1 # Copyright (C) 2018 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # for XOVER, OVER in NNTP, and feeds/homepage/threads in PSGI
5 # Unlike Msgmap, this is an _UNSTABLE_ cache which can be
6 # tweaked/updated over time and rebuilt.
8 # Ghost messages (messages which are only referenced in References/In-Reply-To)
9 # are denoted by a negative NNTP article number.
10 package PublicInbox::OverIdx;
13 use base qw(PublicInbox::Over);
15 use DBI qw(:sql_types); # SQL_BLOB
16 use PublicInbox::MID qw/id_compress mids references/;
17 use PublicInbox::SearchMsg;
18 use Compress::Zlib qw(compress);
19 use PublicInbox::Search;
23 my $dbh = $self->SUPER::dbh_new;
24 $dbh->do('PRAGMA journal_mode = TRUNCATE');
25 $dbh->do('PRAGMA cache_size = 80000');
30 sub get_counter ($$) {
32 my $sth = $dbh->prepare_cached(<<'', undef, 1);
33 SELECT val FROM counter WHERE key = ? LIMIT 1
39 sub adj_counter ($$$) {
40 my ($self, $key, $op) = @_;
41 my $dbh = $self->{dbh};
42 my $sth = $dbh->prepare_cached(<<"");
43 UPDATE counter SET val = val $op 1 WHERE key = ?
47 get_counter($dbh, $key);
50 sub next_tid { adj_counter($_[0], 'thread', '+') }
51 sub next_ghost_num { adj_counter($_[0], 'ghost', '-') }
54 my ($self, $tbl, $id_col, $val_col, $val) = @_;
55 my $dbh = $self->{dbh};
56 my $in = $dbh->prepare_cached(<<"")->execute($val);
57 INSERT OR IGNORE INTO $tbl ($val_col) VALUES (?)
60 my $sth = $dbh->prepare_cached(<<"", undef, 1);
61 SELECT $id_col FROM $tbl WHERE $val_col = ? LIMIT 1
66 $dbh->last_insert_id(undef, undef, $tbl, $id_col);
71 my ($self, $path) = @_;
72 return unless defined $path && $path ne '';
73 id_for($self, 'subject', 'sid', 'path' => $path);
77 my ($self, $mid) = @_;
78 id_for($self, 'msgid', 'id', 'mid' => $mid);
82 my ($self, $num) = @_;
83 my $dbh = $self->{dbh};
84 foreach (qw(over id2num)) {
85 $dbh->prepare_cached(<<"")->execute($num);
86 DELETE FROM $_ WHERE num = ?
91 # this includes ghosts
93 my ($self, $mid, $cols, $cb) = @_;
94 my $dbh = $self->{dbh};
97 I originally wanted to stuff everything into a single query:
99 SELECT over.* FROM over
100 LEFT JOIN id2num ON over.num = id2num.num
101 LEFT JOIN msgid ON msgid.id = id2num.id
102 WHERE msgid.mid = ? AND over.num >= ?
103 ORDER BY over.num ASC
106 But it's faster broken out (and we're always in a
107 transaction for subroutines in this file)
110 my $sth = $dbh->prepare_cached(<<'', undef, 1);
111 SELECT id FROM msgid WHERE mid = ? LIMIT 1
114 my $id = $sth->fetchrow_array;
115 defined $id or return;
118 $cols = join(',', map { $_ } @$cols);
120 my $prev = get_counter($dbh, 'ghost');
122 $sth = $dbh->prepare_cached(<<"", undef, 1);
123 SELECT num FROM id2num WHERE id = ? AND num >= ?
127 $sth->execute($id, $prev);
128 my $nums = $sth->fetchall_arrayref;
129 my $nr = scalar(@$nums) or return;
130 $prev = $nums->[-1]->[0];
132 $sth = $dbh->prepare_cached(<<"", undef, 1);
133 SELECT $cols FROM over WHERE over.num = ? LIMIT 1
136 $sth->execute($_->[0]);
137 my $smsg = $sth->fetchrow_hashref;
138 $cb->(PublicInbox::Over::load_from_row($smsg)) or
141 return if $nr != $lim;
145 # this will create a ghost as necessary
146 sub resolve_mid_to_tid {
147 my ($self, $mid) = @_;
149 each_by_mid($self, $mid, ['tid'], sub {
151 my $cur_tid = $smsg->{tid};
153 merge_threads($self, $tid, $cur_tid);
159 defined $tid ? $tid : create_ghost($self, $mid);
163 my ($self, $mid) = @_;
164 my $id = $self->mid2id($mid);
165 my $num = $self->next_ghost_num;
166 $num < 0 or die "ghost num is non-negative: $num\n";
167 my $tid = $self->next_tid;
168 my $dbh = $self->{dbh};
169 $dbh->prepare_cached(<<'')->execute($num, $tid);
170 INSERT INTO over (num, tid) VALUES (?,?)
172 $dbh->prepare_cached(<<'')->execute($id, $num);
173 INSERT INTO id2num (id, num) VALUES (?,?)
179 my ($self, $winner_tid, $loser_tid) = @_;
180 return if $winner_tid == $loser_tid;
181 my $dbh = $self->{dbh};
182 $dbh->prepare_cached(<<'')->execute($winner_tid, $loser_tid);
183 UPDATE over SET tid = ? WHERE tid = ?
188 my ($self, $refs, $old_tid) = @_;
192 # first ref *should* be the thread root,
193 # but we can never trust clients to do the right thing
194 my $ref = $refs->[0];
195 $tid = resolve_mid_to_tid($self, $ref);
196 merge_threads($self, $tid, $old_tid) if defined $old_tid;
198 # the rest of the refs should point to this tid:
199 foreach my $i (1..$#$refs) {
201 my $ptid = resolve_mid_to_tid($self, $ref);
202 merge_threads($self, $tid, $ptid);
205 $tid = defined $old_tid ? $old_tid : $self->next_tid;
210 sub parse_references ($$$) {
211 my ($smsg, $mid0, $mids) = @_;
212 my $mime = $smsg->{mime};
213 my $hdr = $mime->header_obj;
214 my $refs = references($hdr);
215 push(@$refs, @$mids) if scalar(@$mids) > 1;
216 return $refs if scalar(@$refs) == 0;
218 # prevent circular references here:
219 my %seen = ( $mid0 => 1 );
221 foreach my $ref (@$refs) {
222 if (length($ref) > PublicInbox::MID::MAX_MID_SIZE) {
223 warn "References: <$ref> too long, ignoring\n";
226 next if $seen{$ref}++;
229 $smsg->{references} = '<'.join('> <', @keep).'>' if @keep;
234 my ($self, $mime, $bytes, $num, $oid, $mid0) = @_;
235 my $lines = $mime->body_raw =~ tr!\n!\n!;
242 }, 'PublicInbox::SearchMsg';
243 my $mids = mids($mime->header_obj);
244 my $refs = parse_references($smsg, $mid0, $mids);
245 my $subj = $smsg->subject;
248 $xpath = PublicInbox::Search::subject_path($subj);
249 $xpath = id_compress($xpath);
251 my $dd = $smsg->to_doc_data($oid, $mid0);
254 my $values = [ $smsg->ts, $smsg->ds, $num, $mids, $refs, $xpath, $dd ];
255 add_over($self, $values);
259 my ($self, $values) = @_;
260 my ($ts, $ds, $num, $mids, $refs, $xpath, $ddd) = @$values;
265 $self->delete_by_num($num);
266 foreach my $mid (@$mids) {
268 each_by_mid($self, $mid, ['tid'], sub {
270 my $cur_tid = $cur->{tid};
272 die "num must not be zero for $mid" if !$n;
273 $old_tid = $cur_tid unless defined $old_tid;
274 if ($n > 0) { # regular mail
275 merge_threads($self, $old_tid, $cur_tid);
276 } elsif ($n < 0) { # ghost
277 link_refs($self, $refs, $old_tid);
278 $self->delete_by_num($n);
283 $v > 1 and warn "BUG: vivified multiple ($v) ghosts for $mid\n";
286 my $tid = $vivified ? $old_tid : link_refs($self, $refs, $old_tid);
287 my $sid = $self->sid($xpath);
288 my $dbh = $self->{dbh};
289 my $sth = $dbh->prepare_cached(<<'');
290 INSERT INTO over (num, tid, sid, ts, ds, ddd)
294 my @v = ($num, $tid, $sid, $ts, $ds);
295 foreach (@v) { $sth->bind_param(++$n, $_) }
296 $sth->bind_param(++$n, $ddd, SQL_BLOB);
298 $sth = $dbh->prepare_cached(<<'');
299 INSERT INTO id2num (id, num) VALUES (?,?)
301 foreach my $mid (@$mids) {
302 my $id = $self->mid2id($mid);
303 $sth->execute($id, $num);
307 sub delete_articles {
308 my ($self, $nums) = @_;
309 my $dbh = $self->connect;
310 $self->delete_by_num($_) foreach @$nums;
314 my ($self, $oid, $mid) = @_;
316 each_by_mid($self, $mid, ['ddd'], sub {
318 $self->delete_by_num($smsg->{num}) if $smsg->{blob} eq $oid;
327 CREATE TABLE IF NOT EXISTS over (
328 num INTEGER NOT NULL,
329 tid INTEGER NOT NULL,
333 ddd VARBINARY, /* doc-data-deflated */
337 $dbh->do('CREATE INDEX IF NOT EXISTS idx_tid ON over (tid)');
338 $dbh->do('CREATE INDEX IF NOT EXISTS idx_sid ON over (sid)');
339 $dbh->do('CREATE INDEX IF NOT EXISTS idx_ts ON over (ts)');
340 $dbh->do('CREATE INDEX IF NOT EXISTS idx_ds ON over (ds)');
343 CREATE TABLE IF NOT EXISTS counter (
344 key VARCHAR(8) PRIMARY KEY NOT NULL,
345 val INTEGER DEFAULT 0,
349 $dbh->do("INSERT OR IGNORE INTO counter (key) VALUES ('thread')");
350 $dbh->do("INSERT OR IGNORE INTO counter (key) VALUES ('ghost')");
353 CREATE TABLE IF NOT EXISTS subject (
354 sid INTEGER PRIMARY KEY AUTOINCREMENT,
355 path VARCHAR(40) NOT NULL,
360 CREATE TABLE IF NOT EXISTS id2num (
362 num INTEGER NOT NULL,
366 # performance critical:
367 $dbh->do('CREATE INDEX IF NOT EXISTS idx_inum ON id2num (num)');
368 $dbh->do('CREATE INDEX IF NOT EXISTS idx_id ON id2num (id)');
371 CREATE TABLE IF NOT EXISTS msgid (
372 id INTEGER PRIMARY KEY AUTOINCREMENT,
373 mid VARCHAR(244) NOT NULL,
381 delete $self->{txn} or return;
382 $self->{dbh}->commit;
387 return if $self->{txn};
388 my $dbh = $self->connect or return;
390 # $dbh->{Profile} = 2;
396 delete $self->{txn} or return;
397 $self->{dbh}->rollback;
402 die "in transaction" if $self->{txn};
403 $self->{dbh} = undef;
408 unless (-r $self->{filename}) {
410 require File::Basename;
411 File::Path::mkpath(File::Basename::dirname($self->{filename}));
414 PublicInbox::Over::connect($self);