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);
11 # This becomes PublicInbox::MIME->each_part:
12 # Like Email::MIME::walk_parts, but this is:
14 # * passes depth and indices to the iterator callback
15 sub em_each_part ($$;$$) {
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 # Use this when we may accept Email::MIME from user scripts
40 # (not just PublicInbox::MIME)
41 sub msg_iter ($$;$$) { # $_[0] = PublicInbox::MIME/Email::MIME-like obj
42 my (undef, $cb, $cb_arg, $once) = @_;
43 if (my $ep = $_[0]->can('each_part')) { # PublicInbox::{MIME,*}
44 $ep->($_[0], $cb, $cb_arg, $once);
45 } else { # for compatibility with existing Email::MIME users:
46 em_each_part($_[0], $cb, $cb_arg, $once);
50 sub msg_part_text ($$) {
53 # TODO: we may offer a separate sub for people who need to index
54 # HTML-only mail, but the majority of HTML mail is multipart/alternative
55 # with a text part which we don't have to waste cycles decoding
56 return if $ct =~ m!\btext/x?html\b!;
58 my $s = eval { $part->body_str };
61 # text/plain is the default, multipart/mixed happened a few
62 # times when it should not have been:
63 # <87llgalspt.fsf@free.fr>
64 # <200308111450.h7BEoOu20077@mail.osdl.org>
65 if ($err && ($ct =~ m!\btext/\b!i ||
66 $ct =~ m!\bmultipart/mixed\b!i)) {
67 my $cte = $part->header_raw('Content-Transfer-Encoding');
68 if (defined($cte) && $cte =~ /\b7bit\b/i) {
70 $err = undef if $s =~ /\A[[:ascii:]]+\z/s;
72 # Try to assume UTF-8 because Alpine seems to
73 # do wacky things and set charset=X-UNKNOWN
74 $part->charset_set('UTF-8');
75 $s = eval { $part->body_str };
78 # If forcing charset=UTF-8 failed,
79 # caller will warn further down...
80 $s = $part->body if $@;
85 # returns an array of quoted or unquoted sections
87 # Quiet "Complex regular subexpression recursion limit" warning
88 # in case an inconsiderate sender quotes 32K of text at once.
89 # The warning from Perl is harmless for us since our callers can
90 # tolerate less-than-ideal matches which work within Perl limits.
92 split(/((?:^>[^\n]*\n)+)/sm, shift);