use PublicInbox::MIME;
use PublicInbox::Git;
use PublicInbox::Import;
+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;
# assume 2 cores if GNU nproc(1) is not available
-my $NPROC = int($ENV{NPROC} || `nproc 2>/dev/null` || 2);
+sub nproc () {
+ int($ENV{NPROC} || `nproc 2>/dev/null` || 2);
+}
sub new {
my ($class, $v2ibx, $creat) = @_;
die "$dir does not exist\n";
}
}
+
+ my $nparts = 0;
+ my $xpfx = "$dir/xap" . PublicInbox::Search::SCHEMA_VERSION;
+
+ # always load existing partitions in case core count changes:
+ if (-d $xpfx) {
+ foreach my $part (<$xpfx/*>) {
+ -d $part && $part =~ m!/\d+\z! or next;
+ eval {
+ Search::Xapian::Database->new($part)->close;
+ $nparts++;
+ };
+ }
+ }
+ $nparts = nproc() if ($nparts == 0);
+
my $self = {
-inbox => $v2ibx,
im => undef, # PublicInbox::Import
xap_rw => undef, # PublicInbox::V2SearchIdx
xap_ro => undef,
- partitions => $NPROC,
+ partitions => $nparts,
transact_bytes => 0,
# limit each repo to 1GB or so
rotate_bytes => int((1024 * 1024 * 1024) / $PACKING_FACTOR),
# mimics Import::add and wraps it for v2
sub add {
my ($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}};
+ my ($oid, $len, $msgref) = @{$im->{last_object}};
- $self->idx_init;
- my $num = $self->{skel}->index_mm($mime, 1);
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
+ $$mid0 = $mid;
+ return $num;
+ };
+
+ # crap, Message-ID is already known, hope somebody just resent:
+ $self->barrier;
+ foreach my $m (@$mids) {
+ # read-only lookup now safe to do after above barrier
+ my $existing = $self->lookup_content($mime, $m);
+ if ($existing) {
+ warn "<$m> resent\n";
+ return; # easy, don't store duplicates
+ }
+ }
+
+ # very unlikely:
+ warn "<$mid> reused for mismatched content\n";
+
+ # try the rest of the mids
+ foreach my $i (1..$#$mids) {
+ my $m = $mids->[$i];
+ $num = $self->{skel}->{mm}->mid_insert($m);
+ if (defined $num) {
+ warn "alternative <$m> for <$mid> found\n";
+ $$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);
+ $$mid0 = PublicInbox::Import::digest2mid($dig);
+ my $num = $self->{skel}->{mm}->mid_insert($$mid0);
+ unless (defined $num) {
+ # it's hard to spoof the last Received: header
+ my @recvd = $hdr->header_raw('Received');
+ $dig->add("Received: $_") foreach (@recvd);
+ $$mid0 = PublicInbox::Import::digest2mid($dig);
+ $num = $self->{skel}->{mm}->mid_insert($$mid0);
+
+ # fall back to a random Message-ID and give up determinism:
+ until (defined($num)) {
+ $dig->add(rand);
+ $$mid0 = PublicInbox::Import::digest2mid($dig);
+ warn "using random Message-ID <$$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};
+ 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);
}
sub remove {
- my ($self, $mime, $msg) = @_;
- my $existing = $self->lookup_content($mime) or return;
-
- # don't touch ghosts or already junked messages
- return unless $existing->type eq 'mail';
-
- # always write removals to the current (latest) git repo since
- # we process chronologically
+ my ($self, $mime, $cmt_msg) = @_;
+ $self->barrier;
+ $self->idx_init;
my $im = $self->importer;
- my ($cmt, undef) = $im->remove($mime, $msg);
- $cmt = $im->get_mark($cmt);
- $self->unindex_msg($existing, $cmt);
+ my $ibx = $self->{-inbox};
+ my $srch = $ibx->search;
+ my $cid = content_id($mime);
+ my $skel = $self->{skel};
+ my $parts = $self->{idx_parts};
+ my $mm = $skel->{mm};
+ my $removed;
+ my $mids = mids($mime->header_obj);
+ foreach my $mid (@$mids) {
+ $srch->reopen->each_smsg_by_mid($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 $orig = $$msg;
+ my $cur = PublicInbox::MIME->new($msg);
+ if (content_id($cur) eq $cid) {
+ $mm->num_delete($smsg->num);
+ # $removed should only be set once assuming
+ # no bugs in our deduplication code:
+ $removed = $smsg;
+ $removed->{mime} = $cur;
+ $im->remove(\$orig, $cmt_msg);
+ $orig = undef;
+ $removed->num; # memoize this for callers
+
+ my $oid = $smsg->{blob};
+ foreach my $idx (@$parts, $skel) {
+ $idx->remote_remove($oid, $mid);
+ }
+ }
+ 1; # continue
+ });
+ $self->barrier;
+ }
+ $removed;
}
sub done {
my ($self) = @_;
- my $im = $self->{im};
+ my $im = delete $self->{im};
$im->done if $im; # PublicInbox::Import::done
$self->searchidx_checkpoint(0);
}
$self->searchidx_checkpoint(1);
}
+# issue a write barrier to ensure all data is visible to other processes
+# and read-only ops. Order of data importance is: git > SQLite > Xapian
+sub barrier {
+ my ($self) = @_;
+
+ if (my $im = $self->{im}) {
+ $im->barrier;
+ }
+ my $skel = $self->{skel};
+ my $parts = $self->{idx_parts};
+ if ($parts && $skel) {
+ my $dbh = $skel->{mm}->{dbh};
+ $dbh->commit; # SQLite data is second in importance
+
+ # Now deal with Xapian
+ $skel->barrier_init(scalar(@$parts));
+ # each partition needs to issue a barrier command to skel:
+ $_->barrier foreach @$parts;
+
+ $skel->barrier_wait; # wait for each Xapian partition
+
+ $dbh->begin_work;
+ }
+ $self->{transact_bytes} = 0;
+}
+
sub searchidx_checkpoint {
my ($self, $more) = @_;
# 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 $skel = $self->{skel}) {
- $skel->{mm}->{dbh}->commit;
+ my $dbh = $skel->{mm}->{dbh};
+ $dbh->commit;
if ($more) {
- $skel->{mm}->{dbh}->begin_work;
- }
- $skel->remote_commit;
- unless ($more) {
+ $dbh->begin_work;
+ } else {
$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} = $im;
}
sub lookup_content {
- undef # TODO
+ my ($self, $mime, $mid) = @_;
+ my $ibx = $self->{-inbox};
+
+ my $srch = $ibx->search->reopen;
+ 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 {