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 # like PublicInbox::SearchIdx, but for searching for non-mail messages.
5 # Things indexed include:
8 # * (maybe) git code repository information
9 # Expect ~100K-1M documents with no parallelism opportunities,
10 # so no sharding, here.
12 # See MiscSearch for read-only counterpart
13 package PublicInbox::MiscIdx;
16 use PublicInbox::InboxWritable;
17 use PublicInbox::Search; # for SWIG Xapian and Search::Xapian compat
18 use PublicInbox::SearchIdx qw(index_text term_generator add_val);
19 use PublicInbox::Spawn qw(nodatacow_dir);
22 use PublicInbox::MiscSearch;
23 use PublicInbox::Config;
27 my ($class, $eidx) = @_;
28 PublicInbox::SearchIdx::load_xapian_writable();
29 my $mi_dir = "$eidx->{xpfx}/misc";
30 File::Path::mkpath($mi_dir);
31 nodatacow_dir($mi_dir);
32 my $flags = $PublicInbox::SearchIdx::DB_CREATE_OR_OPEN;
33 $flags |= $PublicInbox::SearchIdx::DB_NO_SYNC if $eidx->{-no_fsync};
34 $json //= PublicInbox::Config::json();
38 indexlevel => 'full', # small DB, no point in medium?
44 croak 'BUG: already in txn' if $self->{xdb}; # XXX make lazy?
45 my $wdb = $PublicInbox::Search::X{WritableDatabase};
46 my $xdb = eval { $wdb->new($self->{mi_dir}, $self->{flags}) };
47 croak "Failed opening $self->{mi_dir}: $@" if $@;
49 $xdb->begin_transaction;
54 croak 'BUG: not in txn' unless $self->{xdb}; # XXX make lazy?
55 delete($self->{xdb})->commit_transaction;
59 my ($self, $eidx_key) = @_;
60 my $xdb = $self->{xdb};
61 my $head = $xdb->postlist_begin('Q'.$eidx_key);
62 my $tail = $xdb->postlist_end('Q'.$eidx_key);
63 my @docids; # only one, unless we had bugs
64 for (; $head != $tail; $head++) {
65 push @docids, $head->get_docid;
67 for my $docid (@docids) {
68 $xdb->delete_document($docid);
69 warn "I: remove inbox docid #$docid ($eidx_key)\n";
73 # adds or updates according to $eidx_key
75 my ($self, $ibx) = @_;
76 my $eidx_key = $ibx->eidx_key;
77 my $xdb = $self->{xdb};
78 # Q = uniQue in Xapian terminology
79 my $head = $xdb->postlist_begin('Q'.$eidx_key);
80 my $tail = $xdb->postlist_end('Q'.$eidx_key);
82 for (; $head != $tail; $head++) {
84 my $i = $head->get_docid;
87 W: multiple inboxes keyed to `$eidx_key', deleting #$i
90 $docid = $head->get_docid;
93 $xdb->delete_document($_) for @drop; # just in case
95 my $doc = $PublicInbox::Search::X{Document}->new;
96 term_generator($self)->set_document($doc);
98 # allow sorting by modified and uidvalidity (created at)
99 add_val($doc, $PublicInbox::MiscSearch::MODIFIED, $ibx->modified);
100 add_val($doc, $PublicInbox::MiscSearch::UIDVALIDITY, $ibx->uidvalidity);
102 $doc->add_boolean_term('Q'.$eidx_key); # uniQue id
103 $doc->add_boolean_term('T'.'inbox'); # Type
105 if (defined($ibx->{newsgroup}) && $ibx->nntp_usable) {
106 $doc->add_boolean_term('T'.'newsgroup'); # additional Type
109 # force reread from disk, {description} could be loaded from {misc}
110 delete $ibx->{description};
111 my $desc = $ibx->description;
113 # description = S/Subject (or title)
115 index_text($self, $desc, 1, 'S');
116 index_text($self, $ibx->{name}, 1, 'XNAME');
120 infourl => 'XINFOURL',
123 while (my ($f, $pfx) = each %map) {
124 for my $v (@{$ibx->{$f} // []}) {
125 index_text($self, $v, 1, $pfx);
129 if (defined(my $max = $ibx->max_git_epoch)) { # v2
130 my $pfx = "/$ibx->{name}/git/";
131 for my $epoch (0..$max) {
132 my $git = $ibx->git_epoch($epoch) or return;
133 if (my $ent = $git->manifest_entry($epoch, $desc)) {
134 $data->{"$pfx$epoch.git"} = $ent;
135 $ent->{git_dir} = $git->{git_dir};
137 $git->cleanup; # ->modified starts cat-file --batch
139 } elsif (my $ent = $ibx->git->manifest_entry) { # v1
140 $ent->{git_dir} = $ibx->{inboxdir};
141 $data->{"/$ibx->{name}"} = $ent;
143 $doc->set_data($json->encode($data));
144 if (defined $docid) {
145 $xdb->replace_document($docid, $doc);
147 $xdb->add_document($doc);