIME::walk_parts, but this is:
14 # * passes depth and indices to the iterator callback
15 sub msg_iter ($$;$$) {
16 my ($mime, $cb, $cb_arg, $do_undef) = @_;
17 my @parts = $mime->subparts;
19 $mime = $_[0] = undef if $do_undef; # saves some memory
21 @parts = map { [ $_, 1, ++$i ] } @parts;
22 while (my $p = shift @parts) {
23 my ($part, $depth, @idx) = @$p;
24 my @sub = $part->subparts;
28 @sub = map { [ $_, $depth, @idx, ++$i ] } @sub;
29 @parts = (@sub, @parts);
35 $cb->([$mime, 0, 0], $cb_arg);
39 sub msg_part_text ($$) {
42 # TODO: we may offer a separate sub for people who need to index
43 # HTML-only mail, but the majority of HTML mail is multipart/alternative
44 # with a text part which we don't have to waste cycles decoding
45 return if $ct =~ m!\btext/x?html\b!;
47 my $s = eval { $part->body_str };
50 # text/plain is the default, multipart/mixed happened a few
51 # times when it should not have been:
52 # <87llgalspt.fsf@free.fr>
53 # <200308111450.h7BEoOu20077@mail.osdl.org>
54 if ($err && ($ct =~ m!\btext/\b!i ||
55 $ct =~ m!\bmultipart/mixed\b!i)) {
56 my $cte = $part->header_raw('Content-Transfer-Encoding');
57 if (defined($cte) && $cte =~ /\b7bit\b/i) {
59 $err = undef if $s =~ /\A[[:ascii:]]+\z/s;
61 # Try to assume UTF-8 because Alpine seems to
62 # do wacky things and set charset=X-UNKNOWN
63 $part->charset_set('UTF-8');
64 $s = eval { $part->body_str };
67 # If forcing charset=UTF-8 failed,
68 # caller will warn further down...
69 $s = $part->body if $@;
74 # returns an array of quoted or unquoted sections
76 # Quiet "Complex regular subexpression recursion limit" warning
77 # in case an inconsiderate sender quotes 32K of text at once.
78 # The warning from Perl is harmless for us since our callers can
79 # tolerate less-than-ideal matches which work within Perl limits.
81 split(/((?:^>[^\n]*\n)+)/sm, shift);