lib/PublicInbox/MID.pm | 24 +++++++++++++++++++++++-
t/mid.t | 22 +++++++++++++++++++++-
diff --git a/lib/PublicInbox/MID.pm b/lib/PublicInbox/MID.pm
index 2c9822f478da32879866065c21a1f1f247297827..786c056f8593cac7d493689a1ce0774500629ff6 100644
--- a/lib/PublicInbox/MID.pm
+++ b/lib/PublicInbox/MID.pm
@@ -6,7 +6,8 @@ package PublicInbox::MID;
use strict;
use warnings;
use base qw/Exporter/;
-our @EXPORT_OK = qw/mid_clean id_compress mid2path mid_mime mid_escape MID_ESC/;
+our @EXPORT_OK = qw/mid_clean id_compress mid2path mid_mime mid_escape MID_ESC
+ mids references/;
use URI::Escape qw(uri_escape_utf8);
use Digest::SHA qw/sha1_hex/;
use constant MID_MAX => 40; # SHA-1 hex length
@@ -47,6 +48,27 @@ "$x2/$x38";
}
sub mid_mime ($) { $_[0]->header_obj->header_raw('Message-ID') }
+
+sub uniq_mids {
+ my ($hdr, @fields) = @_;
+ my %seen;
+ my @raw;
+ foreach my $f (@fields) {
+ push @raw, $hdr->header_raw($f);
+ }
+ my @mids = (join(' ', @raw) =~ /<([^>]+)>/g);
+ my $mids = scalar(@mids) == 0 ? \@raw: \@mids;
+ my @ret;
+ foreach (@$mids) {
+ next if $seen{$_};
+ push @ret, $_;
+ $seen{$_} = 1;
+ }
+ \@ret;
+}
+
+sub mids { uniq_mids($_[0], 'Message-Id') }
+sub references { uniq_mids($_[0], 'References', 'In-Reply-To') }
# RFC3986, section 3.3:
sub MID_ESC () { '^A-Za-z0-9\-\._~!\$\&\';\(\)\*\+,;=:@' }
diff --git a/t/mid.t b/t/mid.t
index 0bf33318e6c8ea120228e913d166b30c195843a9..223be798c47704f7a1823e6f59037ead36c840c6 100644
--- a/t/mid.t
+++ b/t/mid.t
@@ -1,11 +1,31 @@
# Copyright (C) 2016-2018 all contributors
# License: AGPL-3.0+
use Test::More;
-use PublicInbox::MID qw(mid_escape);
+use PublicInbox::MID qw(mid_escape mids references);
is(mid_escape('foo!@(bar)'), 'foo!@(bar)');
is(mid_escape('foo%!@(bar)'), 'foo%25!@(bar)');
is(mid_escape('foo%!@(bar)'), 'foo%25!@(bar)');
+
+{
+ use Email::MIME;
+ my $mime = Email::MIME->create;
+ $mime->header_set('Message-Id', '');
+ is_deeply(['mid-1@a'], mids($mime->header_obj), 'mids in common case');
+ $mime->header_set('Message-Id', '', '');
+ is_deeply(['mid-1@a', 'mid-2@b'], mids($mime->header_obj), '2 mids');
+ $mime->header_set('Message-Id', '', '');
+ is_deeply(['mid-1@a'], mids($mime->header_obj), 'dup mids');
+ $mime->header_set('Message-Id', ' comment');
+ is_deeply(['mid-1@a'], mids($mime->header_obj), 'comment ignored');
+ $mime->header_set('Message-Id', 'bare-mid');
+ is_deeply(['bare-mid'], mids($mime->header_obj), 'bare mid OK');
+
+ $mime->header_set('References', ' ');
+ $mime->header_set('In-Reply-To', '');
+ is_deeply(['hello', 'world', 'weld'], references($mime->header_obj),
+ 'references combines with In-Reply-To');
+}
done_testing();
1;