use warnings;
use Fcntl qw(:flock :DEFAULT);
use PublicInbox::SearchIdxPart;
-use PublicInbox::SearchIdxThread;
+use PublicInbox::SearchIdxSkeleton;
use PublicInbox::MIME;
use PublicInbox::Git;
use PublicInbox::Import;
-use Email::MIME::ContentType;
-$Email::MIME::ContentType::STRICT_PARAMS = 0;
+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
my $PACKING_FACTOR = 0.4;
# mimics Import::add and wraps it for v2
sub add {
my ($self, $mime, $check_cb) = @_;
- my $existing = $self->lookup_content($mime);
- if ($existing) {
- return undef if $existing->type eq 'mail'; # duplicate
+ # spam check:
+ if ($check_cb) {
+ $mime = $check_cb->($mime) or return;
}
- my $im = $self->importer;
+ # All pipes (> $^F) known to Perl 5.6+ have FD_CLOEXEC set,
+ # as does SQLite 3.4.1+ (released in 2007-07-20), and
+ # Xapian 1.3.2+ (released 2015-03-15).
+ # For the most part, we can spawn git-fast-import without
+ # leaking FDs to it...
+ $self->idx_init;
- # im->add returns undef if check_cb fails
- my $cmt = $im->add($mime, $check_cb) or return;
+ my $mid0;
+ my $num = num_for($self, $mime, \$mid0);
+ defined $num or return; # duplicate
+ defined $mid0 or die "BUG: $mid0 undefined\n";
+ my $im = $self->importer;
+ my $cmt = $im->add($mime);
$cmt = $im->get_mark($cmt);
my $oid = $im->{last_object_id};
my ($len, $msgref) = @{$im->{last_object}};
- $self->idx_init;
- my $num = $self->{all}->index_mm($mime);
my $nparts = $self->{partitions};
my $part = $num % $nparts;
my $idx = $self->idx_part($part);
- $idx->index_raw($len, $msgref, $num, $oid);
+ $idx->index_raw($len, $msgref, $num, $oid, $mid0);
my $n = $self->{transact_bytes} += $len;
if ($n > (PublicInbox::SearchIdx::BATCH_BYTES * $nparts)) {
$self->checkpoint;
$mime;
}
+sub num_for {
+ my ($self, $mime, $mid0) = @_;
+ my $mids = mids($mime->header_obj);
+ if (@$mids) {
+ my $mid = $mids->[0];
+ my $num = $self->{skel}->{mm}->mid_insert($mid);
+ if (defined $num) { # common case
+ $$mid0 = $mid;
+ return $num;
+ };
+
+ # crap, Message-ID is already known, hope somebody just resent:
+ $self->done; # write barrier, clears $self->{skel}
+ foreach my $m (@$mids) {
+ # read-only lookup now safe to do after above barrier
+ my $existing = $self->lookup_content($mime, $m);
+ if ($existing) {
+ warn "<$m> resent\n";
+ return; # easy, don't store duplicates
+ }
+ }
+
+ # very unlikely:
+ warn "<$mid> reused for mismatched content\n";
+ $self->idx_init;
+
+ # try the rest of the mids
+ foreach my $i (1..$#$mids) {
+ my $m = $mids->[$i];
+ $num = $self->{skel}->{mm}->mid_insert($m);
+ if (defined $num) {
+ warn "alternative <$m> for <$mid> found\n";
+ $$mid0 = $m;
+ return $num;
+ }
+ }
+ }
+ # none of the existing Message-IDs are good, generate a new one:
+ num_for_harder($self, $mime, $mid0);
+}
+
+sub num_for_harder {
+ my ($self, $mime, $mid0) = @_;
+
+ my $hdr = $mime->header_obj;
+ my $dig = content_digest($mime);
+ $$mid0 = $dig->clone->hexdigest . '@localhost';
+ my $num = $self->{skel}->{mm}->mid_insert($$mid0);
+ unless (defined $num) {
+ # it's hard to spoof the last Received: header
+ my @recvd = $hdr->header_raw('Received');
+ $dig->add("Received: $_") foreach (@recvd);
+ $$mid0 = $dig->clone->hexdigest . '@localhost';
+ $num = $self->{skel}->{mm}->mid_insert($$mid0);
+
+ # fall back to a random Message-ID and give up determinism:
+ until (defined($num)) {
+ $dig->add(rand);
+ $$mid0 = $dig->clone->hexdigest . '@localhost';
+ warn "using random Message-ID <$$mid0> as fallback\n";
+ $num = $self->{skel}->{mm}->mid_insert($$mid0);
+ }
+ }
+ my @cur = $hdr->header_raw('Message-Id');
+ $hdr->header_set('Message-Id', "<$$mid0>", @cur);
+ $num;
+}
+
sub idx_part {
my ($self, $part) = @_;
$self->{idx_parts}->[$part];
}
+# idempotent
sub idx_init {
my ($self) = @_;
return if $self->{idx_parts};
- # first time initialization:
- my $all = $self->{all} =
- PublicInbox::SearchIdxThread->new($self->{-inbox});
+ my $ibx = $self->{-inbox};
+
+ # do not leak read-only FDs to child processes, we only have these
+ # FDs for duplicate detection so they should not be
+ # frequently activated.
+ delete $ibx->{$_} foreach (qw(git mm search));
+
+ # first time initialization, first we create the skeleton pipe:
+ my $skel = $self->{skel} = PublicInbox::SearchIdxSkeleton->new($self);
# need to create all parts before initializing msgmap FD
my $max = $self->{partitions} - 1;
my $idx = $self->{idx_parts} = [];
for my $i (0..$max) {
- push @$idx, PublicInbox::SearchIdxPart->new($self, $i, $all);
+ push @$idx, PublicInbox::SearchIdxPart->new($self, $i, $skel);
}
- $all->_msgmap_init->{dbh}->begin_work;
+
+ # Now that all subprocesses are up, we can open the FD for SQLite:
+ $skel->_msgmap_init->{dbh}->begin_work;
}
sub remove {
sub done {
my ($self) = @_;
- my $im = $self->{im};
+ my $im = delete $self->{im};
$im->done if $im; # PublicInbox::Import::done
$self->searchidx_checkpoint(0);
}
sub searchidx_checkpoint {
my ($self, $more) = @_;
- # order matters, we can only close {all} after all partitions
- # are done because the partitions also write to {all}
-
+ # order matters, we can only close {skel} after all partitions
+ # are done because the partitions also write to {skel}
if (my $parts = $self->{idx_parts}) {
foreach my $idx (@$parts) {
- $idx->remote_commit;
+ $idx->remote_commit; # propagates commit to skel
$idx->remote_close unless $more;
}
delete $self->{idx_parts} unless $more;
}
- if (my $all = $self->{all}) {
- $all->{mm}->{dbh}->commit;
+ if (my $skel = $self->{skel}) {
+ my $dbh = $skel->{mm}->{dbh};
+ $dbh->commit;
if ($more) {
- $all->{mm}->{dbh}->begin_work;
+ $dbh->begin_work;
+ } else {
+ $skel->remote_commit; # XXX should be unnecessary...
+ $skel->remote_close;
+ delete $self->{skel};
}
- $all->remote_commit;
- $all->remote_close unless $more;
- delete $self->{all} unless $more;
}
$self->{transact_bytes} = 0;
}
}
sub lookup_content {
- undef # TODO
+ my ($self, $mime, $mid) = @_;
+ my $ibx = $self->{-inbox};
+
+ my $srch = $ibx->search;
+ my $cid = content_id($mime);
+ my $found;
+ $srch->each_smsg_by_mid($mid, sub {
+ my ($smsg) = @_;
+ $smsg->load_expand;
+ my $msg = $ibx->msg_by_smsg($smsg);
+ if (!defined($msg)) {
+ warn "broken smsg for $mid\n";
+ return 1; # continue
+ }
+ my $cur = PublicInbox::MIME->new($msg);
+ if (content_id($cur) eq $cid) {
+ $smsg->{mime} = $cur;
+ $found = $smsg;
+ return 0; # break out of loop
+ }
+ 1; # continue
+ });
+ $found;
+}
+
+sub atfork_child {
+ my ($self) = @_;
+ if (my $parts = $self->{idx_parts}) {
+ $_->atfork_child foreach @$parts;
+ }
+ if (my $im = $self->{im}) {
+ $im->atfork_child;
+ }
}
1;