Since we'll need to support multiple Message-IDs anyways,
inject a new one if we hit a duplicate (or don't get one at
all).

Try to use a deterministic Message-Id for consistency, but give
up determinism and use a random Message-Id if an "attacker"
wants to prevent their message from being archived.
---
 MANIFEST                      |  1 +
 lib/PublicInbox/V2Writable.pm | 88 +++++++++++++++++++++++++++++++++----------
 t/v2writable.t                | 84 +++++++++++++++++++++++++++++++++++++++++
 3 files changed, 154 insertions(+), 19 deletions(-)
 create mode 100644 t/v2writable.t

diff --git a/MANIFEST b/MANIFEST
index 1aaf8ff..7366aa0 100644
--- a/MANIFEST
+++ b/MANIFEST
@@ -177,5 +177,6 @@ t/spawn.t
 t/thread-all.t
 t/thread-cycle.t
 t/utf8.mbox
+t/v2writable.t
 t/view.t
 t/watch_maildir.t
diff --git a/lib/PublicInbox/V2Writable.pm b/lib/PublicInbox/V2Writable.pm
index 57cb7d3..6d73827 100644
--- a/lib/PublicInbox/V2Writable.pm
+++ b/lib/PublicInbox/V2Writable.pm
@@ -11,8 +11,8 @@ use PublicInbox::SearchIdxSkeleton;
 use PublicInbox::MIME;
 use PublicInbox::Git;
 use PublicInbox::Import;
-use PublicInbox::MID qw(mid_clean mid_mime);
-use PublicInbox::ContentId qw(content_id);
+use PublicInbox::MID qw(mids);
+use PublicInbox::ContentId qw(content_id content_digest);
 use PublicInbox::Inbox;
 
 # an estimate of the post-packed size to the raw uncompressed size
@@ -62,21 +62,8 @@ sub add {
        # leaking FDs to it...
        $self->idx_init;
 
-       my $mid = mid_clean(mid_mime($mime));
-       my $num = $self->{skel}->{mm}->mid_insert($mid);
-       if (!defined($num)) { # mid is already known
-               $self->done; # ensure all subprocesses are done writing
-
-               my $existing = $self->lookup_content($mime);
-               warn "<$mid> resent\n" if $existing;
-               return if $existing; # easy, don't store duplicates
-
-               # reuse NNTP article number?
-               warn "<$mid> reused for mismatched content\n";
-               $self->idx_init;
-               $num = $self->{skel}->{mm}->num_for($mid);
-       }
-
+       my $num = num_for($self, $mime);
+       defined $num or return; # duplicate
        my $im = $self->importer;
        my $cmt = $im->add($mime);
        $cmt = $im->get_mark($cmt);
@@ -95,6 +82,70 @@ sub add {
        $mime;
 }
 
+sub num_for {
+       my ($self, $mime) = @_;
+       my $mids = mids($mime->header_obj);
+       if (@$mids) {
+               my $mid = $mids->[0];
+               my $num = $self->{skel}->{mm}->mid_insert($mid);
+               return $num if defined($num); # common case
+
+               # crap, Message-ID is already known, hope somebody just resent:
+               $self->done; # write barrier, clears $self->{skel}
+               foreach my $m (@$mids) {
+                       # read-only lookup now safe to do after above barrier
+                       my $existing = $self->lookup_content($mime, $m);
+                       if ($existing) {
+                               warn "<$m> resent\n";
+                               return; # easy, don't store duplicates
+                       }
+               }
+
+               # very unlikely:
+               warn "<$mid> reused for mismatched content\n";
+               $self->idx_init;
+
+               # try the rest of the mids
+               foreach my $i (1..$#$mids) {
+                       my $m = $mids->[$i];
+                       $num = $self->{skel}->{mm}->mid_insert($m);
+                       if (defined $num) {
+                               warn "alternative <$m> for <$mid> found\n";
+                               return $num;
+                       }
+               }
+       }
+       # none of the existing Message-IDs are good, generate a new one:
+       num_for_harder($self, $mime);
+}
+
+sub num_for_harder {
+       my ($self, $mime) = @_;
+
+       my $hdr = $mime->header_obj;
+       my $dig = content_digest($mime);
+       my $mid = $dig->clone->hexdigest . '@localhost';
+       my $num = $self->{skel}->{mm}->mid_insert($mid);
+       unless (defined $num) {
+               # it's hard to spoof the last Received: header
+               my @recvd = $hdr->header_raw('Received');
+               $dig->add("Received: $_") foreach (@recvd);
+               $mid = $dig->clone->hexdigest . '@localhost';
+               $num = $self->{skel}->{mm}->mid_insert($mid);
+
+               # fall back to a random Message-ID and give up determinism:
+               until (defined($num)) {
+                       $dig->add(rand);
+                       $mid = $dig->clone->hexdigest . '@localhost';
+                       warn "using random Message-ID <$mid> as fallback\n";
+                       $num = $self->{skel}->{mm}->mid_insert($mid);
+               }
+       }
+       my @cur = $hdr->header_raw('Message-Id');
+       $hdr->header_set('Message-Id', @cur, "<$mid>");
+       $num;
+}
+
 sub idx_part {
        my ($self, $part) = @_;
        $self->{idx_parts}->[$part];
@@ -268,13 +319,12 @@ sub import_init {
 }
 
 sub lookup_content {
-       my ($self, $mime) = @_;
+       my ($self, $mime, $mid) = @_;
        my $ibx = $self->{-inbox};
 
        my $srch = $ibx->search;
        my $cid = content_id($mime);
        my $found;
-       my $mid = mid_mime($mime);
        $srch->each_smsg_by_mid($mid, sub {
                my ($smsg) = @_;
                $smsg->load_expand;
diff --git a/t/v2writable.t b/t/v2writable.t
new file mode 100644
index 0000000..bc2437a
--- /dev/null
+++ b/t/v2writable.t
@@ -0,0 +1,84 @@
+# Copyright (C) 2018 all contributors <meta@public-inbox.org>
+# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
+use strict;
+use warnings;
+use Test::More;
+use PublicInbox::MIME;
+use PublicInbox::ContentId qw(content_digest);
+use File::Temp qw/tempdir/;
+foreach my $mod (qw(DBD::SQLite Search::Xapian)) {
+       eval "require $mod";
+       plan skip_all => "$mod missing for nntpd.t" if $@;
+}
+use_ok 'PublicInbox::V2Writable';
+my $mainrepo = tempdir('pi-v2writable-XXXXXX', TMPDIR => 1, CLEANUP => 1);
+my $ibx = {
+       mainrepo => $mainrepo,
+       name => 'test-v2writable',
+       version => 2,
+       -primary_address => 't...@example.com',
+};
+$ibx = PublicInbox::Inbox->new($ibx);
+my $mime = PublicInbox::MIME->create(
+       header => [
+               From => 'a...@example.com',
+               To => 't...@example.com',
+               Subject => 'this is a subject',
+               'Message-ID' => '<a-mid@b>',
+               Date => 'Fri, 02 Oct 1993 00:00:00 +0000',
+       ],
+       body => "hello world\n",
+);
+
+my $im = PublicInbox::V2Writable->new($ibx, 1);
+ok($im->add($mime), 'ordinary message added');
+{
+       my @warn;
+       local $SIG{__WARN__} = sub { push @warn, @_ };
+       is(undef, $im->add($mime), 'obvious duplicate rejected');
+       like(join(' ', @warn), qr/resent/, 'warned about resent message');
+
+       @warn = ();
+       $mime->header_set('Message-Id', '<a-mid@b>', '<c@d>');
+       ok($im->add($mime), 'secondary MID used');
+       like(join(' ', @warn), qr/mismatched/, 'warned about mismatch');
+       like(join(' ', @warn), qr/alternative/, 'warned about alternative');
+       is_deeply([ '<a-mid@b>', '<c@d>' ],
+               [ $mime->header_obj->header_raw('Message-Id') ],
+               'no new Message-Id added');
+
+       @warn = ();
+       $mime->header_set('Message-Id', '<a-mid@b>');
+       $mime->body_set('different');
+       ok($im->add($mime), 'reused mid ok');
+       like(join(' ', @warn), qr/reused/, 'warned about reused MID');
+       my @mids = $mime->header_obj->header_raw('Message-Id');
+       is($mids[0], '<a-mid@b>', 'original mid not changed');
+       like($mids[1], qr/\A<\w+\@localhost>\z/, 'new MID added');
+       is(scalar(@mids), 2, 'only one new MID added');
+
+       @warn = ();
+       $mime->header_set('Message-Id', '<a-mid@b>');
+       $mime->body_set('this one needs a random mid');
+       my $gen = content_digest($mime)->hexdigest . '@localhost';
+       my $fake = PublicInbox::MIME->new($mime->as_string);
+       $fake->header_set('Message-Id', $gen);
+       ok($im->add($fake), 'fake added easily');
+       is_deeply(\@warn, [], 'no warnings from a faker');
+       ok($im->add($mime), 'random MID made');
+       like(join(' ', @warn), qr/using random/, 'warned about using random');
+       @mids = $mime->header_obj->header_raw('Message-Id');
+       is($mids[0], '<a-mid@b>', 'original mid not changed');
+       like($mids[1], qr/\A<\w+\@localhost>\z/, 'new MID added');
+       is(scalar(@mids), 2, 'only one new MID added');
+
+       @warn = ();
+       $mime->header_set('Message-Id');
+       ok($im->add($mime), 'random MID made for MID free message');
+       @mids = $mime->header_obj->header_raw('Message-Id');
+       like($mids[0], qr/\A<\w+\@localhost>\z/, 'mid was generated');
+       is(scalar(@mids), 1, 'new generated');
+}
+
+$im->done;
+done_testing();
-- 
EW

--
unsubscribe: meta+unsubscr...@public-inbox.org
archive: https://public-inbox.org/meta/

Reply via email to