1 # Copyright (C) 2018-2021 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_ database which can be
6 # tweaked/updated over time and rebuilt.
7 package PublicInbox::Over;
10 use DBI qw(:sql_types); # SQL_BLOB
12 use PublicInbox::Smsg;
13 use Compress::Zlib qw(uncompress);
14 use constant DEFAULT_LIMIT => 1000;
18 my $f = delete $self->{filename};
19 if (!-s $f) { # SQLite defaults mode to 0644, we want 0666
21 require PublicInbox::Spawn;
22 my ($dir) = ($f =~ m!(.+)/[^/]+\z!);
23 PublicInbox::Spawn::nodatacow_dir($dir);
24 open my $fh, '+>>', $f or die "failed to open $f: $!";
25 PublicInbox::Spawn::nodatacow_fd(fileno($fh));
27 $self->{filename} = $f; # die on stat() below:
33 @st = stat($f) or die "failed to stat $f: $!";
34 $st = pack('dd', $st[0], $st[1]); # 0: dev, 1: inode
35 $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
40 sqlite_use_immediate_transaction => 1,
43 @st = stat($f) or die "failed to stat $f: $!";
44 $st = pack('dd', $st[0], $st[1]);
45 } while ($st ne $self->{st} && $tries++ < 3);
46 warn "W: $f: .st_dev, .st_ino unstable\n" if $st ne $self->{st};
49 # TRUNCATE reduces I/O compared to the default (DELETE).
51 # Do not use WAL by default since we expect the case
52 # where any users may read via read-only daemons
53 # (-httpd/-imapd/-nntpd); but only a single user has
54 # write permissions for -watch/-mda.
56 # Read-only WAL support in SQLite 3.22.0 (2018-01-22)
57 # doesn't do what we need: it is only intended for
58 # immutable read-only media (e.g. CD-ROM) and not
59 # usable for our use case described above.
61 # If an admin is willing to give read-only daemons R/W
62 # permissions; they can enable WAL manually and we will
63 # respect that by not clobbering it.
64 my $jm = $dbh->selectrow_array('PRAGMA journal_mode');
65 $dbh->do('PRAGMA journal_mode = TRUNCATE') if $jm ne 'wal';
67 $dbh->do('PRAGMA synchronous = OFF') if $rw > 1;
74 bless { filename => $f }, $class;
79 if (my $dbh = delete $self->{dbh}) {
80 delete $self->{-get_art};
81 $self->{filename} = $dbh->sqlite_db_filename;
85 sub dbh ($) { $_[0]->{dbh} //= $_[0]->dbh_new } # dbh_new may be subclassed
87 sub load_from_row ($;$) {
88 my ($smsg, $cull) = @_;
89 bless $smsg, 'PublicInbox::Smsg';
90 if (defined(my $data = delete $smsg->{ddd})) {
91 $data = uncompress($data);
92 PublicInbox::Smsg::load_from_data($smsg, $data);
94 # saves over 600K for 1000+ message threads
95 PublicInbox::Smsg::psgi_cull($smsg) if $cull;
101 my ($self, $sql, $opts, @args) = @_;
102 my $lim = (($opts->{limit} || 0) + 0) || DEFAULT_LIMIT;
103 $sql .= "LIMIT $lim";
104 my $msgs = dbh($self)->selectall_arrayref($sql, { Slice => {} }, @args);
105 my $cull = $opts->{cull};
106 load_from_row($_, $cull) for @$msgs;
111 my ($self, $beg, $end, $opt) = @_;
112 do_get($self, <<'', $opt, $beg, $end);
113 SELECT num,ts,ds,ddd FROM over WHERE num >= ? AND num <= ?
119 my ($self, $ts, $prev) = @_;
120 do_get($self, <<'', {}, $ts, $prev);
121 SELECT num,ddd FROM over WHERE ts >= ? AND num > ?
128 my $nr = scalar(@_) or return [];
129 my $in = '?' . (',?' x ($nr - 1));
130 do_get($self, <<"", { cull => 1, limit => $nr }, @_);
131 SELECT num,ts,ds,ddd FROM over WHERE num IN ($in)
135 sub nothing () { wantarray ? (0, []) : [] };
138 my ($self, $mid, $prev) = @_;
139 my $dbh = dbh($self);
140 my $opts = { cull => 1 };
142 my $id = $dbh->selectrow_array(<<'', undef, $mid);
143 SELECT id FROM msgid WHERE mid = ? LIMIT 1
145 defined $id or return nothing;
147 my $num = $dbh->selectrow_array(<<'', undef, $id);
148 SELECT num FROM id2num WHERE id = ? AND num > 0
149 ORDER BY num ASC LIMIT 1
151 defined $num or return nothing;
153 my ($tid, $sid) = $dbh->selectrow_array(<<'', undef, $num);
154 SELECT tid,sid FROM over WHERE num = ? LIMIT 1
156 defined $tid or return nothing; # $sid may be undef
158 my $cond_all = '(tid = ? OR sid = ?) AND num > ?';
161 if ($prev) { # mboxrd stream, only
162 $num = $prev->{num} || 0;
166 my $cols = 'num,ts,ds,ddd';
168 return do_get($self, <<"", $opts, $tid, $sid, $num);
169 SELECT $cols FROM over WHERE $cond_all
170 ORDER BY $sort_col ASC
174 # HTML view always wants an array and never uses $prev,
175 # but the mbox stream never wants an array and always has $prev
176 die '$prev not supported with wantarray' if $prev;
177 my $nr = $dbh->selectrow_array(<<"", undef, $tid, $sid, $num);
178 SELECT COUNT(num) FROM over WHERE $cond_all
180 # giant thread, prioritize strict (tid) matches and throw
181 # in the loose (sid) matches at the end
182 my $msgs = do_get($self, <<"", $opts, $tid, $num);
183 SELECT $cols FROM over WHERE tid = ? AND num > ?
184 ORDER BY $sort_col ASC
186 # do we have room for loose matches? get the most recent ones, first:
187 my $lim = DEFAULT_LIMIT - scalar(@$msgs);
189 $opts->{limit} = $lim;
190 my $loose = do_get($self, <<"", $opts, $tid, $sid, $num);
191 SELECT $cols FROM over WHERE tid != ? AND sid = ? AND num > ?
192 ORDER BY $sort_col DESC
194 # TODO separate strict and loose matches here once --reindex
195 # is fixed to preserve `tid' properly
196 push @$msgs, @$loose;
201 # strict `tid' matches, only, for thread-expanded mbox.gz search results
202 # and future CLI interface
203 # returns true if we have IDs, undef if not
205 my ($self, $ctx) = @_;
206 my $dbh = dbh($self);
208 defined(my $num = $ctx->{ids}->[0]) or return;
209 my ($tid) = $dbh->selectrow_array(<<'', undef, $num);
210 SELECT tid FROM over WHERE num = ?
214 SELECT num FROM over WHERE tid = ? AND num > ?
215 ORDER BY num ASC LIMIT 1000
217 my $xids = $dbh->selectcol_arrayref($sql, undef, $tid,
219 if (scalar(@$xids)) {
220 $ctx->{prev} = $xids->[-1];
221 $ctx->{xids} = $xids;
226 shift @{$ctx->{ids}};
231 my ($self, $opts, $after, $before) = @_;
233 if (defined($before)) {
234 if (defined($after)) {
235 $s = '+num > 0 AND ts >= ? AND ts <= ? ORDER BY ts DESC';
236 @v = ($after, $before);
238 $s = '+num > 0 AND ts <= ? ORDER BY ts DESC';
242 if (defined($after)) {
243 $s = '+num > 0 AND ts >= ? ORDER BY ts ASC';
246 $s = '+num > 0 ORDER BY ts DESC';
249 do_get($self, <<"", $opts, @v);
250 SELECT ts,ds,ddd FROM over WHERE $s
255 my ($self, $num) = @_;
256 # caching $sth ourselves is faster than prepare_cached
257 my $sth = $self->{-get_art} //= dbh($self)->prepare(<<'');
258 SELECT num,tid,ds,ts,ddd FROM over WHERE num = ? LIMIT 1
261 my $smsg = $sth->fetchrow_hashref;
262 $smsg ? load_from_row($smsg) : undef;
266 my ($self, $num, $raw) = @_;
267 my $dbh = dbh($self);
268 my $sth = $dbh->prepare_cached(<<'', undef, 1);
269 SELECT ibx_id,xnum,oidbin FROM xref3 WHERE docid = ? ORDER BY ibx_id,xnum ASC
272 my $rows = $sth->fetchall_arrayref;
273 return $rows if $raw;
274 my $eidx_key_sth = $dbh->prepare_cached(<<'', undef, 1);
275 SELECT eidx_key FROM inboxes WHERE ibx_id = ?
279 $eidx_key_sth->execute($r->[0]);
280 my $eidx_key = $eidx_key_sth->fetchrow_array;
281 $eidx_key //= "missing://ibx_id=$r->[0]";
282 "$eidx_key:$r->[1]:".unpack('H*', $r->[2]);
287 my ($self, $mid, $id, $prev) = @_;
288 my $dbh = dbh($self);
290 unless (defined $$id) {
291 my $sth = $dbh->prepare_cached(<<'', undef, 1);
292 SELECT id FROM msgid WHERE mid = ? LIMIT 1
295 $$id = $sth->fetchrow_array;
296 defined $$id or return;
298 my $sth = $dbh->prepare_cached(<<"", undef, 1);
299 SELECT num FROM id2num WHERE id = ? AND num > ?
300 ORDER BY num ASC LIMIT 1
303 $sth->execute($$id, $$prev);
304 my $num = $sth->fetchrow_array or return;
306 get_art($self, $num);
309 # IMAP search, this is limited by callers to UID_SLICE size (50K)
311 my ($self, $beg, $end, $sql) = @_;
312 my $dbh = dbh($self);
313 my $q = 'SELECT num FROM over WHERE num >= ? AND num <= ?';
315 # This is read-only, anyways; but caller should verify it's
316 # only sending \A[0-9]+\z for ds and ts column ranges
318 $q .= ' ORDER BY num ASC';
319 $dbh->selectcol_arrayref($q, undef, $beg, $end);
324 my $sth = dbh($self)->prepare_cached(<<'', undef, 1);
325 SELECT MAX(num) FROM over WHERE num > 0
328 $sth->fetchrow_array // 0;
332 my ($self, $uid_base, $uid_end) = @_;
333 my $sth = dbh($self)->prepare_cached(<<'', undef, 1);
334 SELECT COUNT(num) FROM over WHERE num > ? AND num <= ?
336 $sth->execute($uid_base, $uid_end);
337 $sth->fetchrow_array;
342 my $dbh = $self->{dbh} or return;
343 my $f = $dbh->sqlite_db_filename;
344 if (my @st = stat($f)) { # did st_dev, st_ino change?
345 my $st = pack('dd', $st[0], $st[1]);
347 # don't actually reopen, just let {dbh} be recreated later
348 dbh_close($self) if $st ne ($self->{st} // $st);
350 warn "W: stat $f: $!\n";
355 my ($self, $oidbin) = @_;
357 my $sth = $self->dbh->prepare_cached(<<'', undef, 1);
358 SELECT docid FROM xref3 WHERE oidbin = ? ORDER BY docid ASC
360 $sth->bind_param(1, $oidbin, SQL_BLOB);
362 my $tmp = $sth->fetchall_arrayref;
363 map { $_->[0] } @$tmp;
365 my $sth = $self->dbh->prepare_cached(<<'', undef, 1);
366 SELECT COUNT(*) FROM xref3 WHERE oidbin = ?
368 $sth->bind_param(1, $oidbin, SQL_BLOB);
370 $sth->fetchrow_array;
374 sub blob_exists { oidbin_exists($_[0], pack('H*', $_[1])) }
378 my ($self, $num) = @_;
379 my $ids = dbh($self)->selectcol_arrayref(<<'', undef, $$num);
380 SELECT num FROM over WHERE num > ?
381 ORDER BY num ASC LIMIT 1000
383 $$num = $ids->[-1] if @$ids;