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
# 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 $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}};
+ my ($oid, $len, $msgref) = @{$im->{last_object}};
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];
if ($more) {
$dbh->begin_work;
} else {
- $skel->remote_commit; # XXX should be unnecessary...
$skel->remote_close;
delete $self->{skel};
}
die "$git_dir exists\n" if -e $git_dir;
my @cmd = (qw(git init --bare -q), $git_dir);
PublicInbox::Import::run_die(\@cmd);
- @cmd = (qw/git config/, "--file=$git_dir/config",
- 'repack.writeBitmaps', 'true');
- PublicInbox::Import::run_die(\@cmd);
my $all = "$self->{-inbox}->{mainrepo}/all.git";
unless (-d $all) {
@cmd = (qw(git init --bare -q), $all);
PublicInbox::Import::run_die(\@cmd);
+ @cmd = (qw/git config/, "--file=$all/config",
+ 'repack.writeBitmaps', 'true');
+ PublicInbox::Import::run_die(\@cmd);
}
+ @cmd = (qw/git config/, "--file=$git_dir/config",
+ 'include.path', '../../all.git/config');
+ PublicInbox::Import::run_die(\@cmd);
+
my $alt = "$all/objects/info/alternates";
my $new_obj_dir = "../../git/$new.git/objects";
my %alts;
my ($self, $git, $packed_bytes) = @_;
my $im = PublicInbox::Import->new($git, undef, undef, $self->{-inbox});
$im->{bytes_added} = int($packed_bytes / $PACKING_FACTOR);
- $im->{want_object_id} = 1;
+ $im->{want_object_info} = 1;
$im->{ssoma_lock} = 0;
$im->{path_type} = 'v2';
$self->{im} = $im;
}
sub lookup_content {
- my ($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;