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 open $fh, '+>>', $j or die "failed to open $j: $!";
26 PublicInbox::Spawn::nodatacow_fd(fileno($fh));
28 $self->{filename} = $f; # die on stat() below:
34 @st = stat($f) or die "failed to stat $f: $!";
35 $st = pack('dd', $st[0], $st[1]); # 0: dev, 1: inode
36 $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
41 sqlite_use_immediate_transaction => 1,
44 @st = stat($f) or die "failed to stat $f: $!";
45 $st = pack('dd', $st[0], $st[1]);
46 } while ($st ne $self->{st} && $tries++ < 3);
47 warn "W: $f: .st_dev, .st_ino unstable\n" if $st ne $self->{st};
48 $dbh->do('PRAGMA synchronous = OFF') if ($rw // 0) > 1;
54 bless { filename => $f }, $class;
59 if (my $dbh = delete $self->{dbh}) {
60 delete $self->{-get_art};
61 $self->{filename} = $dbh->sqlite_db_filename;
65 sub connect { $_[0]->{dbh} //= $_[0]->dbh_new }
67 sub load_from_row ($;$) {
68 my ($smsg, $cull) = @_;
69 bless $smsg, 'PublicInbox::Smsg';
70 if (defined(my $data = delete $smsg->{ddd})) {
71 $data = uncompress($data);
72 PublicInbox::Smsg::load_from_data($smsg, $data);
74 # saves over 600K for 1000+ message threads
75 PublicInbox::Smsg::psgi_cull($smsg) if $cull;
81 my ($self, $sql, $opts, @args) = @_;
82 my $dbh = $self->connect;
83 my $lim = (($opts->{limit} || 0) + 0) || DEFAULT_LIMIT;
85 my $msgs = $dbh->selectall_arrayref($sql, { Slice => {} }, @args);
86 my $cull = $opts->{cull};
87 load_from_row($_, $cull) for @$msgs;
92 my ($self, $beg, $end) = @_;
93 do_get($self, <<'', {}, $beg, $end);
94 SELECT num,ts,ds,ddd FROM over WHERE num >= ? AND num <= ?
100 my ($self, $ts, $prev) = @_;
101 do_get($self, <<'', {}, $ts, $prev);
102 SELECT num,ddd FROM over WHERE ts >= ? AND num > ?
109 my $nr = scalar(@_) or return [];
110 my $in = '?' . (',?' x ($nr - 1));
111 do_get($self, <<"", { cull => 1, limit => $nr }, @_);
112 SELECT num,ds,ddd FROM over WHERE num IN ($in)
116 sub nothing () { wantarray ? (0, []) : [] };
119 my ($self, $mid, $prev) = @_;
120 my $dbh = $self->connect;
121 my $opts = { cull => 1 };
123 my $id = $dbh->selectrow_array(<<'', undef, $mid);
124 SELECT id FROM msgid WHERE mid = ? LIMIT 1
126 defined $id or return nothing;
128 my $num = $dbh->selectrow_array(<<'', undef, $id);
129 SELECT num FROM id2num WHERE id = ? AND num > 0
130 ORDER BY num ASC LIMIT 1
132 defined $num or return nothing;
134 my ($tid, $sid) = $dbh->selectrow_array(<<'', undef, $num);
135 SELECT tid,sid FROM over WHERE num = ? LIMIT 1
137 defined $tid or return nothing; # $sid may be undef
139 my $cond_all = '(tid = ? OR sid = ?) AND num > ?';
142 if ($prev) { # mboxrd stream, only
143 $num = $prev->{num} || 0;
147 my $cols = 'num,ts,ds,ddd';
149 return do_get($self, <<"", $opts, $tid, $sid, $num);
150 SELECT $cols FROM over WHERE $cond_all
151 ORDER BY $sort_col ASC
155 # HTML view always wants an array and never uses $prev,
156 # but the mbox stream never wants an array and always has $prev
157 die '$prev not supported with wantarray' if $prev;
158 my $nr = $dbh->selectrow_array(<<"", undef, $tid, $sid, $num);
159 SELECT COUNT(num) FROM over WHERE $cond_all
161 # giant thread, prioritize strict (tid) matches and throw
162 # in the loose (sid) matches at the end
163 my $msgs = do_get($self, <<"", $opts, $tid, $num);
164 SELECT $cols FROM over WHERE tid = ? AND num > ?
165 ORDER BY $sort_col ASC
167 # do we have room for loose matches? get the most recent ones, first:
168 my $lim = DEFAULT_LIMIT - scalar(@$msgs);
170 $opts->{limit} = $lim;
171 my $loose = do_get($self, <<"", $opts, $tid, $sid, $num);
172 SELECT $cols FROM over WHERE tid != ? AND sid = ? AND num > ?
173 ORDER BY $sort_col DESC
175 # TODO separate strict and loose matches here once --reindex
176 # is fixed to preserve `tid' properly
177 push @$msgs, @$loose;
183 my ($self, $opts, $after, $before) = @_;
185 if (defined($before)) {
186 if (defined($after)) {
187 $s = '+num > 0 AND ts >= ? AND ts <= ? ORDER BY ts DESC';
188 @v = ($after, $before);
190 $s = '+num > 0 AND ts <= ? ORDER BY ts DESC';
194 if (defined($after)) {
195 $s = '+num > 0 AND ts >= ? ORDER BY ts ASC';
198 $s = '+num > 0 ORDER BY ts DESC';
201 my $msgs = do_get($self, <<"", $opts, @v);
202 SELECT ts,ds,ddd FROM over WHERE $s
204 return $msgs unless wantarray;
206 my $nr = $self->{dbh}->selectrow_array(<<'');
207 SELECT COUNT(num) FROM over WHERE num > 0
213 my ($self, $num) = @_;
214 # caching $sth ourselves is faster than prepare_cached
215 my $sth = $self->{-get_art} //= $self->connect->prepare(<<'');
216 SELECT num,tid,ds,ts,ddd FROM over WHERE num = ? LIMIT 1
219 my $smsg = $sth->fetchrow_hashref;
220 $smsg ? load_from_row($smsg) : undef;
224 my ($self, $mid, $id, $prev) = @_;
225 my $dbh = $self->connect;
227 unless (defined $$id) {
228 my $sth = $dbh->prepare_cached(<<'', undef, 1);
229 SELECT id FROM msgid WHERE mid = ? LIMIT 1
232 $$id = $sth->fetchrow_array;
233 defined $$id or return;
235 my $sth = $dbh->prepare_cached(<<"", undef, 1);
236 SELECT num FROM id2num WHERE id = ? AND num > ?
237 ORDER BY num ASC LIMIT 1
240 $sth->execute($$id, $$prev);
241 my $num = $sth->fetchrow_array or return;
243 get_art($self, $num);
246 # IMAP search, this is limited by callers to UID_SLICE size (50K)
248 my ($self, $beg, $end, $sql) = @_;
249 my $dbh = $self->connect;
250 my $q = 'SELECT num FROM over WHERE num >= ? AND num <= ?';
252 # This is read-only, anyways; but caller should verify it's
253 # only sending \A[0-9]+\z for ds and ts column ranges
255 $q .= ' ORDER BY num ASC';
256 $dbh->selectcol_arrayref($q, undef, $beg, $end);
261 my $sth = $self->connect->prepare_cached(<<'', undef, 1);
262 SELECT MAX(num) FROM over WHERE num > 0
265 $sth->fetchrow_array // 0;
269 my ($self, $uid_base, $uid_end) = @_;
270 my $sth = $self->connect->prepare_cached(<<'', undef, 1);
271 SELECT COUNT(num) FROM over WHERE num > ? AND num <= ?
273 $sth->execute($uid_base, $uid_end);
274 $sth->fetchrow_array;
279 my $dbh = $self->{dbh} or return;
280 my $f = $dbh->sqlite_db_filename;
281 if (my @st = stat($f)) { # did st_dev, st_ino change?
282 my $st = pack('dd', $st[0], $st[1]);
284 # don't actually reopen, just let {dbh} be recreated later
285 disconnect($self) if $st ne ($self->{st} // $st);
287 warn "W: stat $f: $!\n";