1 # Copyright (C) 2018-2020 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_for_index references/;
17 use PublicInbox::Smsg qw(subject_normalized);
18 use Compress::Zlib qw(compress);
19 use PublicInbox::Search;
24 my $dbh = $self->SUPER::dbh_new($self->{-no_fsync} ? 2 : 1);
26 # 80000 pages (80MiB on SQLite <3.12.0, 320MiB on 3.12.0+)
27 # was found to be good in 2018 during the large LKML import
28 # at the time. This ought to be configurable based on HW
29 # and inbox size; I suspect it's overkill for many inboxes.
30 $dbh->do('PRAGMA cache_size = 80000');
38 my $self = $class->SUPER::new($f);
43 sub get_counter ($$) {
45 my $sth = $dbh->prepare_cached(<<'', undef, 1);
46 SELECT val FROM counter WHERE key = ? LIMIT 1
52 sub adj_counter ($$$) {
53 my ($self, $key, $op) = @_;
54 my $dbh = $self->{dbh};
55 my $sth = $dbh->prepare_cached(<<"");
56 UPDATE counter SET val = val $op 1 WHERE key = ?
60 get_counter($dbh, $key);
63 sub next_tid { adj_counter($_[0], 'thread', '+') }
64 sub next_ghost_num { adj_counter($_[0], 'ghost', '-') }
67 my ($self, $tbl, $id_col, $val_col, $val) = @_;
68 my $dbh = $self->{dbh};
69 my $in = $dbh->prepare_cached(<<"")->execute($val);
70 INSERT OR IGNORE INTO $tbl ($val_col) VALUES (?)
73 my $sth = $dbh->prepare_cached(<<"", undef, 1);
74 SELECT $id_col FROM $tbl WHERE $val_col = ? LIMIT 1
79 $dbh->last_insert_id(undef, undef, $tbl, $id_col);
84 my ($self, $path) = @_;
85 return unless defined $path && $path ne '';
86 id_for($self, 'subject', 'sid', 'path' => $path);
90 my ($self, $mid) = @_;
91 id_for($self, 'msgid', 'id', 'mid' => $mid);
95 my ($self, $num, $tid_ref) = @_;
96 my $dbh = $self->{dbh};
98 my $sth = $dbh->prepare_cached(<<'', undef, 1);
99 SELECT tid FROM over WHERE num = ? LIMIT 1
102 $$tid_ref = $sth->fetchrow_array; # may be undef
104 foreach (qw(over id2num)) {
105 $dbh->prepare_cached(<<"")->execute($num);
106 DELETE FROM $_ WHERE num = ?
111 # this includes ghosts
113 my ($self, $mid, $cols, $cb, @arg) = @_;
114 my $dbh = $self->{dbh};
117 I originally wanted to stuff everything into a single query:
119 SELECT over.* FROM over
120 LEFT JOIN id2num ON over.num = id2num.num
121 LEFT JOIN msgid ON msgid.id = id2num.id
122 WHERE msgid.mid = ? AND over.num >= ?
123 ORDER BY over.num ASC
126 But it's faster broken out (and we're always in a
127 transaction for subroutines in this file)
130 my $sth = $dbh->prepare_cached(<<'', undef, 1);
131 SELECT id FROM msgid WHERE mid = ? LIMIT 1
134 my $id = $sth->fetchrow_array;
135 defined $id or return;
138 $cols = join(',', map { $_ } @$cols);
140 my $prev = get_counter($dbh, 'ghost');
142 $sth = $dbh->prepare_cached(<<"", undef, 1);
143 SELECT num FROM id2num WHERE id = ? AND num >= ?
147 $sth->execute($id, $prev);
148 my $nums = $sth->fetchall_arrayref;
149 my $nr = scalar(@$nums) or return;
150 $prev = $nums->[-1]->[0];
152 $sth = $dbh->prepare_cached(<<"", undef, 1);
153 SELECT $cols FROM over WHERE over.num = ? LIMIT 1
156 $sth->execute($_->[0]);
157 my $smsg = $sth->fetchrow_hashref;
158 $smsg = PublicInbox::Over::load_from_row($smsg);
159 $cb->($self, $smsg, @arg) or return;
161 return if $nr != $lim;
165 sub _resolve_mid_to_tid {
166 my ($self, $smsg, $tid) = @_;
167 my $cur_tid = $smsg->{tid};
169 merge_threads($self, $$tid, $cur_tid);
170 } elsif ($cur_tid > $self->{min_tid}) {
172 } else { # rethreading, queue up dead ghosts
173 $$tid = next_tid($self);
174 my $num = $smsg->{num};
175 push(@{$self->{-ghosts_to_delete}}, $num) if $num < 0;
180 # this will create a ghost as necessary
181 sub resolve_mid_to_tid {
182 my ($self, $mid) = @_;
184 each_by_mid($self, $mid, ['tid'], \&_resolve_mid_to_tid, \$tid);
185 if (my $del = delete $self->{-ghosts_to_delete}) {
186 delete_by_num($self, $_) for @$del;
188 $tid // create_ghost($self, $mid);
192 my ($self, $mid) = @_;
193 my $id = mid2id($self, $mid);
194 my $num = next_ghost_num($self);
195 $num < 0 or die "ghost num is non-negative: $num\n";
196 my $tid = next_tid($self);
197 my $dbh = $self->{dbh};
198 $dbh->prepare_cached(<<'')->execute($num, $tid);
199 INSERT INTO over (num, tid) VALUES (?,?)
201 $dbh->prepare_cached(<<'')->execute($id, $num);
202 INSERT INTO id2num (id, num) VALUES (?,?)
208 my ($self, $winner_tid, $loser_tid) = @_;
209 return if $winner_tid == $loser_tid;
210 my $dbh = $self->{dbh};
211 $dbh->prepare_cached(<<'')->execute($winner_tid, $loser_tid);
212 UPDATE over SET tid = ? WHERE tid = ?
217 my ($self, $refs, $old_tid) = @_;
221 # first ref *should* be the thread root,
222 # but we can never trust clients to do the right thing
223 my $ref = $refs->[0];
224 $tid = resolve_mid_to_tid($self, $ref);
225 merge_threads($self, $tid, $old_tid) if defined $old_tid;
227 # the rest of the refs should point to this tid:
228 foreach my $i (1..$#$refs) {
230 my $ptid = resolve_mid_to_tid($self, $ref);
231 merge_threads($self, $tid, $ptid);
234 $tid = $old_tid // next_tid($self);
239 sub parse_references ($$$) {
240 my ($smsg, $hdr, $mids) = @_;
241 my $refs = references($hdr);
242 push(@$refs, @$mids) if scalar(@$mids) > 1;
243 return $refs if scalar(@$refs) == 0;
245 # prevent circular references here:
246 my %seen = ( $smsg->{mid} => 1 );
248 foreach my $ref (@$refs) {
249 if (length($ref) > PublicInbox::MID::MAX_MID_SIZE) {
250 warn "References: <$ref> too long, ignoring\n";
253 push(@keep, $ref) unless $seen{$ref}++;
255 $smsg->{references} = '<'.join('> <', @keep).'>' if @keep;
259 # normalize subjects so they are suitable as pathnames for URLs
260 # XXX: consider for removal
261 sub subject_path ($) {
263 $subj = subject_normalized($subj);
264 $subj =~ s![^a-zA-Z0-9_\.~/\-]+!_!g;
269 my ($self, $eml, $smsg) = @_;
270 $smsg->{lines} = $eml->body_raw =~ tr!\n!\n!;
271 my $mids = mids_for_index($eml);
272 my $refs = parse_references($smsg, $eml, $mids);
273 my $subj = $smsg->{subject};
276 $xpath = subject_path($subj);
277 $xpath = id_compress($xpath);
279 my $dd = $smsg->to_doc_data;
282 add_over($self, $smsg, $mids, $refs, $xpath, $dd);
286 my ($self, $smsg, $mid, $refs, $old_tid, $v) = @_;
287 my $cur_tid = $smsg->{tid};
288 my $n = $smsg->{num};
289 die "num must not be zero for $mid" if !$n;
290 my $cur_valid = $cur_tid > $self->{min_tid};
292 if ($n > 0) { # regular mail
294 $$old_tid //= $cur_tid;
295 merge_threads($self, $$old_tid, $cur_tid);
297 $$old_tid //= next_tid($self);
299 } elsif ($n < 0) { # ghost
300 $$old_tid //= $cur_valid ? $cur_tid : next_tid($self);
301 link_refs($self, $refs, $$old_tid);
302 delete_by_num($self, $n);
309 my ($self, $smsg, $mids, $refs, $xpath, $ddd) = @_;
312 my $num = $smsg->{num};
315 delete_by_num($self, $num, \$old_tid);
316 $old_tid = undef if ($old_tid // 0) <= $self->{min_tid};
317 foreach my $mid (@$mids) {
319 each_by_mid($self, $mid, ['tid'], \&_add_over,
320 $mid, $refs, \$old_tid, \$v);
321 $v > 1 and warn "BUG: vivified multiple ($v) ghosts for $mid\n";
324 $smsg->{tid} = $vivified ? $old_tid : link_refs($self, $refs, $old_tid);
325 $smsg->{sid} = sid($self, $xpath);
326 my $dbh = $self->{dbh};
327 my $sth = $dbh->prepare_cached(<<'');
328 INSERT INTO over (num, tid, sid, ts, ds, ddd)
332 $sth->bind_param($nc, $num);
333 $sth->bind_param(++$nc, $smsg->{$_}) for (qw(tid sid ts ds));
334 $sth->bind_param(++$nc, $ddd, SQL_BLOB);
336 $sth = $dbh->prepare_cached(<<'');
337 INSERT INTO id2num (id, num) VALUES (?,?)
339 foreach my $mid (@$mids) {
340 my $id = mid2id($self, $mid);
341 $sth->execute($id, $num);
346 my ($self, $smsg, $oid, $removed) = @_;
347 if (!defined($oid) || $smsg->{blob} eq $oid) {
348 delete_by_num($self, $smsg->{num});
349 push @$removed, $smsg->{num};
354 # returns number of removed messages in scalar context,
355 # array of removed article numbers in array context.
356 # $oid may be undef to match only on $mid
358 my ($self, $oid, $mid) = @_;
361 each_by_mid($self, $mid, ['ddd'], \&_remove_oid, $oid, $removed);
365 sub _num_mid0_for_oid {
366 my ($self, $smsg, $oid, $res) = @_;
367 my $blob = $smsg->{blob};
368 return 1 if (!defined($blob) || $blob ne $oid); # continue;
369 @$res = ($smsg->{num}, $smsg->{mid});
373 sub num_mid0_for_oid {
374 my ($self, $oid, $mid) = @_;
377 each_by_mid($self, $mid, ['ddd'], \&_num_mid0_for_oid, $oid, $res);
378 @$res, # ($num, $mid0);
385 CREATE TABLE IF NOT EXISTS over (
386 num INTEGER NOT NULL,
387 tid INTEGER NOT NULL,
391 ddd VARBINARY, /* doc-data-deflated */
395 $dbh->do('CREATE INDEX IF NOT EXISTS idx_tid ON over (tid)');
396 $dbh->do('CREATE INDEX IF NOT EXISTS idx_sid ON over (sid)');
397 $dbh->do('CREATE INDEX IF NOT EXISTS idx_ts ON over (ts)');
398 $dbh->do('CREATE INDEX IF NOT EXISTS idx_ds ON over (ds)');
401 CREATE TABLE IF NOT EXISTS counter (
402 key VARCHAR(8) PRIMARY KEY NOT NULL,
403 val INTEGER DEFAULT 0,
407 $dbh->do("INSERT OR IGNORE INTO counter (key) VALUES ('thread')");
408 $dbh->do("INSERT OR IGNORE INTO counter (key) VALUES ('ghost')");
411 CREATE TABLE IF NOT EXISTS subject (
412 sid INTEGER PRIMARY KEY AUTOINCREMENT,
413 path VARCHAR(40) NOT NULL,
418 CREATE TABLE IF NOT EXISTS id2num (
420 num INTEGER NOT NULL,
424 # performance critical:
425 $dbh->do('CREATE INDEX IF NOT EXISTS idx_inum ON id2num (num)');
426 $dbh->do('CREATE INDEX IF NOT EXISTS idx_id ON id2num (id)');
429 CREATE TABLE IF NOT EXISTS msgid (
430 id INTEGER PRIMARY KEY AUTOINCREMENT,
431 mid VARCHAR(244) NOT NULL,
439 delete $self->{txn} or return;
440 $self->{dbh}->commit;
445 return if $self->{txn};
446 my $dbh = $self->connect or return;
448 # $dbh->{Profile} = 2;
454 delete $self->{txn} or return;
455 $self->{dbh}->rollback;
460 die "in transaction" if $self->{txn};
461 $self->SUPER::disconnect;
466 unless (-r $self->{filename}) {
468 require File::Basename;
469 File::Path::mkpath(File::Basename::dirname($self->{filename}));
472 PublicInbox::Over::connect($self);
476 sub rethread_prepare {
477 my ($self, $opt) = @_;
478 return unless $opt->{rethread};
480 my $min = $self->{min_tid} = get_counter($self->{dbh}, 'thread') // 0;
481 my $pr = $opt->{-progress};
482 $pr->("rethread min THREADID ".($min + 1)."\n") if $pr && $min;
486 my ($self, $opt) = @_;
487 return unless $opt->{rethread} && $self->{txn};
488 defined(my $min = $self->{min_tid}) or croak('BUG: no min_tid');
489 my $dbh = $self->{dbh} or croak('BUG: no dbh');
490 my $rows = $dbh->selectall_arrayref(<<'', { Slice => {} }, $min);
491 SELECT num,tid FROM over WHERE num < 0 AND tid < ?
493 my $show_id = $dbh->prepare('SELECT id FROM id2num WHERE num = ?');
494 my $show_mid = $dbh->prepare('SELECT mid FROM msgid WHERE id = ?');
495 my $pr = $opt->{-progress};
499 $show_id->execute($r->{num});
500 while (defined(my $id = $show_id->fetchrow_array)) {
502 $show_mid->execute($id);
503 my $mid = $show_mid->fetchrow_array;
504 if (!defined($mid)) {
506 E: ghost NUM=$r->{num} ID=$id THREADID=$r->{tid} has no Message-ID
511 I: ghost $r->{num} <$mid> THREADID=$r->{tid} culled
514 delete_by_num($self, $r->{num});
516 $pr->("I: rethread culled $total ghosts\n") if $pr && $total;