1 # Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Combine any combination of PublicInbox::Search,
5 # PublicInbox::ExtSearch, and PublicInbox::LeiSearch objects
7 package PublicInbox::LeiXSearch;
10 use parent qw(PublicInbox::LeiSearch PublicInbox::IPC);
14 PublicInbox::Search::load_xapian();
16 qp_flags => $PublicInbox::Search::QP_FLAGS |
17 PublicInbox::Search::FLAG_PURE_NOT(),
22 my ($self, $ibxish) = @_; # ibxish = ExtSearch or Inbox
24 if (!$ibxish->can('over') || !$ibxish->over) {
25 return push(@{$self->{remotes}}, $ibxish)
27 my $desc = $ibxish->{inboxdir} // $ibxish->{topdir};
28 my $srch = $ibxish->search or
29 return warn("$desc not indexed for Xapian\n");
30 my @shards = $srch->xdb_shards_flat or
31 return warn("$desc has no Xapian shardsXapian\n");
33 if (delete $self->{xdb}) { # XXX: do we need this?
34 # clobber existing {xdb} if amending
35 my $expect = delete $self->{nshard};
36 my $shards = delete $self->{shards_flat};
37 scalar(@$shards) == $expect or die
38 "BUG: {nshard}$expect != shards=".scalar(@$shards);
41 for my $old_ibxish (@{$self->{shard2ibx}}) {
42 next if $prev == $old_ibxish;
44 my @shards = $old_ibxish->search->xdb_shards_flat;
45 push @{$self->{shards_flat}}, @shards;
47 my $nr = scalar(@{$self->{shards_flat}});
49 "BUG: reloaded $nr shards, expected $expect"
51 push @{$self->{shards_flat}}, @shards;
52 push(@{$self->{shard2ibx}}, $ibxish) for (@shards);
55 # returns a list of local inboxes (or count in scalar context)
57 my %uniq = map {; "$_" => $_ } @{$_[0]->{shard2ibx} // []};
61 # called by PublicInbox::Search::xdb
62 sub xdb_shards_flat { @{$_[0]->{shards_flat} // []} }
66 my ($self, $mitem) = @_;
67 # cf. https://trac.xapian.org/wiki/FAQ/MultiDatabaseDocumentID
68 my $nshard = $self->{nshard};
69 my $docid = $mitem->get_docid;
70 my $shard = ($docid - 1) % $nshard;
71 my $num = int(($docid - 1) / $nshard) + 1;
72 my $smsg = $self->{shard2ibx}->[$shard]->over->get_art($num);
73 $smsg->{docid} = $docid;
78 my ($self, $qstr, $opt) = @_;
80 $opt->{relevance} //= -2;
81 $self->mset($qstr //= 'bytes:1..', $opt);
88 my $size = $mset->size;
89 $size && (($mo->{offset} += $size) < ($mo->{limit} // 10000));
92 sub query_thread_mset { # for --thread
93 my ($self, $lei, $ibxish) = @_;
94 my %sig = $lei->atfork_child_wq($self);
95 local @SIG{keys %sig} = values %sig;
97 my ($srch, $over) = ($ibxish->search, $ibxish->over);
98 unless ($srch && $over) {
99 my $desc = $ibxish->{inboxdir} // $ibxish->{topdir};
100 warn "$desc not indexed by Xapian\n";
103 my $mo = { %{$lei->{mset_opt}} };
105 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
109 $mset = $srch->mset($mo->{qstr}, $mo);
110 my $ids = $srch->mset_to_artnums($mset, $mo);
111 my $ctx = { ids => $ids };
113 my %n2item = map { ($ids->[$i++], $_) } $mset->items;
114 while ($over->expand_thread($ctx)) {
115 for my $n (@{$ctx->{xids}}) {
116 my $smsg = $over->get_art($n) or next;
117 next if $dd->is_smsg_dup($smsg);
118 my $mitem = delete $n2item{$smsg->{num}};
119 $each_smsg->($smsg, $mitem);
121 @{$ctx->{xids}} = ();
123 } while (_mset_more($mset, $mo));
124 $lei->{ovv}->ovv_atexit_child($lei);
127 sub query_mset { # non-parallel for non-"--thread" users
128 my ($self, $lei, $srcs) = @_;
129 my %sig = $lei->atfork_child_wq($self);
130 local @SIG{keys %sig} = values %sig;
131 my $mo = { %{$lei->{mset_opt}} };
133 $self->attach_external($_) for @$srcs;
134 my $each_smsg = $lei->{ovv}->ovv_each_smsg_cb($lei);
138 $mset = $self->mset($mo->{qstr}, $mo);
139 for my $it ($mset->items) {
140 my $smsg = smsg_for($self, $it) or next;
141 next if $dd->is_smsg_dup($smsg);
142 $each_smsg->($smsg, $it);
144 } while (_mset_more($mset, $mo));
145 $lei->{ovv}->ovv_atexit_child($lei);
148 sub query_done { # PublicInbox::EOFpipe callback
150 $lei->{ovv}->ovv_end($lei);
155 my ($self, $lei_orig, $srcs) = @_;
156 my ($lei, @io) = $lei_orig->atfork_parent_wq($self);
157 my $remotes = $self->{remotes} // [];
158 pipe(my ($eof_wait, $qry_done)) or die "pipe $!";
159 $io[0] = $qry_done; # don't need stdin
161 if ($lei->{opt}->{thread}) {
162 $lei->{-parallel} = scalar(@$remotes) + scalar(@$srcs) - 1;
163 for my $ibxish (@$srcs) {
164 $self->wq_do('query_thread_mset', \@io, $lei, $ibxish);
167 $lei->{-parallel} = scalar(@$remotes);
168 $self->wq_do('query_mset', \@io, $lei, $srcs);
171 for my $rmt (@$remotes) {
172 $self->wq_do('query_thread_mbox', \@io, $lei, $rmt);
175 close $qry_done; # fully closed when children are done
177 # query_done will run when query_*mset close $qry_done
178 if ($lei_orig->{sock}) { # watch for client premature exit
179 require PublicInbox::EOFpipe;
180 PublicInbox::EOFpipe->new($eof_wait, \&query_done, $lei_orig);
181 $lei_orig->{lxs} = $self;
182 $lei_orig->event_step_init;
185 read($eof_wait, my $buf, 1); # wait for close($lei->{0})
186 query_done($lei_orig); # may SIGPIPE
190 sub ipc_atfork_prepare {
192 $self->wq_set_recv_modes(qw[+<&= >&= >&= +<&=]);
193 $self->SUPER::ipc_atfork_prepare; # PublicInbox::IPC