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_ database which can be
6 # tweaked/updated over time and rebuilt.
7 package PublicInbox::Over;
12 use PublicInbox::Smsg;
13 use Compress::Zlib qw(uncompress);
14 use constant DEFAULT_LIMIT => 1000;
18 my $f = delete $self->{filename};
19 if (!-f $f) { # SQLite defaults mode to 0644, we want 0666
21 require PublicInbox::Spawn;
22 open my $fh, '+>>', $f or die "failed to open $f: $!";
23 PublicInbox::Spawn::nodatacow_fd(fileno($fh));
25 $self->{filename} = $f; # die on stat() below:
31 @st = stat($f) or die "failed to stat $f: $!";
32 $st = pack('dd', $st[0], $st[1]); # 0: dev, 1: inode
33 $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
38 sqlite_use_immediate_transaction => 1,
41 @st = stat($f) or die "failed to stat $f: $!";
42 $st = pack('dd', $st[0], $st[1]);
43 } while ($st ne $self->{st} && $tries++ < 3);
44 warn "W: $f: .st_dev, .st_ino unstable\n" if $st ne $self->{st};
45 $dbh->do('PRAGMA synchronous = OFF') if ($rw // 0) > 1;
51 bless { filename => $f }, $class;
56 if (my $dbh = delete $self->{dbh}) {
57 delete $self->{-get_art};
58 $self->{filename} = $dbh->sqlite_db_filename;
62 sub connect { $_[0]->{dbh} //= $_[0]->dbh_new }
64 sub load_from_row ($;$) {
65 my ($smsg, $cull) = @_;
66 bless $smsg, 'PublicInbox::Smsg';
67 if (defined(my $data = delete $smsg->{ddd})) {
68 $data = uncompress($data);
69 PublicInbox::Smsg::load_from_data($smsg, $data);
71 # saves over 600K for 1000+ message threads
72 PublicInbox::Smsg::psgi_cull($smsg) if $cull;
78 my ($self, $sql, $opts, @args) = @_;
79 my $dbh = $self->connect;
80 my $lim = (($opts->{limit} || 0) + 0) || DEFAULT_LIMIT;
82 my $msgs = $dbh->selectall_arrayref($sql, { Slice => {} }, @args);
83 my $cull = $opts->{cull};
84 load_from_row($_, $cull) for @$msgs;
89 my ($self, $beg, $end) = @_;
90 do_get($self, <<'', {}, $beg, $end);
91 SELECT num,ts,ds,ddd FROM over WHERE num >= ? AND num <= ?
97 my ($self, $ts, $prev) = @_;
98 do_get($self, <<'', {}, $ts, $prev);
99 SELECT num,ddd FROM over WHERE ts >= ? AND num > ?
106 my $nr = scalar(@_) or return [];
107 my $in = '?' . (',?' x ($nr - 1));
108 do_get($self, <<"", { cull => 1, limit => $nr }, @_);
109 SELECT num,ds,ddd FROM over WHERE num IN ($in)
113 sub nothing () { wantarray ? (0, []) : [] };
116 my ($self, $mid, $prev) = @_;
117 my $dbh = $self->connect;
118 my $opts = { cull => 1 };
120 my $id = $dbh->selectrow_array(<<'', undef, $mid);
121 SELECT id FROM msgid WHERE mid = ? LIMIT 1
123 defined $id or return nothing;
125 my $num = $dbh->selectrow_array(<<'', undef, $id);
126 SELECT num FROM id2num WHERE id = ? AND num > 0
127 ORDER BY num ASC LIMIT 1
129 defined $num or return nothing;
131 my ($tid, $sid) = $dbh->selectrow_array(<<'', undef, $num);
132 SELECT tid,sid FROM over WHERE num = ? LIMIT 1
134 defined $tid or return nothing; # $sid may be undef
136 my $cond_all = '(tid = ? OR sid = ?) AND num > ?';
139 if ($prev) { # mboxrd stream, only
140 $num = $prev->{num} || 0;
144 my $cols = 'num,ts,ds,ddd';
146 return do_get($self, <<"", $opts, $tid, $sid, $num);
147 SELECT $cols FROM over WHERE $cond_all
148 ORDER BY $sort_col ASC
152 # HTML view always wants an array and never uses $prev,
153 # but the mbox stream never wants an array and always has $prev
154 die '$prev not supported with wantarray' if $prev;
155 my $nr = $dbh->selectrow_array(<<"", undef, $tid, $sid, $num);
156 SELECT COUNT(num) FROM over WHERE $cond_all
158 # giant thread, prioritize strict (tid) matches and throw
159 # in the loose (sid) matches at the end
160 my $msgs = do_get($self, <<"", $opts, $tid, $num);
161 SELECT $cols FROM over WHERE tid = ? AND num > ?
162 ORDER BY $sort_col ASC
164 # do we have room for loose matches? get the most recent ones, first:
165 my $lim = DEFAULT_LIMIT - scalar(@$msgs);
167 $opts->{limit} = $lim;
168 my $loose = do_get($self, <<"", $opts, $tid, $sid, $num);
169 SELECT $cols FROM over WHERE tid != ? AND sid = ? AND num > ?
170 ORDER BY $sort_col DESC
172 # TODO separate strict and loose matches here once --reindex
173 # is fixed to preserve `tid' properly
174 push @$msgs, @$loose;
179 # strict `tid' matches, only, for thread-expanded mbox.gz search results
180 # and future CLI interface
181 # returns true if we have IDs, undef if not
183 my ($self, $ctx) = @_;
184 my $dbh = $self->connect;
186 defined(my $num = $ctx->{ids}->[0]) or return;
187 my ($tid) = $dbh->selectrow_array(<<'', undef, $num);
188 SELECT tid FROM over WHERE num = ?
192 SELECT num FROM over WHERE tid = ? AND num > ?
193 ORDER BY num ASC LIMIT 1000
195 my $xids = $dbh->selectcol_arrayref($sql, undef, $tid,
197 if (scalar(@$xids)) {
198 $ctx->{prev} = $xids->[-1];
199 $ctx->{xids} = $xids;
204 shift @{$ctx->{ids}};
209 my ($self, $opts, $after, $before) = @_;
211 if (defined($before)) {
212 if (defined($after)) {
213 $s = '+num > 0 AND ts >= ? AND ts <= ? ORDER BY ts DESC';
214 @v = ($after, $before);
216 $s = '+num > 0 AND ts <= ? ORDER BY ts DESC';
220 if (defined($after)) {
221 $s = '+num > 0 AND ts >= ? ORDER BY ts ASC';
224 $s = '+num > 0 ORDER BY ts DESC';
227 my $msgs = do_get($self, <<"", $opts, @v);
228 SELECT ts,ds,ddd FROM over WHERE $s
230 return $msgs unless wantarray;
232 my $nr = $self->{dbh}->selectrow_array(<<'');
233 SELECT COUNT(num) FROM over WHERE num > 0
239 my ($self, $num) = @_;
240 # caching $sth ourselves is faster than prepare_cached
241 my $sth = $self->{-get_art} //= $self->connect->prepare(<<'');
242 SELECT num,tid,ds,ts,ddd FROM over WHERE num = ? LIMIT 1
245 my $smsg = $sth->fetchrow_hashref;
246 $smsg ? load_from_row($smsg) : undef;
250 my ($self, $mid, $id, $prev) = @_;
251 my $dbh = $self->connect;
253 unless (defined $$id) {
254 my $sth = $dbh->prepare_cached(<<'', undef, 1);
255 SELECT id FROM msgid WHERE mid = ? LIMIT 1
258 $$id = $sth->fetchrow_array;
259 defined $$id or return;
261 my $sth = $dbh->prepare_cached(<<"", undef, 1);
262 SELECT num FROM id2num WHERE id = ? AND num > ?
263 ORDER BY num ASC LIMIT 1
266 $sth->execute($$id, $$prev);
267 my $num = $sth->fetchrow_array or return;
269 get_art($self, $num);
272 # IMAP search, this is limited by callers to UID_SLICE size (50K)
274 my ($self, $beg, $end, $sql) = @_;
275 my $dbh = $self->connect;
276 my $q = 'SELECT num FROM over WHERE num >= ? AND num <= ?';
278 # This is read-only, anyways; but caller should verify it's
279 # only sending \A[0-9]+\z for ds and ts column ranges
281 $q .= ' ORDER BY num ASC';
282 $dbh->selectcol_arrayref($q, undef, $beg, $end);
287 my $sth = $self->connect->prepare_cached(<<'', undef, 1);
288 SELECT MAX(num) FROM over WHERE num > 0
291 $sth->fetchrow_array // 0;
295 my ($self, $uid_base, $uid_end) = @_;
296 my $sth = $self->connect->prepare_cached(<<'', undef, 1);
297 SELECT COUNT(num) FROM over WHERE num > ? AND num <= ?
299 $sth->execute($uid_base, $uid_end);
300 $sth->fetchrow_array;
305 my $dbh = $self->{dbh} or return;
306 my $f = $dbh->sqlite_db_filename;
307 if (my @st = stat($f)) { # did st_dev, st_ino change?
308 my $st = pack('dd', $st[0], $st[1]);
310 # don't actually reopen, just let {dbh} be recreated later
311 disconnect($self) if $st ne ($self->{st} // $st);
313 warn "W: stat $f: $!\n";